S.

Palani

Automatic
Control
Systems
With MATLAB
Second Edition
Automatic Control Systems
S. Palani

Automatic Control Systems

With MATLAB
Second Edition
S. Palani
National Institute of Technology
Tiruchirapalli, India

ISBN 978-3-030-93444-6 ISBN 978-3-030-93445-3 (eBook)

https://doi.org/10.1007/978-3-030-93445-3

Jointly published with ANE Books India

The print edition is not for sale in South Asia (India, Pakistan, Sri Lanka, Bangladesh, Nepal and Bhutan)
and Africa. Customers from South Asia and Africa can please order the print book from: ANE Books
Pvt.Ltd.
ISBN of the Co-Publisher’s edition: 978-9-383-65627-1

1st edition: © Author 2016

2nd edition: © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer
Nature Switzerland AG 2022
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publishers, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publishers nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publishers remain neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
This book is
dedicated to My Beloved Teachers
Dr. K. L. P. Mishra
University of Leningrad,
St. Petersburg, Russia
Prof. P. C. Manoharan
P.S.G. College of Technology,
Coimbatore, India
Preface

Automatic Control Engineering is an exciting field of study which gave significant

contributions to the development of latest technologies in diversified applications
which include the very sophisticated aircraft and ship control, control of commu-
nication satellites, control of boilers and nuclear reactors, complex power system
network, war weapons etc. Automatic control theory is divided into two major topics
namely classical control theory and modern control theory. The application of clas-
sical control theory is restricted to single-input and single-output linear time invariant
continuous time systems. Here for the transfer function model of the system, the anal-
ysis and design are carried out in time domain as well as in frequency domain. There
are several limitations in this method. The limitations of classical control theory
were overcome by modern control theory which was introduced in the yearly 1960s.
The method uses state space model and can deal with multi-input and multi-output
system. Many sophisticated design problems are carried out in this method. Non-
linear, multivariable, adaptive and robust control theory comes under modern control
theory.
Originally automatic control theory was applied to linear continuous time system
analysis and design. In recent time, computers and intentionally introduced into
the system to monitor and control all physical variables. Now the system becomes
discrete time system. In all technical universities in India, automatic control
theory course is taught for the undergraduate, postgraduate and doctoral level
students of Electrical and Electronics Engineering, Electronics and Communication
Engineering, Electronics and Instrumental Engineering, Mechanical Engineering,
Computer Science and Engineering, Aeronautical Engineering and Chemical Engi-
neering. To write books to the requirements of students of individual branch is really
a tough task. Therefore, it is decided to write a common book which will cater to the
need of students of all these engineering disciplines of undergraduate level of Indian
universities. I have used all my experiences I gained while teaching this course to
the students of undergraduate, postgraduate and doctor degree for the past four and
a half decades to bring out a good book on this subject.
The book presents a comprehensive coverage of linear system theory which is
intended for undergraduate courses in control system engineering in all technical

vii
viii Preface

universities in India. The concept of each topic is well discussed with a full length
presentation of numerical examples. Each example is unique in its own way and it is
graded in sequential manner. The book highlights and provides with easy methods
of getting solution. Select numerical problems worked out bring a test the capability
of the student’s knowledge and understanding of the basic concepts of the theory.
The organization of the book is as follows:
1. Control systems modelling and their representation
2. Time response analysis
3. Frequency response analysis
4. Stability analysis of linear control system
5. Root locus method for analysis
6. Design of compensators and PID controllers
7. State space modelling and analysis
In Chap. 1, representation of system in open loop and closed loop is discussed and
different terminologies associated with control theory are defined. The concept of
transfer function is explained and system modelling using force-voltage and force-
current analogies is discussed. Physical systems are represented in block diagrams
and signal flow graph and their reduction technique explained.
In control system study, the ultimate objective is to study of nature of variation of
different variables in the system with respect to time. In Chap. 2, transient and steady
state responses of first and second order systems are discussed and expressions for
the time domain specifications of step input are derived. The performance evaluation
of PID controllers when connected in cascade with prototype second order system
is also discussed using the results obtained from MATLAB simulation.
Even though performance of many real time control systems is hedged based on
the time response of the system, we get very valuable information when the system
is tested with sinusoidal input. The test study when the frequency is varied over
a wide frequency range is called frequency response. In Chap. 3, we discuss the
frequency response using the polar plot, Bod plot and Nichols plot. Expressions
for the frequency domain specifications are derived and the use of constant M and
constant N circles and Nichols chart explained.
When a control system is designed, it should be stable. If the system is not stable,
it is no use and causes severe damages not only to the system itself but also to the
environment and human life. In Chap. 4, the concept of linear system stability and
different ways of defining stability are discussed. The general stability conditions
for linear system stability are derived. The stability tests are carried out using Routh-
Hurwitz and Nyquist stability criteria.
It is well established that for the closed loop system to be stable, all the roots of the
characteristic equation should be in the left half of the s-plane. In Chap. 5, the root
locus method which gives the path of the characteristic roots traced out in the s-plane
as the gain of the open loop transfer function is varied is discussed. Root loci are
drawn for different types of transfer functions. From root locus plots determination
of closed loop system absolute stability, transient response, relative stability, phase
margin and gain margin is explained.
Preface ix

Control system design engineer has used his ingenuity to design the system to
give optimum performance. Still if the performance obtained by an individual is not
optimum, a compensator is inserted into a system in order to compensate for the
performance deficiency. In Chap. 6, the design methodologies of lead, lag, lag-lead
compensators, and PI, PD and PID controllers in the time domain as well as in the
frequency domain are discussed.
System representation in state space model has many advantages compared to
transfer function model representation. Recent development in modern control theory
has been possible with the state space model only. Here a system is represented as a set
of n first order differential/difference equation where physical variables are defined as
the state variables. In Chap. 7, continuous and discrete time system modelling in state
space has been derived and the solution of the state question obtained. Further, the
concepts of controllability and observability are explained with numerical examples.
I take this opportunity to thank Shri Sunil Sexana, Managing Director, Ane Books
Pvt. Ltd. India for coming forward to publish the book with the same enthusiasm as
he showed while publishing my other books earlier on. I would like to express my
thanks to Shri A. Rathinam, General Manager (South), Ane Books Pvt. Ltd. who
took the initiatives to publish the book in a short period of time. I am thankful to
my wife Dr. S. Manimegalai M.B.B.S., M.D., who inspires me whenever I write a
book. I also thank Mr. V. Ashok who has done a wonderful job to key a voluminous
book like this with atmost patience and care. Useful suggestions and constructive
criticisms are always welcome from the readers.

Tiruchirapalli, India Dr. S. Palani

Acknowledgements

The author would like to express his gratitude to the members of the teaching faculty
and student community for extending their patronage to the first edition of the book
Automatic Control Systems.
Few errors noticed here and there in the first edition have been corrected in the
present edition. A very exhaustive treatment of automatic control theory, a large
number of variety of memorial problems worked out and the inclusion of a large
number of short and long questions and answers in the exercise increased the volume
of the previous edition. To limit the number of pages in the present edition, some of
the contents such as PID Controller design and MATLAB programs for the design of
compensators and controllers have been shortened. Further, a few numerical prob-
lems worked out in the text have been shifted to the long type questions as exercise.
I hope that the reduced size of the book will be more comfortable for the students to
handle it. Since the first edition was exhaustively informative and contained a variety
of numerical examples, the author did not make any attempt to include new examples
to explain the theory.
The author would like to express his sincere thanks to Shri. Sunil Sexana,
Managing Director, ANE Books Pvt. Ltd., for his encouragement given to the authors.
I would also like to thank Shri. A. Rathinam, General Manager (South), ANE Books
Pvt. Ltd., for taking active steps in promoting the sales of the book. We also thank
Mr. V. Ashok for doing nice work in bringing the book in the present form.

Dr. S. Palani

xi
Contents

1 Control Systems Modelling and Their Representation . . . . . . . . . . . . . 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Basic Concepts and Terminologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Control System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.3 Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.4 Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.5 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.6 System Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.7 Plant or Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.8 Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.9 Servomechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.10 Regulator System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Concept of Feedback-Open Loop and Closed Loop Systems . . . . . 6
1.3.1 Open Loop System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.2 Closed Loop or Feedback System . . . . . . . . . . . . . . . . . . . 7
1.4 Comparison of Open Loop and Closed Loop Systems . . . . . . . . . . . 8
1.4.1 Open Loop Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.2 Closed Loop System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.3 Effects of Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.5 Examples of Closed Loop System . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5.1 Automobile Steering Control System . . . . . . . . . . . . . . . . 10
1.5.2 Regulator System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5.3 Liquid Level System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5.4 Temperature Control System . . . . . . . . . . . . . . . . . . . . . . . 14
1.5.5 Position Control System (Servomechanism) . . . . . . . . . . 15
1.5.6 Computer Controlled System . . . . . . . . . . . . . . . . . . . . . . . 16
1.6 System Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.7 The Concept of Transfer Function . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.7.1 Poles and Zeros of the Transfer Function . . . . . . . . . . . . . 19

xiii
xiv Contents

1.7.2 Complex s-Plane and the Locations of Poles

and Zeros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.7.3 Sinusoidal, Minimum Phase, Non-minimum
Phase and All Pass Transfer Functions . . . . . . . . . . . . . . . 22
1.8 System Representation in Simple Block Diagram . . . . . . . . . . . . . . 26
1.9 Transfer Function Model of Electrical Network . . . . . . . . . . . . . . . . 27
1.10 Transfer Function Model of Mechanical Systems . . . . . . . . . . . . . . 32
1.11 Transfer Function Model for Mechanical Translational
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1.12 Transfer Function Model for Mechanical Rotational System . . . . . 46
1.13 Transfer Function Model of Electro-Mechanical Systems . . . . . . . 54
1.14 Transfer Function Model of D.C. Motor . . . . . . . . . . . . . . . . . . . . . . 60
1.14.1 Transfer Function of Field Controlled D.C. Motor
(No Load) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
1.14.2 Transfer Function of Armature Controlled D.C.
Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
1.14.3 Transfer Function of D.C. Generator (On No Load) . . . . 64
1.14.4 Performance Comparison of Armature and Field
Controlled D.C. Motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
1.15 Two Phase A.C. Servomotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
1.16 A Pair of Synchros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
1.16.1 Synchro Generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
1.16.2 Control Transformer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
1.17 Electrical Analogy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
1.17.1 Force–Voltage Analogy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
1.17.2 Force–Current Analogy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
1.18 Block Diagram Reduction Technique . . . . . . . . . . . . . . . . . . . . . . . . . 86
1.18.1 Component Parts of Block Diagram . . . . . . . . . . . . . . . . . 87
1.18.2 Block Diagram Reduction Rule . . . . . . . . . . . . . . . . . . . . . 88
1.18.3 Block Diagram Reduction for Multiple Inputs . . . . . . . . . 103
1.18.4 Signal Flow Graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
1.18.5 Signal Flow Graph Algebra . . . . . . . . . . . . . . . . . . . . . . . . 110
1.18.6 Definitions of SFG Terminologies . . . . . . . . . . . . . . . . . . . 112
1.18.7 Application of the Gain Formula Between Output
Nodes and Non-input Nodes . . . . . . . . . . . . . . . . . . . . . . . . 127
1.18.8 Applications of the Gain Formula to Block
Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
1.18.9 Signal Flow Graph for the Electrical Network . . . . . . . . . 136
2 Time Response Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
2.2 First Order Continuous Time System . . . . . . . . . . . . . . . . . . . . . . . . . 165
2.2.1 System Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
2.2.2 Time Response of First Order System . . . . . . . . . . . . . . . . 165
2.2.3 Time Domain Specifications . . . . . . . . . . . . . . . . . . . . . . . . 167
Contents xv

2.3 Second Order System Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

2.4 Time Response of a Second Order System . . . . . . . . . . . . . . . . . . . . 174
2.4.1 Impulse Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
2.4.2 Step Response of a Second Order System . . . . . . . . . . . . 179
2.4.3 Time Domain Specifications of a Second Order
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2.5 Steady State Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
2.5.1 Type of the System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
2.5.2 Steady State Error Using Static Error Constants . . . . . . . 230
2.5.3 Steady State Error for Non-unity Feedback
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
2.5.4 Steady State Error for Disturbances . . . . . . . . . . . . . . . . . . 248
2.5.5 The Generalized Error Constants . . . . . . . . . . . . . . . . . . . . 250
2.5.6 Steady State Error When Closed Loop T.F. is Given . . . . 262
2.6 Performance Enhancement By Using Controllers . . . . . . . . . . . . . . 268
2.6.1 Rate Controller or Tachometer Feedback
Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
2.6.2 Proportional Plus Integral Plus Derivative (Three
Term Controllers) Controllers . . . . . . . . . . . . . . . . . . . . . . 274
2.6.3 Integral or Reset Controller . . . . . . . . . . . . . . . . . . . . . . . . 280
2.6.4 Proportional Plus Integral (PI) Controller . . . . . . . . . . . . . 284
2.6.5 Proportional + Derivative (PD) Controller . . . . . . . . . . . . 286
2.6.6 Proportional + Integral + Derivative (PID)
Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
3 Frequency Response Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
3.2 Obtaining Steady State Output to Sinusoidal Input . . . . . . . . . . . . . 329
3.3 Plotting Frequency Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
3.4 Polar Plot (Nyquist Plot) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
3.4.1 General Shape of Polar Plot and Type and Order
of the System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
3.4.2 Polar Plot of Non-minimum Phase Transfer
Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
3.4.3 Polar Plot with Transportation Lag (Time Delay) . . . . . . 349
3.4.4 Polar Plot of Prototype Second Order System . . . . . . . . . 351
3.5 Correlation Between Transient Response and Frequency
Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
3.6 Bandwidth of a Second Order System . . . . . . . . . . . . . . . . . . . . . . . . 354
3.6.1 Cut off Frequency and Cut Off Rate . . . . . . . . . . . . . . . . . 356
3.7 Bode Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
3.7.1 Bode Asymptotic Magnitude Plot . . . . . . . . . . . . . . . . . . . 366
3.7.2 Bode Asymptotic Phase Angle Plots . . . . . . . . . . . . . . . . . 374
3.8 Step by Step Procedure to Draw Bode Magnitude Plot . . . . . . . . . . 378
3.8.1 Transfer Function from Bode Plot . . . . . . . . . . . . . . . . . . . 397
xvi Contents

3.9 Log Magnitude Versus Phase Plot (Nichols Plot) . . . . . . . . . . . . . . . 407

3.10 Frequency Domain Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
3.10.1 Phase Margin and Gain Margin via Nyquist
(Polar) Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
3.10.2 Phase Margin and Gain Margin via Bode Plots . . . . . . . . 424
3.10.3 Phase Margin and Gain Margin Using MATLAB . . . . . . 432
3.10.4 Phase Margin and Gain Margin
via Log-Magnitude-Phase Plot . . . . . . . . . . . . . . . . . . . . . . 435
3.11 Determining Closed Loop Frequency Response from Open
Loop Frequency Response (Unity Feedback System) . . . . . . . . . . . 449
3.11.1 Constant M Circles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
3.11.2 Constant N Circles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
3.11.3 The Nichols Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
3.11.4 Nichols Chart from Constant M and Constant N
Circles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
4 Stability Analysis of Linear Control System . . . . . . . . . . . . . . . . . . . . . . . 501
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
4.2 The Concept of Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502
4.2.1 Asymptotic (Internal) Stability . . . . . . . . . . . . . . . . . . . . . . 503
4.2.2 Marginal (Neutral) Stability . . . . . . . . . . . . . . . . . . . . . . . . 503
4.2.3 Bounded Input Bounded Output (BIBO)
(External) Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
4.2.4 Relative Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
4.2.5 BIBO Stability via Impulse Response Function . . . . . . . 504
4.2.6 BIBO Stability and the Characteristic Roots . . . . . . . . . . 505
4.2.7 Relationship Between BIBO and Asymptotic
Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
4.3 Routh–Hurwitz Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517
4.3.1 Routh–Hurwitz Criterion to Determine Absolute
Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.3.2 Routh–Hurwitz Criterion: Special Cases . . . . . . . . . . . . . 524
4.3.3 Parameters Estimation via Routh–Hurwitz
Stability Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
4.3.4 Relative Stability Using Routh–Hurwitz Stability
Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
4.3.5 Routh’s Test for System with Transportation Lag . . . . . . 543
4.4 Nyquist Stability Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
4.4.1 Concepts Used in Nyquist Stability Criterion . . . . . . . . . 545
4.4.2 The Principle of the Argument—Cauchy’s
Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552
4.4.3 The Nyquist Stability Criterion . . . . . . . . . . . . . . . . . . . . . 553
Contents xvii

5 Root Locus Method for Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
5.1.1 Advantages of Root Locus Method . . . . . . . . . . . . . . . . . . 632
5.2 The Concept of Root Locus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
5.3 Properties of the Root Loci (Rules of the Root Loci) . . . . . . . . . . . . 636
5.4 Procedure to Construct Root Locus . . . . . . . . . . . . . . . . . . . . . . . . . . 643
6 Design of Compensators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723
6.2 Performance Criteria for Compensators . . . . . . . . . . . . . . . . . . . . . . . 724
6.3 Time and Frequency Domain Approaches . . . . . . . . . . . . . . . . . . . . . 726
6.4 Compensator Design in the Frequency Domain . . . . . . . . . . . . . . . . 727
6.4.1 Design of Lead Compensator . . . . . . . . . . . . . . . . . . . . . . . 727
6.4.2 Design Procedure for Lead Compensator
in the Frequency Domain . . . . . . . . . . . . . . . . . . . . . . . . . . 731
6.4.3 Lag Compensator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
6.4.4 Lag–Lead Compensator . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
6.5 Compensator Design in the Time Domain . . . . . . . . . . . . . . . . . . . . . 777
6.5.1 Design of Lead Compensator Using Root Locus . . . . . . . 777
6.5.2 Design of Lag Compensator Using Root Locus . . . . . . . . 787
6.5.3 Design of Lag–Lead Compensator Using Root
Locus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
7 State Space Modelling and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
7.2 The State of a System and State Equation of Continuous
Time System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
7.3 Vector Matrix Differential Equation of Continuous Time
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
7.3.1 State Equations for Mechanical Systems . . . . . . . . . . . . . 810
7.3.2 State Equations for Electrical Circuits . . . . . . . . . . . . . . . 817
7.4 State Equations from Transfer Function . . . . . . . . . . . . . . . . . . . . . . . 824
7.4.1 General Case of Representation—Phase Variable
or Controllable Canonical Form . . . . . . . . . . . . . . . . . . . . . 825
7.4.2 General Case of Representation-Observable
Canonical Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 828
7.5 Transfer Function of Continuous Time System from State
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 839
7.6 Solution of State Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 841
7.6.1 Laplace Transform Solution of State Equations . . . . . . . . 841
7.6.2 Time Domain Solution to State Equations . . . . . . . . . . . . 843
7.6.3 Determination of e At —The Cayley–Hamilton
Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 844
7.7 Controllability of Linear Continuous Time System . . . . . . . . . . . . . 854
7.7.1 State Controllability Condition . . . . . . . . . . . . . . . . . . . . . 854
7.7.2 Output Controllability Condition . . . . . . . . . . . . . . . . . . . . 856
xviii Contents

7.7.3
Controllability and Observability Tests
in the Frequency Domain . . . . . . . . . . . . . . . . . . . . . . . . . . 860
7.8 Observability of Linear Continuous Time System . . . . . . . . . . . . . . 861
7.8.1 Observability Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . 861
7.9 State Equation of Discrete Time System . . . . . . . . . . . . . . . . . . . . . . 865
7.9.1 Discrete Time State Equation in Controllable
Canonical Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 867
7.9.2 Observable Canonical Form Model . . . . . . . . . . . . . . . . . . 870
7.9.3 Diagonal Form (Parallel Form) Model . . . . . . . . . . . . . . . 873
7.9.4 Solution of State Equation . . . . . . . . . . . . . . . . . . . . . . . . . 877
7.10 Controllability and Observability of Discrete Time System . . . . . . 881
7.10.1 Controllability Condition for Discrete Time
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 881
7.10.2 Observability Condition for Discrete Time System . . . . . 882
7.11 Sampled Data System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 886
7.11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 886
7.11.2 Advantages of Sampled Data Control Systems . . . . . . . . 886
7.11.3 Disadvantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 887
7.11.4 A Sampled Data Closed Loop Control System . . . . . . . . 887
7.11.5 Sampling Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 887
7.11.6 Sampled Data System Variables . . . . . . . . . . . . . . . . . . . . . 888
7.11.7 Hold Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 890
7.11.8 Signal Reconstruction Using Zero Order Hold
Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 892
7.11.9 Transfer Function of a ZOH . . . . . . . . . . . . . . . . . . . . . . . . 893
7.11.10 The Sampling Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 893
7.12 MATAB Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 895
7.12.1 Conversion of State Space Model to Transfer
Function Model and Vice Versa . . . . . . . . . . . . . . . . . . . . . 895
7.12.2 Conversion of Transfer Function to State Model . . . . . . . 897
7.12.3 To Write a Program to Obtain the STM . . . . . . . . . . . . . . 898
7.12.4 To Determine Controllability and Observability
of the System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 898

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 905
About the Author

Dr. S. Palani obtained his B.E. degree in Electrical Engi-

neering in the year 1966 from University of Madras,
M.Tech. in Control Systems Engineering from IIT,
Kharagpur in 1968 and Ph.D. in Control Systems
Engineering from University of Madras in 1982 with
academic excellence. He has a wide teaching expe-
rience of over four decades. He started his teaching
career in the year 1968 at the erstwhile Regional Engi-
neering College (Now National Institute of Technology),
Tiruchirappalli in the department of EEE and occu-
pied various positions. As Professor and Head, he took
the initiatives to start the Instrumentation and Control
Engineering Department. After a meritorious service of
more than three decades in REC, Tiruchirappalli, he
joined Sudharsan Engineering College, Pudukkottai as
the Founder Principal. He established various depart-
ments with massive infrastructure. Since 2006, he is the
Director in the same college.
Dr. S. Palani has published more than hundred
research papers in reputed international journals and has
won many cash awards. Under his guidance and super-
vision many research scholars have been awarded Ph.D.
He has carried out several research projects funded by
Government of India and other agencies. As the theme
leader of the Indo - UK, REC Project on energy, he
has visited many Universities and industries in United
Kingdom. He is the author of the popular books titled
Automatic Control Systems, Signals and Systems, Basic
System Analysis, Digital Signal Processing (ECE),
Digital Signal Processing (EEE), and Digital Signal
Processing (CSE).

xix
Chapter 1
Control Systems Modelling and Their
Representation

Learning Objectives

After completing this chapter, you should be able to:

 Represent the system in open loop and closed loop forms.
 Define various terminologies related to control system.
 Write the dynamic equations of mechanical, electrical and electromechan-
ical systems and derive their transfer functions.
 Model mechanical systems by Force–Voltage and Force–Current analogies.
 Derive the transfer function of commonly used control system components.
 Represent the system in block diagram and to study block diagram reduction
technique.
 Represent the system by signal flow graph and to study the application of
Mason’s gain formula.

1.1 Introduction

Automatic control has played a vital role in the advancement of science and engineer-
ing. It finds wide applications in space vehicle systems, missile guidance systems,
robotic systems, anti-aircraft gun control and radar systems. It also finds applica-
tions in manufacturing and industrial process control, autopilot systems, aerospace
industries, design of cars, trucks and war weapons, and in industrial operations such
as controlling pressure, temperature, humidity, viscosity, flow etc. The use of digital
computers as the integral part of modern control system has enhanced its applications
in recent times.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 1

S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_1
2 1 Control Systems Modelling and Their Representation

Thus, control engineering is not limited to any particular engineering but it is

applicable to various fields of engineering such as chemical, aeronautical, mechani-
cal, electrical, civil and environmental engineering. Without the advances in control
system design, space travel would be impossible, music system would not be so
effective, industrial production would be very low and the applications of computers
would be severely limited. The application of control theory and technology is so
widespread today that one can hardly find a working engineering system without
some form of control circuitry or control procedure in it.
In today’s highly competitive global economy control engineers have to produce
the best quality products at the lowest cost. For the industrial control engineers,
improving the product quality and minimizing the cost have been the major tasks.
They look at the control systems, at the latest advances in sensors, actuators, con-
trol methods and computer systems and recommend how plant performance can be
improved for a much lower capital outlay without rebuilding the existing plant which
requires huge amounts of capital expenditure. In recent years modern health services
have undergone a radical change due to advances in control, sensors, actuators and
information technology.
Kidney dialysis, machines, pacemakers, artificial hearts, automation procedures
in surgeries, anaesthesia procedures and radiation treatment all use control methods.
Control engineering also plays an important role in many other areas of industries
such as the processes in pharmaceutical plants, steel works and automobile engine
management system. An aircraft control system is a vital control engineering appli-
cation. Much of our space technology would not be where it is now without control
techniques.
Wind turbine systems are probably the most widely used renewable energy method
for generating electric power. To connect the wind generator to the grid, the speed of
rotation is maintained by the grid frequency. The design and implementation prob-
lems call for control, mechanical and electrical engineering skills. However, it is
the control engineer who will integrate the various aspects to produce a working
system. Many of the most advanced control and instrumentation systems are found
in the aerospace and space industries. Almost all the basic maneuvers performed
by an aircraft, such as landing, take-off and directional control are automated and
implemented by computer systems. A fully integrated flight control system maxi-
mizes aircraft performances while ensuring crew and passenger safety. Such a system
reduces the burden of decision making and ease the task faced by the human crew.
Due to the environmental disturbances, like wind, wave and sea current, a ferry
travelling in a sea experience roll which is unpleasant and unsafe for the travel.
Different types of control systems are designed to compensate for the effects of
these disturbances and make the sea transport more safe and comfortable.
From the many examples described above, we find that control engineering is an
important feature in many aspects of our life. Thus, control is everywhere on which
our society depends. In the life style of today, we rely heavily on technology and its
applications. Control engineers are needed to analyse these technological systems
and provide control systems without which modern society would simply be not
possible.
1.2 Basic Concepts and Terminologies 3

(a)

Input Plant
or Output
Process

(b) Disturbance

Input Plant
Controller or Output
Process

Fig. 1.1 a A plant to be controlled and b A plant with a controller

1.2 Basic Concepts and Terminologies

As mentioned earlier, there are many control systems that we meet in our daily life.
Any portion of a system that we want to control is called a plant or process. The
input signal which is applied to the system interacts with the plant and produces
the output. For the given plant, to get the desired output, which is our objective we
should give appropriate input signal. This is shown in Fig. 1.1a. The control system
designer job is to ensure that the plant operates as required. When disturbances
occur in the system a controller is inserted in between the plant and the input. The
controller output signals produce desired output from the plant. This is represented in
Fig. 1.1b. In the analysis and design of control system, various components which are
interconnected are represented in block diagram form as shown in Fig. 1.1. Arrows
indicate the direction of signal flow. When complex systems are studied, it becomes
easier to understand when they are represented in block diagrams. When we talk
about control systems, we come across several terminologies and some of them are
very commonly used. The definitions of these terminologies are given below.

1.2.1 System

A system is defined as a set of interconnected objects with a definite relationship

between the objects and attributes. It is also defined as the interconnection of elements
and devices for a desired purpose.
4 1 Control Systems Modelling and Their Representation

1.2.2 Control System

Control systems are the systems that implement certain objectives. The interconnec-
tion of connection of components forming a system configuration that provides the
desired output is also called control system.

1.2.3 Variables

Variables are the attributes that are present at different parts of the system.

1.2.4 Input

Input of a system is an attribute of a system which is varied directly and independently

by the operator.

1.2.5 Output

Output of a system is an attribute that varies as a result of variation in the input. It

cannot be independently varied by the operator.

1.2.6 System Parameters

Parameters are the properties of the given system which is inherently present in the
system.

1.2.7 Plant or Process

The device which is under control is called plant or process.

1.2.8 Automation

The control of a process by automatic means is called automation.

1.2 Basic Concepts and Terminologies 5

Fig. 1.2 a Mechanical (a)

system and b Block diagram x(t)
representation of mechanical K
system

M f(t)

(b)

f(t) x(t)
M, B, K
Input Output

1.2.9 Servomechanism

The control system which follows the frequently changing reference signal is called
servomechanism. It is usually referred to system whose output is mechanical position
or velocity.

1.2.10 Regulator System

If the reference or set point signal is kept at a steady value for a long period, and the
output remains at the prescribed set level rejecting disturbance signals, it is called a
regulator system. The above terminologies are explained below.
Consider the mechanical system consisting of mass M, dash-pot B and spring K .
f (t) is the force applied and x(t) is the displacement of mass. The elements M, B
and K are interconnected and form as a translational mechanical system. f (t) is the
input variable (attribute) and x(t) is the output variable (attribute). The following
dynamic equation is written for the above mechanical system:

d 2 x(t) d x(t)
M 2
+B + K x(t) = f (t) (1.1)
dt dt

f (t) is the input attribute and it can be independently varied by the operator as the way
he wants. However, the output attribute x(t) can be varied by varying the input f (t).
M, B, and K are the objects (parameters). These objects are the inherent properties
of the system and are fixed at a particular operating condition. Since there is a definite
relationship between these objects and the attribute x(t) it is called a system. When
6 1 Control Systems Modelling and Their Representation

the input is applied, opposing forces are generated by these objects. The mass M
2
generates the opposing force M d dtx(t)
2 , the dash-pot generates B dt
d x(t)
and the spring
generates K x(t). Hence, the interconnection of M, B and C is called a system
according to the definition. Further, if the objective is to change the displacement
x(t), it can be achieved by changing the input f (t). Hence, it is a control system.

1.3 Concept of Feedback-Open Loop and Closed Loop

Systems

1.3.1 Open Loop System

A simple D.C. motor speed control system is shown in Fig. 1.3a. The speed ω of the
motor depends on the motor armature voltage Va . The armature voltage is adjustable
with a help of a potentiometer. If the load on the motor increases, the speed ω drops.
To maintain the shaft speed constant, the armature voltage Va had to be increased.
If the load on the motor decreases, the shaft speed increases. Now the armature
voltage Va has also to be decreased accordingly. Thus, adjusting the potentiometer
setting according to the load variations requires manual operation which may require
constant vigil on the part of the human operator which may sometime be not accurate.
This type of the system is called open loop system. In the above system Va is the
input and it causes control action. Here ω, the speed is the output. When there is a
change in the output, it does not have any influence on the control action. Thus, the
open loop system is defined as follows.
An open loop system is defined as the system in which the control action
(actuating signal) is independent of the output or the desired result.

(a)

L M E

Va

(b)
Va
M Load
Control action

Fig. 1.3 a D.C. motor speed control system and b Block diagram representation of speed control
system
1.3 Concept of Feedback-Open Loop and Closed Loop Systems 7

1.3.2 Closed Loop or Feedback System

Now consider D.C. motor speed control system shown in Fig. 1.4a. Here, an addi-
tional component, namely a D.C. tachogenerator T is connected to the motor and
load shaft. The tachogenerator generates voltage proportional to speed. The output of
this tachogenerator VT is compared with the reference voltage Va . The difference of
these voltages is Va − VT , as per the connection shown in Fig. 1.4a. Now we say the
output is fedback and compared with the reference input. The system now becomes a
feedback or closed loop system. For the connection shown in Fig. 1.4a, the feedback
is negative. If the polarity of the tachogenerator is reversed the feedback is positive.
The working principle of the feedback is explained below.
With the feedback loop kept open and on no load condition, the reference input
Va is adjusted using the potentiometer such that motor runs at desired speed ω.
When the load is connected to the machine shaft, the speed ω decreases. Since VT
is proportional to ω, this voltage also decreases. Va − VT which is applied to the
armature increases which ultimately increases the shaft speed and brought to the
desired value. Now suppose, a portion of the load is removed. This will increase the

(a)

T L M E

Va

Va VT

VT

(b)
L
Error detector

Va (Va VT)
M
Reference Input Desired output
VT Control action
or actuating
signal
T
Feedback

Fig. 1.4 a Closed loop D.C. motor speed control system and b Block diagram of closed loop D.C.
motor speed control system
8 1 Control Systems Modelling and Their Representation

shaft speed ω which is undesirable when constant speed operation is desired. This
increase in ω increases VT . The control action (Va − VT ) decreases which decreases
the shaft speed ω. Thus, when there is disturbance in the load side, it is rejected
automatically by the feedback connection and the desired output is restored may be
with marginal acceptable deviation. The control action (Va − VT ) is also called by
the name Actuating Signal or Error Signal.
On the other hand, consider the system with a positive feedback. Now the actuating
signal becomes (Va + VT ). When the load is connected to the motor shaft, ω decreases
and also VT . The control action decreases and ω decreases further. This is a cumulative
process and finally the speed becomes zero. Suppose, while running, a part of the
load is removed. Now the speed, ω increases which results in increase of (Va + VT ).
As a result ω also increases in a cumulative manner and may reach a dangerously
high speed causing heavy damage to the entire system. Now we say, the system is in
unstable mode of operation. This is because of positive feedback. For all practical
purposes, only negative feedback is employed and only in selective cases such as
design of oscillators positive feedback is used. Now we give below the definition of
closed loop system and then give the advantages and disadvantages of open loop and
closed loop systems.
A closed loop system is defined as a system in which the measured output is
compared with the reference input. Here the control action depends upon the
changes in the output.

1.4 Comparison of Open Loop and Closed Loop Systems

1.4.1 Open Loop Systems

1.4.1.1 Advantages

1. Design and construction are simple and economical.

2. Almost all physically realizable open loop system are stable.

1.4.1.2 Disadvantages

1. They are less accurate and due to manual operation sometimes become unreliable.
2. They do not compensate for the effect of internal and external disturbances which
result in corrupted output and degrade the performance of the system.
1.4 Comparison of Open Loop and Closed Loop Systems 9

1.4.2 Closed Loop System

1.4.2.1 Advantages

1. Closed loop systems have much greater accuracy than the open loop systems.
2. Due to feedback, they do not require manual supervision. Further, they are fast
acting in removing the error.
3. They reject internal and external disturbances and minimize their effects.
4. Closed loop systems have higher bandwidth. This property gives better transient
response in the time domain. The range of frequencies over which the system
responds is increased.
5. The sensitivity of the system for parameter variation, noise, disturbances and
environmental changes are made small by increasing the gain of the feedback loop.
6. If the open loop system is stable by proper choice of feedback gain the system
can be made stable.

1.4.2.2 Disadvantages

1. Because of feedback, the overall gain of the system is reduced.

2. It requires more components compared to open loop system. This increases the
cost.
3. Since more components are connected the system becomes more complex. Sys-
tem design and analysis become more difficult.
4. In general closed loop system is less stable. But by proper design, it can be well
stabilized with faster response and higher accuracy. In view of these important
properties, closed loop systems are invariably used and not the open loop system.

1.4.3 Effects of Feedback

The effects of feedback can be summarized as given below.

1. Negative feedback reduces the overall gain of the system while positive feedback
increases it. Further, in one frequency range, the overall gain is increased while
in another frequency range, the overall gain is reduced for the given feedback.
2. Feedback, if properly designed will improve the stability. Improper design will
make the system unstable.
3. The sensitivity function can be made small by increasing the feedback gain. This
will remove the harmful effect of parameter variation.
4. Feedback reduces the effect of noise and disturbances.
5. Feedback increases the bandwidth of the system. This increases the fastness of
time response of the system in the time domain and increases the frequency range
of operation in the frequency domain.
10 1 Control Systems Modelling and Their Representation

1.5 Examples of Closed Loop System

1.5.1 Automobile Steering Control System

A closed loop automobile steering system controlled by a human operator is shown

in Fig. 1.5. Here θi is the desired direction of the automobile. The actual direction of
the car is judged by the naked eyes of the driver. The eyes compare the difference
between the desired direction and the actual direction of the vehicle and this message
is sent to the brain of the driver. The arms of the driver steers the wheel in such a
direction such that (θi − θ0 ), the error is zero or minimum. Thus, the above system
is an example of a closed loop system with man-made controllers.

1.5.2 Regulator System

Consider the voltage regulator system shown in Fig. 1.6 where a D.C. generator run
by a prime mover supplies electrical load. G is the gain of the D.C. generator which is
expressed in volts per field ampere at a particular speed condition. The block diagram
of the system is shown in Fig. 1.6b. Here the reference voltage is Vr and is applied
to the field winding of the generator. The system operates in open loop mode since
there is no feedback provided. Under no load condition, the reference voltage Vr is
adjusted such that the induced voltage Vg in the generator is the same as the load
rated voltage Vt . If the generator is loaded, the current through the armature of the
generator Ia also increases. At this stage the following equation applies:

Vg − Ia Ra = Vt (1.2)

As the load current (same as armature current) increases, Ia Ra drop increases which
results in decrease of load terminal voltage Vt , which is undesirable. To solve the
problem, the reference voltage is adjusted (increased) using the potential divider.
The flux in the field winding of the D.C. generator is thus increased and also emf Vg .
Thus, the Ia Ra drop is compensated. Suppose, while operating on load, a portion of
it is switched off. This results in decrease of Ia which as per Eq. (1.2) increases the

Error dectector (eyes)

i Error Automobile 0
Driver Steering wheels and
Desired i 0 engine Actual
direction direction
of travel of travel

Fig. 1.5 A closed loop automobile steering system

1.5 Examples of Closed Loop System 11

If Ia
(a)

Ra

Rf

E G Vg Vt Load
Vf

Vr

IaRa
(b)

Va=Vf If Vg Vt
1 G Load
Rf

Ra Ia
(c) If

Rf
R1
E G Vg Vt Load
Vf
R2
Vr

Vb S

IaRa
(d)

Vr Vf If Vg Vt
1 G Load
Rf
Vb

R2
R1+R2

Fig. 1.6 Voltage regulator system. a Schematic diagram of open loop regulator system; b Block
diagram of open loop regulator system; c Schematic diagram of closed loop regulator system and
d Block diagram of closed loop regulator system
12 1 Control Systems Modelling and Their Representation

load terminal voltage. If this condition is allowed to continue, due to over voltage
the load will be damaged.
Now, according to the changes in the load current, the reference voltage Vr is
decreased such that, the load terminal voltage is kept at the required rated voltage.
Thus, all the time close inspection of the variation of the load is necessitated and the
potentiometer setting has to be adjusted manually. Thus, the accuracy and fastness
of maintaining the voltage regulation depend upon the manual skill, which on many
occasions may not be so accurate.
Now consider the voltage regulator system where feedback is provided which is
shown in Fig. 1.6c and its block diagram in Fig. 1.6(d). A portion of the signal Vt is
taken and compared with the reference input Vr . They are related by the following
equation
R2
Vb = Vt
(R1 + R2 )
V f = (Vr − Vb ) (1.3)

The above feedback is negative since the negative of the feedback voltage is connected
to the negative of the reference voltage while their positive points are connected
when comparison is done. The system becomes positive feedback if the polarities of
the feedback voltage are reversed. The operation of the negative feedback voltage
regulator system is explained below.
To start with, under no load condition the feedback loop is broken by using the
switch S. The reference voltage Vr is adjusted until the load terminal voltage is at
its rated value. Now the feedback loop is closed using the switch S and the load
is connected to the main generator. When there is no disturbance, the load terminal
voltage will be retained at the prescribed set level. Suppose more load is connected to
the generator. This causes increase in the load current and hence the armature current.
Since R1 and R2 are very high value resistors, the current drawn in that branch is
negligible and for all practical purposes the load current and generator current are
same.
When the armature current increases when the load in increased, the armature
voltage drop Ia Ra also increases. This results in drop in Vt which is not desir-
able. However, the drop reflects in Vb also. Now, V f = (Vr − Vb ) increases which
increases the excitation of the generator and hence the induced emf Vg . Thus, the
fall in Vt is automatically corrected without any manual supervision. On the other
hand, if some load is dropped, Vt and hence Vb increases and V f decreases. Thus,
Vg is decreased and hence Vt is retained at the prescribed set level. Thus, the system
becomes automatically a controlled one.
1.5 Examples of Closed Loop System 13

1.5.3 Liquid Level System

Consider the open loop liquid level system shown in Fig. 1.7a. The liquid level in
the tank is adjusted by manual operation of the inlet valve attached to the supply
pipe. Through exit pipe, the water is consumed. If the consumption is more than the
supply, the tank becomes empty. On the other hand if the consumption is less than
the supply, the water overflows. Thus, according to the consumption rate, the inlet
valve has to be adjusted manually so that a desired level of water is maintained in
the supply. This is an example of open loop regulator system.
Now, consider the same system shown in Fig. 1.7b with a float and a pivot mech-
anism incorporated and made as a closed loop system. Let us assume that, initially,
the water in the tank is in the prescribed set level. If water is consumed through exit
pipe, the water level falls and the float comes down. Due to the pivot action, the arm
of the rod connected to float comes down while the arm connected to the control
valve goes up. This allows more opening for the inflow and fills the water tank until
it reaches the set level. If the liquid level goes up, the arm A of the pivot goes up
and arm B goes down. This pushes the control valve down and narrows the gap of
incoming flow of the liquid. Once, the level of the liquid in the tank reaches the
desired value, the control valve is completely closed and the inflow of the liquid in
the tank is stopped. At this level, pivot arm is horizontal. This is the mechanism used
in toilet bowel flushing system.

Fig. 1.7 Liquid level system. a Open loop system and b Closed loop system
14 1 Control Systems Modelling and Their Representation

1.5.4 Temperature Control System

A simple closed loop temperature control system is shown in Fig. 1.8a and its block
diagram in Fig. 1.8b. A process industry requires a constant temperature θ of material
being mixed in a tank be held constant. The input is ‘r ’ a voltage and is calibrated
in terms of the desired temperature. The actual temperature of the process in the
tank is ‘θ ’ and is measured by means of a thermocouple immersed in the tank. The
voltage produced in the thermocouple is amplified to produce a voltage b = K b θ .
This voltage is fedback and compared with the reference voltage r . The error signal
(actuating signal) e = (r − b) and being a weak signal it is amplified by an amplifier
to get the output voltage ei = K a e. This voltage energizes the solenoid whose current
i s = (ei /R). The solenoid produces the force f s = K s i s and pulls the mass M which
includes the mass of the boiler valve, spring K and viscous friction damping B.
This activates the boiler valve which lets the steam out. The position of the valve is
denoted as x. The valve position in turn controls the flow q of the hot steam from
the boiler into the tank. The temperature θ of the tank is related as θ = K c q.
The block diagram of the temperature controlled process is shown in Fig. 1.8b.
It is desired to keep the temperature θ of the tank constant by feedback. When the
temperature falls due to disturbance or load, it is sensed by the thermocouple and
its output is now reduced than what it was and b is reduced. The actuating signal
e = r − b is increased which increases the excitation current in the solenoid. The
force f s generated by the solenoid is increases the displacement x of the boiler valve
opening. Now more steam is let into the tank which increases the temperature θ of the
process in the tank. The reverse process happens when θ is increased. This time the
actuating signal is reduced which in turn reduces the pull generated by the solenoid

(a) Tank
Boiler Valve
q

E x
Amp ei is
Steam
r e
Ka exhaust
b
Solenoid M, B, K
Thermo
Amp couple
(b)

r e ei is fs x q
Ka 1 Ks M, B, K Kq Kc
R
b

Kb

Fig. 1.8 Temperature control system. a Process and b Block diagram

1.5 Examples of Closed Loop System 15

and hence x. This restricts the flow of steam into the tank which ultimately reduces
its temperature to the prescribed set value which is performed automatically.

1.5.5 Position Control System (Servomechanism)

A position control of an anti-aircraft gun system is shown in Fig. 1.9a and its block
diagram in Fig. 1.9b. θi is the angular position of the input shaft and θ0 the angular
position of the output shaft. These shafts are connected to the movable arms of the
potentiometer which sends out signal according to the following equation:

Vi = K p (θi − θ0 ) (1.4)

where K p is expressed in volts per radian error. This signal is weak and it is amplified
by an amplifier with a gain A whose output is v0 . This voltage provides excitation to
the field winding of a D.C. motor whose electromagnetic torque is given by

Td = K T I f (1.5)

This electromagnetic torque is used to drive the load which is the anti-aircraft gun.
The gun is connected to a worm-gear arrangement. To increase the load torque, the
output shaft of the motor and the worm gear are connected with gears of N teeth
on the motor shaft and N L teeth on the load side where N L > N . The output arm
of the potentiometer is connected to the motor shaft through gear Nc . The input

(b)

i i 0 Vi Vo Motor m N L
Kp A Load
KT NL
0

N
NC

Fig. 1.9 Position control of anti-aircraft gun system. a Schematic diagram and b Block diagram
16 1 Control Systems Modelling and Their Representation

arm of the potentiometer is controlled by the radar signal which is converted into
angular position. Suppose, the radar signal makes the input arm rotate through 30◦
in the clockwise direction. An error voltage Vi proportional to 30◦ appears and is
amplified by the amplifier. The excitation current in the field winding of the D.C.
motor interacts with the current carrying conductors of the D.C. motor armature and
produces electromechanical torque which drives the anti-aircraft gun. The objective
is to make the load also move through 30◦ in the same clockwise direction. Once
θ L = 30◦ , the difference of (θi − θ0 ) = 0 and the motor stops its activation. Similar
operation is carried out for anti-clock wise rotation. Since the output is mechanical
positioning the system is also called by the name servomechanism.

1.5.6 Computer Controlled System

Computer controlled system have now become very popular and intentionally intro-
duced to control multi-variables in the plant. For large scale systems and systems
with more complexity, the digital computer is successfully used as a controller. The
computer can also perform supervisory functions such as scheduling many required
applications in addition to providing closed loop control. The block diagram of a
typical computer control system to control plant variables is shown in Fig. 1.10. The
input r (t), output c(t) and error e(t) are all continuous signals. The error signal
e(t) is converted into a sequence of numbers m(kT ) by means of analog to digi-
tal (A/D) converter where T is the sampling time. The A/D converter samples e(t)
into e(kt). This process is called discritization. The discritized signal is converted
into m(kT ) and applied to the digital computer. The output of the digital computer
appears as u(kT ). Since the plant requires continuous time signal u(t), a digital to
analog converter (D/A) is used between the digital computer and the plant. In the
computer hardware, there is a clock which sends a pulse every T seconds to A/D
and D/A converters. The A/D converter sends a number to the digital computer. The
A/D converter, digital computer, D/A converter and the clock are all represented by a
sampling switch followed by a data extrapolator (hold circuit). Appropriate software
is developed to modify the control law in the digital computer. If the output deviates
from the desired one, an error appears as e(t), and this error is processed in the digital
computer which sends out the signals which correct the deviating variables.

r (t) e (t) A/D m (kT) Digital u (kT) D/A u (t) c (t)

Computer Plant
Converter Converter
c (t)

Clock

Fig. 1.10 Block diagram of a computer controlled system

1.6 System Classification 17

1.6 System Classification

For convenience of description and mathematical analysis, systems are classified

according to the nature of inputs, number of inputs/outputs inherent characteristics,
design solution components used etc. Systems are broadly classified as given below:
1. Linear and non-linear.
2. Time varying and time invariant.
3. Continuous time system and discrete time system.
4. Single input–single output (SISO) and multi-input–multi-output (MIMO).
5. Deterministic and stochastic system.
6. Lumped parameter and distributed parameter system.
7. Optimal/adaptive/fuzzy type/robust control systems.

1.7 The Concept of Transfer Function

In Sect. 1.3, the dynamic behaviour of a simple mechanical system was represented
by the differential Eq. (1.1) relating the input and output of the system. However,
it is not a so satisfying representation from system perspective. The time domain
representation of Eq. (1.1) is not a convenient way when subsystems are connected
in cascade. Some analysis are easier to perform in the frequency domain. Converting
the time domain model to frequency domain model of a linear time invariant con-
tinuous time system is done using Laplace transform. This transformation converts
differential equation into algebraic expressions which are easier to manipulate. The
Laplace transform converts functions with a real dependent variable such as time
into functions with a complex dependent variable such as frequency which is often
represented by s which is a complex variable and expressed as s = σ + jω. The
transfer function is defined as follows.
The transfer function of a linear time invariant system is defined as the ratio
of the Laplace transform of the output variable to the Laplace transform of the
input variable with all initial conditions assumed to be zero.
The transfer function is not applicable (defined) to non-linear and time varying
systems. Since the transfer function gives the description relating the input-output of
the system, the information about the internal structure and its behaviour is missing
which is the major disadvantage of this model. Now taking Laplace transform on
both sides of the Eq. (1.1) we get

(Ms 2 + Bs + K )X (s) = F(s)

X (s)
= G(s)
F(s)
X (s) 1
= (1.6)
F(s) (Ms + Bs + K )
2
18 1 Control Systems Modelling and Their Representation

Fig. 1.11 Block diagram

representation of transfer F (s) X (s)
G (s)
function

The transfer function G(s) of system described by Eq. (1.1) is given in Eq. (1.6) and
is represented in the block diagram as shown in Fig. 1.11.
From Fig. 1.11, it is to be noted that the transfer function is independent of the
input. Further, it is not a function of the real variable, time or any other independent
variable. While transfer function of a continuous data system is expressed as the
function of the complex variable s together with the system parameter, the transfer
function of a discrete time system (described by difference equation) is the function
of z.
In Eq. (1.6), the denominator is a polynomial in s. System transfer function may
have polynomial in the numerator also. If the order of the numerator polynomial
is less than that of the denominator polynomial the transfer function is said to be
strictly proper. Transfer function whose order of numerator polynomials is the
same as that of the denominator polynomial, is called proper transfer function. If
the order of the numerator polynomial is greater than that of the denominator, such
a transfer function is called improper. If the denominator polynomial is set to zero,
the equation so obtained is called characteristic equation. Transfer function is also
defined for multi-variable system which has multiple inputs and outputs. Hence, the
relationship is between an input–output pair when all other inputs are set to zero. The
total effect on any output due to all the inputs acting simultaneously is obtained by
adding up the outputs due to each input acting alone. The principle of superposition
is applied since the system considered is linear.
Consider the following transfer function whose numerator and denominator are
the polynomials in s.

bm s m + bm−1 s m−1 + · · · + b0
G(s) = (1.7)
an s n + an−1 s n−1 + · · · + a0

In Eq. (1.7) if n > m, the transfer function is said to be strictly proper, if n = m, the
transfer function is proper and if n < m, the transfer function is improper. Further if
the denominator polynomial is equated to zero, we get

an s n + an−1 s n−1 + · · · + a0 = 0 (1.8)

Equation (1.8) is called the characteristic equation. In Eq. (1.7) if the coefficient
an = 1, the denominator polynomial is called monic polynomial. Equation (1.8)
is called as characteristic equation because it decides the stability of the system.
Further, the transient and steady state responses, to a large extend depends upon the
characteristic equation.
1.7 The Concept of Transfer Function 19

A Historical Note: Pierre-Simon Laplace (1749–1827)

Pierre-Simon, Marquis de Laplace was born on

23rd March, 1749 in France to Pierre Laplace and
Marie-Ancesochan, who were farm laborers. Hav-
ing his school education in his native village nor-
mondy, he was sent to university of Caen for higher
studies in theology. Laplace did not study theology
but was sent to Paris to work under D’Alembert
by his enthusiastic teachers who awoke his zeal
for mathematics. D’Alembert a mathematic genius
received Laplace very poorly and wanted to get
rid of him. He gave him a very thick mathematics
book and asked him to come back when he had
read it. Laplace came back over night. D’Alembert
was very impressed by the way Laplace answered
all his queries which were difficult to answer and
recommended him for a teaching place in a military college. For the next seven-
teen years from 1771–1787, he devoted his time for the original research work in
astronomy.
Laplace was elected as the associated member in the French academy of science
at the age of 24. During the period 1799–1825 he published his works in celestial
mechanics in five volumes. He formulated Laplace’s equations, Laplace transform
and the Laplacian differential operator which are widely used now in college level
mathematics. One of the biggest contributions of Laplace is the development of neb-
ular hypothesis of the origin of the solar system and developed the theory of black
holes and the development of gravitational collapse. He is remembered as one of
the greatest scientists of all time and sometimes referred to as a French Newton. He
became a Count of the French empire in 1806 and was named a Marquis in 1817.
Laplace contemporaries include the great mathematical wizards like Lagrange, Leg-
endre, Monge, Lacorse, Fourier and French ruler Napoleon Bonaparte who himself
was great scientist. Laplace married Marie-Charlotte in his late thirties and the cou-
ple had a daughter, Sophie and a son, Charles-Emile. The all time great scientist died
on 5 March, 1827 in Paris at the age of 77.

1.7.1 Poles and Zeros of the Transfer Function

Consider the transfer function given in Eq. (1.7). The numerator and denominator
polynomials can be written in factors as given below:
20 1 Control Systems Modelling and Their Representation

(s + z 1 )(s + z 2 ) · · · (s + z m )
G(s) = (1.9)
(s + p1 )(s + p2 ) · · · (s + pn )

In Eq. (1.9) the factors in the numerator are called zeros of the transfer function and
the factors in the denominator are called poles of the transfer function. The poles
and zeros can be located in the complex s-plane at s = − p1 , s = − p2 etc., and the
zeros at s = −z 1 , s = −z 2 etc. The poles locations are marked with a cross × and
the zero locations are marked with a small circle . It is also noticed in Eq. (1.9),
at s = − p1 or s = − p2 etc., the T.F. (transfer function) becomes infinitive and at
s = −z 1 or s = −z 2 etc. the T.F. becomes zero. p1 , p2 , . . . , pn and z 1 , z 2 , . . . , z n
may be real or complex. Thus, the factors of the denominator of the transfer function
which are called zeros are defined as “the points in the complex s-plane at which
the transfer function becomes zero”. Similarly the factors of the denominator of
the transfer function which are called poles are defined as “the points in the complex
s-plane at which the transfer function becomes infinitive”.

1.7.2 Complex s-Plane and the Locations of Poles and Zeros

The complex variable s = σ + jω and the s-plane is shown in Fig. 1.12. The hori-
zontal axis represents the positive and negative real parts and the vertical axis repre-
sents positive and negative imaginary parts. It contains four quadrants as shown in
Fig. 1.12. The infinite space complex s-plane is divided by the vertical imaginary as
Left Half Plane (LHP) and Right Half Plane (RHP). The importance of this division
plays significant role in the system stability and the transient response studies which
will be dealt with in the chapters to follow.
Now let us consider the location of the poles and zeros in the complex s-plane.
Consider the following transfer function:

Positive imaginary
j
LHP
RHP
s-plane
2nd quadrant
1st quadrant

Negative real Positive real

0

LHP RHP
3rd quadrant 4th quadrant
j
Negative imaginary

Fig. 1.12 Complex s-plane diagram

1.7 The Concept of Transfer Function 21

j
2 j3

3 j s-plane

0 2
3 2 1

3 j

2 j3

Fig. 1.13 Pole-zero diagram of Eq. (1.10)

10(s − 2 + j3)(s − 2 − j3)(s + 1)

G(s) = (1.10)
s(s + 2)2 (s + 3 + j)(s + 3 − j)

In Eq. (1.10) there are three zeros and five poles. Thus, the order of the numerator
polynomial is three and that of the denominator is five. In all practically realizable
system the order of the denominator polynomial is higher than that of the numerator
polynomial. Hence, system Eq. (1.10) is strictly proper. The characteristic equation is
s(s + 2)2 (s + 3 + j)(s + 3 − j) = 0. System Eq. (1.10) is called fifth order system.
The order of the system is defined as follows.
The order of the system is defined as the highest order of the denominator
polynomial or the numerator polynomial whichever is greater.
Now we place the poles and zeros of Eq. (1.10) in the s-plane as shown in Fig. 1.13.
The poles and zeros are located in the s-plane by equating the pole factors and the
zero factors to zero. For example consider the zero (s + 1). When this is equated to
zero we get s = −1. At negative real with one magnitude, this zero is located with the
marking 0. It is to be noted that the complex poles and zeros which will occur only in
pair are also located in the same way. For example the complex pole s + 3 + j = 0 or
s = −3 − j is located in LHP in the third quadrant and its compliment (s + 3 − j)
which is the mirror image is located in LHP in the second quadrant. Also note that
there are two poles at s = −2 and they are called repeated poles marked twice at
s = −2. Further, the pole s = 0, which is called a pure integrator is located at the
origin in the s-plane. By the number of poles located at the origin, the systems are
designated by its Type. Since there is one pole located at the origin of the s-plane,
system Eq. (1.10) is called Type one system. If no pole is located at the origin the
system is designated as Type zero and so on. Hence, the Type of the system is defined
as follows.
22 1 Control Systems Modelling and Their Representation

The Type of the system is defined by the number which represents the pres-
ence of pure integrators in the transfer function. This is equivalent to saying the
number of poles of the T.F. located at the origin of the s-plane.
The number of zeros located at the origin does not define the Type of the system.

1.7.3 Sinusoidal, Minimum Phase, Non-minimum Phase and

All Pass Transfer Functions

1.7.3.1 Sinusoidal Transfer Function

Consider system shown in Fig. 1.14a where R(s) is the Laplace transform of input
variable r (t) and C(s) is the Laplace transform of the output variable. G(s) is the
transfer function of the system which is related by the following equation:

C(s)
= G(s) (1.11)
R(s)

If the input r (t) = A sin ωt, a sinusoid, for a linear system, that the output is also a
sinusoid with a phase shift of φ. This is obtained after the input signal is processed
by the system. For Fig. 1.14b, the following equation is obtained:

C( jω)
= G( jω) (1.12)
R( jω)

Equation (1.12) is obtained from Eq. (1.11) simply replacing s by jω.

Fig. 1.14 Representation of

sinusoidal transfer function
1.7 The Concept of Transfer Function 23

Fig. 1.15 Representation of complex minimum phase function

From transfer function G(s) sinusoidal T.F G( jω) is obtained replacing s

by jω. Sinusoidal T.F. is very useful in the study of frequency response method of
analysis which will be discussed later in detail.

1.7.3.2 Minimum Phase Transfer Function

Consider the following T.F.

10(2s + 3)
G(s) =
(5s + 4)

The sinusoidal T.F. is obtained replacing s by jω.

10( j2ω + 3)
G( jω) =
( j5ω + 4)

In the complex plane a complex function is represented with its real and imaginary
parts as shown in Fig. 1.15.
From Fig. 1.15, the magnitude and phase angle of G( jω) are obtained as,

10 (9 + 4ω2 )
G( jω) =  ∠θ1 − θ2 (1.13)
(16 + 25ω2 )

Form Fig. 1.15, it is evident that θ1 and θ2 when measured from positive real axis
(reference axis) for any particular ω, they give minimum phase shift. This is possible
if the poles and zeros are in the LHP of the s-plane. Hence, the following definition
follows.
A minimum phase transfer function is defined as the transfer function whose
poles and zeros are located in the left half s-plane.
Here the zero is located at s = (−3/2) and pole at s = (−4/5).
24 1 Control Systems Modelling and Their Representation

Fig. 1.16 Representation of non-minimum phase transfer function

1.7.3.3 Non-minimum Phase Transfer Function

Consider the following transfer function:

(2s − 3)
G(s) =
(−5s + 4)

The sinusoidal transfer function is given by

( j2ω − 3)
G( jω) =
(− j5ω + 4)
(4ω2 + 9) ∠180◦ − θ1
= √
25ω2 + 16 ∠360◦ − θ2

The phase angle contribution of the poles and zero of the given T.F. is not minimum.
Maximum value of θ1 and θ2 is 90◦ . However, the zero contributes more than 90◦
and the pole contributes more than 270◦ . Hence, they are called non-minimum phase
T.F. The pole and zero are located in RHP of the s-plane. Hence, the non-minimum
phase transfer is defined as follows.
The non-minimum phase transfer function is defined as the transfer function
which has atleast a pole or a zero located in the RHP of the s-plane.
In the above T.F. the zero is located at s = 23 and the pole is located at s = 45 .
which are points on the positive real in RHP.

1.7.3.4 All Pass Transfer Function

Consider, for example the following transfer functions:

1.7 The Concept of Transfer Function 25

(a) (b) (c)

j j j
1 j 1 j

3 3 3 2 2 3

1 j 1 j

Fig. 1.17 Representation of all pass transfer function

(s − 3)
(a) G(s) =
(s + 3)
(s + 2)(s − 3)
(b) G(s) =
(s − 2)(s + 3)
(s 2 − 2s + 2)
(c) G(s) = 2
(s + 2s + 2)

The pole-zero locations of the above transfer functions are shown in Fig. 1.17a–c
respectively. Here, corresponding to a pole, a zero is symmetrically placed with
respect to the jω axis. Consider the following T.F.

(s − 3)
G(s) =
(s + 3)
( jω − 3)
G( jω) =
( jω + 3)

(ω2 + 9) ∠180◦ − tan−1 ω/3
=
(ω2 + 9) ∠tan−1 ω/3
= 1∠180◦ − 2 tan−1 ω/3

Form the above expression, it is evident that the input signal is processed by the
system and the output is sent with a gain of 1. This means that the system passes the
signal without amplification or atteneuation. However, there is a phase shift which
varies from 180◦ to 0◦ for 0 ≤ ω ≤ ∞. Hence, the all pass transfer function is defined
as follows.
An all pass transfer function is defined as the T.F. which has poles and zeros
symmetrically placed with respect to the jω axis. Such systems do not alter the
amplitude of the input signal.
26 1 Control Systems Modelling and Their Representation

1.8 System Representation in Simple Block Diagram

Systems are often represented in a simple block diagram as shown in Fig. 1.18a where
the symbols have the following meaning:
R(s) = Laplace transform of the input variable r (t).
C(s) = Laplace transform of the output variable c(t).
E(s) = Laplace transform of the error signal e(t).
G(s) = Forward path or open loop transfer function.
H (s) = Feedback path transfer function.
G H (s) = Loop transfer function.
From Fig. 1.18a, the following equations are written:

E(s) = R(s) − C H (s)

E G(s) = C(s)
[R(s) − C H (s)]G(s) = C(s)

Re-arranging the terms, we get

C(s) G(s)
= (1.14)
R(s) 1 + G H (s)

Equation (1.14) is called the closed loop transfer function of the system. For positive
feedback it can be shown as

(a)

R(s) E(s) C(s)

G(s)

H(s)

(b)

R(s) G(s) C(s)

1+GH(s)

Fig. 1.18 System representation in simple block diagram

1.8 System Representation in Simple Block Diagram 27

Table 1.1 Voltage-current and impedance relationship

Component Voltage-current Current-voltage Impedance
V (s)
Z (s) =
I (s)
Capacitor, C 1  dv(t) 1
v(t) = i(t)dt i(t) = C
C dt Cs
Inductor, L di(t) 1  Ls
v(t) = L i(t) = v(t)dt
dt L
Resistor, R 1 R
v(t) = i(t)R i(t) = v(t)
R

C(s) G(s)
= (1.15)
R(s) 1 − G H (s)

Unless otherwise, it is mentioned, the system is always given negative feedback.

From Eq. (1.14), the characteristic equation is written as,

F(s) = 1 + G H (s) = 0 (1.16)

For example, if G(s) = [10/s(s + 2)] and H (s) = (s + 5) the following closed loop
T.F. is obtained.
C(s) 10
= 2
R(s) s + 12s + 50

The characteristic equation is written as

s 2 + 12s + 50 = 0.

1.9 Transfer Function Model of Electrical Network

In this section, we determine the transfer function of electric circuits from the mathe-
matical model. The electric circuit consists of three passive linear components namely
resistor, inductor and capacitor. Mathematical equations are written connecting the
input and output following Kirchhoff’s laws. Differential equations are written either
for the sum of the voltages around loops or sum of currents at nodes and then equated
to zero. By taking Laplace transforms of the differential equations, the transfer func-
tion is obtained. Alternatively, the transfer function can also be obtained by transfer
method. Table 1.1 gives the components and their relationship between voltage and
current under zero initial conditions and also the impedance relationship.
28 1 Control Systems Modelling and Their Representation

Fig. 1.19 Electrical network R1 i(t) i(t)=0

for Example 1.1

R2
vi(t) vo(t)

Example 1.1
V0
Determine the T.F. (s) for the electrical network shown in Fig. 1.19.
Vi
Solution For the circuit shown in Fig. 1.19, the following Kirchhoff’s loop equation
is written

1
vi (t) = (R1 + R2 )i(t) + i(t)dt
C

1
v0 (t) = R2 i(t) + i(t)dt
C

Taking Laplace transform on both sides of the above equations we get

1
Vi (s) = (R1 + R2 )I (s) + I (s)
Cs
[(R1 + R2 )Cs + 1]
= I (s)
 Cs
1
V0 (s) = R2 + I (s)
Cs
[(R2 Cs + 1)]
= I (s)
Cs
Dividing the above equations one by the other we get

V0 (1 + R2 Cs)
(s) =
Vi (1 + (R1 + R2 )Cs)

It is to be noted, the time function is written with lower case letters and the s functions
are written with upper case letters.
1.9 Transfer Function Model of Electrical Network 29

R1 i1(t) i2(t) R2 i(t)=0

vi(t) C vo(t)
L

Fig. 1.20 Electrical network for Example 1.2

Example 1.2
Find the T.F. of the network shown in Fig. 1.20.

Solution For the (vi (t) − R1 − C) loop we write the following equation.

1
vi (t) = R1 i 1 (t) + (i 1 (t) − i 2 (t))dt
C

Taking Laplace transform on both sides we get


1 1
Vi (s) = R1 + I1 (s) − I2 (s) (1.17)
Cs Cs

For the loop consisting of R2 − L − C, the following equation is written


di 2 (t) 1
R2 i 2 (t) + L + (i 2 (t) − i 1 (t))dt = 0
dt C

Taking Laplace transform on both sides we get


1 1
R2 + Ls + I2 (s) = I1 (s)
Cs Cs

Solving for I1 (s) we get

I1 (s) = (R2 Cs + LCs 2 + 1)I2 (s) (1.18)

Substituting for I1 (s) in (1.17) we get the following equation

30 1 Control Systems Modelling and Their Representation

i(t)
C
i(t)

R1 R2
vi(t) Z1(s) vo(t)
Z2(s)
L

Fig. 1.21 Electrical network for Example 1.3

[(R2 Cs + LCs 2 + 1)(1 + R1 Cs) − 1]

Vi (s) = I2 (s)
Cs
di 2 (t)
v0 (t) = L
dt
V0 (s) = L s I2 (s)
LsVi (s)Cs
=
[(R2 Cs + LCs 2 + 1)(1 + R1 Cs) − 1]

V0 Ls
(s) =
Vi [L R1 Cs 2 + (R1 R2 C + L)s + (R1 + R2 )]

The problem of this nature can be solved easily using signal flow graph which is
explained later.

Example 1.3
Find the T.F. of the electrical network shown in Fig. 1.21 using impedance transform
method.

Solution Let Z 1 (s) be the impedance function for R1 − C parallel combination and
Z 2 (s) for R2 − L series combination.
1.9 Transfer Function Model of Electrical Network 31

v(t) i1(t)

i(t) R1 L R2 vo(t)

Fig. 1.22 Electrical network for Example 1.4

1
R1
Z 1 (s) = Cs
+ R1
1
Cs
R1
=
1 + R1 Cs
Z 2 (s) = R2 + Ls
Vi (s) = I (s)[Z 1 (s) + Z 2 (s)]
Vi (s)
I (s) =
Z 1 (s) + Z 2 (s)
V0 (s) = I (s)Z 2 (s)
V0 (s) Z 2 (s)
=
Vi (s) Z 1 (s) + Z 2 (s)
R2 + Ls
= R1
(1+R1 Cs)
+ R2 + Ls
V0 (R2 + Ls)(1 + R1 Cs)
(s) =
Vi R1 + (R2 + Ls)(1 + R1 Cs)

V0 R1 LCs 2 + (R1 R2 C + L)s + R2

(s) =
Vi R1 LCs 2 + (R1 R2 C + L)s + R1 + R2

Example 1.4
For the electrical network shown in Fig. 1.22, find the transfer function V0
I
(s) by
impedance transform method.

Solution For the electrical network shown in Fig. 1.22 the following equation in
terms of Laplace transform variable is written at node V (s).
32 1 Control Systems Modelling and Their Representation

V (s) V (s)
I (s) = + + (V (s) − V0 (s)Cs) (1.19)
R1 Ls
V0 (s) = I1 (s)R2
= (V (s) − V0 (s)Cs R2
V0 (s)(1 + R2 Cs) = V (s)R2 Cs
(1 + R2 Cs)V0 (s)
V (s) = (1.20)
R2 Cs

Substituting Eq. (1.20) in (1.19) for V (s) we get


1 1
I (s) = V (s) + + Cs − V0 Cs
R1 Ls
  
1 + R2 Cs 1 1
= V0 (s) + + Cs − Cs
R2 Cs R1 Ls
 
(1 + R2 Cs)(Ls + R1 + R1 LCs 2 )
= V0 (s) − Cs
R1 R2 C Ls 2

V0 (R1 R2 LCs 2 )
(s) =
I [(R1 + R2 )LCs 2 + (R1 R2 C + L)s + R1 ]

1.10 Transfer Function Model of Mechanical Systems

Mechanical systems, like electrical networks, have three passive linear components.
They are classified as mechanical translational and mechanical rotational systems. In
the translational mechanical system, linear displacement, linear velocity and linear
acceleration are the variables of motion while force is the input. The mass, spring and
dash-pot which offers viscous friction or damping are the three passive elements. The
spring and the mass are the energy storage elements while viscous damper dissipates
the energy. A lever a matching device rested on a pivot, transmits energy from one
part of the system to another so that the force, speed and displacement may be altered
and maximum power transfer obtained.
In the rotational mechanical system, the angular displacement, angular velocity
and angular acceleration are the variables of motion and torque is the input applied to
the system. Like the translational system, there are three elements in rotational system
also. While the inertia and the spring store the energy, the viscous damper (dash-pot)
dissipates energy. Like a lever in the translational system, gear arrangements are
used as a matching device in rotational system which is used to alter torque, angular
speed and displacement. Thus, energy is transmitted from one part of the system to
another part and maximum power transfer is obtained. Transfer function model of
1.10 Transfer Function Model of Mechanical Systems 33

these mechanical system is obtained in the same way used for the electrical network.
This is discussed below.

1.11 Transfer Function Model for Mechanical

Translational System

The basic elements of the translational systems are mass, spring, dash-pot and lever.
Newton law of motion is applied according to which the algebraic sum of the forces
acting at a point is zero. This law, as applied to the above elements is discussed below.
The units of the above elements and the associated variables are given below:
1. Mass, M = kilogram (kg).
2. Viscous friction coefficient, B = N/m/s.
3. Spring constant, K = N/m.
4. Force, f (t) = Newton (N).
5. Displacement, x(t) = metre (m).
6. Velocity, v(t) = m/s.
7. Acceleration, a(t) = m/s2 .
In Fig. 1.23, the components of mechanical translational system and their free-body
diagrams are represented. Form the free-body diagrams the following equations are
written as per Newton’s law of motion.
For mass M,
d 2 x(t)
M = f (t)
dt 2
For dash-pot B,
d x(t)
B = f (t)
dt
For spring K ,
K x(t) = f (t)

For lever

f a (t)l1 = f b (t)l2
x y
= (1.21)
l1 l2

Now consider the dash-pot B whose one end is neither connected to any mass
nor to any fixed reference frame. Force f (t) is applied to the other end as shown in
Fig. 1.24. Its free-body diagram is also shown. At the point where the force f (t) is
applied, the following equation is written
34 1 Control Systems Modelling and Their Representation

Mass
x(t) x(t)

d 2 x(t)
M f(t) M M f(t)
dt2

Dash-Pot
x(t) f(t)
d x(t)
f(t) B
dt
B

Spring
x(t) f(t)

f(t) K x(t)
K

y
Lever
l1 l2 y l1
A
a c b C l2 B
x
x fa(t) fb(t)

Fig. 1.23 Forces acting on mechanical translational system components and their free-body dia-
gram

x2(t)
x1(t) f(t)
B d (x1(t) x2(t))
f(t) dt

x2(t)
x1(t) f(t)
K (x1(t) x2(t))
f(t)
K

Fig. 1.24 Dash-pot and spring whose ends are not connected to the fixed frame
1.11 Transfer Function Model for Mechanical Translational System 35

d
B (x1 (t) − x2 (t)) = f (t) (1.22)
dt

Here d x1 (t)/dt is treated as the final velocity and d xdt2 (t) as the initial velocity. At
the point of application of the applied force, the opposing force generated by the
dash-pot is proportional to the final velocity minus initial velocity.
Similarly consider the spring whose one end is neither connected to any mass nor
to the reference frame. At the point where the force is applied from Fig. 1.24, the
following equation is written

K (x1 (t) − x2 (t)) = f (t) (1.23)

K (Final displacement-initial displacement) = Applied force

In the above two cases, the final velocity and the final displacement are taken at the
point of application of the force f (t). The initial velocity and initial displacement
are taken from the other end of the dash-pot and spring respectively. If any end of
these elements is connect6ed to the mass, that end moves with the displacement and
velocity of that particular mass. It is to be noted that each mass has to be given
individual displacement. If there are n masses in the system, there should be n
displacements to be identified in addition to other displacements depending upon the
connections made by the spring, dash-pot and levers. Then only complete description
is possible. The following step by step procedure is followed while writing equation
of motion for translational mechanical system.
Step by Step Procedure

1. For each mass give one independent displacement x. If there are n masses con-
nected in the system, there will be n degree of freedom and their displacements
are identified as x1 , x2 . . . . . . xn .
2. Identify the ends of the spring or dash-pot which are neither connected to any
mass nor to any reference frame. These ends move with additional degree of
freedom and hence with different displacement. Include the lever arrangement
connected if any.
3. Draw the free-body diagram for each degree of freedom and mark the various
forces acting.
4. Write down the equation for each free-body diagram using D’Alembert’s principle
according to which  forces = 0. (The algebraic sum of the forces acting at a
point is zero).
5. Take Laplace transform for equations written in step 4. By solving simultaneous
equations, retain only the Laplace transform of the output variable and Laplace
transform of the input variable and eliminate intermittent variables.
6. The transfer function of the given system is obtained by the ratio of the Laplace
transform of output variable to the Laplace transform of the input variable.
36 1 Control Systems Modelling and Their Representation

(a) (b)

x(t) x(t)
K
K x(t)
2
M f(t) M d x(t) M f(t)
dt 2
B d x(t)
B dt

Fig. 1.25 a Translational mechanical system and b Free-body diagram

The following examples illustrate the above procedure.

Example 1.5
Consider the mechanical system shown in Fig. 1.25a. The applied force f (t) is the
input and the displacement x(t) of mass M is the output variable. Obtain the T.F.

Solution

1. The mass M is given a displacement x(t) which acts in the direction of the applied
force f (t). The opposing force generated is Md 2 x(t)/dt 2 and acts opposite to
the applied force.
2. One end of the spring K is connected to the mass M. This end moves with
a displacement x(t). Its other end is connected to the reference frame. The
opposing force generated by the spring is K x(t) and acts opposite to the applied
force.
3. The behavior of the dash-pot is similar to that of the spring except that the
opposing force generated is equal to B d x(t)
dt
.
4. The free body diagram is shown in Fig. 1.25b and various forces acting in the
system are marked. Using D’Alembert’s principle, the sum of the forces acting
towards the left is equated to the sum of the forces acting towards the right. Thus,
the following equation is obtained:

d 2 x(t) d x(t)
M 2
+B + K x(t) = f (t)
dt dt
5. Taking Laplace transform on both sides of the above equation, the following
equation is obtained:

[Ms 2 + Bs + K ]X (s) = F(s)

1.11 Transfer Function Model for Mechanical Translational System 37

(a) (b)
x1 x
x(t)
2
Md x
dt2
M f(t)
K A M f(t)
B
B d (x x1)
dt

(c) B d (x1 x)
dt
A
K x1

Fig. 1.26 a Mechanical translational system b and c Free-body diagrams

X 1
(s) =
F Ms + Bs + K
2

The system has only a single degree of freedom. The notation x(t), for simplicity
is written as x hereafter.

Example 1.6
Consider the mechanical system shown in Fig. 1.26a, where all the three elements
namely mass, damper and spring are connected in cascade. Determine the T.F. with
the displacement of mass M as the output variable and force f (t) as the input variable.

Solution
1. The mass M is given independent displacement x. One end of the dash-pot is
connected to the mass M and the other and to the spring K . Hence, one end of the
dash-pot is neither connected to the mass nor to any reference frame. Hence, this
point is marked as A and let it move with a displacement x1 . Thus, the system
has two degrees of freedom, one for the mass and another at point A.
2. When we apply D’Alembert’s principle at mass M, x is taken as the final dis-
placement and x1 as the initial displacement. However, at point A, the final
displacement is taken as x1 and initial displacement as x. The free-body dia-
grams are shown in Fig. 1.26a, b. From these free-body diagrams, the following
equations are written

d2x d
M + B (x − x1 ) = f (t)
dt 2 dt
38 1 Control Systems Modelling and Their Representation

Taking Laplace transform on both sides, we get

(Ms 2 + Bs)X (s) − Bs X 1 (s) = F(s) (1.24)

For Fig. 1.26b the following equation is written

d
B (x1 − x) + K x1 = 0
dt
Taking Laplace transform on both sides we get

(Bs + K )X 1 (s) = Bs X (s) (1.25)

Substituting for X 1 (s) from (1.25) in (1.24) we get

B2s2
(Ms 2 + Bs)X (s) − X (s) = F(s)
(Bs + K )
[(Ms 2 + Bs)(Bs + K ) − B 2 s 2 ]
X (s) = F(s)
(Bs + K )

X (Bs + K )
(s) =
F s(M Bs 2 + M K s + B K )

Example 1.7
For the mechanical translational systems shown in Fig. 1.27a, b find the transfer
function relating the displacement of the mass M and the applied force f (t).

Solution Consider the mechanical system shown in Fig. 1.27a. Its free-body dia-
gram is shown in Fig. 1.27c. From free-body diagram, the following equations are
written
d2x dx
M 2 +B + K 2 x + K 1 (x − x1 ) = 0
dt dt
Taking Laplace transform on both sides we get

[Ms 2 + Bs + K 2 ]X (s) = K 1 [X 1 (s) − X (s)] (1.26)

But

K 1 (x1 − x) = f (t)
K 1 [X 1 (s) − X (s)] = F(s) (1.27)
1.11 Transfer Function Model for Mechanical Translational System 39

(a)

x x1

K1
M f(t)
A
K2
B

(b)
x x1

B1

M f(t)
A
K
B2

(c)
x
K2 x
f(t)
2
Md x M K1(x x1) A
dt2
B dx K1(x1 x)
dt

(d)
x(t)
Kx
f(t)
2
Md x M B1d (x x1)
A
dt2 dt
B1d (x1 x)
B2 d x
dt dt

Fig. 1.27 Mechanical systems for Example 1.7. c Free-body diagrams of Fig. 1.27a, d Free-body
diagram of Fig. 1.27b
40 1 Control Systems Modelling and Their Representation

Substituting Eq. (1.27) in (1.26) we get

X 1
(s) =
F (Ms + Bs + K 2 )
2

Note: When the force f (t) is applied to the spring K 1 , it is transformed

and two displacements x1 and x must be assigned to the end points of
the spring. Further it is to be noted that the above T.F. determined is
independent of K 1 . Here K 1 just transforms f (t) from one point to another
point. However, if XF1 (s) is determined it will contain all system elements
including K 1 .
Now consider the mechanical system shown in Fig. 1.27b. The free-body diagram
of this is shown in Fig. 1.27d. From the free-body diagram, the following equations
are written.
d2x dx d
M 2 + B2 + K x + B1 (x − x1 ) = 0
dt dt dt
Taking Laplace transform on both sides we get

[Ms 2 + B2 s + K ]X (s) = B1 s[X 1 (s) − X (s)] (1.28)

Also,

d
B1 (x1 − x) = f (t)
dt
B1 s[X 1 (s) − X (s)] = F(s) (1.29)

Substituting Eq. (1.29) in Eq. (1.28) we get

X (s) 1
=
F(s) (Ms 2 + B2 s + K )

Example 1.8
Consider the mechanical system shown in Fig. 1.28a. Determine the T.F. relating the
displacement of mass M1 and the force f (t).

Solution The free-body diagram for mass M1 and mass M2 of Fig. 1.28a are shown
in Fig. 1.28b, c respectively. For Fig. 1.28b the following equation is written.

d 2 x1 d x1 d
M1 + B1 + B (x1 − x2 ) + K (x1 − x2 ) = 0
dt 2 dt dt
Taking Laplace on both sides, we get
1.11 Transfer Function Model for Mechanical Translational System 41

(a)
x1 x2
B

M1 M2 f(t)

B1 K B2

(b) (c)
x1 x2
M1 d 2x1 B d (x1 x2) B d (x2 x1)
dt2 dt dt
M1 M2d 2x2 M2 f(t)
dt2
B1 d x1 K (x1 x2) K(x2 x1) B2 dx2
dt
dt

Fig. 1.28 a Mechanical System for Example 1.8. b and c Free-body diagram of mechanical system
of Fig. 1.28a

[M1 s 2 + (B1 + B)s + K ]X 1 (s) = [Bs + K ]X 2 (s) (1.30)

From the free-body diagram of mass M2 , the following equation is written.

d 2 x2 d x2 d
M2 + B2 + B (x2 − x1 ) + K (x2 − x1 ) = f (t)
dt 2 dt dt
Taking Laplace transform on both sides, we get,

[M2 s 2 + (B2 + B)s + K ]X 2 (s) − [Bs + K ]X 1 (s) = F(s) (1.31)

Substituting for X 2 (s) from Eq. (1.30) in (1.31) we get

F(s) =
 
[M1 s 2 + (B1 + B)s + K ]
[M2 s + (B2 + B)s + K ]
2
− (Bs + K ) X 1 (s)
(Bs + K )
X 1 (s) Bs + K
=
F(s) [[M2 s 2 + (B2 + B)s + K ][M1 s 2 + (B1 + B)s + K ] − (Bs + K )2 ]
42 1 Control Systems Modelling and Their Representation

Fig. 1.29 a Mechanical (a)

system for Example 1.9. b
Free-body diagrams of
A2 A1
Fig. 1.29a x
K2 K1
B
y x1

(b) B d (x1 y) (c) B d (y x1)

dt dt
A1 A2

K1(x1 x) K2 y

X1
= (Bs + K ) s[M1 M2 s 3 + {M2 (B1 + B) + M1 (B2 + B)}s 2
F(s)
+{K (M1 + M2 ) + B1 B2 + B(B1 + B2 )}s + K (B1 + B2 )]

Example 1.9
Y
Consider the mechanical system shown in Fig. 1.29a. Find the T.F. (s).
X
Solution
For Fig. 1.29a, x is the input and output is taken at point A2 where the displacement
is y. The end point of the spring K 1 , at A1 moves with the displacement x1 . The free-
body diagrams are shown in Fig. 1.29b, c. From Fig. 1.29b the following equations
are written:
d
B (x1 − y) + K 1 (x1 − x) = 0
dt
Taking Laplace transform on both sides we get

(Bs + K 1 )X 1 (s) = BsY (s) + K 1 X (s) (1.32)

From Fig. 1.29c, the following equation is written

d
B (y − x1 ) + K 2 y = 0
dt
Taking Laplace transform on both sides we get

(Bs + K 2 )Y (s) = Bs X 1 (s) (1.33)

1.11 Transfer Function Model for Mechanical Translational System 43

Substituting for X 1 (s) in Eq. (1.32) we get

(Bs + K 1 )(Bs + K 2 )
Y (s) − BsY (s) = K 1 X (s)
Bs
[B 2 s 2 + Bs(K 1 + K 2 ) + K 1 K 2 − B 2 s 2 ]Y (s) = s B K 1 X (s)

Y (s) B K1s
=
X (s) [B(K 1 + K 2 )s + K 1 K 2 ]

Example 1.10
Consider the mechanical system with lever arrangement which is shown in Fig. 1.30a.
Y
Find the T.F (s).
F
Solution
The free-body diagrams of Fig. 1.30a are shown in Fig. 1.30b, c. Form Fig. 1.30b,
the net force acting downwards on the lever at point a is,

d2x dx
f a = f (t) − M1 2 + B1 + K1 x (1.34)
dt dt

f b is the force acting downwards at point c which opposes motion. It is given as

d2 y dy
f b = M2 + B2 + K2 y (1.35)
dt 2 dt
These forces and the displacement of the lever are shown in Fig. 1.30d.
Now taking moment at the pivot point b, the following equation is obtained.

f a l1 − f b l2 = 0
f a l1 = f b l2
   
d2x dx d2 y dy
f (t) − M1 2 + B1 + K 1 x l1 = M2 2 + B2 + K 2 y l2
dt dt dt dt

Taking Laplace transform on both sides we get

l2
F(s) − (M1 s 2 + B1 s + K 1 )X (s) = (M2 s 2 + B2 s + K 2 )Y (s) (1.36)
l1
44 1 Control Systems Modelling and Their Representation

F
(a)
l1 l2

a y
b c
f(t)
x

M1 M2

B1 B2
K1 K2

(b) (c)
M1 d 2x f(t) M2 d 2 y y
dt2 dt2

x
M1 M2

K2 y B2dy
K1 x B1dx dt
dt

Fig. 1.30 a Mechanical system with lever arrangement. b Free-body diagrams of Fig. 1.30a. c
Forces acting on the lever

From Fig. 1.30e we write,

l1 l2
=
x y
l1
x= y (1.37)
l2

Substituting for x from Eqs. (1.37) in (1.36) we get

l1 l2
F(s) − (M1 s 2 + B1 s + K 1 ) Y (s) = (M2 s 2 + B2 s + K 2 )Y (s)
l2 l1
 
l 1 l2
F(s) = (M1 s + B1 s + K 1 ) + (M2 s + B2 s + K 2 )
2 2
Y (s)
l2 l1
1.11 Transfer Function Model for Mechanical Translational System 45

(a)

K
B1

M2 x2

M2g B2

M1 x1

f(t) M 1g

(b) B dx1 M d12x1 B d (c) 2

M2 d x22
1 1 2 (x −x )
dt dt2 dt 1 2 Kx2 dt

M1 x1 M2 x2

f(t) M1 g M2g B2 d (x2−x1)

dt

Fig. 1.31 a Mechanical system for Example 1.11. b Free-body diagrams of Fig. 1.31a

Y (s) l1 l2
= 2
F(s) l1 (M1 s + B1 s + K 1 ) + l22 (M2 s 2 + B2 s + K 2 )
2

Example 1.11
Consider the mechanical system shown in Fig. 1.31a. Obtain the T.F. connecting the
displacement of mass M2 and the applied force f (t).

(Anna University, May, 2010)

Solution
The free-body diagrams of Fig. 1.31a are shown in Fig. 1.31b, c. It is to be noted
that the gravitational forces M1 g and M2 g are acting on mass M1 and mass M2
respectively. If the system is initially at the equilibrium, the gravitational forces
acting in the system are balanced by preloading of the spring and hence g is set
to zero. Hence, the following equations are written for Fig. 1.31b, c.
46 1 Control Systems Modelling and Their Representation

d 2 x1 d x1 d
M1 + B1 + B2 (x1 − x2 ) = f (t)
dt 2 dt dt
Taking Laplace transform on both sides we get

[M1 s 2 + (B1 + B2 )s]X 1 (s) − B2 s X 2 (s) = F(s) (1.38)

d 2 x2 d
M2 2 + B2 (x2 − x1 ) + K x2 = 0
dt dt
[M2 s 2 + B2 s + K ]X 2 (s) = B2 s X 1 (s)
(M2 s 2 + B2 s + K )
X 1 (s) = X 2 (s) (1.39)
B2 s

Substituting Eq. (1.39) in (1.38) we get

 
[M1 s 2 + (B1 + B2 )s][M2 s 2 + B2 s + K ]
F(s) = − B2 s X 2 (s)
B2 s

X2 B2
(s) =
F [M1 M2 s +{M1 B2 +M2 (B1 +B2 )}s 2 +(M1 K +B1 B2 )s+K (B1 +B2 )]
3

1.12 Transfer Function Model for Mechanical Rotational

System

The behaviour of mechanical rotational system resembles with that of translational

system. The basic difference between the two is that rotational system has angular
displacement while translational system has linear displacement. Further, instead of
force, torque is applied in the rotational system. Mass is replaced by inertia. The
basic elements of the rotational system consists of inertia, rotational dash-pot and
spring. Instead of lever, gear arrangements are used as matching device in rotational
system which is used to alter torque, angular speed, angular displacement and transfer
maximum power. The dynamic behaviour of these elements due to the torque applied
are explained below in Fig. 1.32.
In the mechanical rotational system energy is stored in inertia J and spring K .
Energy is dissipated in the dash-pot B. Power transfer is made through gear system.
The units of the above elements and the associated variables are given below:
1. Inertia, J = Kg-m2 .
2. Viscous friction coefficient, B = N-m/rad/s.
3. Torsional spring, K = N-m/rad.
4. Torque, T (t) = N-m.
5. Angular displacement, θ (t) = radians (rad).
1.12 Transfer Function Model for Mechanical Rotational System 47

Inertia
θ

Inertia 2
J d θ2 J T(t)
J dt
T(t) θ(t)
Dash-Pot
θ

B dθ B T(t)
dt
B T(t) θ(t)

Torsional Spring
θ

Kθ K T(t)
K
T(t) θ(t)

Gear system
N1
N1θ 1 = N2θ 2
T1θ 1 = T2θ 2
T1 θ 2 N1
T1(t) θ 1(t) = =
T2 θ 1 N2
θ 2(t) T2(t)

N2

Fig. 1.32 Torque acting on mechanical rotation system components and their free-body diagrams

6. Angular velocity, ω(t) = rad/s.

7. Angular acceleration, a(t) = rad./s2 .

Example 1.12
For the mechanical rotational system shown in Fig. 1.33a, find θ1 /F(s).

Solution The Free-body diagrams for Fig. 1.33a are shown in Fig. 1.33b, c. For
Fig. 1.33b, the following equation is written
48 1 Control Systems Modelling and Their Representation

(a) θ1 θ2

J1 J2
K1 K2
T(t)
B1 B2

(b) (c)
θ1 θ2
d 2θ 1 2
J1
dt J2 d θ 2
dt
T(t) J1 K1(θ 1− θ2) K1(θ 2 − θ1) J2 K2θ 2
dθ 1 dθ 2
B1 B2
dt dt

Fig. 1.33 a Mechanical rotational system for Example 1.12. b Free-body diagram of Fig. 1.33a

d 2 θ1 dθ1
J1 + B1 + K 1 (θ1 − θ2 ) = T (t)
dt 2 dt
Taking Laplace transform on both sides we get,

(J1 s 2 + B1 s + K 1 )θ1 (s) − K 1 θ2 (s) = T (s) (1.40)

For Fig. 1.33c the following equation is written:

d 2 θ2 dθ2
J2 2
+ B2 + K 2 θ2 + K 1 (θ2 − θ1 ) = 0
dt dt
Taking Laplace transform on both sides we get

[J2 s 2 + B2 s + (K 2 + K 1 )]θ2 (s) = K 1 θ1 (s) (1.41)

Substituting for θ2 (s) from Eq. (1.41) in (1.40) we get

K 12 θ1 (s)
(J1 s 2 + B1 s + K 1 )θ1 (s) − = T (s)
(J2 s 2 + B2 s + K 1 + K 2 )

[(J1 s 2 + B1 s + K 1 )(J2 s 2 + B2 s + K 1 + K 2 ) − K 12 ]
θ1 (s) = T (s)
(J2 s 2 + B2 s + K 1 + K 2 )
1.12 Transfer Function Model for Mechanical Rotational System 49

T
θ1 θ2 θ3
K1 K2
J1 J2 J3

T(t)
B

(b) (c)
θ1 θ2
d2θ
J2 22 T(t)
dt
d 2θ
J1 21 J1 J2
dt K1(θ 1− θ2) K2(θ 2− θ3)

K1(θ 2− θ1) dθ 2
B
dt

(d)
θ3

2
K2(θ 3−θ 2) J3 J3 d θ 3
dt2

Fig. 1.34 a Mechanical translational system for Example 1.13. b Free-body diagrams of Fig. 1.34a

θ1
(s) = (J2 s 2 + B2 s + K 1 + K 2 ) J1 J2 s 4 + (J1 B2 + J2 B1 )s 3
T

+(J1 (K 1 + K 2 ) + B1 B2 + J2 K 1 )s 2 + (K 1 + K 2 + K 1 B2 )s + K 1 K 2

Example 1.13
θ2
For the mechanical rotational system shown in Fig. 1.34a, find (s).
T
Solution
The free-body diagrams of Fig. 1.34a are shown in Fig. 1.34b–d. For Fig. 1.34b the
following equation is written:
50 1 Control Systems Modelling and Their Representation

d 2 θ1
J1 + K 1 (θ1 − θ2 ) = 0
dt 2
Taking Laplace transform on both sides we get

(J1 s 2 + K 1 )θ1 (s) = K 1 θ2 (s) (1.42)

For Fig. 1.34c, following equation is written

d 2 θ2 dθ2
J2 2
+B + K 1 (θ2 − θ1 ) + K 2 (θ2 − θ3 ) = T (t)
dt dt
Taking Laplace transform on both sides we get

[J2 s 2 + Bs + (K 1 + K 2 )]θ2 (s) − K 1 θ1 (s) − K 2 θ3 (s) = T (s) (1.43)

For Fig. 1.34d, the following equation is written:

d 2 θ3
J3 + K 2 (θ3 − θ2 ) = 0
dt 2
Taking Laplace transform on both sides we get

[J3 s 2 + K 2 ]θ3 (s) = K 2 θ2 (s) (1.44)

Substituting for θ1 (s) and θ2 (s) from Eqs. (1.42) and (1.44) respectively in (1.43)
we get the following equation:
 
K 12 K2
J2 s 2 + Bs + (K 1 + K 2 ) − − θ2 (s) = T (s)
(J1 s 2 + K 1 ) (J3 s 2 + K 2 )

On simplifying the above equation we get

θ2
(s) = (J1 s 2 + K 1 )(J3 s 2 + K 2 ) s J1 J2 J3 s 5 + J1 J3 Bs 4 + (J1 J2 K 2 + J2 J3 K 1
T
+J1 J3 (K 1 + K 2 ))s 3 + B(J1 K 2 + J3 K 1 )s 2

+(K 1 K 2 (J1 + J2 + J3 ))s + B1 K 1 K 2
1.12 Transfer Function Model for Mechanical Rotational System 51

(a) K θ
N2
J
T1(t) θ1 T(t)
B

N1

(b) (c)
θ

Kθ
T(t) N2
2 T1(t)
J d θ2 J T1(t) N1
dt

B dθ
dt

Fig. 1.35 a Mechanical system with gear arrangement for Example 1.14. b Free-body diagrams
of Fig. 1.35a

Example 1.14
Consider the mechanical rotational translational system shown in Fig. 1.35a. Find
the T.F. relating the angular displacement of inertia J and the applied torque T (t)
and angular displacement of θ1 and T (t).

Solution The free-body diagrams for Fig. 1.35a are shown in Fig. 1.35b, c.
 
d 2θ dθ
J + B + K θ = T1 (t)
dt 2 dt

Taking Laplace transform on both sides we get

[J s 2 + Bs + K ]θ (s) = T1 (s)

But T1 N1 = T N2

N2
T1 = T
N1

Substituting this in the above equation we get

52 1 Control Systems Modelling and Their Representation

N1
(J s 2 + Bs + K )θ (s) = T (s)
N2


θ N2 1
(s) =
T N1 (J s 2 + Bs + K )

Also we have

N1 θ1 = N2 θ
N1
θ= θ1
N2

substituting this in the above equation we get

 2
θ1 N2 1
(s) =
T N1 (J s 2 + B + K)

Example 1.15
Consider the mechanical rotational system shown in Fig. 1.36a. Find the T.F. relating
the position θ4 of J2 and the input torque T (t).

Solution
The free-body diagrams of the given mechanical system are shown in Fig. 1.36b–d.
For Fig. 1.36b, the following equation is written:

d
B (θ1 − θ2 ) + K 1 (θ1 − θ2 ) = T (t)
dt
Taking Laplace transform on both sides we get

(Bs + K 1 )θ1 (s) − (Bs + K 1 )θ2 (s) = T (s) (1.45)

For Fig. 1.36c the following equation is written:

d 2 θ2 d
J1 + B (θ2 − θ1 ) + K 1 (θ2 − θ1 ) + K 2 (θ2 − θ3 ) = 0
dt 2 dt
d 2 θ2
J1 2 + K 2 (θ2 − θ3 ) = T (t)
dt
Taking Laplace transform on both sides we get

(J1 s 2 + K 2 )θ2 (s) − K 2 θ3 (s) = T (s) (1.46)

1.12 Transfer Function Model for Mechanical Rotational System 53

B N1
J1
K2
θ 1 T(t) θ2 θ3
K1 θ4
J2

N2

(b) (c)
(θ 1−θ2) θ2
d (θ −θ )
Bdt d
1 2 B dt 2 1)
(θ −θ
2 K2(θ 2−θ3)
T(t) BK1 J1 d θ 2 J1
dt2
K1(θ 1−θ2)
K1(θ 2−θ1)

(d) θ3
N 2 2
( N1 ( J2 d θ 3
N 2 2 dt2
K2(θ 3−θ2) ( N1 ( J2
2

Fig. 1.36 a Mechanical rotational system for Example 1.15. b Free-body diagrams of Fig. 1.35a

For Fig. 1.36d, the following equation is written:

 2
N1 d 2 θ3
J2 + K 2 (θ3 − θ2 ) = 0
N2 dt 2

Taking Laplace transform on both sides of the above equation we get

 
2
N1
J2 s 2 + K 2 θ3 = K 2 θ2 (1.47)
N2

Substituting for θ2 from Eq. (1.47) in (1.46) we get

   
2
N 1 1
(J1 s 2 + K 2 ) J2 s 2 + K 2 − K 2 θ3 (s) = T (s)
N2 K2
  
2
N1 N1 2 2 θ3 (s)
J1 J2 s + J1 K 2 s + K 2
4 2
J2 s + K 2 − K 2
2 2
= T (s)
N2 N2 K2
54 1 Control Systems Modelling and Their Representation

θ3 K2
(s) =  2   2
T N1
N2
J1 J2 s 4 + K 2 J1 + N1
N2
J2 s 2

But

N1 θ3 = N2 θ4
N2
θ3 = θ4
N1
 2
N1
Je = J2
N2

θ4 K NN21
(s) = 2
T s [J1 Je s 2 + K 2 (J1 + Je )]

1.13 Transfer Function Model of Electro-Mechanical

Systems

Real world electromechanical problems require the ability to design magnetic and
electric components to produce needed forces and torques, such as motor and solenoid
design. Application examples include actuators, alternators and generators, elec-
tromagnetic brakes and clutches loudspeakers, magnetic bearings, micro-electro
mechanical system (MEMS), motors, relays and contactors, sensors and solenoids.
These devices are very extensively used in defence, manufacturing, communication,
commercial air travel, energy products etc. The transfer functions of some of the
commonly used electromechanical devices are derived in this section.

Example 1.16
Consider the solenoid shown in Fig. 1.37. The force of attraction generated is K f i(t)
Newtons. The back emf constant of the solenoid coil is K b volts/m/s. Derive the T.F.
relating the displacement x(t) of the M of the solenoid core and the applied voltage
e(t).

Solution
The electrical circuit of the solenoid coil is shown in Fig. 1.37b. For Fig. 1.37b the
following equations are written:
1.13 Transfer Function Model of Electro-Mechanical Systems 55

(a) x(t)
R, L, Kb, Kf
K i(t)

M
e(t)

(b) L (c) x(t)

R
d2x(t)
+ M
i(t) dt2

e(t) eb(t) Kf i(t) = f(t) dx(t)

M B
dt

_ Kx(t)

Fig. 1.37 a Solenoid coil for Example 1.16. b Equivalent electrical circuit and mechanical free-
body diagram of Fig. 1.37a

di(t)
e(t) = L + Ri(t) + eb (t)
dt
di(t) d x(t)
=L + Ri(t) + K b
dt dt
Taking Laplace transform on both sides of the above equation we get

E(s) = [Ls + R]I (s) + K b s X (s) (1.48)

For Fig. 1.37c, the following equations are written:

d 2 x(t) d x(t)
M 2
+B + K x(t) = K f i(t)
dt dt
Taking Laplace transform on both sides of the above equation we get

[Ms 2 + Bs + K ]X (s) = K f I (s) (1.49)

Substituting Eq. (1.49) in (1.48) for I (s) we get

 
(Ms 2 + Bs + K )
E(s) = (Ls + R) + K b s X (s)
Kf
56 1 Control Systems Modelling and Their Representation

(a)
x(t)
R, L, M, Kb, Kf
K i(t) R1 i1(t)

M
C e(t)

(b) R1 e(t) R

i1(t) L
dx(t)
e(t) C eb(t)=Kb
dt

Fig. 1.38 a Solenoid coil for Example 1.17. b Electrical circuit part of Fig. 1.13a

X (s) Kf
=
E(s) [(Ls + R)(Ms 2 + Bs + K ) + K b K f s]

Example 1.17
Consider the solenoid coil shown in Fig. 1.38a. The coil is connected to the voltage
source via an R-C filter as shown in Fig. 1.38a. Other parameters remain the same as
X
in Example 1.16. Derive the T.F. relating (s).
E
Solution
By independence transform method, the following equations are written for
Fig. 1.38b:
 
1 1
E(s) = R1 + I1 (s) − I (s)
Cs Cs
I (s) 1
I1 (s) = + Cs E(s) (1.50)
(1 + R1 Cs) (1 + R1 Cs)
 
1 1
Ls + R + I (s) − I1 (s) + K b s X (s) = 0
Cs Cs
(LCs 2 + RCs + 1)I (s) − I1 (s) + K b Cs 2 X (s) = 0 (1.51)

Substituting Eq. (1.50) for I1 (s) in (1.51) we get

1.13 Transfer Function Model of Electro-Mechanical Systems 57

[(LCs 2 + RCs + 1)(1 + R1 Cs) − 1]I (s) − Cs E(s)

+K b Cs 2 (1 + R1 Cs)X (s) = 0 (1.52)

For the mechanical system, the following equation written

K f I (s) = [Ms 2 + Bs + K ]X (s) (1.53)

Substituting for I (s) from Eq. (1.53) in (1.52) we get

(Ms 2 + Bs + K )
Cs E(s) = [(LCs 2 + RCs + 1)(1 + R1 Cs) − 1]
Kf
+K b Cs 2 (1 + R1 Cs)]X (s)

X
(s) = K f (Ms 2 + Bs + K )(LC R1 s 2
E

+(R1 RC + L)s + (R1 + R) + K f K b s(1 + R1 Cs)) (1.54)

Example 1.18
The solenoid shown in Fig. 1.39a produces a magnetic force proportional to the
current in the coil with proportionality constant being K i N/ampere. The coil has
resistance R and inductance L. Determine the transfer function XE2 (s). Assume the
back emf constant of the solenoid coil as K b volts/meter.

(Anna University, December, 2004)

Solution
The portion of electrical circuit for Fig. 1.39a is shown in Fig. 1.39b. For Fig. 1.39b
the following equations is written:

di d x1
e(t) = L + Ri + K b
dt dt
Taking Laplace transform on both sides of the above equation we get

E(s) = (R + Ls)I (s) + K b s X 1 (s)

E(s) − K b s X 1 (s)
I (s) = (1.55)
(R + Ls)

The electromechanical force developed by the solenoid coil is,

58 1 Control Systems Modelling and Their Representation

(a) l1 l2
x2(t)
a b c
x1(t)

i(t) M2
e(t) R, L, Ki, Kb
M1
K2 B2
B1

(b) (c)
R L
+ fa(t) x2
i(t)
l1
dx1 a c
e(t) eb=Kb dt b L2
x1 fc(t)

Fig. 1.39 a Electromechanical system for Example 1.18. b Electrical circuit and free-body diagram
of mechanical system for Example 1.18

f (t) = K i i(t)
F(s) = K i I (s)

Substituting Eq. (1.55) in the above equation we get

[E(s) − K b s X 1 (s)]
F(s) = K i (1.56)
(R + Ls)

The force acting at point a of the lever is


d 2 x1 d x1
f a (t) = f (t) − M1 + B1
dt 2 dt

Taking Laplace transform on both sides we get

Fa (s) = F(s) − (M1 s + B1 )s X 1 (s) (1.57)

The force acting at point c on the lever is,

d 2 x2 d x2
f e (t) = M2 + B2 + K 2 x2
dt 2 dt
Taking Laplace transform on both sides we get
1.13 Transfer Function Model of Electro-Mechanical Systems 59

Fc (s) = [M2 s 2 + B2 s + K 2 ]X 2 (s) (1.58)

From Fig. 1.39c we get

x1 x2
=
l1 l2

The Laplace transformed equation is

l1
X 1 (s) = X 2 (s)
l2

Substituting the above equation in (1.57) we get

l1
Fa (s) = F(s) − (M1 s + B1 )s X 2 (s)
l2
K i E(s) l1 X 2 (s)
F(s) = − Kb Ki (1.59)
R + Ls l2 R + Ls

Taking moment at point b on the pivot in Fig. 1.39c we get

Fa (s)l1 = Fc (s)l2
l2
Fa (s) = Fc (s)
l1

Substituting for Fa (s) in Eq. (1.59) we get

 
E(s) l1 l l1
Ki − Kb Ki X 2 (s) − (M1 s + B1 )s X 2 (s)
(R + Ls) l2 (R + Ls) l2
l2
= (M2 s 2 + B2 s + K )X 2 (s)
l1

Rearranging the like terms we get

 
E(s) l1 1 l1 l2
Ki = Kb Ki + (M1 s + B1 )s + (M2 s 2 + B2 s + K ) X 2 (s)
(R + Ls) l2 (R + Ls) l2 l1
X2 Ki
(s) =  
E (R + Ls) (M2 s 2 + B2 s + K ) + (M1 s + B1 )s ll21 + K b K i ll21 (R+Ls)
1

X2 Ki
(s) = 
E (R + Ls) (M2 s 2 + B2 s + K ) l1 + (M1 s + B1 )s ll21 +
l 2 l1 1
K b K i (R+Ls)
l2
60 1 Control Systems Modelling and Their Representation

1.14 Transfer Function Model of D.C. Motor

The D.C. motor is a power actuator device that delivers to the load. It converts direct
current electrical energy into rotational mechanical energy. The torque generated in
the rotor shaft is used to drive an external load connected to the shafts. The operation
of D.C. motor is based on the principle that if a current carrying conductor is placed in
a magnetic field, there will be a force exerted on the conductor. The reverse process,
namely if a conductor is moved through a magnetic field there will be a voltage
induced in the conductor. This principle is made use of in generators. The D.C.
motor consists of a rotating cylinder called armature in which copper conductors are
placed in the slots made in the armature. The armature is placed in the magnetic field
between stationary magnetic poles, called the field. D.C. motors have traditionally
been used for variable speed applications. There are two methods of controlling the
speed of a D.C. motor.
1. The armature current is kept constant and the torque is controlled via the field
voltage. Such a control is called field controlled D.C. motor.
2. The field is connected to a constant voltage source to produce constant field flux
and the torque is controlled via the armature voltage. Such a control is called
armature controlled D.C. motor.

In both the cases the electromagnetic torque produced is proportional to the armature
current and the field magnetic flux. The D.C. motors have the following advantages:
1. Torque developed is very high.
2. A wide range of speed controllability.
3. Easily portable.
4. Good torque-speed characteristics.
5. Adaptability to various types of control methods.
The control applications include the following:
1. Machine tools
2. Servo valve actuators
3. Robatic manipulators
4. Tape transport mechanisms
5. Disk drives.

1.14.1 Transfer Function of Field Controlled D.C. Motor (No

Load)

The field controlled D.C. motor is shown in Fig. 1.40. The symbols and notations
have the following meanings:
1.14 Transfer Function Model of D.C. Motor 61

if (t)
Rf, Lf ia(t)
Ra La

M Constant
vf (t) J current
f source

dθ(t)
ω(t)= dt

Fig. 1.40 Field controlled D.C. motor

1. J = Moment of inertia of the motor in Kg-m2 .

2. f = Friction coefficients N-m/rad/s.
3. R f , L f = Field coil resistance and inductance respectively.
4. Ra , L a = Armature coil resistance and inductance respectively.
5. i a (t) = Armature current
6. v f (t) = Voltage applied across the field coil to produce flux.
7. ω(t) = dθ(t)
dt
= Motor shaft speed in rad/s.
8. θ (t) = Motor shaft position in rad.
If the field is unsaturated the air-gap flux of the motor is proportional to the field
current i f (t) and is given by

 = K f i f (t) (1.60)

The electromagnetic torque developed by the motor which is proportional to the field
flux and the armature current is given by

Td (t) = K a i a (t)
= K a K f i f (t)i a (t) (1.61)

From Eq. (1.61) it is evident that any one of i f (t) or i a (t) must be maintained constant
while the other current becomes the input so that Eq. (1.61) is linear. If i a (t) is kept
constant the motor is field controlled one and Eq. (1.61) becomes

Td (t) = K t i f (t) (1.62)

where K t is expressed in N-m field ampere. If i f (t) is kept constant, the motor is
armature controlled one and Eq. (1.61) becomes

Td (t) = K t i a (t) (1.63)

where K t is expressed in N-m armature ampere. The developed torque is used to drive
the system having the total inertia J and to overcome the friction f . This torque is
expressed as
62 1 Control Systems Modelling and Their Representation

d 2 θ (t) dθ (t)
TL (t) = J + f (1.64)
dt 2 dt

Equating the developed torque Td (t) to the torque used TL (t) we get, for a field
controlled D.C. motor

d 2 θ (t) f dθ (t)
K t i f (t) = J +
dt 2 dt
Taking Laplace transform on both sides we get

K t I f (s) = (J s + f )sθ (s) (1.65)

From Fig. 1.40 we get

V f (s)
I f (s) = (1.66)
R f + sL f

Substituting Eq. (1.66) in (1.65) we get

θ (s) Kt
=
V f (s) s(R f + s L f )(J s + f )

θ (s) K
= (1.67)
V f (s) s(1 + sTe )(1 + sTm )

where
Kt
K =
Rf f
Lf
Te = = Electrical time constant
Rf
J
Tm = = Mechanical time constant
f

The block diagram representation of field controlled D.C. motor is shown in Fig. 1.41.

Vf (s) If (s) Td (s) (s) (s)

1 1 1
Kf
Rf +Lf s Js +f s

Fig. 1.41 Block diagram of field controlled D.C. motor

1.14 Transfer Function Model of D.C. Motor 63

Ra La ia(t)

M dθ(t)
J eb=Kb va(t)
dt
f θ(t)

Fig. 1.42 Armature controlled D.C. motor

1.14.2 Transfer Function of Armature Controlled D.C. Motor

For the armature controlled D.C. motor, the output variable is angular displacement
θ (t) and the input variable is the applied voltage va (t). The ratio of the Laplace
transform of these variables which is defined as the T.F is obtained as follows. The
electromagnetic torque developed by the motor is given by Eq. (1.63) while Eq. (1.64)
gives the torque used to drive the combined inertia and friction connected to the motor
shaft. Equating these two equations we get

d 2 θ (t) dθ (t)
K t i a (t) = J 2
+ f
dt dt
Taking Laplace transform on both sides of the above equation we get

K t Ia (s) = (J s + f )sθ (s) (1.68)

The following equation for the armature circuit is written

di a (t)
va (t) = L a + Ra i a (t) + eb (t)
dt
di a (t) dθ (t)
= La + Ra i a (s) + K b
dt dt
Taking Laplace transform on both sides of the above equation we get

Va (s) = (Ra + L a s)Ia (s) + K b sθ (s) (1.69)

Substituting Eq. (1.68) in (1.69) for Ia (s) we get

 
(J s + f )
Va (s) = (Ra + L a s) + K b sθ (s)
Kt
64 1 Control Systems Modelling and Their Representation

θ Kt
(s) =
Va s[(Ra + L a s)(J s + f ) + K b K t ]

θ Kt
(s) = (1.70)
Va s[(1 + sTe )(1 + sTm )Ra f + K b K t ]

where
La
Te = = Electrical time constant
Ra
J
Tm = = Mechanical time constant
f

To Prove K b = K t

When the motor operates at steady-state and if we neglect the losses in the rotor
resistances, the following equations are written. The power input to the rotor is

Pr = eb i a
= K b ωi a

The power delivered to the shaft

Ps = Torque × speed
= Td ω
= K t ia ω

Since Pr = Ps

K b ωi a = K t ωi a

Kb = Kt (1.71)

The block diagram of armature controlled D.C. motor is shown in Fig. 1.43.

1.14.3 Transfer Function of D.C. Generator (On No Load)

The schematic diagram of a D.C. generator run by a constant speed prime mover is
shown in Fig. 1.44a. Due to the cutting of the flux produced by the electromagnet, emf
is induced in the armature which has constant K g expressed in volts per field ampere
1.14 Transfer Function Model of D.C. Motor 65

Va (s) Kt Td (s) 1 (s ) θ(s)

1
+ (Ra +La s) (Js +f) s

Kb

Fig. 1.43 Block diagram of armature controlled D.C. motor

(a) if (t) Rf, Lf

(b)

G eg (t) Ef (s) Kg Eg (s)

ef (t)
=Constant (Rf +Lf s)
Kg=Volts/Field amp

Fig. 1.44 Field excited D.C. generator and its T.F

when speed is constant. Under this condition the following equation is written:

eg (t) = K g i f (t)
K g E f (s)
E g (s) = K g I f (s) =
Rf + L fs

Eg K
(s) = (1.72)
Ef (1 + sT f )

where
Kg
K =
Rf
Lf
Tf = = Electrical time constant.
Rf

D.C. Tachogenerator
Instead of electromagnet, if a permanent magnet is used in Fig. 1.44a, the generator
becomes a D.C. tachogenerator. Now, ω the speed is the input variable and eg (t) is
the output variable. If K T is the constant of the tachogenerator expressed in V per
rad. per s; the T.F. is derived as
66 1 Control Systems Modelling and Their Representation

Fig. 1.45 D.C.

Tachogenerator +

N T S eg(t) Output

−
(t) Input
KT =volts/rad/sec

Eg
(s) = K T (1.73)
ω

The tachogogenerator is shown in Fig. 1.45.

1.14.4 Performance Comparison of Armature and

Field Controlled D.C. Motors

The performance comparison of armature and field controlled D.C. motors are given
in Table 1.2.

1.15 Two Phase A.C. Servomotor

The two phase servomotors are basically, two phase, reversible induction motors
modified for servo operations. These motors are used in applications which require
rapid accurate response characteristics. Therefore, they are designed with small diam-
eter rotor with high resistance rotor coils. The small diameter provides faster response
and the high resistance in the rotor provides linear speed torque characteristics with
a negative slope which provides positive damping and hence the stability.
The schematic diagram of a two phase servomotor is shown in Fig. 1.46a. It
consists of a stator and a rotor. In the rotor slots heavy copper bars of high resistance
are inserted and their ends are shorted by end rings. The stator consists of a pair
of winding on the inner periphery of the stator with an angle space of 90◦ with
reference to their axes. One winding is called the reference winding while the other
is called the control winding. The voltages applied to these windings also differ in
phase by 90◦ . Usually, the reference winding is excited by the voltage vr (t) from
a single phase A.C. supply. The control voltage vc (t) after sending through a 90◦
phase shifter is applied to the control winding. The direction of rotation of the rotor
depends on the phase relationship of these two voltages. These two phase voltages
1.15 Two Phase A.C. Servomotor 67

Table 1.2 Comparison of armature and field controlled D.C. motors

Armature controlled D.C. Field controlled D.C. motor
motor
1. Electrical time constant of the Since the field coil inductance
motor is very
is very small and hence the high the electrical time
time constant is
response is faster high and the time response is
sluggish
2. Efficiency is high Efficiency is not so good
3. Due to back emf developed There is no additional damping
there is here and the oscillations are
inherent feedback inside the high
system when the system is disturbed
which increases the damping.
This avoids unwanted
oscillations in the
system variables due to
disturbances
4. Power amplifiers are required Power amplifies are used in the
while field
the D.C. motors are used. circuit where the current and
Design of power
power amplifiers which are requirements are low.
incorpo- Designing a
rated in the armature circuit is power amplifier does not pose
little any
combursone and the cost is problem and the cost is also
high less
5. Armature controlled D.C. Field controlled D.C. motor
motor requires
requires constant voltage constant current source in the
source in arma-
the field circuit which can be ture circuit which is difficult to
easily get
obtained

establish a rotating magnetic field of synchronous speed Ns and sweeps over the
rotor conductors. Voltages are induced in these stationary conductors. Since they are
short circuited a heavy current is circulated in the rotor conductors. These conductors
which are carrying current when situated in a magnetic field experience a force which
moves them in the same direction as that of the rotating magnetic field with a speed
slightly less than the synchronous speed. The relative speed of the rotor and that of
the notating magnetic is thus reduced and rotates at steady speed ω(t).
The dynamic behavior of the two phase servomotor is decided if the transfer
function relating ω(t) = dθ (t)/dt where θ (t), the angular position of the rotor is
known. This can be obtained from the torque speed characteristic of the motor. The
68 1 Control Systems Modelling and Their Representation

(a)
Rotor

vc(t) Control
Input Winding J

(t)
Output

f
Reference
Winding
vr(t)

(b) (c) T0
Torque T

Torque T

R1 Vc1>Vc2>Vc3
R2 Vc1
Vc2
R3>R2>R1
R3 Vc3

Speed s Speed 0

Fig. 1.46 a Two phase A.C. servomotor; b Torque-speed characteristic of ordinary induction motor
and c Torque-speed characteristic of two phase A.C. servomotor

torque speed characteristic of the ordinary two phase induction motor is shown in
Fig. 1.46b. If these characteristic curves have positive slope, while starting, it can
be proved that it will result in unstable operation. By intentionally introducing, high
resistance rotor conductors, the torque-speed characteristic can be linearized with a
negative slope which ensures stability to the motor but at the cost of efficiency. In
control applications, stability and fastness of response of the motor are given priority
over efficiency. The following parameters and variables are defined as given below:

Block rotor torque at the rated voltage vc

K = in N-m/V
Rated control phase voltage
−T0
m= = Slope of the torque − Speed curve
ω0
Rotor stand still torque
= N-m/rad/s
No load speed

The torque Td developed by the motor is given by

1.15 Two Phase A.C. Servomotor 69

Td = Static torque + Dynamic torque

dθ
= K vc (t) + m
dt
This torque is used to accelerate the motor and overcome friction and any load
connected.
d 2θ dθ
TL = J 2 + f
dt dt

where J is combined inertia connected to the motor shaft in kg-m2 and f is combined
friction connected to the motor shaft in N-m/rad/s.

Td = TL
dθ d 2θ dθ
K vc (t) + m =J 2 + f
dt dt dt
Taking Laplace transform on both sides we get,

K Vc (s) = s[J s + ( f − m)]θ (s)

θ K
(s) = (1.74)
Vc s[J s + ( f − m)]

θ Km
(s) =
Vc s[1 + sTm ]

where K m = K /( f − m) is motor gain constant and Tm is mechanical time constant

Tm = J/( f − m).
Since m is negative, in Eq. (1.74), the poles are in the left half of the s-plane
and the motor is stable with added damping. In case of ordinary induction motor
m is positive and m  f . Now one pole of Eq. (1.74) is in the right half of s-plane
and hence the motor dances and jumps when the supply is initially given until its
slope of the torque-speed curve becomes negative. That is why ordinary two phase
induction motors are unsuitable for servo applications.

A.C. Technogenerator
In Fig. 1.46a if mechanical power is given as the input to the shaft, the machine will
give out voltage vc (t) in the control winding whose magnitude is proportional to the
speed ω(t) and its frequency is the same as that of the reference voltage frequency.
The T.F of the A.C. tachogenerator is given as
70 1 Control Systems Modelling and Their Representation

Fig. 1.47 Synchro generator

Vc
(s) = K g (1.75)
ω

where K g is expressed in V/rad./s. For purpose of feedback in A.C. control system

which requires device which will generate A.C. voltage proportional to speed but at
constant frequency, A.C tachogenarators are used.

1.16 A Pair of Synchros

A synchro is a type of electrical transformer that is used for measuring the angle
of a rotating machine such as antenna platform. In its general construction, it is
much like a three phase alternator. In servo applications it is mainly used as an error
detector. It consists of a pair of synchro which are named as synchro transmitter
and synchro receiver. The synchro transmitter is also called as synchro generator
(SG) and synchro receiver or control transformer (CT). Their basic construction and
principles of operation are explained below.

1.16.1 Synchro Generator

A synchro transmitter consists of a stator and a rotor. In the stator, a Y connected

three phase winding is provided. It resembles that of a three phase synchronous
generator. The synchro generator has a salient pole rotor over which a coil is wound.
Single phase excitation is given to the rotor coil through slip rings R1 R2 . The flux
produced by this electro magnet links the stator windings. The rotor shaft is normally
connected to any external rotation which is the input to the system. When the rotor
rotates, emf’s are induced in the stator windings S1N , S2N and S3N . Eventhough
1.16 A Pair of Synchros 71

Fig. 1.48 Schematic

diagram of synchro control
transformer

the construction of the stator winding resembles that of a three phase synchronous
generator, the voltages induced in these coils are only single phase voltages.

1.16.2 Control Transformer

The schematic diagram of control transformer is shown in Fig. 1.48. The construction
is similar to that of a synchro generator except that the rotor is a bell shaped cylindrical
one. This special construction is meant for minimizing the change in rotor impedance
when it is rotated. This is necessary because the output of the rotor is connected to
the servo-amplifier.
For most of the servo applications, the synchro pair acts as an error detector. The
stator coils of SG and CT are connected with the corresponding phase windings. (S1
to S1 , S2 to S2 and S3 to S3 ). Single phase supply is given to the rotor of SG, which
produces magnetic flux. When SG rotor is rotated, there will be emf induced in the
stator coils of SG. These voltages are now applied to the CT stator windings. These
voltages again set up their own flux and induce a voltage in the CT rotor coil. The
flux pattern in the air gap of CT rotor is same as that of the SG rotor air gap.

1.17 Electrical Analogy

Mechanical translational and rotational systems could be represented and studied by

their equivalent electrical circuits which are more easily constructed than the corre-
sponding models. Similarity between the equations resulting from Kirchhoff’s laws
for electrical systems and equations of motion of mechanical systems obtained using
D’Alembert’s principle exist and hence analogous electric circuits can be just con-
structed by inspection. The variables of the electric circuit so obtained behave exactly
as the analogous variables of the mechanical systems. Conversion of mechani-
72 1 Control Systems Modelling and Their Representation

(a) v(t) (b) L M R B

K
e(t) i(t) C
M f(t) +
−
f(t) v(t) 1
B K

Fig. 1.49 a Mechanical translational system and b Electrical analogous circuit

cal systems to electrical network is done before writing the describing equa-
tions. However, the one to one correspondence between the equations describing the
dynamic behaviour of both these systems can be verified by writing the equations
corresponding to these two systems. An electric circuit that is analogous to a mechan-
ical system is called electric circuit analog. The equation of motion of a mechanical
system when compared to mesh equation of electric circuit it is called a series analog
and when compared to nodal equations it is called parallel analog. Series analog
is called Force-Voltage (Torque-Voltage in rotational system) analog. Parallel
analog is called Force-Current (Torque-Current in rotational system) analog.

1.17.1 Force–Voltage Analogy

Consider the mechanical system shown in Fig. 1.49a. As discussed before, the fol-
lowing equation of motion is written

dv(t) 1
M + Bv(t) + v(t)dt = f (t) (1.76)
dt K

For the electric circuit of Fig. 1.49b the following mesh equation is written

di(t) 1
L + Ri(t) + i(dt) = e(t) (1.77)
dt C

The electrical circuit of Fig. 1.49b is said to be analogous to the mechanical system
of Fig. 1.49a with the following identities:

1. Inductor L is analogous to mass M.

2. Resistor R is analogous to viscous friction B.
3. Capacitor C is analogous to reciprocal of spring K .
4. Voltage e(t) is analogous to the applied force f (t).
5. Mesh-current i(t) is analogous to velocity v(t).
1.17 Electrical Analogy 73

Fig. 1.50 Force-current e(t) v(t)

electric analogous circuit
ic iR iL

i(t) C R L

f(t) M 1 1
B K

The above analogy also holds good for mechanical rotational system. Here e(t) is
analogous to the torque applied and the mesh current i(t) is analogous to angular
velocity ω(t). Other things remain the same with appropriate equivalent units.

1.17.2 Force–Current Analogy

Consider the electric circuit shown in Fig. 1.50, where the current source i(t) and the
circuit elements R, L and C are connected to the voltage node e(t). At node e(t),
the sum of the current is equal to zero and hence the following equation is written:

i c + i R + i L = i(t)

de(t) e(t) 1
C + + e(t)dt = i(t) (1.78)
dt R L

Now compare Eq. (1.78) with Eq. (1.76). It is found C is analogous to the mass
M, 1/R is analogous to viscous friction B and 1/L is analogous to the spring K .
The current i(t) is analogous to the applied force f (t) and the nodal voltage e(t)
is analogous to the linear velocity v(t). The same analogy can also be extended to
the mechanical translational system. The summary of analogy of mechanical and
electrical systems are given in Table 1.3.

Example 1.19
Consider the mechanical system shown in Fig. 1.51a. Draw the electrical circuits
based on both (a) Force-voltage and (b) Force-current analogies. Write the differential
equations governing the dynamic behaviour of these systems.

(Anna University, May, 2011)

Solution Force-Voltage Analogy
1. Since f (t) is applied to mass M1 , the voltage source e(t) in Fig. 1.51b is con-
nected to the series connection of L 1 → M1 and R1 → B1 which move with
velocity v1 . The current flow in this series circuit is therefore i 1 → v1 .
74 1 Control Systems Modelling and Their Representation

Table 1.3 Analogy of mechanical and Electrical systems

Mechanical system Electrical system
S. no. Items Translational Rotational Loop system Nodal system
system system (Force- (Force-
voltage) current)
1. Independent Force f (t) Torque T (t) Voltage e(t) Current i(t)
variable
2. Dependent Linear Angular Current i(t) Voltage e(t)
variable velocity v(t) velocity ω(t)
3. Power Linear Rotational Resistance Conductance
dissipation dash-pot dash-pot B G
component
where where where where
f (t) = Bv(t) T = Bω e(t) = i(t)R i(t) = Ge(t)
4. Energy storing Mass M Inertia J Inductance L Capacitance C
element
where where where where
dv dω di de(t)
f (t) = M T (t) = J e(t) = L i(t) = C
dt dt dt dt
5. Energy storing Spring K Spring K Capacitance C Inductance L
element
 f (t) =
where  T (t) = where 1 
where where I =
1 
K vdt K ωdt e(t) = idt e(t)dt
C L
6. Physical laws D’Alembert’s D’Alembert’s Kirchhoff’s Kirchhoff’s
principle principle voltage law current law
   
f =0 T =0 e(t) = 0 i(t) = 0
7. Changing the Lever Gear Voltage Current
level of the f1 I2 T1 N1 transformer transformer
= = e1 N1 i1 N2
independent f2 I1 T2 N2 = =
variable e2 N2 i2 N1

2. The mass M2 , spring K 2 and dash-pot B2 move with a velocity v2 . So in the

electrical circuit a branch is created where the current flow is i 2 → v2 . In this
branch L 2 → M2 , R2 → B2 and C2 → (1/K 2 ) are connected in series.
3. The inductance K 1 and dash-pot B12 move with a velocity difference (v1 − v2 ).
Hence, in the electrical circuit a branch in which (i 1 − i 2 ) → (v1 − v2 ) flows is
created. In this branch C1 → (1/K 1 ) and R12 → B12 are connected in series.
This completes the force-voltage analogous circuit. This circuit equation are
verified with that of the mechanical system.

From Fig. 1.51b, for the loop consisting of L 1 , R1 , C1 and R12 the following equation
is written:

dt1 1
L1 + i 1 R1 + (i 1 − i 2 )dt + R12 (i 1 − i 2 ) = e(t) (1.79)
dt C1
1.17 Electrical Analogy 75

(a) v1(t) v2(t)

K1 K2

f(t) M1 M2

B12 B2
B1

(b)
i1 v1 i2 v2 R2 B2

L1 R1 L2 M2
C1 1
f(t) K1
M1 B1
i1−i2
e(t) C2 1
R12 B12 K2

(c)
v1 v2

M1 dv1 K1∫(v1−v2)dt
dt K1∫(v2−v1)dt K2∫v2dt

dv2
f(t) M1 M2 M2
dt

B1v1 B12(v1−v2) B12(v2−v1)

B2v2

(d) L1 1
K1

e1 v1 e2 v2

R12 1
f(t) R1 B12 R2 L2
C1 C2
1 1 1
i(t) M1 M2 K2
B1 B2

Fig. 1.51 a Mechanical translational system for Example 1.19. b Force-voltages analogy of exam-
ple. c Free-body diagram of Fig. 1.51a. d Force -current analogous circuit of Fig. 1.51a
76 1 Control Systems Modelling and Their Representation

For the loop consisting of L 2 , R2 , C2 , R12 and C1 , the following equation is written:

di 2 1
L2 + i 2 R2 + R12 (i 1 − i 2 ) + (i 1 − i 2 )dt
dt C1

1
+ i 2 dt = 0 (1.80)
C2

The free-body diagram of the given mechanical system given in Fig. 1.51a is
shown in Fig. 1.51c. For Fig. 1.51c the following equations are written:

dv1
M1 + B1 v1 + B12 (v1 − v2 ) + K 1 (v1 − v2 )dt = f (t) (1.81)
dt
 
dv2
M2 + B2 v2 + B12 (v2 − v1 ) + K 1 (v2 − v1 )dt + K 2 v2 dt = 0 (1.82)
dt

From Table 1.3, Eq. (1.79) is analogous to Eqs. (1.81) and (1.80) is analogous too
Eq. (1.82). Hence, force-voltage analogy is verified.
Force-Current Analogy
1. In Fig. 1.51a, mass M1 and dash-pot B1 move with a velocity v1 . The force is
applied to mass M1 . Hence, the one end of current sources i(t) → f (t), C1 →
M1 and R1 → (1/B1 ) are connected to the node e1 → v1 . Their other ends are
connected to the common point.
2. Mass M2 , dash-pot B2 and spring K 2 move with velocity v2 . Hence, the one end of
C2 → M2 , R2 → (1/B2 ) and L 2 → (1/K 2 ) are connected to the mode e2 → v2 .
Their other ends are connected to the common points.
3. The dash-pot B12 and spring K 1 move with a velocity difference (v1 − v2 ). There-
fore, R12 → (1/B12 ) and L 1 → (1/K 1 ) are connected in parallel and in between
the nodes e1 → v1 and e2 → v2 . This completes the force-current analogy. This
is verified by writing down the electric circuit equations.

For the node e1 , the following equation is written:


di 1 1 1 1
C1 + e1 + (e1 − e2 )dt + (e1 − e2 ) = i(t) (1.83)
dt R1 L1 R12

For the node e2 , the following equation is written:

 
de2 1 1 1 1
C2 + e2 + e2 dt + (e2 − e1 ) + (e2 − e1 )dt = 0 (1.84)
dt R2 L2 R 12 L2

From Table 1.3, it is evident that Eq. (1.83) is analogous to Eqs. (1.81) and (1.84) is
analogous to Eq. (1.80). Thus, force-current analogy is verified.
1.17 Electrical Analogy 77

(a) (b) i1 v1 i2 v2 L2 M2

B1 L1 R1 i1−i2
K
f(t)
M1 B1 C 1
R2 B2 K
e(t)
M2 v2

(c)
f(t) v1 v2
B2 dv2
M1 dv1 M2
dt dt

M1 v1 f(t) M1 B1v1 M2 K2∫v2dt

B2 (v1−v2) B2 (v2−v1)

(d) e1 v1 e2 v2

R2 1
B2
f(t) C1 R1 C2
1
L
1 M2 K
i(t) M1 B1

Fig. 1.52 a Mechanical system for Example 1.20. b Force-voltage analogous circuit of Fig. 1.52a.
c Free-body diagram of Fig. 1.52a. d Force-current analogy of Fig. 1.52a

Example 1.20
Draw the force voltage and force-current analogous circuits for the mechanical trans-
lational system given in Fig. 1.52a.

(Anna University, May, 2010)

Solution Force-Voltage Analogy

1. Force f (t) is applied to mass M1 which moves with velocity v1 . Dash-pot B1

is connected to mass M1 and it also moves with the same velocity. A voltage
source e(t) → f (t) is created and in series with e(t), L 1 → M1 and R1 → B1
are connected in series as shown in Fig. 1.52b.
2. Mass M2 moves with velocity v2 and also the spring K . A loop wherein the
current flow is i 2 is created and the series combination of L 2 → M2 and C →
(1/K ) is connected.
3. The velocity difference across B2 is (v1 − v2 ). A branch current (i 1 − i 2 ) is
opened from i 1 and i 2 and R2 → (1/B2 ) is connected in that branch. The force
voltage analogy is now completed and is shown in Fig. 1.52b.
4. The free-body diagram of the mechanical system is shown in Fig. 1.52c.
78 1 Control Systems Modelling and Their Representation

For Fig. 1.52b the following equations are written:

di 1
L1 + R1 i 1 + R2 (i 1 − i 2 ) = e(t) (1.85)
dt 
di 2 1
L2 + i 2 dt + R2 (i 2 − i 1 ) = 0 (1.86)
dt C

For Fig. 1.52c, the following equations are written:

dv1
M1 + B1 v1 + B2 (v1 − v2 ) = f (t) (1.87)
dt 
dv2
M2 + K v2 dt + B2 (v2 − v1 ) = 0 (1.88)
dt

Equation (1.85) is analogous to Eqs. (1.87) and (1.86) to (1.89) as seen from Table 1.3.
Hence, force-voltage analogy is verified.

Force-Current Analogy
1. Mass M1 and dash-pot B1 move with velocity v1 and the force is applied to mass
M1 . A node e1 → v1 is created and to this node e(t) → f (t), C1 → M1 and
R1 → (1/B1 ) are connected as shown in Fig. 1.52d.
2. Mass M2 and spring K move with velocity v2 . A node e2 → v2 is created and to
this node C2 → M2 and L → (1/K ) are connected as shown in Fig. 1.52d.
3. The dash-pot B2 move with a velocity difference of (v1 − v2 ). Hence, R2 →
(1/B2 ) is connected in between the nodes e1 and e2 . This completes the force–
current analogy.
For Fig. 1.52d the following equation is written at node e1

de1 1 1
C1 + e1 + (e1 − e2 ) = i(t) (1.89)
dt R1 R2

de2 1 1
C2 + e2 dt + (e2 − e1 ) = 0 (1.90)
dt L R2

Equation (1.89) is analogous to Eqs. (1.87) and (1.90) is analogous to Eq. (1.89) and
hence force-current analogy is verified.

Example 1.21
Consider the mechanical system shown in Fig. 1.53a. Draw the force-voltage and
force-current analogous circuits.

Solution Force-Voltage Analogy

1. In Fig. 1.53a the junction point A of dash-pot B and spring K is neither connected
to any mass nor to any reference frame. Hence, that point is assumed to move
with a velocity v3 .
1.17 Electrical Analogy 79

(a)
v1 v3 v2
K1 K2
K B
M1 M2 f(t)
A
B1 B2

(b) C1 1
K1 L2 M2 R2 B2 i2 v2

L1 R1 (i2−i1) C2

i1 M1 B1 R 1
v1 C 1 K2 e(t) = f(t)
K
B (i2−i3)
(i3−i1)

(c)
v1 v3

K1∫v1dt B (v3−v1)

M1 dv1 M1 B (v1−v3) A
dt

B1v1 K∫(v3−v2)dt

v2

K∫(v2−v3)dt K2∫v2dt

M2 f(t)

dv2
M2 B2v2
dt

Fig. 1.53 a Mechanical system for Example 1.21. b Force-voltage analogy of Fig. 1.53a. c Free-
body diagram of Fig. 1.53a. d Force-current analogy of Fig. 1.53a

2. Force f (t) is applied to mass M2 . The spring K 2 , dash-pot B2 and mass M2 all
move with velocity v2 . The voltage source e(t) → f (t), L 2 → M2 , R2 → B2
and C2 → (1/K 2 ) are all connected in series and the current flow in this branch
is i 2 → v2 .
3. Mass M1 , dash-pot B1 and spring K 1 all move with velocity v1 . A branch current
i 1 is opened and in this branch L 1 → M1 , R1 → B1 and C1 → (1/K 1 ) are
connected in series.
4. The velocity difference across the cascade connected B and K is (v1 − v3 ). From
i 1 and i 2 branches, (i 2 − i 1 ) branch is created. The velocity difference across K
80 1 Control Systems Modelling and Their Representation

is (v2 − v3 ). From (i 2 − i 1 ) branch, a new branch with (i 2 − i 3 ) current is created

and C → (1/K ) is connected in that branch. Similarly, the velocity difference
across B is (v3 − v1 ). The branch current (i 3 − i 1 ) is created from (i 2 − i 1 ) and
in this branch R → B is connected. The series connected mechanical elements
B and K in Fig. 1.53a become parallel connected R and C in Fig. 1.53b. This
completes the force-voltage analogy. The free-body diagram of Fig. 1.53a is
represented in Fig. 1.53b. For Fig. 1.53a the following equation is written:

di 1 1
L1 + R1 i 1 + i 1 dt + R(i 1 − i 3 ) = 0 (1.91)
dt C1
 
di 2 1 1
L2 + R2 i 2 + i 2 dt + (i 2 − i 3 )dt = e(t) (1.92)
dt C2 C
1
R(i 3 − i 1 ) + (i 3 − i 2 )dt = 0 (1.93)
C
From free-body diagram of Fig. 1.53c the following equations are written:

dv1
M1 + B1 v1 + K 1 v1 dt + B(v1 − v3 ) = 0 (1.94)
dt
 
dv2
M2 + B2 v2 + K 2 v2 dt + K (v2 − v3 )dt = f (t) (1.95)
dt

B(v3 − v1 ) + K (v3 − v2 )dt = 0 (1.96)

Equations (1.91), (1.94), (1.92), (1.95), (1.93) and (1.96) are analogous as seen
from Table 1.3. Hence, force-voltage analogy is proved.

Force-Current Analogy
1. Corresponding to the velocities v1 , v2 and v3 in the mechanical system, nodal
voltage points e1 , e2 and e3 are created for force-current analogy as shown in
Fig. 1.53d.
2. Mass M1 , dash-pot B1 and spring K 1 all move with velocity v1 and therefore the
one ends of C1 → M1 , R1 → (1/B1 ) and L 1 → (1/K 1 ) are connected to the
node e1 while their other ends are connected to the common point.
3. The force f (t) is applied to mass M2 . Mass M2 , spring K 2 and the dash-pot B2
all move with velocity v2 and therefore the one ends of i(t) → f (t), C2 → M2 ,
R2 → (1/B2 ) and L 2 → (1/K 2 ) are connected to the node e2 and their other
ends are connected to the common point.
4. The velocity difference across B is (v1 − v3 ) and therefore R → B is connected
between e1 and e3 nodes. Similarly the velocity difference across K is (v2 − v3 ).
The inductance L → (1/K ) is connected between the nodes e2 and e3 . This
completes the force-current analogy. The following equations are written for
Fig. 1.53d and verified with those obtained for the given mechanical system
shown in Fig. 1.53a.
1.17 Electrical Analogy 81

At node e1 , the following equation for the algebraic sum of the current entering and
leaving is written.

de1 1 1 1
C1 + e1 + e1 dt + (e1 − e3 ) = 0 (1.97)
dt R1 L1 R

At node e2 , the following equation is written:

 
de2 1 1 1
C2 + e2 + e2 dt + (e2 − e3 )dt = i(t) (1.98)
dt R2 L2 L

At node e3 , the following equation is written:


1 1
(e3 − e1 ) + (e3 − e2 )dt = 0 (1.99)
R L

Equations (1.94), (1.97), (1.95), (1.99), (1.91) and (1.99) are as per Table 1.3 are
analogous. Hence, force-current analogy is verified.

Example 1.22
For the rotational mechanical system shown in Fig. 1.54a draw the torque-voltage
and torque-current analogous circuits.

Solution
Torque-Voltage Analogy
1. In Fig. 1.54, torque T (t) is applied to inertia J1 . Inertia J1 and damper B1 expe-
rience the same angular velocity ω1 . Therefore, the voltage source e(t) → T (t),
inductor L 1 → J1 and resistor R1 are all connected in series and the current flow
in this branch is i 1 → ω1 .
2. A branch with i 2 current is created. J2 and B2 move with a velocity ω2 . Therefore,
in the branch where i 2 → ω2 flows, the series combination of L 2 → J2 and
R2 → B2 are connected.
3. From branches where i 1 and i 2 flow, a third branch (i 1 − i 2 ) → (ω1 − ω2 ) is
created. The velocity difference across the spring K and dash-pot B12 is (ω1 −
ω2 ). In the branch where the current flow is (i 1 − i 2 ). R12 → B12 and C = 1/K
are connected in series. This completes the torque-voltage analogous circuit. The
following equations are written for Fig. 1.54b.

di 1 1
L1 + R1 i 1 + R12 (i 1 − i 2 ) + (i 1 − i 2 )dt = e(t) (1.100)
dt C

di 2 1
L2 + R2 i 2 + R12 (i 2 − i 1 ) + (i 2 − i 1 )dt = 0 (1.101)
dt C
82 1 Control Systems Modelling and Their Representation

(a) K
T(t) ω1

J1 J2

ω2
B12
B1 B2

(b) L2 J2
i1 1 i2 2

L1 J1 R1 R2
T(t) R12 B12
B1 B2
e(t) 1
C K

(c) 1 2
T(t) K∫( 1− 2)dt K∫( 2− 1) J2 d 2
dt
J1 d 1
dt J1 J2

B1 B12 ( B12 (
1 1− 2) 2− 1) B2 2

(d)
e1 L1 1 e2
K

R1 R2
f(t) C1 R12 C2

e(t) J1 J2 1
1 1
B1 B12
B12

Fig. 1.54 a Mechanical rotational system for Example 1.22. b Torque-force analogous circuit of
Fig. 1.54a. c Free-body diagram of Fig. 1.54a. d Torque-current analogous circuit of Fig. 1.54a

For the free-body diagram of Fig. 1.54c, the following equations are written:

dω1
J1 + B1 ω1 + B12 (ω1 − ω2 ) + K (ω1 − ω2 )dt = T (t) (1.102)
dt

dω2
J2 + B2 ω2 + B12 (ω2 − ω1 ) + K (ω2 − ω1 )dt = 0 (1.103)
dt

Equations (1.100), (1.102), (1.101) and (1.103) are analogous as seen from
Table 1.3. Hence, the torque-voltage analogy is verified.
1.17 Electrical Analogy 83

Torque-Current Analogy
The following equations are written for Fig. 1.54d. At node e1 we equate the current
entering to the current leaving as given below:

de1 e1 1 1
C1 + + (e1 − e2 ) + (e1 − e2 )dt = i(t) (1.104)
dt R1 R12 L

At node e2 we write,

de2 1 1 1
C2 + e2 + (e2 − e1 ) + (e2 − e1 )dt = 0 (1.105)
dt R2 R12 L

Equations (1.102), (1.104), (1.103) and (1.105) are seen to be analogous as per
Table 1.3. Hence, torque-current analogy is verified.

Example 1.23
Consider the mechanical system shown in Fig. 1.55a. Draw the torque-voltage and
torque-current analogous circuits.

Solution
Torque-Voltage Analogy

1. In the mechanical system of Fig. 1.55a there are four angular displacements ω1 ,
ω2 , ω3 and ω4 . Corresponding to these displacements, branch currents i 1 , i 2 , i 3
and i 4 are created. Torque T (t) is applied across the spring K 1 and dash-pot B1 .
A voltage source e(t) → T (t) is created. From this source i 1 → ω1 flows. From
i 1 branch i 2 branch and (i 1 − i 2 ) branch are created.
2. The velocity difference across K 1 and B1 is (ω1 − ω2 ) and therefore R1 → B1
and C1 → (1/K 1 ) are connected in series with the branch where the current
flow is (i 1 − i 2 ). J1 moves with angular velocity ω2 . L 1 → J1 is connected in
the branch where the current flow is i 2 .
3. From i 2 branch, two more branches are created, one branch where i 3 current flows
and another branch where (i 2 − i 3 ) flows. Spring K 2 moves with angular speed
difference of (ω2 − ω3 ) and therefore, C2 → (1/K 2 ) is connected where the
current flow is (i 2 − i 3 ) → (ω2 − ω3 ). The current i 3 → ω3 which corresponds
to primary gear. i 3 is connected to the primary of the voltage transformer whose
turns ratio is N2 : N1 .
4. The secondary gear moves with angular speed ω4 . Hence, in the secondary of
the transformer i 4 current flows.
5. Inertia J2 and dash-pot B2 rotate with the angular speed ω4 and therefore L 2 →
J2 and R2 → B2 are connected in series in the secondary of the transformer.
This completes the torque-voltage analogy.

The following equations for mechanical and electrical systems are written and the
analogy verified. The free-body diagram for Fig. 1.55a is shown in Fig. 1.55c. For
84 1 Control Systems Modelling and Their Representation

(a) K1 T1(t)
K2 N1
A J1

T(t) ω 1 B1 ω2 ω3 ω4

J2

T2(t) B2
N2

(b) L1 J1 i2 i4
i1 1 2 N2 : N1 4 L2 J2
(i1i2) R1
i3 3
T(t) C2 (i2i3)
B1 e1 e2 R2 B2
N C1
e(t) 1
K2
1
K1

(c) 1
2
K2∫( 1− 2) dt
J1 d 2
T(t) A
dt
K1∫( 2− 1)dt J1 K2∫( 2− 3)
B1 ( 1− 2)
B1 ( 2− 1)
4

J2 d 4
dt
J2 T2(t)

B2 4

(d) R1 1
B1
L2 1
e1 e2 K2 e3 e4
i4
L1 1
K1
T(t) C1 C2
R2 1
i(t) J1 B2
J2

Fig. 1.55 a Mechanical rotational system for Example 1.23. b Torque-voltage analogy of Fig. 1.55a.
c Free-body diagram of Fig. 1.55a. d Torque-current analogous circuit for Fig. 1.55a
1.17 Electrical Analogy 85

Fig. 1.55b, the following equations are written:


1
R1 (i 1 − i 2 ) + (i 1 − i 2 )dt = e(t) (1.106)
C1
 
di 2 1 1
L1 + (i 2 − i 3 )dt + (i 2 − i 1 )dt + R1 (i 2 − i 1 ) = 0 (1.107)
dt C2 C1
di 4
L2 + R2 i 4 = e2 (1.108)
dt

T1 N1 ω4
= =
T2 N2 ω3
N1 ω4 e2
N1 ω3 = N2 ω4 or = =
N2 ω3 e1

Hence, the turns ratio of the transformer is N2 : N1 . For Fig. 1.55c the following
equations are written:

B1 (ω1 − ω2 ) + K 1 (ω1 − ω2 )dt = T (t) (1.109)
 
dω2
J1 + K 2 (ω2 − ω3 )dt + K 1 (ω2 − ω1 )dt + B1 (ω2 − ω1 ) = 0 (1.110)
dt

dω4
J2 + B2 ω4 = T2 (t) (1.111)
dt
Equations (1.106), (1.109), (1.107), (1.110), (1.108) and (1.111) are analogous as
per Table 1.3. Hence, torque-voltage analogy is verified.
Torque-Current Analogy
Consider Fig. 1.55d. The following equations are written. At node e1 we get,

1 1
(e1 − e2 ) + (e1 − e2 )dt = i(t) (1.112)
R1 L1

At node e2 we get,
 
de2 1 1 1
C1 + (e2 − e3 )dt + (e2 − e1 )dt + (e2 − e1 ) = 0 (1.113)
dt L2 L1 R1

At e4 node we get,
de4 1
C2 + e4 = i 4 (1.114)
dt R2
86 1 Control Systems Modelling and Their Representation

Equations (1.106), (1.112), (1.107), (1.113), (1.108) and (1.114) are analogous as per
Table 1.3. Hence, torque-current analogy is verified. It is to be noted that the current
transformer will have a turns ratio N1 : N2 .

1.18 Block Diagram Reduction Technique

A simple system is described as the operation which transforms one input signal to
produce one output signal as shown in Fig. 1.56. The system in general has the cause
and effect relationship. The block diagram represents the cause and effect relationship
or simply the functional relationship between the input signal and output signal. The
input signal r (t) can be altered or manipulated to achieve a desired system behaviour.
The output signal c(t) is a variable which is to be controlled in some desired manner.
The dynamic systems are very often represented by a set of differential equations.
The application of Laplace transform to these set of differential equations results in a
set of linear algebraic equations. Thus, a functional relationship between the output
variable and input variable is established which is defined as the transfer function.
The cause and effect relationship of the transfer function can be represented by
diagrammatic means as in Fig. 1.56 which is called the block diagram. Block dia-
grams consist of unidirectional operational blocks that represent the transfer function
of the variables of interest. It is a pictorial representation of individual components
and of the flow of the signal. In the block diagram representation while the blocks
represent the transfer function, the direction of the signal flow is represented by
arrow. The signal can only pass in the direction of the arrows. The dimensions of the
output signal from the block are the dimensions of the input signals multiplied by
the dimension of the transfer function G(s).
In Fig. 1.56 individual system is represented by a single block with its input and
output. In practice, many such systems are interconnected. Such multiple subsys-
tems can be represented with a single transfer function. The technique to reduce the
interconnected subsystem to a single block diagram as in Fig. 1.56 is known as block
diagram reduction technique and is discussed below.
Advantages of Block Diagram Representation
The following are the advantages of block diagram representation:
1. It is easy to form the overall block diagram for the entire system by merely
connecting the blocks of the components according to the signal flow.
2. It is easy to evaluate the contribution of each component to the overall perfor-
mance of the system.

Fig. 1.56 System

representation in block Input System functional Output
diagram relationship
r(t) G(s) c(t)
1.18 Block Diagram Reduction Technique 87

3. In block diagram, the functional operation of the system can be visualized more
readily by examining the individual blocks rather than examining the physical
system itself.
Disadvantage of Block Diagram Representation
The following are the disadvantages of block diagram representation:
1. The main source of energy flow in the system is not shown in block diagram.
2. For a particular system, different block diagram is drawn. It is not unique.
3. It is difficult to identify the nature of physical elements connected in the system.
The second order electrical or mechanical or hydraulic systems are all repre-
sented by the same block diagram.
4. Important functions are concealed within the walls of block diagram.
5. Block diagram representation does not take into account the effect of interaction
between blocks.
6. Passive symmetrical pi networks are reciprocal since the input and output are
inter-changeable. However, in an amplifier the input and output are not inter-
changeable. This property is not considered in block diagram representation.

1.18.1 Component Parts of Block Diagram

When multiple subsystems are interconnected new elements such as junction and
pick off points are added to the system which are shown in Fig. 1.57.
Few topologies for interconnecting subsystems are available so that a single trans-
fer function, using these topologies are obtained. These topologies include the fol-
lowing:

R(s) C(s) R(s) C ( s)

G(s)
Input Output

(c) Summing junction (d) Pick off poin

R2(s)
R(s)
+
R1(s)
R(s) = R1(s) + R2(s) − R3(s)
+
−
R3(s)

Fig. 1.57 Components of block diagram

88 1 Control Systems Modelling and Their Representation

(a)
R(s) C(s) R(s) C(s)
G1(s) G2(s) G3(s) G1(s)G2(s)G3(s)

(b)

G1(s)

+
R(s) + C(s) R(s) C(s)
G2(s) G1(s)+G2(s)−G3(s)
−

G3(s)

(c)

R(s) + C(s) R(s) G(s) C(s)

G (s )
− 1+
− GH(s)
+

H(s)

Fig. 1.58 Block diagram topology. a Cascade connection, b Parallel connection and c Feedback
connection

1. Cascade form
2. Parallel form
3. Feedback form
These connections and their equivalence are shown in Fig. 1.58.

1.18.2 Block Diagram Reduction Rule

The reduction of block diagram to a single block representation can be derived

by simple algebraic manipulation of the equations representing the blocks. These
diagram transformations are known as block diagram reduction rules. These rules
are given in Table 1.4. System analysis by the method of block diagram reduction
technique gives a better understanding of the contribution of each component than
is possible to obtain by the manipulation of equations.
The following examples illustrate the method of block diagram reduction. Table 1.4
is used whereever necessary.
Table 1.4 Block diagram reduction rules

Transformation Original Block Diagram Equivalent Block Diagram

R C
R C G1G2
1. Combining blocks in Cascade G1 G2
C = (G1G2)R

R + C R
2. Combining blocks in parallel G1 G1 G2
_
+ C
or eliminating a forward loop
1.18 Block Diagram Reduction Technique

C = (G1 G2)R
G2

R + C R G1 + C
G1 G2
_
+ G2 _
+
3. Removing a block from a
forward path G1
G2 C=G2 _ R
+1
(G 2
(
R C R G C
G1 _ 1
4. Eliminating a feedback loop _
+ 1+G1G2
G1
G2 C= (_ R
(
1+G1G2

(continued)
89
Table 1.4 (continued)
90

R C
G1 R 1 + C
_
+ G1G2
5. Removing a block from a G2 _
+
feedback loop
G2

R R X C R R X C
6. Rearranging summing points
X Y
Y X
C R X Y C R X Y

R R X C R C
7. Rearranging summing points
X Y
X
Y X
C (R X) Y
C=R ( X Y)
=R ( X Y) Y

R C R C
8. Moving a take-off point ahead G G
of a block
C C G
C GR C GR
1 Control Systems Modelling and Their Representation

(continued)
Table 1.4 (continued)

R C R C
G G
9. Moving a take-off point
beyond a block C C 1
G
C GR
C GR
R C R C
G G

10. Moving a summing point C 1 X

ahead of a block G
X
C=RG X C=G(R (1/G))X=RG X
1.18 Block Diagram Reduction Technique

R C R
G G C

11. Moving a summing point

behind a block X G
X
C=RG XG=G(R X)
C=RG XG=G(R X)

R C R C

12. Moving a take-off point

ahead of a summing point C C
X
C=R X X
91

(continued)
 92

Table 1.4 (continued)

R C
R C
13. Moving a take-off point
R X
beyond a summing point
X R
C=R X

R X
14. Moving a take-off point R R X R
behind a summation R X
R X
X

(continued)
1 Control Systems Modelling and Their Representation
1.18 Block Diagram Reduction Technique 93

Example 1.24
Reduce the block diagram shown in figure and obtain its closed loop transfer function
C
(s).
R
(Anna University, May, 2009)
Solution The successive reductions of blocks using Table 1.4 are explained below.

1. In Fig. 1.59b G 2 and H2 blocks are connected in feedback mode and their equiv-
alence is G 2 /1 + G 2 H2 (Rule 4). The summers S1 and S2 are interchanged using

(a)
R(s) + + − +
G1 G2 C(s)
− −

H1 H2

(b)
Feedback
S1 − +
R(s) + + S2
G1 G2 C(s)
− −

H1 H2

(c)
−
R(s) + G2
G1 C(s)
S2 + (1+G2H2)
S1 −
H1

(d)
R(s) + G1 G2 C(s)
− 1+G1H1 1+G2H2

(e)
R(s) + G1G2
C(s)
− (1+G1H1)(1+G2H2)

Fig. 1.59 a Block diagram for Example 1.24. c Reduced block diagram. d Reduced block diagram
(e)
94 1 Control Systems Modelling and Their Representation

Rule 6 so that the blocks G 1 and H1 come under feedback mode. The reduced
block diagram is shown in Fig. 1.59c.
2. G 1 and H1 blocks are in feedback mode and they are transformed as
G 1 /1 + G 1 H1 . The reduced block diagram is shown in Fig. 1.59d.
3. The two blocks in the forward path are in cascade and they are transformed as

G1G2
(1 + G 1 H1 )(1 + G 2 H2 )

and is shown in Fig. 1.59e.

4. Figure 1.59e is in feedback mode and is reduced using Rule 4 with the following
result.

C G1G2
(s) =
R (1 + G 1 H1 )(1 + G 2 H2 ) + G 1 G 2

Example 1.25
Using block diagram reduction technique find the closed loop T.F. of the system
whose block diagram is given in Fig. 1.60a.

(Anna University, December, 2010)

(a) (d)
H2 Feedback
R(s) + + G2G3 + C(s)
R(s) + + − + G1
G1 G2 G3 C(s) R1 (1+G2G3H2) C1
+ − + + − +
H1
H1
G3
G4 H1
G3

(b) G4
H2

R(s) − + C(s)
+ +
G1 G2 G3 (e)
+ + Feedback
−
R(s) R2(s) + G1G2G3 + C(s)
H1
+ (1+G2G3H2)+G2H1 C1
+
G4 H1
G3
(c) Feedback
H1 G4
−
R(s) + + + C(s)
G1 G2G3
+ (f)
− +
H1 R(s) G1G2G3 + C(s)
G3 (1+G2G3H2+G2H1−G1G2H1) +
H1
G3
G4
G4

Fig. 1.60 a Block diagram for Example 1.25

1.18 Block Diagram Reduction Technique 95

Solution
1. In Fig. 1.60a, using Rule 9, the take off point of H1 block is moved beyond G 3
block. This enables G 2 and G 3 blocks in cascade. The reduced block diagram is
shown in Fig. 1.60c.
2. The blocks G 2 G 3 and H2 are in feedback mode and they are reduced using
Rule 4 as
G2G3
.
(1 + G 2 G 3 H2 )

The reduced block diagram is shown in Fig. 1.60d.

3. The blocks inside the dotted line are in feedback mode. Using Rule 4, they are
reduced as,

C1 G 2 G 3 /(1 + G 2 G 3 H2 )
=
R1 1 + (G 2 G 3 H1 /(1 + G 2 G 3 H2 )G 3 )
G2G3
=
(1 + G 2 G 3 H2 ) + G 2 H1

The reduced block diagram is shown in Fig. 1.60e. Coupled with cascaded block
G1.
4. The blocks inside the dotted line are in positive feedback mode and they are
reduced using Rule 4 as,
G1 G2 G3
C1 (1+G 2 G 3 H2 +G 2 H1 ) G1G2G3
= =
R2 1 − (1+G 2GG13GH22G+G
3 H1 1 + G 2 G 3 H2 + G 2 H1 − G 1 G 2 H1
2 H1 ) G 3

The reduced block diagram is shown in Fig. 1.60f.

5. The two blocks shown in Fig. 1.60f are in parallel connection. Rule 2 is used to
reduce these blocks and thus we get,

C(s) G1G2G3
= + G4
R(s) (1 + G 2 G 3 H2 + G 2 H1 − G 1 G 2 H1 )

C(s) G 1 G 2 G 3 + G 4 (1 + G 2 G 3 H2 + G 2 H1 − G 1 G 2 H1 )
=
R(s) (1 + G 2 G 3 H2 + G 2 H1 − G 1 G 2 H1 )
96 1 Control Systems Modelling and Their Representation

Example 1.26
The block diagram of a closed loop system is shown in Fig. 1.61a. Using block
diagram reduction technique determine the closed loop T.F.

(Anna University, December, 2009)

Solution
1. The blocks G 2 and H2 are in negative feedback mode and using Rule 4,
their equivalence is G 2 /(1 + G 2 H2 ). The reduced block diagram is shown in
Fig. 1.61c.
2. The block H1 in the feedback path is shifted after the summer S2 . Rule 11 is
used and the modified block diagram is shown in Fig. 1.61d.
3. From Fig. 1.61d, between R(s) and R1 variable, a parallel connection exists and
this part is written as
R1
= (1 + G 1 )
R
Between R1 and C(s), the two blocks are connected in negative feedback mode
and using Rule 4 we get,

C G 2 /(1 + G 2 H2 ) G2
= =
R1 1 + (G 1 G 2 H1 /(1 + G 2 H2 )) (1 + G 2 H2 + G 1 G 2 H1 )

This is represented in Fig. 1.61e.

4. From Fig. 1.61e the final input/output ratio is obtained as,

C(s) G 2 (1 + G 1 )
=
R(s) (1 + G 2 H2 + G 1 G 2 H1 )

Example 1.27
C(s)
For the block diagram shown in Fig. 1.62a obtain the overall transfer function .
R(s)
1.18 Block Diagram Reduction Technique 97

(a)

+
R(s) + +
G1 G2 C(s)
− −

H2

H1

(b)

+
R(s) + + C(s)
G1 G2
− −
H2

H1

(c)

R(s) + + G2 C(s)
G1
− S2 (1+G2H2)
S1

H1

(d)
Feedback
+ C(s
R(s) + R1 + G2
G1
(1+G2H2)
S2 − S1

G1H1

(e)

R(s) R1 G2 C(s)
1G1 (1G2H2G1G2H1)

Fig. 1.61 Block diagram for Example 1.26

98 1 Control Systems Modelling and Their Representation

(a)

R(s) + + + C(s)
G1 G2 G3
− − −

H3 H2 H1

(b )
R(s) + + +
G1 G2 G3 C
− −
S1 − S2 S3

H3 H2 H1
A B

(c)
Cascade Feedback
R(s) + + + C(s)
G1 G2 G3
S1 − S2 −
−
R1
S3
1
G1 H2
1
H2
H3 H2 H1
B
A

(d)

R(s) + + R2 R1 G3 C(s)
G1 G2
− − (1+G3H1H2)

Cascade
H3 1
G1 H2

A H1 H2

Parallel

(e)

R(s) + R2 G1 G2 G3 C(s)
− (1+G3H1H2)

(1+G1H2H3)
H1 H2
G1 H2

Fig. 1.62 Block diagram for Example 1.27

1.18 Block Diagram Reduction Technique 99

Solution
1. The summer S2 is shifted before block G 1 . Rule 10 is applied. Take off point
A is shifted to take off point B and vice-versa. Rules 8 and 9 are respectively
applied. The modified block diagram is shown in Fig. 1.62c.
2. From Fig. 1.62c, for the negative feedback indicated,

C(s) G3
=
R1 (1 + G 3 H1 H2 )

The blocks G 1 and G 2 are in cascade and can be combined as G 1 G 2 using Rule 1.
The block H2 is shifted to C(s) using Rule 8. The reduced block diagram is shown
in Fig. 1.62d.
3. The blocks H3 and 1/G 1 H2 are in parallel and their combined block is

1 (1 + G 1 H2 H3 )
H3 + =
G 1 H2 G 1 H2

The cascaded blocks are combined as,

C(s) G1G2G3
=
R2 (1 + G 3 H1 H2 )

The modified block is shown in Fig. 1.62e.

4. The forward path T.F. of Fig. 1.62e is,

G1G2G3
G=
(1 + G 3 H1 H2 )

The feedback loop T.F. is,

(1 + G 1 H2 H3 )H1
H=
G1

Using Rule 4 we get,

G1 G2 G3
C(s) G (1+G 3 H1 H2 )
= = G 1 G 2 G 3 (1+G 1 H2 H3 )
R(s) 1 + GH 1+ (1+G 3 H1 H2 ) G1
H1

C(s) G1G2G3
=
R(s) (1 + G 3 H1 H2 ) + G 2 G 3 H1 (1 + G 1 H2 H3 )
100 1 Control Systems Modelling and Their Representation

(a)
G3
+ C(s )
R(s) +
G1 G2
+
−

G1

(b)
G3

S1 +
R(s) + +
G1 G2 C(s)
− S2

H1

(c)

G3
+ C(s)
R(s) + +
G1 G2
S2 −
S1

Parallel G 1 H1

Feedback

(d)

R(s) G2 C(s)
(G1G3) 1G1 G2 H1

Fig. 1.63 Block diagram for Example 1.28

Example 1.28
Consider the block diagram shown in Fig. 1.63a. Determine the overall T.F. using
block diagram reduction technique.

Solution
1. In Fig. 1.63a the summer S1 is shifted after the block G 1 using Rule 11. Now the
block H1 is multiplied by G 1 . Further the summer S1 and S2 are interchanged
using Rule 6. The reduced block diagram is redrawn as shown in Fig. 1.63c.
1.18 Block Diagram Reduction Technique 101

2. The parallel connection in Fig. 1.63c is reduced, using Rule 2 as G 1 + G 3 . The

feedback connection is reduced, using Rule 4 as G 2 /1 + G 1 G 2 H1 . The reduced
block diagram is shown in Fig. 1.63d.
3. In Fig. 1.63d the two blocks are in cascade. Using Rule 1, the overall T.F. is
obtained as,

C(s) G 2 (G 1 + G 3 )
=
R(s) (1 + G 1 G 2 H1 )

Example 1.29
For the block diagram shown in Fig. 1.64a, determine the overall T.F. by block dia-
gram reduction technique.

Solution
1. Three points are shifted in the given block diagram. The block G 3 is shifted
behind the block G 1 using Rule 8. Block G 3 gain is changed as G 3 /G 1 . Summer
S3 is shifted behind the block G 2 using Rule 11. This makes the H2 block as
G 2 H2 . Further summer S3 and S4 are interchanged using Rule 7. The input to
summer S5 when H2 block shifted is H1 H2 block. The resultant block diagram
is shown in Fig. 1.64c.
2. In Fig. 1.64c, the blocks G 1 and H1 are in feedback mode and is converted to
G 1 /1 + G 1 H1 as per Rule 4. The parallel blocks shown in Fig. 1.64c is combined
as
G3 (G 3 + G 2 G 1 )
+ G2 =
G1 G1

as per Rule 2. The feedback loop 2, is converted as 1/1 + G 2 H2 . The summer

S1 and S5 are combined together with a summer S1 with the input from C(s) as
(1 + H1 H2 ). The complete reduced block diagram is shown in Fig. 1.64d.
3. From Fig. 1.64d three blocks are in cascade. Their equivalence is,

(G 1 G 2 + G 3 )
G=
(1 + G 1 H1 )(1 + G 2 H2 )

The feedback loop gain is H = (1 + H1 H2 ). From Fig. 1.64d using feedback

Rule 4, we get,

C(s) G
=
R(s) 1 + GH
102 1 Control Systems Modelling and Their Representation

(a)
G3
+
R(s) + + + +
G1 G2 C(s)
− − −
+
H2
+
H1

(b)
G3

R(s) + + +
+
G1 G2 C(s)
− S1 − S2 S4
S3
+
S5 H2
+
H1

(c)
G3
Feedback 1 Feedback 2
G1
+
R(s) + + + + + C(s)
G1 G2
− S1 − − S2 + S3
S5 S4
Parallel G2H2
H1

H1H2

(d)
Cascade G

R(s) + G1 (G3+G2G1) 1 C(s)

− 1+G1H1 G1 (1+G1H2)

(1+H1H2)

H
Feedback

Fig. 1.64 Block diagram for Example 1.29

1.18 Block Diagram Reduction Technique 103

C(s) (G 1 G 2 + G 3 )
=
R(s) [(1 + G 1 H1 )(1 + G 2 H2 ) + (G 1 G 2 + G 3 )(1 + H1 H2 )]

1.18.3 Block Diagram Reduction for Multiple Inputs

Systems which are given more than one input simultaneously are called multiple
input system. If the output taken is more than one, such systems are called multiple
output systems. When the input is more than one and output is also more than one such
systems are called multiple input and multiple output systems or simply multi-input,
multi-output (MIMO) systems. The block diagram reduction is done by following
the steps given below:
1. Except one input, assume all the outputs are absent and put them zero.
2. For the particular output and the input considered in step 1, reduce the block
diagram using rules given in Table 1.4.
3. Repeat steps one and two for the remaining inputs but for a particular output.
4. Algebraically add the responses (outputs) obtained in steps 1–3 which is the
output of the system when all the inputs act simultaneously.
5. Steps 1 to 4 are repeated for some other output.
The following examples illustrate the above procedure.

Example 1.30
For the block diagram shown in Fig. 1.65a determine C(s) for the inputs R(s) and
D(s) when they act simultaneously.

Solution

1. R(s) alone is present and D(s) = 0. The block diagram is shown in Fig. 1.65b.
The G 1 and G 2 are in cascade and the whole block diagram is in negative feed-
back mode. Let C1 (s) be the output when R(s) alone is present. From Fig. 1.65b,
the output is easily obtained as,

G1G2
C1 (s) = R(s)
(1 + G 1 G 2 H1 )

2. D(s) alone is present and the corresponding output is C2 (s) · R(s) = 0. For
Fig. 1.65c the forward path T.F is G 2 and the feedback path T.F. is G 1 H1 which
is negative feedback mode. Using Rule 4, we derive the output C2 (s) as,

G2
C2 (s) = D(s)
(1 + G 1 G 2 H1 )
104 1 Control Systems Modelling and Their Representation

Fig. 1.65 Block diagram for D(s)

(a)
Example 1.30 +
R(s) + + C(s
G1 G2
−

H1

(b)
R(s) +
G1 G2 C1(s)
−

H1

(c) D(s)
+ C2(s)
G1 − G2

H1

3. When R(s) and D(s) are simultaneously present, using superposition theorem
we get,

C(s) = C1 (s) + C2 (s)

[R(s)G 1 G 2 + D(s)G 2 ]
C(s) =
[1 + G 1 G 2 H1 ]

Note: The denominator for C1 (s) and C2 (s) is same which is called the
determinant of the block diagram.

Example 1.31
For the block diagram shown in Fig. 1.66a determine C(s) when the input R(s) and
disturbance D(s) are simultaneously present.

Solution

1. D(s) = 0. The output C1 (s) due to R(s) is shown in Fig. 1.66b. The take off
point of block H1 is shifted after the block G 3 . Applying Rule 9, H1 block is
modified as H1 /G 3 and is shown in Fig. 1.66c.
1.18 Block Diagram Reduction Technique 105

(a)
H1 D(s)
R(s) + − + C(s)
+ + +
G1 G2 G3
− −

H2

H3

(b)
H1

R(s) + − C1(s)
+ +
G1 G2 G3
− −

H2

H3

(c) H1
G3 Feedback
R(s) + + − +
G1 G2 G3 C1(s)
−
−

H2

H3

(d)
H1
G3
−
R(s) + + G2G3
G1 C1(s)
− R1 (1+G3H2)

Feedback
H3

Fig. 1.66 Block diagram for Example 1.31

2. From Fig. 1.66c the equivalence of feedback blocks is obtained as

G 3 /(1 + G 3 H2 ) and this is in cascade with G 2 . The reduced block diagram
is shown in Fig. 1.66d.
3. The feedback mode between R1 and C1 (s) is reduced as,
106 1 Control Systems Modelling and Their Representation

(e)
R(s) + G1G2G3
1+G3 H2 +G2 H1 C1(s)
−

H3

(f)
H1
D(s)
− +
− + +
G1 G2 G3
S1 − S2 S3
H2

H3

(g)
H1H2

H1
D(s)

+ − + + +
G2 G3 C2(s
− − S2
S1 S3
Feedback
H2

G1H3

(h)
D(s)

+ G2 + + C2(s)
G3
(1+G2 H1 ) S3 −
S2
H2
Cascade

Parallel
(H1H2−G1H3)

(i)
D(s) + G3 C2(s)
+ (1+G3H2)
S3

G2(H1H2−G1H3)
(1+G2 H1)

Fig. 1.66 (continued)

1.18 Block Diagram Reduction Technique 107

G2G3
C1 (s) (1 + G 3 H2 )
=
R1 G2G3 H1
1+
(1 + G 3 H2 ) G 3
G2G3
=
[(1 + G 3 H2 ) + G 2 H1 ]

The reduced block diagram is shown in Fig. 1.66e after cascading it with the G 1
series block.
4. Figure 1.66e is in feedback mode. Using Rule 4, C1 (s) is obtained as,

G1G2G3
R(s)
(1 + G 3 H2 + G 2 H1 )
C1 (s) =  
G 1 G 2 G 3 H3
1+
(1 + G 3 H2 + G 2 H1 )

R(s)G 1 G 2 G 3
C1 (s) =
(1 + G 3 H2 + G 2 H1 + G 1 G 2 G 3 H3 )

5. Assume R(s) = 0 and D(s) alone is present. Now the output is C2 (s). The block
diagram is shown in Fig. 1.66f.
6. Shifting the take off point of the H1 block ahead of the summer S2 , using Rule 12,
the block diagram is represented as shown in Fig. 1.66g.
7. The blocks G 1 H3 and H1 H2 are in parallel and their equivalence is (H1 H2 −
G 1 H3 ). The blocks G 2 and H1 are in negative feedback mode and their equiv-
alence is G 2 /(1 + G 2 H1 ). Incorporating these changes and interchanging sum-
mers S2 and S3 , the block diagram is represented as shown in Fig. 1.66h.
8. The cascade and feedback connections are converted using Rules 1 and 4 respec-
tively. The reduced block diagram is shown in Fig. 1.66h.
9. Figure 1.66i is in positive feedback mode and using Rule 4, the following expres-
sion is obtained for C2 (s).
G3
D(s) (1+G 3 H2 )
C2 (s) = 
G 2 (H1 H2 −G 1 H3 )
1− G3
(1+G 3 H2 ) (1+G 2 H1 )

D(s)G 3 (1 + G 2 H1 )
C(s) =
[1 + G 3 H2 + G 2 H1 + G 1 G 2 G 3 H3 ]

10. C(s) = C1 (s) + C2 (s)

108 1 Control Systems Modelling and Their Representation

[R(s)G 1 G 2 G 3 + D(s)G 3 (1 + G 2 H1 )]
C(s) =
[1 + G 3 H2 + G 2 H1 + G 1 G 2 G 3 H3 ]

Example 1.32
For the block diagram shown in Fig. 1.67a determine the outputs C1 and C2 due to
the inputs R1 and R2 .

Fig. 1.67 Block diagram for (a)

Example 1.32 R1 +
G1
−

H2

H1

R2 −
+ G2

(b)
R1 +
G1 C11
+

H1G2H2

(c) R2 +
−G2H1G1 C12
−

H2

(d) R1 +
−G1H2G2 C21
−

H1

(e) R2 
G2 C22


H1G1H2
1.18 Block Diagram Reduction Technique 109

Solution
1. Let R2 = 0.
Let the output C1 due to R1 be C11 . The block diagram is shown in Fig. 1.67b.
From Fig. 1.67b, the output C11 is obtained using Rule 4 for positive feedback
mode as
R1 G 1
C11 =
(1 − G 1 G 2 H1 H2 )

2. Let R1 be zero.
Let output C1 be C12 . The block diagram for the above condition is shown in
Fig. 1.67c.
Figure 1.67c is in negative feedback mode and using Rule 4 we get

R2 (−G 2 H1 G 1 )
C12 =
(1 − G 1 G 2 H1 H2 )

The output C1 when R1 and R2 are simultaneously applied we get

C1 = C11 + C12

(R1 G 1 − R2 G 1 G 2 H1 )
C1 =
(1 − G 1 G 2 H1 H2 )

3. Let R1 be present and R2 be zero.

Let the output C2 due to R1 be C21 . For this condition the block diagram is shown
in Fig. 1.67d.
Figure 1.67d is in feedback mode and using Rule 4, the following equation for
C21 is obtained.
−R1 G 1 G 2 H2
C21 =
(1 − G 1 G 2 H1 H2 )

4. Finally, R1 is assumed to be zero.

Let the output C2 due to R2 be C22 . The block diagram for this condition is
shown in Fig. 1.67e.
From Fig. 1.67e, using Rule 4 for a positive feedback connection, the following
equation is obtained for C22

R2 G 2
C22 =
(1 − G 1 G 2 H1 H2 )

5. The total output C2 due to R1 and R2 when they are simultaneously applied is
110 1 Control Systems Modelling and Their Representation

C2 = C21 + C22

−R1 G 1 G 2 H2 + G 2 R2
C2 =
(1 − G 1 G 2 H1 H2 )

Note: The denominators in all the output equations are the same which
is the determinant of the block diagram.

1.18.4 Signal Flow Graph

Signal flow graph like block diagram is a graphical model representation of feedback
control system. It is a pictorial representation of simultaneous equations describing
the system. It has a distinct advantage over block diagram in that it is easier to
draw and manipulate. It primarily consists of nodes and branches to represent these
simultaneous equations. While nodes represent the variables, the branch which is
connected between the nodes represents the gain or transmission function.
Representation of basic signal flow graph is shown in Fig. 1.68. x1 and x2 are the
variables and the line drawn between x1 and x2 is the branch. G is the gain or the
transmission function. The arrow indicates the direction of signal flow.

1.18.5 Signal Flow Graph Algebra

1. The Addition Rule

Like the rules used in block diagram reduction we have few rules in signal flow graph
(SFG) manipulation also. Consider Fig. 1.69, where the variables x1 , x2 , x3 and x4

Fig. 1.68 Basic signal flow x1 G x2

graph
node branch node

Fig. 1.69 Illustration of x1

G1
addition rule of signal flow
graph x2 G2
G3
x3 G4

x4
1.18 Block Diagram Reduction Technique 111

Fig. 1.70 Illustration of x2

transmission rule of SFG G1
G2 x3

x1 −G3
x4
G4
x5

are directed towards the nodes x5 with their respective branch gain. For Fig. 1.69, the
following equation is written:

G 1 x1 + G 2 x2 + G 3 x3 + G 4 x4 = x5 (1.115)

Equation (1.115) illustrates the addition rule of SFG. According to the addition rule,
the value of the variable represented by the node is equal to the sum of all signals
entering that node.
2. The Transmission Rule
According to the transmission rule, the value of the variable designated by a node is
transmitted on every branch leaving that node. In Fig. 1.70, x1 is the designated node
and it transmits signals to other nodes x2 , x3 , x4 and x5 . For Fig. 1.70, the following
equations are written:

x2 = G 1 x1
x3 = G 2 x1
x4 = −G 3 x1
x5 = G 4 x1 (1.116)

3. Multiplication Rule Consider Fig. 1.71 in which the nodes x1 to x5 are con-
nected in cascade. The cascaded connection of these four branches with branch gain
G 1 , G 2 , G 3 and G 4 can be replaced by a single branch that has the branch gain which
is the product of the old one.

G1 G2 G3 G4 G1G2G3G4
x1 x2 x3 x4 x5 x1 x5

Fig. 1.71 Illustration of multiplication rule of SFG

112 1 Control Systems Modelling and Their Representation

−H2

−H4 −H3

G1 G2 G3 G4 G5 G6
x1 x2 x3 x4 x5 x6 x7

−H1
G6

Fig. 1.72 Illustration of terminologies of SFG

1.18.6 Definitions of SFG Terminologies

The following terminologies which are commonly used in SFG are defined as follows:
1. Node: A node represents the variable and are represented by a dot 0. x1 , x2 , x3 ,
x4 , x5 , x6 and x7 in Fig. 1.72 are the nodes.
2. Source: A source or input node has only outgoing branches. x1 node is the
source node.
3. Sink: A sink or output node has only incoming branches. x7 in Fig. 1.72 is a
sink node.
4. Branch: Branch represents the transmission of signal from one node to the
other node and is represented by a line with an arrow and the transmittance.
The arrow represents the direction of transmission between nodes. In Fig. 1.72,
x1 − x2 is a branch with a branch gain G 1 . Similarly x2 − x3 is a branch with a
branch gain G 2 and so on.
5. Mixed nodes: Nodes which do not belong to either the input node or the output
node are called mixed nodes.
6. Path: A path is a continuous unidirectional succession of branches along which
no node is passed more than once. In Fig. 1.72, x1 to x2 is a path, x1 − x2 − x3 −
x4 − x5 − x6 − x7 is a path, and x5 − x6 − x5 is a path. There are several such
paths in Fig. 1.72.
7. Forward path: A forward path is a path from the input node to the output
node. In Fig. 1.72, x1 − x2 − x3 − x4 − x5 − x6 − x7 is a forward path. Similarly
x1 − x2 − x3 − x5 − x6 − x7 is a forward path.
8. Path gain: Path gain is the product of the gains of all the branches encountered
in a path. In Fig. 1.72 for the path x1 − x2 − x3 the path gain is G 1 G 2 G 3 .
9. Loop: Loop is a closed path which originates from a node and terminates at
the same node without touching a node more than once. In Fig. 1.72, x4 − x5 −
x6 − x4 is a loop. Similarly x5 − x6 − x5 is a loop.
10. Loop gain: Loop gain is the product gain of the branches forming a loop. In
Fig. 1.72, for the loop x4 − x5 − x6 − x4 , the loop gain is −G 4 G 5 H2 . For the
loop x5 − x6 − x5 , the loop gain is −G 5 H1 .
1.18 Block Diagram Reduction Technique 113

11. Self loop: A self loop is a feedback loop consisting of a single branch. It starts
and terminates at the same node. In Fig. 1.72, x2 − x2 is a self loop with a loop
gain −H4 . Similarly, x3 − x3 is a self loop with a loop gain −H3 .
12. Non-touching loops: Loops which do not have a common node are said to be
non-touching. In Fig. 1.72 there are four feedback loops

L 1 = x4 − x5 − x6 − x4 (−G 4 G 5 H2 loop gain)

L 2 = x5 − x6 − x5 (−G 5 H1 loop gain)
L 3 = x3 − x3 (−H3 loop gain)
L 4 = x2 − x2 (−H4 loop gain)

L 1 and L 4 are non-touching loops L 4 and L 2 are non-touching and so on. When
two loops are not touching, they are called two, non-touching loops. In Fig. 1.72,
L 1 L 3 , L 1 L 4 , L 2 L 3 , L 2 L 4 and L 3 L 4 are two non-touching loops. The loops
L 1 L 3 L 4 and L 2 L 3 L 4 are called three non-touching loops.

1.18.6.1 Application of Mason’s Gain Formula for SFG

To find the overall gain between input node and output node it is possible to write
down the input-output relationship from SFG and obtain the same. However, the
procedure is somewhat tedious and cumbersome. But this can be accomplished using
Mason’s gain formula which is explained below. Using this formula, sometimes, the
results can be obtained just by inspection of SFG. Because of the simplicity of the
formula, SFG has been very extensively used for the system analysis. The Mason’s
general gain formula for any signal flow graph is


n
Tm m
m=1
A= (1.117)

where
A is the total gain between the input node and output node.
Tm is the gain of the mth forward path.

  
=1− Pn1 + Pn2 − Pn3 + · · ·
n n n

where
114 1 Control Systems Modelling and Their Representation

Pnr = gain product of ‘n’ possible combinations of ‘r ’ non-touching loops (or)

 = 1 − (sum of individual loop gains)
+(sum of gain product of all possible combinations of two non-touching loops)
−(sum of gain product of all possible combinations of three non-touching loops) + · · ·
= Determinant of SFG.
m = The value of  for that part of all graph with the feedback loops touching
the m th forward path being broken = co-factor of m th forward path.

For simplicity, the individual loop gains in Eq. (1.117) are denoted by L 1 , L 2 , L 3
etc. The forward path gains are denoted by T1 , T2 , T3 etc. The following methods
illustrate the application of Mason’s gain formula to solve SFG.

Example 1.33
For the signal flow graph shown in Fig. 1.73a using Mason’s gain formula, determine
C(s)/R(s).

Solution
The forward path of Fig. 1.73a is shown in Fig. 1.73b. There is only one forward path.
Equation (1.117) for this problem becomes

C(s) T1 1
=
R(s) 

From Fig. 1.73b,

T1 = G

(a) 1 x1 G
C(s)
R(s)

−H
(b) 1 G (c) x1 G
R(s) C(s) C(s)
x1

−H
(d) 1 G
C(s)
R(s)

−H

Fig. 1.73 a SFG for Example 1.33. b Forward path; c Feedback loop and d Feedback loop touching
the forward path when removed it is broken
1.18 Block Diagram Reduction Technique 115

There is only one feedback loop namely x1 − C(s) − x1 as shown in Fig. 1.73c.
Feedback loop gain L 1 is
L 1 = −G H

To calculate 1 , the co-factor of the graph, the feedback loop touching the forward
path T1 at two points when broken is shown in Fig. 1.73d. This means there is no
feedback when 1 is calculated for the forward path T1 .
From Eq. (1.117),

 = 1 − L 1 = 1 − (sum of individual loop gain)

= 1 + GH
1 = 1

C(s) G
=
R(s) 1 + GH

Example 1.34
C(s)
The SFG of a system is shown in Fig. 1.74a. Determine .
R(s)

(a) 1 1
1 x1 s x2 x3 s x4
R(s)
6
1 −2 −3

C(s) −4

1
(b) 1 x1 1 (c) x2 6 x3 s x4
R(s) C(s)
T1=1
L1= −24
s
−4
1 1
(d) x3 s x4 (e) x1 s x2 (f)
1 x1 s1 x2 x3 s1 x4
R(s) 6
−3 −2 1 −2 −3
L2 = −3 L3 = −2
s −4
s C(s)

Fig. 1.74 a SFG for Example 1.34. b Forward path; c Feedback loop 1; d Feedback loop 2;
e Feedback loop 3; and f feedback touching the forward path when broken
116 1 Control Systems Modelling and Their Representation

Solution

1. There is only one forward path for the SFG which is shown in Fig. 1.74b. The
forward path gain T1 = 1.
2. There are three feedback loops as shown in Fig. 1.79c–e. The feedback loop gains
are L 1 = −24/s and L 2 = −3/s and L 3 = −2 s
. However, L 1 and L 2 feedback
loops are touching each other at x3 and x4 nodes. The feedback loops L 2 and L 3
are not touching each other.
3. The feedback loops have connections at node x2 and when broken at x1 , where
it is touching the forward path they are shown in Fig. 1.74f.
4. The overall T.F. for a single forward path of SFG is

T1 1
A=

 = 1 − (L 1 + L 2 + L 3 ) + L 2 L 3

24 3 2 6
= 1+ + + + 2
s s s s
(s 2 + 29s + 6)
=
s2
5. 1 is calculated from Fig. 1.74f.  for T1 when the feedback loops touching the
forward path T1 gives 1 .

1 = 1 − (L 1 + L 2 )
24 3
= 1+ +
s s
(s + 27)
=
s
6. The overall gain is given by

C(s) T1 1
=
R(s) 

(s + 27) s2
=
s (s 2 + 29s + 6)

C(s) s(s + 27)

= 2
R(s) (s + 29s + 6)
1.18 Block Diagram Reduction Technique 117

Example 1.35
Find C(s)/R(s) by using Manson’s gain formula for the SFG shown in Fig. 1.75a.

(Anna University, May, 2009)

Solution

1. There are two forward paths T1 and T2 as shown in Fig. 1.75b, c respectively.
The overall gain is therefore given by

(a)
G5

x1 G2 x2 x3 G4 x4
R(s) G1 G3 1 C(s)

−H1

−H2

(b) 1 G1 G2 G3 G4 1
R(s) x1 x2 x3 x4 x5 C(
T1 = G1 G2 G3 G4

(c)
G5

R(s) 1 x1 x2 x3 x4 G4 x5 1 C(s)
T2 = G5 G4

(d) x1 G1 x2 G2 x3 G3 x4 G4 x5

−H2

L1 = −G1 G2 G3 G4 H2
(e) x2 x3
G2
L2= −G2H1
−H1

Fig. 1.75 a SFG for Example 1.35. b Forward path T1 ; c Forward path T2 ; d Feedback loop L 1 and
e Feedback loop L 2 . f Feedback loop L 3 ; g Two non-touching loops; h Feedback loops touching
the forward path T1 , when broken; and i Feedback loops touching the forward path T2 , when broken
118 1 Control Systems Modelling and Their Representation

(f)
G5

x1 x2 x3 x4 G4 x5

L3 G5 G4 H2
H2

(g)
G5

x1 x2 x3 x4 G4 x5

H2

L2 L3 G2 H1 G5 H4 H2
H2
(h)
G5

G1 G2 G3 G4 1
R(s) c(g)
x1 x2 x3 x4 x4

H1
L1 0

H2

(i)
G5

G2 G4 1
R(s) c(s)
x1 x2 x3 x4 x5

H1

L2 G2H1

H2

Fig. 1.75 (continued)

1.18 Block Diagram Reduction Technique 119

C(s) T1 1 + T2 2
=
R(s) 

2. From Fig. 1.75b,

T1 = G 1 G 2 G 3 G 4

From Fig. 1.75c,

T2 = G 5 G 4

3. There are three feedback loops for the given SFG. They are shown in Fig. 1.75d–f.
The feedback loop gains are

L 1 = −G 1 G 2 G 3 G 4 H2
L 2 = −G 2 H1
L 3 = −G 5 G 4 H2

While loops 1 and two are touching each other, loops 2 and three are non-
touching. Hence, the gain product of these non-touching loops are G 2 H1 G 5 G 4 H2

 = 1 − (L 1 + L 2 + L 3 ) + (L 2 L 3 )
= 1 + G 1 G 2 G 3 G 4 H2 + G 2 H1 + G 5 G 4 H2 + G 2 H1 G 5 G 4 H2

4. Fig. 1.75h represents the feedback loop touching the forward path T1 when it is
broken. From this figure there is no feedback loop. Hence, 1 = 1.
5. Fig. 1.75i represents the feedback loop touching the forward path T2 when it is
broken. There is only one feedback loop and its loop gain is L 2 = −G 2 H1 .

2 = 1 − (L 2 )
= 1 + G 2 H1

6.
C(s) T1 1 + T2 2
=
R(s) 

C(s) G 1 G 2 G 3 G 4 + G 5 G 4 (1 + G 2 H1 )
=
R(s) [1 + G 1 G 2 G 3 G 4 H2 + G 2 H1 + G 5 G 4 H2 + G 2 H1 G 5 G 4 H2 ]
120 1 Control Systems Modelling and Their Representation

Note: Just by inspection of SFG, it is possible to identity the forward

paths and feedback loops. Hence, the detailed representation of these
figures are avoided in future explanation.

Example 1.36
For the system represented by the following equations find the T.F. X
U
(s) by signal
flow graph technique:

x = x1 + β3 u
ẋ1 = −a1 x1 + x2 + β2 u
ẋ2 = −a2 x1 + β1 u

(Anna University, December, 2009)

Solution

x = x1 + β3 u or X (s) = X 1 (s) + β3 U (s) (a)

ẋ1 = −a1 x1 + x2 + B2 u

Taking Laplace transform on both sides we get

(s + a1 )X 1 (s) = X 2 (s) + β2 U (s)

Solving for X 1 (s) we get

1 β2
X 1 (s) = X 2 (s) + U (s) (b)
(s + a1 ) s + a1

ẋ2 (t) = −a2 x1 + B1 u

Taking Laplace transform on both sides we get

a2 β1
X 2 (s) = − X 1 (s) + U (s) (c)
s s
Equations (a), (b) and (c) are represented in Fig. 1.76. From Fig. 1.76, the following
forward paths are identified:
1.18 Block Diagram Reduction Technique 121

X1(s)
U(s) X2(s)
2
−a2 X(s
(s+a1) s

1
(s+a1)
1
s

Fig. 1.76 SFG for Example 1.36

Forward path Forward path gain

1. U−X T1 = β3
β2
2. U − X1 − X T2 =
s + a1
β1
3 U − X2 − X1 − X T3 =
s(s + a1 )

The following feedback loops are identified:

Feedback loop Loop gain

−a2
1 X1 − X2 − X1 L1 =
s(s + a1 )

 = 1 − (L 1 )
a2
= 1+
s(s + a1 )
(s 2 + a1 s + a2 )
=
s(s + a1 )

For the forward path one, the feedback loop X 1 − X 2 − X 1 is not touching the
forward path T1 when broken.
122 1 Control Systems Modelling and Their Representation

x2 G4 x4
G1 G7
x1 x6 1
G3 G6
R(s) C(s
G2 G8
x3 G5 x5

−H2

Fig. 1.77 SFG for Example 1.37

1 = 1 − L 1
a2
= 1+
s(s + a1 )
(s 2 + a1 s + a2 )
=
s(s + a1 )

For the forward paths T2 and T3 , the feedback loops are touching these forward paths
and when broken from the forward path, there is no feedback loop.
Hence, 2 = 1 and 3 = 1

X (s)
= T1 1 + T2 2 + T3 3
U (s)
β3 (s 2 +a1 s+a2 ) β2 β1
s(s+a1 )
+ (s+a 1)
+ s(s+a1 )
= (s 2 +a1 s+a2 )
s(s+a1 )

X β3 (s 2 + a1 s + a2 ) + β1 + β2 s
(s) =
U (s 2 + a1 s + a2 )

Example 1.37
C(s)
For the SFG shown in Fig. 1.77 determine using Mason’s gain formula.
R(s)
1.18 Block Diagram Reduction Technique 123

Solution

Forward path Forward path gain

1. R(s) − x1 − x2 − x4 − x6 − C(s) T1 = G 1 G 4 G 7
2. R(s) − x1 − x2 − x4 − x5 − x6 − C(s) T2 = G 1 G 4 G 6 G 8
3. R(s) − x3 − x2 − x4 − x6 − C(s) T3 = G 2 G 3 G 4 G 7
4. R(s) − x1 − x3 − x2 − x4 − x5 − x6 − C(s) T4 = G 2 G 3 G 4 G 6 G 8
5 R(s) − x1 − x3 − x5 − x6 − C(s) T5 = G 2 G 5 G 8
Feedback loop Feedback loop gain
1. x2 − x4 − x2 L 1 = −G 4 H1
2 x5 − x5 L 2 = −H2

C(s) T1 1 + T2 2 + T3 3 + T4 4 + T5 5
=
R(s) 
 = 1 − (L 1 + L 2 ) + L 1 L 2
= 1 + G 4 H1 + H2 + G 4 H1 H2

For forward path T1 , L 1 , is broken and L 2 alone is not touching

1 = 1 + H2 .

For forward path T2 , both L 1 and L 2 are touching and hence when removed from the
forward path, they are broken. Hence,

2 = 1.

For forward path T3 , L 1 is touching L 2 is not touching. When the touching portion
of L 1 is removed, it is broken. L 2 is the only non-touching loop. Hence,

3 = 1 + H2 .

For the forward path T4 , both L 1 and L 2 are touching the forward path and when they
are removed from the forward path, they are broken. Hence, there are no feedback
loops.
4 = 1

For the forward path T5 , loop L 2 is touching and loop L 1 is not touching. Thus, L 2
is broken when removed.
5 = 1 + G 4 H1

C(s)
= G 1 G 4 G 7 (1 + H2 ) + G 1 G 4 G 6 G 8 + G 2 G 3 G 4 G 7 (1 + H2 )
R(s)
+G 2 G 3 G 4 G 6 G 8 + G 2 G 5 G 8 (1 + G 4 H1 )
124 1 Control Systems Modelling and Their Representation

Example 1.38
For the SFG shown in Fig. 1.78a determine the outputs due to the inputs R1 and R2 .

Solution The multi-input-multi-output problems are solved in similar way as we

followed for block diagram reduction. When the input R1 is applied the input R2 = 0
and the output is C11 . This is shown in Fig. 1.78b.
For Fig. 1.78b,

T1 = R1 − x1 − x2 − C11
= G1
T2 = R1 − x1 − x3 − x4 − x2 − C11
= G3G2G4
L 1 = −G 1 H1 ;
L 2 = −G 2 H2 ;
L 3 = −G 3 G 2 G 4 H1
 = 1 − (L 1 + L 2 + L 3 ) + L 1 L 2
= 1 + G 1 H1 + G 2 H2 + G 2 G 3 G 4 H1 + G 1 G 2 H1 H2

For T1 , L 1 and L 3 are broken and L 2 is not broken

1 = 1 + G 2 H2

For T2 , L 1 , L 2 and L 3 are broken 2 = 1

T1 1 + T2 2
C11 =

R1 [G 1 (1 + G 2 H2 ) + G 2 G 3 G 4 ]
∴ C11 =
[1 + G 1 H1 + G 2 H2 + G 1 G 2 H1 H2 ]

When R2 is applied, R1 = 0 and the output at C1 is C12 . For Fig. 1.78c we get

T1 = R2 − x3 − x4 − x2 − C12
= G2G4

Feedback loops L 1 and L 2 and L 3 are in touch with T1 and when removed they are
broken. Hence, 1 = 1
1.18 Block Diagram Reduction Technique 125

(a) −H1
(b) −H1

R1 1 x1 G1 x2 1 C1 R1 1 x1 G1 x2 1 C11

G3 G4 G3 G4

G2 G2
x3 x4 1 x3 x4
R2 1 C2

−H2 −H2

(c) −H1 (d) −H1

x1 G1 x2 1 C12 R1 1 x1 G1

G3 G4 G3 G4

1 G2 G2
R2 x3 x4 x3 x4 C21

−H2 −H2

(e) −H1

G1

G3 G4

G2
R2 1 1 C22

−H2

Fig. 1.78 a SFG for Example 1.38

126 1 Control Systems Modelling and Their Representation

T1 1 R2
C12 =

R2 G 2 G 4
=
[1 + G 1 H1 + G 2 H2 + G 2 G 3 G 4 H1 + G 1 G 2 H1 H2 ]
C1 = C11 + C12

R1 [G 1 (1 + G 2 H2 ) + G 2 G 3 G 4 ] + R2 G 2 G 4
C1 =
[1 + G 1 H1 + G 2 H2 + G 2 G 3 G 4 H1 + G 1 G 2 H1 H2 ]

Note:  is same for the given SFG.

Now let us calculate C2 due to R1 and R2 . First we assume R1 is applied and R2 = 0.

The output now is C21 . SFG is shown in Fig. 1.78d. There is only on forward path

T1 = R1 − x1 − x3 − x4 − C21
T1 = G 3 G 2

The feedback loops touching T1 when removed are broken. Hence, 1 = 1

R1 T1 1
C21 =

R1 G 2 G 3
=
[1 + G 1 H1 + G 2 H2 + G 2 G 3 G 4 H1 + G 1 G 2 H1 H2 ]

Now R2 is applied and R1 =0. The output is C22 . The SFG is shown in Fig. 1.78e.
From Fig. 1.78e
T1 = G 2

The feedback loop L 1 is not touching the forward path. Hence,

L 1 = −G 1 H1
1 = 1 − (L 1 )
= 1 + G 1 H1
T1 1
C22 = R2

R2 G 2 (1 + G 1 H1 )
C22 =

when R1 and R2 are simultaneously applied,
1.18 Block Diagram Reduction Technique 127

Fig. 1.79 SFG between −H1

output node and non-input
node x1 1 x 2 G1 x3 G2 x4 G3 x5 G4 x6

−H2

C2 = C21 + C22
= R1 G 2 G 3 + R2 G 2 (1 + G 1 H1 )

R1 G 2 G 3 + R2 G 2 (1 + G 1 H1 )
C2 =
[1 + G 1 H1 + G 2 H2 + G 2 G 3 G 4 H1 + G 1 G 2 H1 H2 ]

1.18.7 Application of the Gain Formula Between Output

Nodes and Non-input Nodes

Mason’s gain formula was applied to find the overall gain between the input and
output nodes. However, it is also applicable to find the relation between an output
node variable and a non-input variable.
Consider the SFG shown in Fig. 1.79. The output node is x6 . We find the ratio
x6 /x2 where x2 is a non-input node. x6 /x2 can be written in the following form:
x6 x6 x1
=
x2 x1 x2
x6 1
=
x1 x2 /x1
n
Tma ma
x6 m=1
=
x1 
n
Tmb mb
x2 m=1
=
x1 


n
Tma ma
x6 m=1
= n (1.118)
x2 
Tmb mb
m=1
128 1 Control Systems Modelling and Their Representation

Fig. 1.80 SFG for Example L1

1.39 −H3 −H2 −H1
L3
L2

x1 G1 x2 G2 x3 G 3 x4 G4 x5 G5 x6

G6

Equation (1.118) suggests that there is no necessity to determine , the determinant

of SFG in the above case. The following example illustrates this:

Example 1.39
For the figure shown in Fig. 1.80. Find x6 /x3 .

Solution For Fig. 1.80, between x1 and x6 nodes,

Ta1 = G 1 G 2 G 3 G 4 G 5
Ta2 = G 1 G 2 G 6 G 5
x6 Ta1 a1 + Ta2 a2
=
x1 

There are three feedback loops. They are all touching the forward path Ta1 and
when they are all removed they are all broken. Hence, a1 = 1.
Now consider the forward path Ta2 , the feedback loop with gain −H2 is not
torching Ta2 and it is not broken when removed from Ta2 . However, the other two
feedback loops are in touch with Ta2 and when removed they are broken. Therefore,
a2 = (1 + H2 )

x6 Ta1 + Ta2 (1 + H2 )
=
x1 
G 1 G 2 G 5 [G 3 G 4 + G 6 (1 + H2 )]
=

Now consider the forward path x1 − x2 − x3 . This is the only forward path between
the nodes x1 and x3 . The forward path gain is

Tb1 = G 1 G 2
x3 Tb1 b1
=
x1 

The feedback loops are

1.18 Block Diagram Reduction Technique 129

L 1 = −G 2 G 3 H3
L 2 = −H2
L 3 = −G 5 H1

The loops L 2 and L 3 are not touching the forward path Tb1 . Further they are touching
themselves and form two non-touching loops. However, loop L 1 is in touch with Tb1
and when removed it is broken.

b1 = 1 − (L 2 + L 3 ) + L 2 L 3
= 1 + H2 + G 5 H1 + G 5 H1 H2
x3 Tb1 b1
=
x1 
G 1 G 2 [1 + H2 + G 5 H1 + G 5 H1 H2 ]
=

x6 x6 1
=
x3 x1 x3 /x1
G 1 G 2 G 5 [G 3 G 4 + G 6 (1 + H2 )]
=
G 1 G 2 [1 + H2 + G 5 H1 + G 5 H1 H2 ]

x6 G 5 [G 3 G 4 + G 6 (1 + H2 )]
=
x3 [1 + H2 + G 5 H1 + G 5 H1 H2 ]

1.18.8 Applications of the Gain Formula to Block Diagrams

Mason’s gain formula can be directly applied to block diagram and the input–output
gain can be easily determined. However, it becomes easier if an equivalent SFG is
drawn for the block diagram before applying the gain formula.
1. The input and the output in the block diagram are identified and are designated
with input node and output node respectively in SFG.
2. All take off points in the block diagram are designated with nodes in SFG.
3. The output of the summers in the block diagram are identified and they are also
designed with nodes in SFG.
Consider the block diagram shown in Fig. 1.81a. The input and output are R(s)
and C(s) respectively and they are represented as input nodes and output nodes
in SFG as shown in Fig. 1.81b. There is a take off point after the block G and a
node should be created to represent this. But this point is already represented by the
output node and hence it coincides with output node. After the summer, a variable e
is created and a node is identified for e. The following equations hold good both for
block diagram as well as SFG.
130 1 Control Systems Modelling and Their Representation

(a) (b)

R(s) + e C(s) R(s) e G

G C(s)
− 1

H
−H

Fig. 1.81 a Block diagram and b Equivalent SFG

e = R(s) − H C(s)
C(s) = eG

The following examples illustrate the above procedure.

Example 1.40
For the block diagram shown in Fig. 1.82a draw the SFG and using Mason’s gain
formula determine C/R.

(Anna University, May, 2009)

Solution
In Fig. 1.82a the output of the summers and the take off points are identified and
marked as x1 , x2 and x3 . The SFG with these nodes and the input and output nodes
is drawn and shown in Fig. 1.82b. The following forward paths and feedback loops
are identified with their gains.

Forward path Forward path gain

1. R − x1 − x2 − C T1 = −G 1 G 2
2. R − x1 − x2 − C T2 = G 1 G 3
3. R − x3 − x1 − x2 − C T3 = −G 4 H2 G 1 G 2
4. R − x3 − x1 − x2 − C T4 = G 4 H2 G 1 G 3
Feedback loop Loop gain
1. x1 − x2 − C − x3 − x1 L 1 = G 1 G 2 H1 H2
2 x1 − x2 − C − x3 − x1 L 2 = −G 1 G 3 H1 H2

The two feedback loops are touching with each other and therefore there are no
two non-touching loops

 = 1 − (L 1 + L 2 )
= 1 + G 1 H1 H2 (G 3 − G 2 )

These feedback loops L 1 and L 2 are touching all the four forward paths T1 , T2 , T3 and
T4 . When they are removed from the forward path, they are broken. Thus, 1 = 1;
2 = 1; 3 = 1 and 4 = 1
1.18 Block Diagram Reduction Technique 131

(a)
G2

R + + −
x1 G1 G3 C
x2
−

x3 +
H2 H1
−

G4

(b)
−G4

−H2

x3 C
R 1 x1 G1 x2
H1

−G2

G3

Fig. 1.82 a Block diagram for Example 1.40. b SFC of Fig. 1.82a

C T1 1 + T2 2 + T3 3 + T4 4
=
R 
−G 1 G 2 + G 1 G 3 − G 4 H2 G 1 G 2 + G 4 H2 G 1 G 3
=
1 + G 1 H1 H2 (G 3 − G 2 )

C G 1 (G 3 − G 2 )(1 + G 4 H2 )
=
R [1 + G 1 H1 H2 (G 3 − G 2 )]

Example 1.41
Reduce the block diagram using reduction rules and obtain C(s)/R(s) for Fig. 1.83a.
Verify by SFG.

(Anna University, May, 2010)

132 1 Control Systems Modelling and Their Representation

(a)
R(s) + + + C(s)
G1 G2 G3
− − −

H3 H2 H1

(b)
R(s) + + C(s)
G1 G2 G3
x1 − x2 − x3

H3 H2 H1
x5 x4
(c)
R(s) C(s)
1 x1 G1 x2 G2 x3 G3
−1
x4
−H3 H2 H1
x5
−1

Fig. 1.83 Block diagram for Example 1.41

Solution The block diagram is solved in Example 1.27. To solve the problem by
SFG, the output of the summers and the take off points are assigned variables as
given below in Fig. 1.83b.
The SFG, with the variables marked in Fig. 1.83b is shown below in Fig. 1.83c.
The forward path T1 = R(s) − x1 − x2 − x3 − C(s) = G 1 G 2 G 3

Feedback loop Loop gain

1. x1 − x2 − x3 − C − x4 − x5 − x1 L 1 = −G 1 G 2 G 3 H1 H2 H3
2. x2 − x3 − C − x4 − x2 L 2 = −G 2 G 3 H1
3 x3 − C − x4 − x5 − x3 L 3 = −G 3 H1 H2

All the feedback loops are touching with each other. Hence, there is no two non-
touching loops

 = 1 − (L 1 + L 2 + L 3 )

 = 1 + G 1 G 2 G 3 H1 H2 H3 + G 2 G 3 H1 + G 3 H1 H2
1.18 Block Diagram Reduction Technique 133

All the feedback loops are touching the forward path T1 . When they are removed
from the forward path all the three feedback paths are broken. Hence, 1 = 1

C(s) T1 1
=
R(s) 

C(s) G1G2G3
=
R(s) [1 + G 1 G 2 G 3 H1 H2 H3 + G 2 G 3 H1 + G 3 H1 H2 ]

Example 1.42
Consider the block diagram shown in Fig. 1.84a. Find C/R using block diagram
reduction method. Verify the result using SFG.

Solution Block Diagram Reduction Method

The given block diagram is redrawn and shown in Fig. 1.84b. The outputs of the
summers and the take off points are identifies as x1 , x2 , x3 and x4 . The take off
point x4 is shifted ahead of the block G 2 . The modified block diagram is shown in
Fig. 1.84c (Rule 8).
The take off point at x3 is shifted ahead of the summer S2 . Rule 12 is followed
(Fig. 1.84d).
Now the summer S2 is shifted behind the block G 2 using Rule 11. This is shown in
Fig. 1.84e.
Interchanging the summers S2 and S3 , the block diagram is reduced as shown in
Fig. 1.84f.
By shifting the take off point ahead of block H1 and combining the parallel connection
the block diagram is reduced as shown in Fig. 1.84g.
Converting the feedback connection and cascading it with (G 2 + G 3 ) block, the
block diagram is reduced as shown in Fig. 1.84h.
The above block diagram is redrawn as shown in Fig. 1.84i.
The feedback is reduced and cascaded. This is shown in Fig. 1.84j.
The above block diagram is reduced using feedback rule.

C G 1 (G 2 + G 3 )
=
R (1 + G 2 H1 + G 1 G 2 H2 − G 1 G 2 G 3 H1 H2 )

Signal Flow Graph Method

For the block diagram shown in Fig. 1.84b, the SFG is drawn as shown in Fig. 1.84k.
From Fig. 1.84k,
134 1 Control Systems Modelling and Their Representation

Fig. 1.84 Block diagram for (a)

Example 1.42 G3

+
R + + + C
G1 G2
− −

H2

H1

(b)
G3

+
R + + + C
G1 G2
x1 x2 x3 x4
− s1 −s s3
2

H2

H1

(c)
G3

+
R + + + C
G1 G2
x1 x2 x3 x4
− s1 −s s3
2

G2H2

H1

(d)
G3

+
R + + + C
G1 G2
x1 x2 − x3 x4
− s1 s2 s3

+
G2H2 H1
s4

(e)
G3

+
R + + + C
G1 G2
x1 x2 s2
− s1 − s3

G2

s4 +
G2H2 H1
−
1.18 Block Diagram Reduction Technique 135

Fig. 1.84 (continued) (f)

G3

+
R + + + C
G1 G2
x1 x2
− s1 s3 − s2
Parallel
G2

s4 +
G2H2 H1
−

(g)
R + + C
G1 (G2+G3)
− −

G2H1
+
G2H2 H1
− Feedback

(h) C
R + 1
G1 (G2+G3)
(1+G2H1)
−

+
G2H2 H1
−

(i) Feedback
R + (G2+G3) C
G1
(1+G2H1)
+ −

G2H2

G2H1H2

(j)
R  G1 (G2G3) C
(1G1G2H2) (1G2H1)


G2H1H2

(k)
−H2

R C
1 x1 G1 x2 1 x3 G2 x4 1

−H1

G3
136 1 Control Systems Modelling and Their Representation

T1 = R − x1 − x2 − x3 − x4 − C
= G1G2
T2 = R − x1 − x2 − C
= G1G3
L 1 = x1 − x2 − x3 − x4 − x1
= −G 1 G 2 H2
L 2 = x3 − x4 − C − x3
= −G 2 H1
L 3 = x1 − x2 − C − x3 − x4 − x1
= G 1 G 3 H1 G 2 H2

All the three loops are touching with each other. Hence, there is no two non-touching
loops. Further, all the three feedback loops are touching both the forward paths T1 and
T2 and the portion touching the forward paths when removed, the loops are broken.
Hence, 1 = 1 and 2 = 1

 = 1 − (L 1 + L 2 + L 3 )
= 1 + G 1 G 2 H2 + G 2 H1 − G 1 G 2 G 3 H1 H2
C T1 1 + T2 2
=
R 
G1G2 + G1G3
=


C G 1 (G 2 + G 3 )
=
R (1 + G 1 G 2 H2 + G 2 H1 − G 1 G 2 G 3 H1 H2 )

The same result is obtained using block diagram reduction method with much more
tediousness. However, SFG method is much simpler.

1.18.9 Signal Flow Graph for the Electrical Network

The signal flow graph can be easily extended to determine the transfer function of
electrical network and other system that are commonly used. In electrical networks,
the mesh and nodal equations are written using Kirchhoff’s laws. The variables
include the mesh currents and nodal voltages and these variables form the nodes in
SFG in addition to nodes due to input and output variables. By applying Mason’s
1.18 Block Diagram Reduction Technique 137

(a) v(t) i2(t)

R1 i1(t) R2

vi(t) C L v0(

(i1−i2)
i1(t) i2(t)

(b) 1 1 1
R1 I1 Cs V (R2+Ls) I2 Ls
Vi(s) Vo(s)

−1 −1
R1 Cs

Fig. 1.85 Electrical network for Example 1.43

gain formula, the T.F. of the network is obtained. The following examples illustrate
the above procedure.

Example 1.43
For the electrical network shown in Fig. 1.85a determine the T.F.

(Anna University, May, 2010)

Solution
The following transform equations for Fig. 1.85a are written

Vi (s) = R1 I1 (s) + V (s)

Vi (s) 1
I1 (s) = − V (s) (a)
R1 R1

1
V (s) = [I1 (s) − I2 (s)] (b)
Cs
Also,

V (s) = R2 I2 (s) + Ls I2 (s)

1
I2 (s) = V (s) (c)
(R2 + Ls)
138 1 Control Systems Modelling and Their Representation

V0 (s) = Ls I2 (s) (d)

In the above equations, Vi (s) is assigned the input mode and V0 (s), the output node.
I1 (s), V (s) and I2 (s) are assigned intermittent nodes. Equations (a), (b), (c) and
(d) are represented in SFG as shown in Fig. 1.85b. The following forward paths and
feedback loops are identified in Fig. 1.85b.

L
Forward path, T1 =
R1 C(R2 + Ls)
1
Feedback loop, L 1 = I1 − V − I1 = −
R1 Cs
−1
Feedback loop, L 2 = V − I2 − V =
Cs(R2 + Ls)

L 1 and L 2 are touching with each other and therefore there is no two non-touching
loops.

 = 1 − (L 1 + L 2 )
1 1
= 1+ +
R1 Cs Cs(R2 + Ls)
L R1 Cs 2 + (R1 R2 C + L)s + (R1 + R2 )
=
R1 C(R2 + Ls)s

Since both the feedback loops L 1 and L 2 are touching the forward path T1 and when
their touching part is removed, both the feedback loops are broken. Hence, for the
forward path T1 , there is no feedback loops and hence 1 = 1.

V0 T1 1 L
= =
Vi  R1 C(R2 + Ls)

V0 Ls
(s) =
Vi L R1 Cs 2 + (R1 R2 C + L)s + (R1 + R2 )

Example 1.44
For the electrical network shown in Fig. 1.86a draw the SFG and using Mason’s gain
formula determine V4 /V1 .

Solution For Fig. 1.86a, the following equations are written:

V1 V2
I1 = − (a)
R1 R1
1.18 Block Diagram Reduction Technique 139

(a) V2 V3
I1 I2 I3
R1 R2 R3
I1−I2 I2−I3
V1 C1 C2 C3 V4

(b) 1 1 1 1 1
I1 C1s V2 R2 I2 C2s V3 R3 I3 C3s
V1 V4
1 L1 L2 L3 L4 L5
R1
− 1 − 1 − 1 − 1 − 1
R1 C1S R2 C2 S R3

Fig. 1.86 a Electrical network for Example 1.44. b SFG for Fig. 1.86a

I1 I2
V2 = − (b)
C1 s C1 s

V2 V3
I2 = − (c)
R2 R2

I2 I3
V3 = − (d)
C2 s C2 s

V3 V4
I3 = − (e)
R3 R3

I3
V4 = (f)
C3 s

There are seven variables including the input and output variables and hence seven
nodes are created in SFG. The SFG is drawn which satisfies equations (a) to (f)
derived above and shown in Fig. 1.86b.
For Fig. 1.86b there is only one forward path with a gain,

1
T1 =
R1 R2 R3 C 1 C 2 C 3 s 3
140 1 Control Systems Modelling and Their Representation

There are five feedback paths. The feedback loops and the loop gains are given
below:

Feedback loop Loop gain

1
1. I1 − V2 − I1 L1 = −
R1 C 1 s
1
2. V2 − I2 − V2 L2 =−
R2 C 1 s
1
3. I2 − V3 − I2 L3 =−
R2 C 2 s
1
4. V3 − I3 − V3 L4 =−
R3 C 2 s
1
5 I3 − V4 − I3 L5 =−
R3 C 3 s

The following pair of loops form two non-touching loops:

(a) L 1 L 3 (d) L 2 L 4
(b) L 1 L 4 (e) L 2 L 5
(c) L 1 L 5 (f) L 3 L 5
There is only one combination of three non-touching loops namely L 1 L 3 L 5 .

 = 1 − (L 1 + L 2 + L 3 + L 4 + L 5 )
+(L 1 L 3 + L 1 L 4 + L 1 L 5 + L 2 L 4 + L 2 L 5 + L 3 L 5 ) − L 1 L 3 L 5
1 1 1 1 1 1
 = 1+ + + + + +
R1 C 1 s R2 C 1 s R2 C 2 s R3 C 2 s R3 C 3 s R2 C 1 R3 C 3 s 2
1 1 1 1
+ + + +
R2 C 2 R3 C 3 s 2 R1 R2 C 1 C 2 s 2 R1 C 1 R3 C 2 s 2 R1 C 1 R3 C 3 s 2
1 1
+ 2
+
R C R C s R1 R2 R3 C 1 C 2 C 3 s 3
 2 1 3 2
= R1 R2 R3 C1 C2 C3 s 3 + (R2 R3 C2 C3 + R1 R3 C2 C3

+R1 R3 C1 C3 + R1 R2 C1 C2 + R1 R2 C1 C3 )s 2 + (R3 C3

+R2 C3 + R2 C2 + R1 C3 + R1 C2 + R1 C1 )s + 1 R1 R2 R3 C 1 C 2 C 3 s 3

When the feedback loops touching the forward path T1 are removed all the five
feedback loops are broken. Hence, 1 = 1.

V4 T1 1
=
V1 
1.18 Block Diagram Reduction Technique 141

(a)
i

rp

eg eg

ei

rk eo

(b)
ei eg μ i eo
1 μeg 1 rk
(rp+rk)

−1

Fig. 1.87 a Cathode follower for Example 1.45. b SFG for Fig. 1.87a

V4
=1 R1 R2 R3 C1 C2 C3 s 3 + (R2 R3 C2 C3 + R1 R3 C2 C3 + R1 R3 C1 C3
V1
+R1 R2 C1 C2 + R1 R2 C1 C3 )s 2 + (R3 C3 + R2 C3 + R2 C2 + R1 C3

+R1 C2 + R1 C1 )s + 1

Example 1.45
Determine the SFG and T.F. of the cathode follower circuit shown in Fig. 1.87a.

(Anna University, December, 2006)

Solution For Fig. 1.87a, the following equations are written:

ei = eg + eo

eg = ei − eo (a)

μeg
i= (b)
r p rk
142 1 Control Systems Modelling and Their Representation

eo = i rk (c)

The variables are ei , eg , μeg , i and eo and they form the nodes in the SFG. Which is
shown in Fig. 1.87b.
There is only one forward path. The forward path gain
μrk
T1 =
r p + rk

There is only one feedback loop. The loop gain is

μrk
L1 = −
r p + rk

When the feedback loop touching the forward path is removed it is broken. There is
no feedback loop for the forward path T1 . Hence, 1 = 1.

 = 1 − L1
μrk
= 1+
(r p + rk )
r p + rk (1 + μ)
=
(r p + rk )
eo T1 1
=
ei 

eo μrk
=
ei r p + rk (1 + μ)

Example 1.46
For the electrical network shown in Fig. 1.88a draw the SFG and determine V3 /V1
using Mason’s gain formula.

(Anna University, April, 2004)

Solution Assume that the current in the branch where R4 is connected is I2 . The
current from the voltage source V1 is I1 (given) and that of the current source is α I1
(given) where α is any constant. Using Kirchhoff’s law the current through R2 is
(α I1 − I2 ) and that of R3 is (I1 − I2 ). The following loop and nodal equations are
written for Fig. 1.88a.
1.18 Block Diagram Reduction Technique 143

I1
(a)

V2 ( I1− I2)

I1 R1 I2
R2

+ R3 R4 V3
V1
−
I1− I2

(b)
− 1 − 1
R1 −R3 R2
L1 L2 L3

V1 V3
1 I1 R3 V2 1 I2 R4
R1 R2

L4
α

Fig. 1.88 a Electrical network for Example 1.46. b SFG for Fig. 1.88a

V1 V2
I1 = − (a)
R1 R1

V2 = I1 R3 − I2 R3 (b)

V3 V2
α I1 − I2 = −
R2 R2

V2 V3
I2 = α I1 + − (c)
R2 R2

V3 = I2 R4 (d)

V1 , I1 , V2 , I2 and V3 are represented as nodal variables in SFG. Equations (a), (b),

(c) and (d) are used to construct SFG which is shown in Fig. 1.88b.
144 1 Control Systems Modelling and Their Representation

The forward path and the path gain, and the feedback loop and loop gains are
given in the following tables:

Forward path Forward path gain

R3 R4
1. V1 − I1 − V2 − I2 − V3 T1 =
R1 R2
R4
2. V1 − I1 − I2 − V3 T2 = α
R1
Feedback loop Loop gain
R3
1. I1 − V2 − I1 L1 = −
R1
R3
2. V2 − I2 − V2 L2 = −
R2
R4
3. I2 − V3 − I2 L3 = −
R2
R3
4 I1 − I2 − V2 − I1 L4 = α
R1

L 1 and L 3 are the only two non-touching loops. Their gain product is given by,

R3 R4
L1 L3 =
R1 R2
 = 1 − (L 1 + L 2 + L 3 + L 4 ) + L 1 L 3
R3 R3 R4 R3 R3 R4
= 1+ + + −α +
R1 R2 R2 R1 R1 R2
R1 R2 + R2 R3 + R3 R1 + R1 R4 − α R2 R3 + R3 R4
=
R1 R2

All the feedback loops touching the forward paths T1 and T2 when removed are
broken. Hence, 1 = 1 and 2 = 1.

V3 T1 1 + T2 2
=
V1 
(R3 R4 /R1 R2 ) + α(R4 /R1 )
=


V3 R3 R4 + α R2 R4
=
V1 [R1 R2 + R2 R3 + R3 R1 + R1 R4 + R3 R4 − α R2 R3 ]

Example 1.47
V3
For the electrical network shown in Fig. 1.89a, draw the SFG and determine (s),
V1
using Mason’s gain formula.
1.18 Block Diagram Reduction Technique 145

(a)
I2
C
V2
I1 L1 I1− I2 L2 I2

V1 V3
R

I1

(b)
− 1 1
L1s L 2s
L1 L3

V1 V3
1 I1 R V2 1 I2
L1s L2s − 1
Cs
L2
1

Fig. 1.89 a Electrical network for Example 1.47. b SFG for Fig. 1.89a

Solution The Laplace transformed network equations for Fig. 1.89a are written as
given below.

V1 − V2 = L 1 s(I1 − I2 ) (a)
V1 V2
I1 = − + I2
L 1s L 1s

V2 = I1 R (b)

1
V1 − V3 = I2 (c)
Cs
1
V3 = V1 − I2
Cs

V3 − V2 = L 2 s I2 (d)
V3 V2
I2 = −
L 2s L 2s
146 1 Control Systems Modelling and Their Representation

The variables V1 , I1 , V2 , I2 and V3 form the nodes in SFG. Using equations (a), (b),
(c) and (d) SFG is drawn as shown in Fig. 1.89b.
From Fig. 1.89b, the forward paths and their gains, the feedback loops and the
loop gains are determined as given below:

Forward path Forward path gain

R
1. V1 − I1 − V2 − I2 − V3 T1 =
L 1 L 2 Cs 3
2 V1 − V3 T2 = 1

Feedback loop Loop gain

R
1. I1 − V2 − I1 L1 = −
L 1s
R
2. I1 − V2 − I2 − I1 L2 = −
L2s
1
3 I2 − V3 − I2 L3 = −
L 2 Cs 2

Feedback loops L 1 and L 3 are not touching with each other. Hence, there are two
non-touching loops.

 = 1 − (L 1 + L 2 + L 3 ) + L 1 L 3
R R 1 R
= 1+ + + +
L 1s L 2s L 2 Cs 2 L 1 L 2 Cs 3
L 1 L 2 Cs 3 + R L 2 Cs 2 + R L 1 Cs 2 + L 1 s + R
=
L 1 L 2 Cs 3

All the three feedback loops are touching the forward path T1 and when removed
from the touching part, they are broken. Hence, 1 = 1. The feedback loops L 1 and
L 2 are not touching the forward path T2 . However, loop L 3 is touching the forward
path T2 and when the touching portion is removed, this feedback loop is broken.
Feedback loops L 1 and L 2 are touching with each other. Hence,

2 = 1 − (L 1 + L 2 )
R R
= 1+ +
L 1s L 2s
L 1 L 2 s 2 + R(L 1 + L 2 )s
=
L 1 L 2s2
V3 T1 1 + T2 2
=
V1 
R L 1 L 2 s + R(L 1 + L 2 )
+
V3 L 1 L 2 Cs 3 L 1 L 2s
=
V1 L 1 L 2 Cs 3 + R L 2 Cs 2 + R L 1 Cs 2 + L 1 s + R
L 1 L 2 Cs 3
1.18 Block Diagram Reduction Technique 147

V3 [L 1 L 2 s + R(L 1 + L 2 )]Cs 2 + R
=
V1 [L 1 L 2 Cs 3 + (L 1 + L 2 )RCs 2 + L 1 s + R]

SUMMARY

1. Control systems are generally classified as open loop and closed loop systems.
Each one has its own advantages and disadvantages. However, because of fast-
ness of response and high accuracy feedback (or closed loop) systems are invari-
ably used in almost all applications.
2. To analyse the dynamic behaviour of a system, systems are modelled in the
frequency and time domains. Transfer function model is used in the frequency
domain while state space model is used in the time domain. Transfer function
model is described in this chapter.
3. Transfer function model, using Laplace transform, is obtained for mechanical,
electrical and electromechanical systems.
4. The transfer function of mechanical systems are obtained by writing the dynamic
equations using D’Alembert’s principle. It is also possible to convert the given
mechanical systems to its electrical equivalent and obtain the T.F from the equa-
tions written by following Kirchhoff’s laws. This is called electrical analog.
Force-voltage and force-current analogies are available and one can follow any
of them. Since there exists a one to one relationship between the components
and variables in the mechanical and electrical systems, the results obtained from
one system can be easily transferred to the other system.
5. Components are interconnected to form a system. Each component is represented
by a block with its functional relationship kept inside the box. The input/ output
relationship of the entire system is obtained by reducing the block diagram to
a single block. This is known as block diagram reduction technique. This is
obtained using a set of rules.
6. Block diagram reduction method is somewhat tedious and as the complex-
ity of the system increases, it becomes more cumbersome. However, systems
can be represented by a graphical model which is known as signal flow graph
model. Using Mason’s gain formula, the input-output relationship can be easily
obtained. Even the block diagram can be converted into its equivalent SFG and
by using Mason’s gain formula, the over all gain between any two variables can
be obtained.
148 1 Control Systems Modelling and Their Representation

EXERCISE

Short Answer Type Questions

1. Define a system.
A system is defined as the interconnections of objects which have a definite
relationship between objects and attributes (variables).
2. What is a control system?
A system, which is formed by interconnection of components when it provides
a desired response is called a control system.
3. What is an open loop system?
An open loop system is one in which the control action is independent of the
output.
4. What is a closed loop system?
A closed loop system is one in which the control action depends upon the output.
5. What is control action?
Actuating signal also otherwise called error signal is called control action which
actuates the power device directly or sometimes indirectly. The actuating signal
is the difference of the reference input and part of the output or full output.
6. What are the advantages of open loop system?
Open loop systems are simple to design. Component requirements are less and
therefore less costly. Almost all physically realizable open loop systems are
stable.
7. What are the drawbacks of open loop systems?
Open loop systems are manually controlled and hence accuracy is less or it
depends upon the ingenuity of the human operator. The time response of the
system is poor (sluggish operation). Its operation requires close manual inspec-
tion during any internal and external disturbances.
8. What are the advantages of closed loop systems?
Closed loop systems are very fast acting and more accurate. It rejects internal
and external disturbances automatically without human involvement. Since the
bandwidth is high, the range of frequency over which the system responds is
high. If the open loop system is unstable, it is possible to make the closed loop
system stable by proper feedback.
9. What are the drawbacks of closed loop systems?
Closed loop systems tempt to oscillate. System complexity is increased by addi-
tion of components. This increases the cost and reduces the stability limit. The
over all gain of the system is reduced.
10. What is an input?
Input is an attribute of the system which can be varied independently.
11. What is an output?
Output is an attribute which can be varied by varying the input.
12. What is servomechanism? Give examples.
Servomechanism is a branch of engineering study if the output is mechanical
1.18 Block Diagram Reduction Technique 149

position or velocity. Examples include control of position of anti-aircraft gun,

radar antenna, rudders in ships etc.
13. What is a regulator system? Give examples.
A regulator system is a system which maintains the system output at a prescribed
set level. Temperature, pressure, level control of a process, voltage control of
an electrical generator, speed control of electric motors, oil engines etc. are
examples of regulator system.
14. Why positive feedback is not used in closed loop system?
For positive feedback systems, the characteristic equation is 1 − G H (s) = 0
which will yield poles in the right half s-plane. This makes the closed loop
system unstable. That is why positive feedback is not employed in closed loop
system. However, positive feedback may be employed in the design of oscillators
which has sustained oscillations.
15. What is characteristic equation and why is it so called?
The denominator of a transfer function when equated to zero is called the char-
acteristic equation. Since the roots of the characteristic equation determine the
character of the transient response and the stability character it is called charac-
teristic equation.
16. Define transfer function of a system. Does it exit for non-linear systems?
Transfer function is defined as the ratio of the Laplace transform of output
variable to the Laplace transform of input variable with all initial conditions
being zero. Laplace transform does not exist for non-linear system.
17. What do you understand by the T.F being strictly proper?
A transfer function is said to be strictly proper if the order of the denominator
polynomial is greater than the order of the numerator polynomial. For example,

10(s 2 + 2s + 4)
G(s) =
(s 3 + 5s 2 + 7s + 8)

is called a strictly proper T.F.

18. What is a proper T.F?
A T.F where numerator and denominator polynomials have the same order is
called a proper T.F. For example,

10(s 2 + 2s + 4)
G(s) =
(5s 2 + 7s + 8)

is called a proper transfer function.

19. What is an improper T.F?
A transfer function whose order of the numerator polynomial is greater than the
order of the denominator polynomial is called improper T.F. For example,

10(s 2 + 2s + 4)
G(s) =
(2s + 5)
150 1 Control Systems Modelling and Their Representation

is called an improper T.F.

20. How T.F is defined in terms of impulse response of a system?
The transfer function of a linear system is defined as the inverse Laplace trans-
form of the impulse response of the system with all the initial conditions set to
zero.
21. What is a minimum phase T.F? Give example.
A transfer function which has all the poles and zeros placed in the left of the
s-plane is called a minimum phase T.F. For example,

10(s + 2)
G(s) =
(s + 3)(s + 9)

is a minimum phase T.F.

22. What is a non-minimum phase T.F?
A T.F which has atleast one pole or a zeros placed in the right half s-plane is
called non-minimum phase T.F. For example,

(s − 2)
G(s) =
s(s + 2)

is a non-minimum phase T.F. A zero is present in the right half s-plane.

23. What is an all pass T.F.?
A T.F. which has its poles and zeros present symmetrically with respect to the
imaginary axis is called all pass T.F. For example

(s + 10)
G(s) =
(s − 10)

is an all pass T.F.

24. Define zero of a T.F.?
The points at which the T.F. becomes zero in the s-plane are called zeros of the
T.F. They are also the roots of the numerator polynomial of the T.F.
25. Define pole of a T.F.?
The points at which the T.F. becomes infinity in the s-plane are called poles of
the T.F. They are also the roots of the denominator polynomial of the T.F.
26. What is the use of a lever in mechanical system?
Lever is used as a matching device in a mechanical translational system to attain
maximum power transfer. It transmits translational motion and forces from one
point to another point in the system.
27. What is the use of a gear in mechanical system?
A gear is a matching device in a mechanical rotational system to attain maximum
power transfer. It transmits rotational motion and torque from one point to another
point in the system.
28. What are the energy storing elements in the mechanical system?
In the translational system, kinetic energy is stored in mass and potential energy
1.18 Block Diagram Reduction Technique 151

is stored in the spring. In the rotational system, kinetic energy is stored in inertia
and potential energy is stored in the spring.
29. Distinguish between ordinary D.C motor and D.C servomotor.
For fastness of response, the time constant of a D.C servomotor should be small.
To achieve this, for the given friction the inertia of the servomotor is made
small so that the mechanical time constant is small. To reduce the electrical time
constant, the inductance of the field is made small if it is field controlled motor
and the inductance of armature coil is small if it is armature controlled motor.
Further rotor inertia should be low.
30. What are the advantages of armature controlled D.C motor?
The electrical time constant of the armature controlled D.C motor is small and
hence its time response is very fast. The back emf developed in the armature
circuits provides additional damping which eliminates unwanted oscillations
during transients.
31. What are the advantages of field controlled D.C motor?
In field controlled D.C motor, the power requirement is less and there is no
need for any power amplifier and it requires voltage amplifier only which is not
difficult to design.
32. What are the disadvantages of armature controlled D.C motor?
Armature controlled D.C motor requires power amplifier which is somewhat
difficult to design and costly too.
33. What are the disadvantages of field controlled D.C motor?
The electrical time constant of field circuit of the D.C motor is high and hence
the time response is very sluggish. Further it requires a constant current source
in the armature circuit which is some what difficult to get.
34. What is the necessacity of negative slope torque-speed characteristic of a
two phase A.C servomotor?
The negative torque-speed characteristic of a two phase A.C servomotor provides
positive damping to the motor which ensures stable operation. In other words,
the roots of the characteristic equation in the T.F of the motor will lie in the left
half of the s-plane.
35. Why an ordinary two phase induction motor is not useful for control appli-
cation?
The torque-speed characteristic of an ordinary two phase induction motor has
positive slope at low speed which provides negative damping to the motor opera-
tion and is unstable. Hence, ordinary two phase induction motors are not suitable
for control application.
36. What change is made in the design of two phase servomotor from that of
the ordinary two phase A.C motor?
The rotor resistance of the two phase A.C motor is considerably increased. This
makes the torque-speed characteristic to have negative slope for entire speed
of operation without losing stability but at the cost of efficiency. For control
applications stability is more important than efficiency.
37. Can two phase A.C servomotor be operated as A.C technogenerator?
Similar to D.C motor being converted to D.C technogenerator, two phase A.C
152 1 Control Systems Modelling and Their Representation

servomotor can be used as A.C technogenerator where output is proportional to

the speed at constant frequency which is the frequency of the excitation voltage
of the reference field winding.
38. What is a pair of synchros?
A pair of synchros is used as an error detector in servo mechanism. It consists
of a synchro generator and a control transformer.
39. What is the constructional difference between synchro generator and con-
trol transformer?
The synchro generator and control transformer have identical construction of
stators where three phase winding is provided. However, there is a major differ-
ence in the construction of the rotors. While the rotor of the synchro generator
is a salient pole one, the rotor of the control transformer is dump bell shaped
cylindrical type.
40. Why the rotor of control transformer is cylindrically shaped?
The CT rotor is cylindrically shaped so that the air gap flux is uniformly dis-
tributed to the servo amplifier and hence the change in rotor impedance with the
rotor position is minimized.
41. In the stators of S.G and C.T three phase A.C windings are provided. What
is the nature of induced voltage in the stator windings?
Even though the stators of S.G and C.T are provided with three phase A.C
windings, the induced voltages in the windings are single phase voltages.
42. What is a block diagram?
The pictorial representation of a system is called a block diagram. System inputs
and outputs are represented by drawing a single line towards the block and away
from the block while the block represents the functional relationship between
input and output.
43. What are the basic connections of block diagram?
Series (cascade), parallel and feedback connections of block diagram are the
basic connections.
44. What are the basic operations performed in block diagram reduction?
The basic operations performed in block diagram reduction include shifting the
blocks before and after the summer, moving a take off point before and after
a block and before and after summers, interchanging of summers and reducing
blocks in parallel, cascade and feedback.
45. State the block diagram simplification rule for removing the feedback.
If G(s) is the T.F of the forward path and H (s) is that of the feedback path, than
the overall T.F is simplified for negative feedback as

C(s) G(s)
=
R(s) 1 + G H (s)

The sign of G H (s) will be negative for positive feedback.

46. What are the advantages of block diagram representation?
By connecting the blocks of individual components, the block diagram for the
1.18 Block Diagram Reduction Technique 153

overall system can be easily formed. Further, it is easy to visualize the actual
system by examining the block diagram.
47. What are the disadvantages of block diagram representation?
The disadvantages of block diagram are that it is not unique for a given system.
The main source of energy flow in the system is not shown in the block diagram.
Some important functions are often concealed inside the four walls of the blocks.
From the block diagram it is difficult to understand what type of system the block
diagram represents since no information about the construction is known. Fur-
ther, block diagram is drawn under the assumption that what is inside the block
is not affected by what is outside of the block which is not true.
48. What is the signal flow graph?
Signal flow graph is a graphical method of signal modelling in which the nodes
represent the different variables in the system. The nodes are connected by
branches whose gain give functional relationship between any two nodes.
49. Explain Mason’s gain formula for signal flow graphs.
According to Mason’s gain formula, the total gain between the output variable
and the input variable is given by:

m=1
n
Tm m
A=

where A is total gain and Tm is gain of m th forward path.
  
 = 1− Pn1 + Pn2 − Pn3 + · · ·
n n n
= Determinant of the graph

Pnr is the gain product of n possible combination of r non-touching loops. m

is  for mth forward path with the feedback loops touching that forward path
when removed is equal to cofactor of the graph.
50. What are input and output and mixed nodes of SFG?
A node which has only outgoing branches is called input while node which has
only incoming branches are called output node or sink nodes. A node which is
neither an input node nor an output node is called a mixed node.
51. What is path and what is a forward path of SFG?
The traversal of branches connected by nodes is called a path. A path which orig-
inates from the input node and terminates at the output node is called a forward
path.
52. What is path gain?
The product of the branch gains while traversing the path is called the path gain.
154 1 Control Systems Modelling and Their Representation

53. What is a feedback loop and loop gain?

If the path which originates from a node and ends at the same node without
encountering the same node more than once is known as feedback loop. The
product of all the branch gains of the feedback loop is called loop gain.
54. What is a self loop?
The path starting from a node if it ends in the same node without touching any
other node is called self node.
55. What is non-touching loop?
Two feedback loops are said to be non-touching if they do not have any common
node.
56. What are basic rules of SFG?
The basic rules of SFG include the following. A signal always flows along the
branch in the direction of arrow. Its strength gets multiplied by the branch gain.
The value of the signal at any node is given by the sum of all the signals entering
that node.
57. How block diagram is converted into SFG?
In addition to input and output variables in the block diagram, the outputs of
the summers and also the take off points are identified with variables. These
variables become nodes is SFG. These nodes are connected by branches which
are obtained by inspection of block diagram.

Long Answer Type Questions

1. Find the T.F. of the circuit shown in Fig. 1.90. The initial conditions are zero.

V0 106
(s) = 2
Vi s + 106 s + 106

X
2. For the translational mechanical system shown in Fig. 1.91, determine (s).
F

X (s) (B1 s + K 1 )
=
F(s) [M B1 s + (M K 1 + B1 B2 )s + (B1 K 1 + B2 K 1 + B1 K 2 )s + K 1 K 2 ]
3 2

Fig. 1.90 Electrical network

for Question 1 10KΩ 10mH

Vi(s) Vo(s)
100μF
1.18 Block Diagram Reduction Technique 155

Fig. 1.91 Mechanical

system for Question 2 x(t)
f(t)

M2
K1 K2
B1 B2

3. Consider the electro mechanical system shown in Fig. 1.92. The force of attrac-
tion of the solenoid coil is directly proportional to the current in the coil and the
proportionality constant is K f N /amp. Determine the T.F. XE (s). Neglect back
emf effect of the solenoid coil. The coil has a resistance R and inductance L.
X K f (Bs + K )
(s) =
E s(R + Ls)[M Bs 2 + M K s + B K ]

4. For the block diagram shown in Fig. 1.93 find C/R using block diagram reduction
method.

C G1G2G3G4
=
R [1 + G 2 G 3 H2 + G 3 G 4 H1 + G 1 G 2 G 3 G 4 H3 ]

5. For the block diagram shown in Fig. 1.94, find C/R using block diagram reduc-
tion method.

x(t)
i(t)

M R L
e(t )
K
B

Fig. 1.92 Electromechanical system for Question 3

H2
R + + − +
G1 G2 G3 G4 C
− −
H1

H4

Fig. 1.93 Block diagram for Question 4

156 1 Control Systems Modelling and Their Representation

G5

R + + + +
G1 G2 G3 G4 C
− −
H1

H2

Fig. 1.94 Block diagram for Question 5

Fig. 1.95 Block diagram for

G3
Question 6

+ + C
R G1 G2
− +

H1

C G 1 G 2 (G 3 G 4 + G 5 )
=
R [1 + G 2 G 3 H1 + G 1 G 2 G 3 G 4 H2 + G 1 G 2 G 5 H2 ]

6. For the block diagram shown in Fig. 1.95, obtain C/R using block diagram
reduction method.

C G1G2 + G3
=
R 1 + G 1 H1

7. For the block diagram shown in Fig. 1.96, determine C/R using block diagram
reduction method.

C G 1 G 2 (G 3 + G 4 )
=
R [1 + G 1 G 2 H1 + G 2 G 3 H2 + G 2 G 4 H2 + G 1 G 2 (G 3 + G 4 )]

8. For the block diagram shown in Fig. 1.97, determine C/R using block diagram
reduction method.

C G 2 [G 1 + G 3 (1 + G 1 H1 )]
=
R [1 + G 1 (H1 + G 2 )]
1.18 Block Diagram Reduction Technique 157

G4

+ +
R + + +
G1 G2 G3 C
− − −

H2

H1

Fig. 1.96 Block diagram for Question 7

Fig. 1.97 Block diagram for

G3
Question 8
+
R + + + C
G G2
− −1

H1

9. For the block diagram shown in Fig. 1.98, determine C when the inputs R and
D are simultaneously present.

[R(G 1 + G 2 )G 3 G 4 ] + D[G 4 (1 + G 3 H1 )]
C=
[1 + G 3 H1 + (G 1 + G 2 )G 3 G 4 H2 ]

10. For the block diagram shown in Fig. 1.99a, draw the equivalent SFG and using
Mason’s gain formula determine C/R.

T1 = G 1 G 2 G 3 ; T2 = G 1 G 4 ;
L 1 = −G 4 H2 ; L 2 = −G 1 G 2 H1 ;
L 3 = −G 2 G 3 H2 ; L 4 = −G 1 G 2 G 3 ; L 5 = −G 1 G 4 ;
1 = 1; 2 = 1;  = 1 − (L 1 + L 2 + L 3 + L 4 + L 5 )

C (G 1 G 2 G 3 + G 1 G 4 )
=
R [1 + G 2 G 3 H2 + G 1 G 2 G 3 + G 4 H2 + G 1 G 2 H1 + G 1 G 4 ]

11. Using Mason’s gain formula, obtain the overall T.F. of the SFG shown in
Fig. 1.100.
158 1 Control Systems Modelling and Their Representation

G2
D
+ +
R + + + C
G1 G3 G4
− −

H1

H2

Fig. 1.98 Block diagram for Question 9

(a) G4

+ x4 +
R + + +
G1 G2 G3
x1 x2 x3
− − −

H2

H1

(b) −H2
G4

1 x2 G1 G2 G3
R x1 x3 x4 C
1

−H1

−1

Fig. 1.99 a Block diagram for Question 10. b SFG for Fig. 1.99a

T1 = G 1 G 2 G 3 G 4 G 5 G 8 ; T2 = G 1 G 6 G 4 G 5 G 8 ; T3 = G 1 G 2 G 7 ;
L 1 = −G 4 G 5 H1 ; L 2 = −G 2 G 3 G 4 G 5 G 8 H2 ;

L 3 = −G 6 G 4 G 5 G 8 H2 ; L 4 = −G 2 G 7 H2 ;
1 = 1; 2 = 1; 3 = 1 + G 5 G 4 H1 ;
 = 1 − (L 1 + L 2 + L 3 + L 4 ) + L 1 L 4
1.18 Block Diagram Reduction Technique 159

G6

x3 x4
R C
G1 x1 G 2 x2 G3 G4 G5 x5 G8

−H1

G7

−H2

Fig. 1.100 SFG for Question 11

C
=
R
G 1 G 2 G 3 G 4 G 5 G 8 + G 1 G 6 G 4 G 5 G 8 + G 1 G 2 G 7 (1 + G 4 G 5 H1 )
[1+G 4 G 5 H1 +G 2 G 4 G 5 G 8 H2 +G 6 G 4 G 5 G 8 H2 +G 2 G 7 H2 +G 2 G 4 G 5 G 7 H1 H2 ]

12. For the SFG shown in Fig. 1.101, evaluate C/R using Mason’s gain formula.

T1 = G 1 G 2 G 3 G 4 ; T2 = G 1 G 2 G 5 ;
L 1 = −G 2 H4 ; L 2 = −G 2 G 3 H2 ;
L 3 = −H1 ; L 4 = −G 2 G 3 G 4 H3 ;
1 = 1; 2 = (1 + H1 );
 = 1 − (L 1 + L 2 + L 3 + L 4 ) + (L 1 L 3 )

C G 1 G 2 G 3 G 4 + G 1 G 2 G 5 (1 + H1 )
=
R [1 + G 2 H4 + G 2 G 3 H2 + H1 + G 2 G 3 G 4 H3 + G 2 H1 H4 ]

13. Using Mason’s gain formula obtain the overall T.F. of the SFG given in Fig. 1.102.

T1 = G 1 G 2 G 3 G 4 ; T2 = G 1 G 2 G 5 G 4 ;
L 1 = −G 4 H2 ; L 2 = −G 1 G 2 G 3 G 4 ;
L 3 = −G 1 G 2 G 5 G 4 ; L 4 = G 2 H1 ;
1 = 1; 2 = 1;
 = 1 − (L 1 + L 2 + L 3 + L 4 ) + L 1 L 4
160 1 Control Systems Modelling and Their Representation

−H2

R G2 x2
x3 C
G1 x1 G3 G4

−H4 −H1
G5

−H3

Fig. 1.101 SFG for Question 12

G5

1 G2 G3 G4
R G1 1 C
−H1 −H2

−1

Fig. 1.102 SFG for Question 13

C G 1 G 2 G 4 (G 3 + G 5 )
=
R [1 + G 4 H2 + G 1 G 2 G 4 (G 3 + G 5 ) − G 2 H1 (1 + G 4 H2 )]

14. For the SFG shown in Fig. 1.103 obtain the values of C/R.

T1 = 400; T2 = 200;
L 1 = −4; L 2 = −4; L 3 = −1;
1 = 1 + 1 = 2; 2 = 1 + 4 = 5;
 = 1 − (L 1 + L 2 + L 3 ) + L 1 L 3 + L 2 L 3 ;
 = 1 + 4 + 4 + 1 + 4 + 4 = 18

C 400 × 2 + 200 × 5
= = 100
R 18

15. For the electrical network shown in Fig. 1.104a, draw the SFG and using Mason’s
gain formula determine VV31 (s).
1.18 Block Diagram Reduction Technique 161

x1
2
10 −1

x4 10 C
R x5
1 x2 5 x3 4 2

−1 −2

Fig. 1.103 SFG for Question 14

1
T1 = ; T2 = 1;
R1 R2 C 1 C 2 s 2
1 1 1
L1 = − ; L2 = − ; L3 = − ;
R1 C 2 s R2 C 2 s R2 C 1 s
 = 1 − (L 1 + L 2 + L 3 ) + L 1 L 3 ; 1 = 1;
2 = 1 + L 1 + L 2

(a) C1
I2

I1-I2 L
I1 V2
I2
R1
R2
I1
V1 C2 V3

(b) 1
1
R2
L2
L3

V1 V3
1 I1 − 1 V2 I2 − 1
R1 C 2s − 1 C1s
R2
L1
− 1
R1

Fig. 1.104 a Electrical network for Question 15. b SFG for Fig. 1.104a
162 1 Control Systems Modelling and Their Representation

V3 [R1 R2 C1 C2 s 2 + (R1 + R2 )C1 s + 1]

=
V1 [R1 R2 C1 C2 s 2 + (R1 C1 + R2 C1 + R1 C2 )s + 1]
Chapter 2
Time Response Analysis

After completing this chapter, you should be able to:

 Model first and second order systems.

 Find the time response of first order systems for impulse and step inputs.
 Derive expressions for time domain specifications of first order system.
 Derive the transfer function of second order system in terms of damping
ratio and natural frequency of oscillation.
 Classify the second order systems as under damped, critically damped and
over damped systems.
 Find the impulse response of a second order system for different damping
ratios.
 Find the step response of second order system for different damping ratios.
 Derive expressions for time domain specifications of second order system
for step input.
 Find the steady state error using static and dynamic error constants for unity
and non-unity feedback systems.
 Find the dynamic behaviour of the system with PID controllers.
 Plot the dynamic response of the system using MATLAB.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 163
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_2
164 2 Time Response Analysis

2.1 Introduction

When the control system is excited by the input r (t), the output c(t) which is
expressed as a function of time t is known as time response of the system. When time
t tends to infinity, the output response reaches the steady state. Such a response is
called the steady state response. The dynamic behaviour of continuous time system
is described by differential equation. The functional relationship between the output
and the input of a linear time invariant system is described by transfer function which
is defined as the ratio of the Laplace transform of the output variable to the Laplace
transform of the input variable with all initial conditions being zero. The system
function of linear continuous time system is expressed in terms of poles and zeros.
The values of s at which the transfer function becomes infinity are called poles of
Linear Time Invariant Continuous system (LTIC). The poles are also known as the
factors of the denominator polynomial. Similarly the values of s at which the transfer
function becomes zero are called zeros of the transfer function of the systems. They
are also the factors of the numerator polynomial of the transfer function. For system
performance analysis the transfer function is represented in the form of a block. A
complex control system when inter-connected by numerous blocks can be ultimately
reduced to a single block with the system required output and the input. This is done
by what is known as block diagram reduction technique as described in Chap. 1.
The Laplace transform of the output variable when inverse transformed gives
time response of the system. In this chapter, time response of first and second order
systems is obtained for impulse and step inputs. Further, expressions for the time
domain specifications of these systems are derived in terms of system parameters.
The steady state errors for typical inputs such as step, ramp and parabolic are derived
in terms of error constants. Expressions for the steady state error are also determined
for inputs other than the above inputs. Steady state error expressed as a function of
time is derived in terms of generalized error constants. To obtain desired steady state
and transient response, PID controllers are introduced and connected in cascade with
the plant. For performance evaluation, a prototype second order system is considered.
The study includes P, I, D, PI, PD, ID, and PID controllers. The transient response of
the system with the above controllers is obtained by simulation study in the digital
computer using MATLAB program. Extensive numerical examples are worked out
to explain different concepts in the study of transient and steady state responses of
the system with and without controllers. The simulation study is carried out in digital
computer using MATLAB program.
2.2 First Order Continuous Time System 165

2.2 First Order Continuous Time System

2.2.1 System Modelling

Consider the following first order differential equation with input r (t) and output
c(t).

dc(t)
+ ac(t) = K r (t) (2.1)
dt
Taking Laplace transform on both sides, we get

(s + a)C(s) = K R(s)
C(s)
= G(s)
R(s)
K
= (2.2)
(s + a)

Equation (2.2) gives the transfer function of a first order system. This is represented
in block diagram and is shown in Fig. 2.1.

2.2.2 Time Response of First Order System

2.2.2.1 Impulse Response of a First Order System

The impulse input is defined as


1, t = 0
δ(t) =
0, otherwise

Here r (t) = δ(t)

R(s) = 1

From Eq. (2.2), for impulse input the output is written as

K
C(s) = (2.3)
(s + a)

The impulse response of a first order system is obtained by taking inverse Laplace
transform. Thus
c(t) = K e−at
166 2 Time Response Analysis

R(s) K C(s)
G(s)
(s a)

Fig. 2.1 Block diagram representation of first order system

c(t)
K

0 t

Fig. 2.2 Impulse response of a first order system

K
a

c(t)

63.2%
86.5%

98.2%

99.3%

50%
95%

0.693T T 1 2T 3T 4T 5T t
a

Fig. 2.3 Step response of first order system

The impulse response curve is plotted and is shown in Fig. 2.2.

2.2.2.2 Step Response of First Order System

The unit step input is defined as follows:

⎧
⎪
⎨u(t),
r (t) = 1, t ≥0
⎪
⎩
0, t ≤0
2.2 First Order Continuous Time System 167

The Laplace transform of unit step input is

1
R(s) =
s
Substituting the above in Eq. (2.2), we get

K
C(s) = (2.4)
s(s + a)

The above equation is put into partial fraction as given below:

A1 A2
C(s) = +
s s+a
 
K 1 1
= −
a s s+a

Taking inverse Laplace transform, we get

K
c(t) = [1 − e−at ] (2.5)
a

The transient response curve of Eq. (2.5) is shown in Fig. 2.3.

Note: The time response for ramp input r (t) = t can be obtained by integrating
Eq. (2.5) and is given below.

K
c(t) = −1 + at + e−at (2.6)
a2

2.2.3 Time Domain Specifications

The following time domain specifications are defined for a first order system:
(a) Time constant T .
(b) Rise time tr .
(c) Time delay td .
(d) Settling time ts .

2.2.3.1 Time Constant

The time constant T is defined as the time taken for the step response to reach 63.2%
of its final value for the first time. From Fig. 2.3, the time constant T is obtained as
168 2 Time Response Analysis

1
T = (2.7)
a

It is to be noted here that the tangential line of the exponential curve has the slope at
t = 0 as K . Thus, if the time constant T is small, the response is fast and vice versa.
The above equation can be analytically derived as follows. From Eq. (2.5), 6.3.2%
output is 0.632(K /a) and t = T .

K K
0.632 = [1 − e−at ]
a a
e−at = 0.368

Taking loge on both sides and putting t = T , we get

−aT = −1.0

1
T =
a

2.2.3.2 Rise Time tr

The rise time tr is defined as the time taken for the response to go from 10% to 90%
of its final value K T . From Eq. (2.5), at t = T1 , let the output be 10% and t = T2
the output be 90%. Thus,

K K
0.1 = [1 − e−aT1 ]
a a
K K
0.9 = [1 − e−aT2 ]
a a
Dividing the second equation by the first equation, we get

(1 − e−aT2 )
9=
(1 − e−aT1 )

Taking loge on both sides, we get

2.2 = a(T2 − T1 )

Substituting tr = (T2 − T1 ), the rise time is obtained as

tr = 2.2T (2.8)
2.2 First Order Continuous Time System 169

2.2.3.3 Time Delay td

Time delay td is defined as the time taken for the response c(t) to reach 50% of
its final value for the first time. From Eq. (2.5), for 50% output, we may write the
following equation with t = td .

0.5 = 1 − e−(td /T )
e−(td /T ) = 0.5

Taking loge on both sides, we get

td = 0.693T (2.9)

2.2.3.4 Settling Time ts

Settling time ts is defined as the time taken for the response c(t) to reach and stay
within 2% of its final value for the first time.
From Eq. (2.5) for 98% output we may write the following equation with t = ts .
−ts
0.98 = 1 − e T

Taking loge on both sides, we get

ts = 3.91T (2.10)

For 5% error tolerance

ts = 3T

For 7% error tolerance

ts = 2.66T

It is to be noted that error tolerance may be given as 5, 7%, etc., and the corresponding
settling time is determined. Unless otherwise the error tolerance is specified it is
always taken as 2% error.
170 2 Time Response Analysis

Summary of Time Domain Specifications of First Order System

1 Time constant T = a1
2 Rise time tr = 2.2T
3 Time delay td = 0.693T
4 Settling time ts = 3.91T

Example 2.1
Consider the following T.F. of a certain first order system:

10
G(s) =
(s + 10)

Derive an expression for the response of the system for r (t) = 5u(t). Find the time
constant, settling time, time delay and rise time.

Solution
C(s) 10
= G(s) =
R(s) (s + 10)
5
R(s) =
s
50
C(s) =
s(s + 10)
 
1 1
=5 −
s s + 10

Taking inverse Laplace transform, we get

c(t) = 5[1 − e−10t ]

1
Time constant T = = 0.1 s
10
Settling time ts = 3.91T = 0.391 s
Rise time tr = 2.2T = 0.22 s
Time delay td = 0.693T = 0.0693 s
2.2 First Order Continuous Time System 171

T = 0.1 s
ts = 0.391 s
tr = 0.22 s
td = 0.0693 s

Example 2.2
A glass bulb thermometer reads 98.2% of its final value of temperature 1 min after
immersing it in hot water. Determine the time constant, rise time, time delay and
settling time for 5% error tolerance.

Solution The mercury thermometer is a first order system. From Eq. (2.5), for
98.2% output, the time taken is 60 s.

K −60 K
1 − e T = 0.982
a a
4T = 60 s
T = 15 s

Rise time

tr = 2.2T
= 2.2 × 15
= 33 s

Time delay

td = 0.693T
= 0.693 × 15
= 10.4 s

For 5% error tolerance, the settling time is

ts = 3T
= 3 × 15
= 45 s
172 2 Time Response Analysis

T = 15 s
tr = 33 s
td = 10.4 s
ts = 45 s

2.3 Second Order System Modelling

For model development of a general second order system, consider the electrical
motor represented in block diagram form as shown in Fig. 2.4a. The system param-
eters are as follows:

J = Moment of inertia of motor in Kg.m2

B = Motor frictional coefficient in N-m/rad/s
K = Error detector constant in N-m/rad error

Torque developed by the motor is

Td (t) = K e(t)
= K (r (t) − c(t))

This torque is to overcome the torque opposed by J and B under no load condition.
Thus

d 2 c(t) dc(t)
Tu (t) = J 2
+B
dt dt
Tu (t) = Td (t)
d 2 c(t) dc(t)
J 2
+B = K (r (t) − c(t)) (2.11)
dt dt
Taking Laplace transform on both sides and expressing the ratio of the output variable
to the input variable we get the following equation:

C(s) K
= (2.12)
R(s) (J s 2 + Bs + K )

The denominator of Eq. (2.12) is a second degree polynomial in s and therefore

Eq. (2.12) describes the dynamics of a second order system. Equation (2.12) can be
written as
2.3 Second Order System Modelling 173

C(s) K /J
= 2 B (2.13)
R(s) (s + J s + K
J
)

Equation (2.13) can be written in a generalized form as given below:

C(s) ωn2
= 2 (2.14)
R(s) (s + 2ζ ωn s + ωn2 )

where
K
ωn = = Natural frequency of oscillation in rad/s.
J
B
ζ = √ = Damping ratio and ζ ωn = Damping factor
2 KJ

Equation (2.14) is called prototype second order system equation. Thus, natural
frequency of oscillation and damping ratio are the two parameters of a generalized
second order system. The systems may be electrical, mechanical, thermal, hydraulic,
biological or in any form. Equation (2.14) is represented in block diagram form as
shown in Figs. 2.4b and c which are in the form of open loop and closed loop forms
respectively. Equation (2.14) is also called standard equation for a second order
system.
The natural frequency ωn is defined as the frequency of oscillation of a second
order system without damping
If the damping is provided to the system, the system time response contains damped
oscillations with exponential decay. Now consider Eq. (2.14) which can be written
as follows:

C(s) ωn2
= (2.15)
R(s) (s + a1 )(s + a2 )

where a1 and a2 are the pole locations of Eq. (2.14) and they are expressed in terms
of ζ and ωn . For 0 ≤ ζ ≤ 1.

a1 = −ζ ωn + jωn 1 − ζ 2

a2 = −ζ ωn − jωn 1 − ζ 2 (2.16)

In the above case the system is said to be under damped and the pole locations are
as shown in Fig. 2.5. 
From Fig. 2.5, ωd = ωn 1 − ζ 2 is called damped frequency of oscillation. For
ζ = 1, the system is said to be critically damped and the pole locations are shown in
Fig. 2.6. which are repeated poles at s = −ωn .
If ζ > 1, the system is said to be over damped. In this case the poles are at,
174 2 Time Response Analysis

a1 = −ζ ωn + ωn ζ 2 − 1

a2 = −ζ ωn − ωn ζ 2 − 1 (2.17)

The pole locations are shown in Fig. 2.7.

2.4 Time Response of a Second Order System

2.4.1 Impulse Response

Consider Equation (2.14),

C(s) ωn2
= 2
R(s) (s + 2ζ ωn s + ωn2 )

(a) Error detector

r(t) e(t) MOTOR c(t)

J, B

Error

(b)

R(s) n
2
C(s)
(s2 2 ns
2)
n

(c)

R(s) E(s) 2 C(s)

n
G(s)
s(s 2 n)

Fig. 2.4 a Block diagram representation of a second order systems. Second order system repre-
sentation. b open loop form, c closed loop form
2.4 Time Response of a Second Order System 175

n j d j
s-plane
n
1 2 2
d n tan 1 1

n j d

Fig. 2.5 Pole location in the s-plane for complex conjugate poles for 0 ≤ ζ ≤ 1 (under damped)

s-plane

Fig. 2.6 Pole location of a critically damped system (ζ = 1)

a1 n n
2 1)

0
a2 n n
2 1)

Fig. 2.7 Pole location of an over damped system (ζ > 1)

For an impulse input,

R(s) = 1

The above equation is written as

176 2 Time Response Analysis

ωn2
C(s) = (2.18)
(s 2 + 2ζ ωn s + ωn2 )

The impulse response for the following cases are determined:

1. Under damped case (ζ < 1).
2. Over damped case (ζ > 1).
3. Critically damped case (ζ = 1).

2.4.1.1 Under Damped Case (ζ < 1)

For ζ < 1, the second degree denominator polynomial of Eq. (2.18) is written as
follows:

ωn2
C(s) =
(s + ζ ωn + jωd )(s + ζ ωn − jωd )

where

ωd = ωn 1 − ζ 2 = Damped frequency of oscillation

Putting the above equation into partial fraction we get,

A1 A2
C(s) = +
s + ζ ωn + jωd s + ζ ωn − jωd
A1 (s + ζ ωn − jωd ) + A2 (s + ζ ωn + jωd )
=
(s 2 + 2ζ ωn s + ωn2 )
ωn2 = A1 (s + ζ ωn − jωd ) + A2 (s + ζ ωn + jωd )

Put

s = −ζ ωn + jωd
ωn2 = A2 (2 jωd )
ωn
A2 = 
j2 1 − ζ 2
A1 = Conjugate of A2
−ωn
= 
j2 1 − ζ 2
 
ωn −1 1
C(s) =  +
j2 1 − ζ 2 (s + ζ ωn + jωd ) (s + ζ ωn − jωd )
2.4 Time Response of a Second Order System 177

Fig. 2.8 Impulse response c(t)

curves of a second order
system
<1

>1

0
t

Taking inverse Laplace transform we get,

ωn
c(t) =  −e−(ζ ωn + jωd )t + e−(ζ ωn − jωd )t
j2 1 − ζ 2
 jωd t 
ωn e − e− jωd t
=  e−ζ ωn t
1 − ζ2 2j

ωn 
c(t) =  e−ζ ωn t sin ωn 1 − ζ 2 t (2.19)
1− ζ2

2.4.1.2 Critically Damped Case (ζ = 1)

Equation (2.18) for ζ = 1 is written as

ωn2
C(s) =
s 2 + 2ωn s + ωn2
ωn2
=
(s + ωn )2

Taking inverse Laplace transform we get,

c(t) = ωn2 te−ωn t (2.20)

2.4.1.3 Over Damped Case (ζ > 1)

Equation (2.18) for ζ > 1 is written as

178 2 Time Response Analysis

ωn2
C(s) =  
(s + ζ ωn + ωn ζ 2 − 1)(s + ζ ωn − ωn ζ 2 − 1)
A1 A2
=  + 
(s + ζ ωn + ωn ζ − 1) (s + ζ ωn − ωn ζ 2 − 1)
2
 
ωn2 = A1 (s + ζ ωn − ωn ζ 2 − 1) + A2 (s + ζ ωn + ωn ζ 2 − 1)

Put s = −ζ ωn − ωn ζ 2 − 1

ωn2 = A1 (−2ωn ζ 2 − 1)
−ωn
A1 = 
2 ζ2 − 1

Put s = −ζ ωn + ωn ζ 2 − 1

ωn2 = A2 (2ωn ζ 2 − 1)
ωn
A2 = 
2 ζ2 − 1
 
ωn −1 1
C(s) =   + 
2 ζ2 − 1 (s + ζ ωn + ωn ζ 2 − 1) (s + ζ ωn − ωn ζ 2 − 1)

Taking inverse Laplace transform we get,

ωn √2 √2
c(t) =  e−ζ ωn t −e−ωn ζ −1t + eωn ζ −1t (2.21)
2 ζ2 − 1

The time response curves of Eqs. (2.19), (2.20) and (2.21) are shown in Fig. 2.8.

2.4.1.4 Importance of Impulse Response

1. If the area under the impulse response curve is finite, then the system is said to
be Bounded Input, Bounded Output (BIBO) stable.
2. From impulse response, by taking inverse Laplace transform the system transfer
function is obtained.
3. If impulse response is known, step response can be obtained by integrating it.

2.4.1.5 Importance of Step Response

Step response of a system is important for the following reasons:

2.4 Time Response of a Second Order System 179

1. It is easy to generate step signal and test the system in the laboratory.
2. The step signal is sufficiently drastic and if satisfactory step response is obtained,
then the system is likely to give satisfactory performance for other types of inputs.
3. From step response impulse response can be obtained by differentiating it and
useful information may be derived. Similarly from step response, ramp response
can be obtained by integrating it.
4. The application of step input is equivalent to the application of numerous sinu-
soidal signals with a wide range of frequencies.

2.4.2 Step Response of a Second Order System

Consider Eq. (2.14).

C(s) ωn2
= 2
R(s) (s + 2ζ ωn s + ωn2 )

For a step input of height R, the Laplace transform

R
R(s) =
s
Substituting this in the above equation we get,

Rωn2
C(s) = (2.22)
s(s 2 + 2ζ ωn s + ωn2 )

2.4.2.1 Under Damped Response (ζ < 1)

For ζ < 1, Eq. (2.22) is written in the following form:

Rωn2
C(s) =
s(s + ζ ωn + jωd )(s + ζ ωn − jωd )

A1 A2 A3
C(s) = + + (2.23)
s (s + ζ ωn + jωd ) (s + ζ ωn − jωd )

Analytical Method of Determining the Residues A1 , A2 and A3

Rωn2 = A1 (s 2 + 2ζ ωn s + ωn2 ) + A2 s(s + ζ ωn − jωd ) + A3 s(s + ζ ωn + jωd )

Putting s = 0 in the above equation we get,

180 2 Time Response Analysis

n j d
tan 1 2 A3

s-plane
n

A1

n
A2

n j d

Fig. 2.9 Poles and residue locations of Eq. (2.23)

Rωn2 = A1 ωn2
A1 = R

Putting s = (−ζ ωn − jωd ) we get,

Rωn2 = A2 (−ζ ωn − jωd )(− jωd − jωd )

R
A2 =  
j2 1 − ζ (ζ + j 1 − ζ 2 )
2

R∠−φ − π2
= 
2 1 − ζ2
√
1−ζ 2
where tan φ = ζ

R∠φ + π2
A3 = A∗2 = 
2 1 − ζ2

Graphical Method of Determining the Residues

The residues of Eq. (2.23) can also be determined as explained below. The poles are
located along with their residues as shown in Fig. 2.9.
2.4 Time Response of a Second Order System 181

Constant × Product of directed vector distances from all zeros to point A1

A1 =
Product of directed vector distances from all poles to point A1
Rωn2
A1 = =R
ωn ∠φ ωn ∠−φ
Rωn2 R∠ π2 + φ
A3 = = 
ωn ∠π − φ 2ωd ∠ π2 2 1 − ζ2
R∠−( π2 + φ)
A2 = A∗3 = 
2 1 − ζ2

The residues determined by analytical method are the same as obtained by graphical
method. However, the graphical method is simpler and quicker. Substituting the
above residues in Eq. (2.23) we get,
  π π 
1 e− j ( 2 +φ) e j ( 2 +φ)
C(s) = R 1 +  +
2 1 − ζ2 (s + ζ ωn + jωd ) (s + ζ ωn − jωd )

Taking inverse Laplace transform we get,

 
e−ζ ωn t  − j ( π +φ+ωd t) π 
c(t) = R 1 +  e 2 + e j ( 2 +φ+ωd t)
2 1−ζ 2
 
e−ζ ωn t π 
= R 1+  cos + φ + ωd t
1 − ζ2 2

 
e−ζ ωn t 
c(t) = R 1 −  sin(ωn 1 − ζ 2 t + φ) (2.24)
1 − ζ2

where φ is in radians and ωn = rad/s.

2.4.2.2 Critically Damped Response (ζ = 1)

For ζ = 1,
s 2 + 2ζ ωn s + ωn2 = (s + ωn )2

Equation (2.14) can be written as

182 2 Time Response Analysis

R
C(s) =
s(s + ωn )2
A1 A2 A3
= + +
s (s + ωn )2 (s + ωn )
R = A1 (s + ωn )2 + A2 s + A3 s(s + ωn )

Put s = 0
R
A1 =
ωn2

Put s = −ωn
−R
A2 =
ωn

Equating the coefficients of s 2 terms we get,

0 = A1 + A3
−R
A3 = 2
ωn
 
R 1 ωn 1
C(s) = 2 − −
ωn s (s + ωn )2 s + ωn

Taking inverse Laplace transform we get,

R
c(t) = 1 − tωn e−ωn t − e−ωn t (2.25)
ωn2

2.4.2.3 Over Damped Response (ζ > 1)

For over damped case the time response of a second system for step input can be
derived following the method described above. The time response is given below:

 √2 √2 
e−ωn (ζ + ζ −1)t e−ωn (ζ − ζ −1)t
c(t) = R 1 +  +  (2.26)
2{ζ 2 − 1 + ζ ζ 2 − 1} 2{ζ 2 − 1 − ζ ζ 2 − 1}

Using Eqs. (2.24), (2.25) and (2.26), the transient response curves are plotted as
shown in Fig. 2.10 for R = 1.
The poles locations in the s-plane and the transient response curves are shown in
Fig. 2.11a. The effect of roots on damping ratio is shown in Fig. 2.11b.
2.4 Time Response of a Second Order System 183

Fig. 2.10 Transient response curves of a second order system

2.4.3 Time Domain Specifications of a Second Order System

The performance of a second order system is measured by the following specifica-

tions:
1. Peak overshoot M p and % peak overshoot % M p .
2. Time at which the peak overshoot occurs peak time t p .
3. Time constant T .
4. Rise time tr .
5. Settling time ts .
6. Time delay td .
Expressions for the above specifications are derived in terms of the second order
system parameters ζ and ωn . The transient response curve shown in Fig. 2.12 is
followed.
184 2 Time Response Analysis

(a) c(t)
j
0
(undamped) j n
s-plane

j n

0 t
(b)
j
j c(t)
n d

0< <1 s-plane

(underdamped)
1

n j d
0
t
(c)
c(t)
j
1
(critically damped) s-plane

0 t
(d)

j c(t)
>1
s-plane
(over damped)

( 2 1)
n n

( 2 1)
n n
0 t

Fig. 2.11 a Pole locations and transient response of a second order system. b Effect of roots on
damping ratio
2.4 Time Response of a Second Order System 185

0
0< <1
1< <0

>1 1
>1

1 < 1

0< <1

Fig. 2.11 (continued)

2.4.3.1 Peak Over Shoot M p

Peak overshoot is defined as the amount by which the transient response waveform
overshoots the steady value or the final value.
The peak overshoot is denoted by M P . It is expressed as

M P = Cmax − Cfinal

The percentage overshoot is expressed as

(Cmax − Cfinal )
% MP = × 100
Cfinal

Peak overshoot can be expressed in terms of the system parameters ζ and ωn . This
occurs for the under damped system response. Consider Eq. (2.24).
 
e−ζ ωn t 
c(t) = R 1 −  sin(ωn 1 − ζ 2 t + φ)
1 − ζ2

Cmax is obtained by differentiating c(t) with respect to t which gives t p , the time at
which the maxima occurs and substituting in Eq. (2.24).
186 2 Time Response Analysis

1
1
1 2

e nt
Cmax Upper envelope 1
1 2

c(t) Tolerance error

Mp

Cfinal R

e n
t
Lower envelope 1
1 2

0.5R

Transient response
Steady
Transient response curve
state
response
0 t
td tr tp ts

1
1
1 2

Fig. 2.12 Time domain specifications of a second order system


dc(t) e−ζ ωn t 
= R 0−  ωn 1 − ζ 2
dt 1 − ζ2

× cos(ωn 1 − ζ 2 t + φ)
 
e−ζ ωn t (ζ ωn ) sin(ωn 1 − ζ 2 t + φ)
+  =0
1 − ζ2
  
1 − ζ 2 cos(ωn 1 − ζ 2 t + φ) = ζ sin(ωn 1 − ζ 2 t + φ)

tan φ = tan(ωn 1 − ζ 2 t + φ)

ωn 1 − ζ 2 t = nπ where n = 0, 1, 2, . . .

For n = 1 maxima occurs and t = t p

2.4 Time Response of a Second Order System 187

π
tp =
ωd

Substituting t = t p and c(t) = Cmax in Eq. (2.24) we get

⎡ ⎤
e √−ζ π 2
1−ζ
Cmax = R ⎣1 −  sin(π + φ)⎦
1 − ζ2
  
√−ζ π
= R 1 + e 1−ζ 2 ∵ sin φ = 1 − ζ 2

M P = Cmax − Cfinal
√−ζ π
= R + Re 1−ζ 2 −R

√−ζ π
M P = Re 1−ζ 2 (2.27)

MP
% MP = × 100
R

√−ζ π
% MP = e 1−ζ 2 × 100 (2.28)

2.4.3.2 Peak Time t p

The peak time t p is defined as the time at which the first maximum of transient
response waveform occurs. It is expressed as

π
tp =  (2.29)
ωn 1 − ζ 2

2.4.3.3 Settling Time ts

The settling time ts is defined as the time required for the transient response to reach
and stay within the prescribed percentage error. The expression for the settling time
of a second order system is derived as follows.
188 2 Time Response Analysis

Let

± e−m × 100 where m = 1, 2, 3 . . . (2.30)

be the prescribed error within which the system transient response settles down.
Equation (2.28) represents the error and can be written as follows:

−ζ π n
% MP = e  × 100 (2.31)
1 − ζ2

where n = 0, 1, 2, . . .. For odd values of n overshoots occur and for even values of
n under-shoots occur. For any error, Eqs. (2.30) and (2.31) are same. Thus,

− √ζ πn
e−m = e 1−ζ 2
m 1 − ζ2
n= (2.32)
ζπ

From Eq. (2.29), for any peak value of the response t p can be written as follows:

nπ
tp =  (2.33)
ωn 1 − ζ 2

For the given error e−m , t p = ts . Substituting this in Eq. (2.33) and also for n from
Eq. (2.32), we get the following expression for the settling time:

m 1 − ζ 2π
ts = 
ζ π ωn 1 − ζ 2
m
ts =
ζ ωn

ts = mT (2.34)

In Eq. (2.34), T = ζ ω1 n is called the time constant of the system.

Normally the permissible error prescribed for the settling time is 2%. In this case

e−m = 0.02
loge e−m = loge (0.02)
= −3.91
−m = −3.91
2.4 Time Response of a Second Order System 189

m = 3.91
ts = 3.91T (2.35)

For 5% error, m = 3 and

ts = 3T

For 7% error, m = 2.66

ts = 2.66T

2.4.3.4 Time Constant T

Time constant is defined as the time taken for the transient response to reach 63.2%
of its final value for the first time. From Eq. (2.24) and Fig. 2.12 either from the upper
envelope 1 + e−ζ ωn t or from lower envelope 1 − e−ζ ωn t can be obtained as

1
T = (2.36)
ζ ωn

2.4.3.5 Rise Time tr

Rise time tr is defined as the time taken for the transient response to go from 10%
to 90% of the final value. Sometimes, the rise time is also defined as the time taken
for the transient response to reach the final value for the first time. The expression
for the rise time is derived as follows using the second definition stated above. For
c(t) = R, Eq. (2.24) is written as follows:
 
e−ζ ωn t
R = R 1−  sin(ωd t + φ) (2.37)
1 − ζ2

Equation (2.37) gives the solution as

sin(ωd t + φ) = 0
= sin nπ where n = 0, 1, 2 . . .
ωd t + φ = nπ = π for n = 1

Substituting t = tr in the above equation we get

190 2 Time Response Analysis

(π − φ)
tr =  (2.38)
ωn 1 − ζ 2

In Eq. (2.38) it is to be noted that φ is in radians.

2.4.3.6 Time Delay td

Time delay is defined as the time taken for the transient response to reach 50% of
the final value for the first time. The expression for the time delay is given by the
following empirical formula:

(1 + 0.7ζ )
td = (2.39)
ωn

2.4.3.7 The Period of Oscillation T0

The period of oscillation of the transient response curve of a second order system
is defined as the time taken for the response to complete one cycle of the damped
oscillation and it is given as

2π
T0 = (2.40)
ωd

2.4.3.8 The Number of Oscillations Before Settling Time is Reached T0s

The number of oscillations before settling time is reached is given by

ts
T0s = (2.41)
T0
2.4 Time Response of a Second Order System 191

Summary of Time Domain Specifications of Second Order System

1 Time constant T = 1
ζ ωn

(π
√−φ)
2 Rise time tr =
ωn 1−ζ 2

(1+0.7ζ )
3 Time delay td = ωn
4 Settling time ts = mT
√−ζ π
5 % Peak overshoot % Mp = e 1−ζ 2 × 100
π
6 Time at which the peak overshoot occurs tp = ωd

7 The period of oscillations t0 = 2π

ωd

ts
8 The number of oscillations before settling Tos = T0
Time is reached

Example 2.3

Obtain the impulse and step responses of the following unity feedback control system
with open loop transfer function

6
G(s) =
s(s + 5)

(Anna University, December, 2009)

Solution The given system is represented in block diagram form as shown in
Fig. 2.13. From Fig. 2.13, the following equation is derived.

C(s) G(s)
=
R(s) 1 + G(s)
6
= 2
s + 5s + 6
C(s) 6
=
R(s) (s + 2)(s + 3)

1
For unit step input R(s) = .
s
192 2 Time Response Analysis

6
C(s) =
s(s + 2)(s + 3)
A1 A2 A3
= + +
s s+2 s+3
6 = A1 (s + 2)(s + 3) + A2 s(s + 3) + A3 s(s + 2)

For s = 0
A1 = 1

For s = −2
A2 = −3

For s = −3
A3 = +2

1 3 2
C(s) = − +
s s+2 s+3

Taking inverse Laplace transform we get

c(t) = 1 − 3e−2t + 2e−3t

The impulse response is obtained by differentiating the step response. If we denote

impulse response as h(t), then

dc(t)
h(t) =
dt

h(t) = 6[e−2t − e−3t ]

Alternatively, the impulse response is obtained from first principle.

C(s) 6
=
R(s) (s + 2)(s + 3)

For an impulse R(s) = 1

6
C(s) = H (s) =
(s + 2)(s + 3)
 
1 1
=6 −
s+2 s+3
2.4 Time Response of a Second Order System 193

R(s) 6 C(s)
G(s)
s(s 5)

Fig. 2.13 System block diagram for Example 2.3

h(t) = 6[e−2t − e−3t ]

It is to be noted that the poles of the closed loop transfer function lie on the −ve real
axis of the s-plane and hence the system is over-damped.

Example 2.4
A certain unity negative feedback control system has an open loop transfer function.

10
G(s) =
s(s + 2)

Find the rise time, percentage overshoot, peak time, td , T0 , Tos and settling time for
a step input of 12 units.

(Anna University, December, 2009)

Solution
10
G(s) =
s(s + 2)
C(s) G(s) 10
= = 2
R(s) 1 + G(s) s + 2s + 10

Comparing the above equation with the following standard second order equation.

C(s) ωn2
= 2
R(s) s + 2ζ ωn s + ωn2
√
ωn = 10
2ζ ωn = 2
ζ = 0.3162

Given R = 12 units. Using Eq. (2.27),

−ζ π √−0.3162π
overshoot M p = Re  = 12e 1−0.31622
1 − ζ2
194 2 Time Response Analysis

M p = 4.2 units

Mp 4.2
% overshoot = × 100 = × 100
R 12

% M p = 35%

Using Eq. (2.29), the peak time is obtained.

π π
tp =  =√ 
ωn 1 − ζ 2 10 (1 − 0.31622 )

t p = 1.05 s

The rise time is obtained using Eq. (2.38)

(π − φ)
tr = 
ωn 1 − ζ 2
 √ 
ωn 1 − ζ 2 = 10 1 − 0.31622 = 3
cos φ = ζ
φ = cos−1 0.3162
= 1.25 radians
(π − 1.25)
tr =
3

tr = 0.63 s

The settling time is obtained using Eq. (2.34). Here 2% error tolerance is taken.
Hence, m = 3.91.
1 1
Time constant, T = = √
ζ ωn 0.3162 10

T = 1s

ts = mT
2.4 Time Response of a Second Order System 195

ts = 3.91 s

The time delay is obtained using Eq. (2.39)

1 + 0.7ζ 1 + 0.7 × 0.3162

td = = √ = 0.3862
ωn 10

td = 0.3862

The period of oscillation is obtained using Eq. (2.40)

2π 2π
T0 =  =√  = 2.0943
ωn 1 − ζ 2 10 1 − (0.3162)2

T0 = 2.0943

The number of oscillations before the settling time is reached is obtained using Eq.
(2.41)

ts 3.91
T0s = = = 1.867
T0 2.0943

T0s = 1.867

Answers:
M p = 4.2 units; % M p = 35%; t p = 1.05 s
tr = 0.63 s; T = 1 s; ts = 3.91 s
td = 0.3862 s; T0 = 2.0943 s; Tos = 1.867 s

Example 2.5

Figure 2.14 shows a unity feedback system. Calculate ζ and ωn when K = 0. Also
calculate K when ζ = 0.6.

(Anna University, December, 2009)

Solution The given block diagram is reduced and represented as shown in Fig. 2.15.
From Fig. 2.15, the following equation is written:
196 2 Time Response Analysis

R(s) 1 C(s)
64
s(s 4)

Ks

Fig. 2.14 Block diagram of Example 2.5

Fig. 2.15 Reduced block

R(s) 64 C(s)
diagram of Fig. 2.14
s(s K 4)

C(s) 64
= 2
R(s) s + (K + 4)s + 64

Compare this with the following standard second order equation:

C(s) ωn2
= 2
R(s) (s + 2ζ ωn s + ωn2 )

For K = 0;
√
ωn = 64 = 8
2ζ ωn = 4
ζ = 0.25

For ζ = 0.6;

2ζ ωn = K + 4
K = 2 × 0.6 × 8 − 4
K = 5.6

Answers:
For K = 0; ζ = 0.25
ωn = 8
For ζ = 0.6; K = 5.6
2.4 Time Response of a Second Order System 197

Example 2.6

The forward path T.Fs of certain unity negative feedback control systems are given
below. Determine the damping ratio and natural frequency of oscillation and the
nature of damping of each system.

(3s + 13)
(a) G(s) =
(s + 3)(s + 4)
2(s + 2)
(b) G(s) =
s(s + 4)
(s + 9)
(c) G(s) =
s(s + 2)

Solution
(a)

(3s + 13)
G(s) =
(s + 3)(s + 4)

The closed loop T.F. of the system is

C(s) G(s)
=
R(s) 1 + G(s)
(3s + 13)
= 2
(s + 10s + 25)

The characteristic equation of the above system is

s 2 + 10s + 25 = 0

The characteristic equation of the prototype second order system is

s 2 + 2ζ ωn s + ωn2 = 0

Comparing the given characteristic equation with that of the prototype second
order system equation, we get

ωn2 = 25 or ωn = 5 rad/s
2ζ ωn = 10 or ζ = 1

Since ζ = 1, the system is critically damped.

198 2 Time Response Analysis

(b)

2(s + 2)
G(s) =
s(s + 4)

The closed loop T.F. of the system is

C(s) 2(s + 2)/(s(s + 4)) 2(s + 2)

= = 2
R(s) 1 + [2(s + 2)/(s(s + 4))] s + 6s + 4

The characteristic equation is

s 2 + 6s + 4 = 0

Comparing the above equation with the standard second order system charac-
teristic equation, we get

ωn2 = 4 or ωn = 2 rad/s
2ζ ωn = 6 or ζ = 1.5

Since ζ > 1, the system is over damped.

(c)

(s + 9)
G(s) =
s(s + 2)

The closed loop T.F. is

C(s) (s + 9) (s + 9)
= = 2
R(s) s(s + 2) + s + 9 s + 3s + 9

The characteristic equation is

s 2 + 3s + 9 = 0

Comparing this with the characteristic equation of a standard second order sys-
tem, we get

ωn2 = 9 or ωn = 3 rad/s
2ζ ωn = 3 or ζ = 0.5

Since ζ < 1, the system is under damped.

2.4 Time Response of a Second Order System 199

Example 2.7

Consider the following systems with their open loop T.F. Determine the value of K
of the unity negative feedback system to give the required damping ratio and natural
frequency of oscillation.

(s + K )
(a) G(s) =
s(s + 3)

The required damping ratio ζ = 0.5

(2s + 1)
(b) G(s) =
(3s + 4)(s + K )

The required natural frequency of oscillation is 4 rad/s.

Solution
(a)

(s + K )
G(s) =
s(s + 3)

The closed loop T.F. is

C(s) (s + K )/s(s + 3) (s + K )
= = 2
R(s) 1 + [(s + K )/s(s + 3)] s + 4s + K

The characteristic equation is

s 2 + 4s + K = 0

In the above equation √

ωn = K and 2ζ ωn = 4

Given ζ = 0.5, 2ζ ωn = 4

K = 16

(b)

(2s + 1)
G(s) =
(3s + 4)(s + K )
C(s) (2s + 1)
= 2
R(s) 3s + (3K + 6)s + (4K + 1)
200 2 Time Response Analysis

The characteristic equation is

(3K + 6) (4K + 1)
s2 + s+ =0
3 3
(4K + 1)
ωn2 = =4×4
3

K = 11.75

(3K + 6)
2ζ ωn = = 13.75
3

ζ = 1.7188

Example 2.8

A unity feedback control system has the following open loop transfer function:

2(s + 8)
G(s) =
s(s + 2)

(a) Write the closed loop system T.F. C(s)/R(s) and from this find damping ratio,
natural frequency and peak overshoot.
(b) Calculate c(t) for unit step input.
(c) Calculate the steady state error for unit step input and the rise time.

(Anna University, December, 2010)

Solution
(a)

C(s) G(s) 2(s + 8)

= = 2
R(s) 1 + G(s) (s + 4s + 16)

The characteristic equation is

s 2 + 4s + 16 = 0
2.4 Time Response of a Second Order System 201

Comparing this with the standard prototype second order system characteristic
equation

s 2 + 2ζ ωn s + ωn2 = 0
√
ωn = 16 = 4
2ζ ωn = 4 or ζ = 0.5

ωn = 4 rad/s
ζ = 0.5

(b) Performance specifications have been derived for the transfer function of the
closed loop system which has no zero. However, there is a zero at s = −8 for
the given closed loop system. To find the peak overshoot, the use of expression
Eq. (2.27) will give erroneous result. Hence, it is to be derived from first principle.
The characteristic equation is factorised as

(s 2 + 4s + 16) = (s + 2 + j3.465)(s + 2 − j3.465)

The closed loop T.F. is put into partial fraction as

C(s) 2(s + 8)
=
R(s) (s + 2 + j3.465)(s + 2 − j3.465)

For unit step input R(s) = 1/s

A1 A2 A3
C(s) = + +
s (s + 2 + j3.465) (s + 2 − j3.465)
2(s + 8) = A1 (s 2 + 4s + 16) + A2 s(s + 2 − j3.465) + A3 s(s + 2 + j3.465)

For s = 0,

16 = A1 × 16 + 0 + 0
A1 = 1

For s = −2 − j3.465,
202 2 Time Response Analysis

2(−2 − j3.465 + 8) = A1 × 0 + A2 (−2 − j3.465)(−2 − j3.465 + 2 − j3.465)

+A3 × 0
2(6 − j3.465) = A2 (−2 − j3.465)(− j6.93)
2 × 6.93∠−30◦ = A2 4∠60◦ 6.93∠90◦
A2 = −0.5
A3 = Conjugate of A2
A3 = −0.5
 
1 1 1
C(s) = − 0.5 +
s (s + 2 + j3.465) (s + 2 − j3.465)

Taking inverse Laplace transform, we get

c(t) = 1 − 0.5[e−(2+ j3.465)t + e−(2− j3.465)t ]

c(t) = 1 − e−2t cos 3.464t

To find the maximum overshoot the above equation is differentiated with respect
to t and equated to zero

dc(t)
= e−2t [(−2) cos 3.465t − 3.465t sin 3.465t] = 0
dt
−2
tan 3.465t =
3.465
 π
= tan nπ −
6
The first maximum overshoot occurs for n = 1 and correspondingly t = t p .
 π
tan 3.465t p = tan π −
6
5 π
tp = × = 0.7556 s
6 3.465
Substituting t = t p in the equation for c(t), we get

Cmax = 1 − e−2×0.7556 cos(3.465 × 0.7556) = 1.191

Peak value

M p = Cmax − 1
= 1.191 − 1 = 0.191
2.4 Time Response of a Second Order System 203

%M p = 19.1%

(c)

c(t) = 1 − e−2t cos(3.465t)

C(∞) = 1

Steady state error

ess = r (∞) − C(∞)

= 1−1

ess = 0

For the rise time, c(t) = 1 or cos 3.465 tr = cos π/2

tr = 0.4533

Example 2.9
The open loop T.F. of a certain unity negative feedback control system is given by

K
G(s) =
s(T + s)

where K and T are constants and greater than zero.

The system overshoot for unit step input is 60%. It is desired to bring the over-
shoot to 20% by reducing the gain of the amplifier in the system. Determine by
what factor the amplifier gain is to be reduced to meet the above requirement. K 1
corresponds to 60% overshoot and K 2 corresponds to 20% overshoot.

Solution The closed loop T.F. of the given system is

C(s) G(s)
=
R(s) 1 + G(s)
K
= 2
s + Ts + K

Comparing the characteristic equation with that of the standard second order equa-
tion, we get
204 2 Time Response Analysis
√
ωn = K
2ζ ωn = T
T
ζ = √
2 K

Let the peak overshoot corresponding to ζ = ζ1 be M p1 and ζ2 for M p2 . Using Eq.

(2.27), we get

− √1
ζ π
1−ζ12
e = 0.6
−√
ζ2 π
1−ζ22
e = 0.2

Taking loge on both sides of the above equations, we get

−ζ1 π
 = −0.5108
1 − ζ12
ζ1 = 0.1605
√−ζ2 π 2
e 1−ζ2 = 0.2

Taking loge on both sides, we get

ζ2 π
− = −1.6094
1 − ζ22
ζ2 = 0.456
T
ζ1 = √ = 0.1605
2 K1
T
ζ2 = √ = 0.456
2 K2

Dividing one by the other and squaring, we get

K2 1
=
K1 8.07

1
K2 = K1
8.07
2.4 Time Response of a Second Order System 205

The amplifier gain has to be reduced by a factor 8.07.

Example 2.10
A certain unity feedback control system when subjected to step input of size 10 gave
the response as
c(t) = 10 − 30e−2t + 20e−3t

Determine the closed loop T.F. Also find the damping ratio and undamped natural
frequency of oscillation.

Solution

c(t) = 10 − 30e−2t + 20e−3t

Taking Laplace transform on both sides of the above equation, we get

 
1 3 2
C(s) = 10 − +
s s+2 s+3
10[(s + 2)(s + 3) − 3s(s + 3) + 2s(s + 2)]
=
s(s + 2)(s + 3)
10[s 2 + 5s + 6 − 3s 2 − 9s + 2s 2 + 4s]
=
s(s + 2)(s + 3)
10 × 6
=
s(s 2 + 5s + 6)

For a step size of 10,

10
R(s) =
s
6
C(s) = R(s)
(s 2 + 5s + 6)

C(s) 6
= 2
R(s) (s + 5s + 6)

The characteristic equation is

206 2 Time Response Analysis

Fig. 2.16 Block diagram for

Example 2.11 R(s) K1 C(s)
s(s 1.2)

(1 K2s)

s 2 + 5s + 6 = 0
√
ωn = 6
= 2.45 rad/s
2ζ ωn = 5
5
ζ = = 1.02
2 × 2.45

ζ = 1.02
ωn = 2.45

Example 2.11
The block diagram of a certain control system is shown in Fig. 2.16. Determine the
value of K 1 and K 2 so that the maximum overshoot for step response is 25% which
occurs at t = 2 s. For these values, find the time domain specifications.

Solution From Fig. 2.16, the closed loop T.F. is obtained as

C(s) K 1 /s(s + 1.2)

=
R(s) 1 (1+K 2 s)
1 + Ks(s+1.2)

C(s) K1
= 2
R(s) [s + (K 1 K 2 + 1.2)s + K 1 ]

The characteristic equation is

s 2 + (K 1 K 2 + 1.2)s + K 1 = 0

ωn = K 1
2ζ ωn = (K 1 K 2 + 1.2)

For 25% overshoot, the damping ratio is calculated using Eq. (2.28) as
2.4 Time Response of a Second Order System 207

√ζ π
e 1−ζ 2 = 0.25

Taking loge on both sides, we get

ζπ
− = −1.3863
1 − ζ2

ζ = 0.404

Using Eq. (2.29), we get

π
 = tp
ωn 1 − ζ 2
π π
ωn =  =  = 1.717
tp 1 − ζ 2 2 1 − (0.404)2
K 1 = ωn2 = (1.717)2

K 1 = 2.95

(K 1 K 2 + 1.2) = 2ζ ωn
= 2 × 0.404 × 1.717

K 2 = 0.0635

(π − )
tr =  where  = cos−1 ζ = 1.155 rad
ωn 1 − ζ 2
(π − 1.155)
= 
1.717 1 − (0.404)2

tr = 1.265 s

1 1
T = =
ζ ωn 0.404 × 1.717
208 2 Time Response Analysis

T = 1.442 s

ts = 3.91T = 3.91 × 1.442

ts = 5.638 s

(1 + 0.7ζ ) (1 + 0.7 × 0.404)

td = =
ωn 1.717

td = 0.747

2π 2π
To = = 
ωd 1.717 1 − (0.404)2

To = 4.0 s

ts 5.638
Tos = =
To 4

Tos = 1.41

T = 1.442 s; ts = 5.638 s; td = 0.747 s

tr = 1.265 s; M p = 25%; t p = 2 s
To = 4.0; Tos = 1.41

Example 2.12
The closed loop T.F. of a unity feedback control system is
2.4 Time Response of a Second Order System 209

C(s) 20s 2
=
R(s) (s + 1)(s + 3)(s + 5)

Determine the response of the system when the excitation applied to the input terminal
is
3
r (t) = 1 + 2t + t 2
2

(Anna University, December, 2010)

Solution
3
r (t) = 1 + 2t + t 2
2
1 2 3 (s 2 + 2s + 3)
R(s) = + 2 + 3 =
s s s s3
2
C(s) 20s
=
R(s) (s + 1)(s + 3)(s + 5)
20(s 2 + 2s + 3) A1 A2 A3 A4
C(s) = = + + +
s(s + 1)(s + 3)(s + 5) s s+1 s+3 s+5
20(s 2 + 2s + 3) = A1 (s + 1)(s + 3)(s + 5) + A2 s(s + 3)(s + 5)
+ A3 s(s + 1)(s + 5) + A4 s(s + 1)(s + 3)

20 × 3
For s = 0; =⇒ A1 = = 4.
3×5
20(1 − 2 + 3)
For s = −1; =⇒ A2 = = −5.
(−1)(2)(4)
20(9 − 6 + 3)
For s = −3; =⇒ A3 = = 10.
(−3)(−2)(2)
20(25 − 10 + 3)
For s = −5; =⇒ A4 = = −9.
(−5)(−4)(−2)

4 5 10 9
C(s) = − + −
s s+1 s+3 s+5

c(t) = 4 − 5e−t + 10e−3t − 9e−5t

210 2 Time Response Analysis

Fig. 2.17 Block diagram for R(s) C(s)

Example 2.13 Gc(s) 5/(s2)

Example 2.13
A double integrator plant has the transfer function G(s) = 5/s 2 with H (s) = 1. It
is to be compensated to achieve the damping ratio ζ = 0.5 and undamped natural
frequency ωn = 5 rad/s. What is the T.F. of the series compensator?

Solution The system with the series compensator G c (s) is shown in Fig. 2.17. From
Fig. 2.17, we get

C(s) G c (s)[5/s 2 ] 5G c (s)

= = 2
R(s) 1 + G c (s)[5/s 2 ] s + 5G c (s)

The standard second order system to have ζ = 0.5 and ωn = 5 rad/s will have the
following characteristic equation:

s 2 + 5s + 25 = 0

Comparing the above two characteristic equations, we get

s 2 + 5s + 25 = s 2 + 5G c (s)
G c (s) = (s + 5)

G c (s) = (s + 5)
C(s) 5(s + 5)
=
R(s) (s 2 + 5s + 25)

Example 2.14

The time response curve of a certain second system for a step input of size 4 is
shown in Fig. 2.18. Determine the T.F. of the closed loop system and all time domain
specifications.
2.4 Time Response of a Second Order System 211

c(t)

2 sec

Fig. 2.18 Time response of a second order system for Example 2.14

Solution From Fig. 2.18, % peak overshoot is

(6 − 4)
%M p = × 100 = 50%
4
√−ζ π
e 1−ζ 2 = 0.5

Taking ln on both sides, we get

−ζ π
 = −0.6931
1 − ζ2
ζ = 0.215

From Fig. 2.18, t p = 2 s.

π
tp = 
ωn 1 − ζ 2
π
ωn =  = 1.6084 rad/s.
2 1 − (0.215)2

Time constant T is given by

1 1
T = =
ζ ωn 0.215 × 1.6084

T = 2.892 s

The rise time tr is given by

212 2 Time Response Analysis

(π − φ)
tr =
ωd

where φ = cos−1 0.215 = 1.354 rad

(π − 1.354)
tr = 
1.6084 1 − (0.215)2

tr = 1.138 s

The settling time ts for 2% tolerance is given by

ts = 3.91T = 3.91 × 2.892

ts = 11.3 s

The time delay is given by

1 + 0.7ζ 1 + 0.7 × 0.215

td = =
ωn 1.6084

td = 0.7153 s

The period of oscillation is given by

2π 2π
To = = 
ωd 1.6084 1 − (0.215)2

To = 4 s

The number of oscillation is given by

ts 11.3
Tos = =
To 4

Tos = 2.825
2.4 Time Response of a Second Order System 213

Fig. 2.19 Block diagram of

second order system for (s 1) 3e 4s C(s)
Example 2.15 R(s) (s 2) (s 3)

T = 2.892 s; tr = 1.138 s; ts = 11.3 s

td = 0.7153 s; %M p = 50%; t p = 2 s
To =4 s; Tos = 2.825

Example 2.15
A certain second order system is represented in block diagram shown in Fig. 2.19.
Derive an expression for the unit step response of the system.

Solution For the system represented in Fig. 2.19, the T.F. is

C(s) (s + 1) + 3e−4s
=
R(s) (s + 2)(s + 3)

For unit step input R(s) = 1/s and the above equation is written as

[(s + 1) + 3e−4s ]
C(s) =
s(s + 2)(s + 3)
(s + 1) 3e−4s
= +
s(s + 2)(s + 3) s(s + 2)(s + 3)
= C1 (s) + C2 (s)

where
(s + 1)
C1 (s) =
s(s + 2)(s + 3)
3e−4s
C2 (s) =
s(s + 2)(s + 3)

Now consider C1 (s) which can be put into partial fraction as given below.

A1 A2 A3
C1 (s) = + +
s s+2 s+3
(s + 1) = A1 (s + 2)(s + 3) + A2 s(s + 3) + A3 s(s + 2)

For s = 0 =⇒ A1 = 1/(2 × 3) = 1/6

214 2 Time Response Analysis

For s = −2 =⇒ A2 = (−1)/(−2) = 1/2

For s = −3 =⇒ A3 = (−2)/[(−3)(−1)] = −2/3

1 1 2
C1 (s) = + −
6s 2(s + 2) 3(s + 3)
 
1 1 −2t 2 −3t
c1 (t) = + e − e u(t)
6 2 3

where u(t) indicates that c(t) exits for t ≥ 0. Now consider C2 (s) without delay e−4s
which can be put into partial fraction as given below. This is denoted as C3 (s)

3
C3 (s) =
s(s + 2)(s + 3)
A1 A2 A3
= + +
s s+2 s+3
3 = A1 (s + 2)(s + 3) + A2 s(s + 3) + A3 s(s + 2)

3
For s = 0 =⇒ A1 = = 21
2×3
3
For s = −2 =⇒ A2 = = −3
(−2)(1) 2

3
For s = −3 =⇒ A3 = =1
(−3)(−1)
Thus, C2 (s) without time delay is written as

1 3 1
C3 (s) = − +
2s 2(s + 2) s + 3

Taking inverse Laplace transform we get.

 
1 3 −2t
c3 (t) = − e + e−3t u(t)
2 2

The output c2 (t) with delay is written as

c2 (t) = c3 (t − 4)u(t − 4)
 
1 3 −2(t−4) −3(t−4)
c2 (t) = − e +e u(t − 4)
2 2

The symbol u(t − 4) indicates that the response starts from t ≥ 4. For t < 4 it is
zero. The total response is
2.4 Time Response of a Second Order System 215

c(t) = c1 (t) + c2 (t)

   
1 1 −2t 2 −3t 1 3 −2(t−4)
c(t) = + e − e u(t) + − e + e−3(t−4) u(t − 4)
6 2 3 2 2

Example 2.16
The forward path T.F. of a certain unity negative feedback system is given by

K
G(s) =
s(1 + 0.1s)

The overshoot to unit step is 10%, and the settling time for 2% error tolerance is 2 s.
Determine the value of K .

Solution
K
G(s) =
0.1s 2 + s
C(s) K 10K
= = 2
R(s) 0.1s + s + K
2 s + 10s + 10K

The characteristic equation is

s 2 + 10s + 10K = 0

For 10% overshoot,

√−ζ π
e 1−ζ 2 = 0.1

Taking loge on both sides, we get

−ζ π
 = −2.3026
1 − ζ2
ζ = 0.5916

The required settling time is ts = 2.

ts = 3.91T
2
T = = 0.5115
3.91
1
T =
ζ ωn
1 1
ωn = = = 3.3 rad/s
ζ ×T 0.5916 × 0.5115
216 2 Time Response Analysis

From the characteristic equation

√
ωn = 10K
√
3.3 = 10K

K = 1.089

Example 2.17
A certain unity feedback control system has the following open loop T.F.

K
G(s) =
(s + A)(s + 2)

Find the values of K and A so that the damping ratio is 0.707 and the peak time for
unit step response is 1.8 s.

Solution
K
G(s) =
s2 + (A + 2)s + 2 A

The closed loop T.F. is

C(s) G(s) K
= = 2
R(s) 1 + G(s) s + (A + 2)s + (2 A + K )

Given

ζ = 0.707
t p = 1.8 s

But
π
tp = 
ωn 1 − ζ 2
π π
ωn =  = 
tp 1 − ζ 2 (1.8) 1 − (0.707)2
ωn = 2.4683

The characteristic equation of the system is

2.4 Time Response of a Second Order System 217

s 2 + (A + 2)s + (2 A + K ) = 0
2ζ ωn = A + 2
A = 2ζ ωn − 2
= 2 × 0.707 × 2.4683 − 2
= 3.49 − 2
A = 1.49
(2 A + K ) = ωn2
K = ωn2 − 2 A
= (2.4683)2 − 2 × 1.49
K = 6.0925 − 2.98
K = 3.1125

A = 1.49
K = 3.1125

Example 2.18
A certain unity negative feedback control system has the following open loop transfer
function
K
G(s) = 2
s + As + K

Determine the values of K and A so that the peak overshoot of the step response is
20% and the rise time is 1.2 s.

Solution For the given system, the closed loop T.F. is

C(s) G(s)
=
R(s) 1 + G(s)
K
= 2
s + As + 2K

For 20% peak overshoot, the damping ratio is obtained from

√−ζ π
%M p = e 1−ζ 2 × 100
√−ζ π
e 1−ζ 2 = 0.2

Taking loge on both sides, we get

218 2 Time Response Analysis

−ζ π
 = log 0.2 = −1.609
1 − ζ2
ζ = 0.456

The rise time is given by

(π − )
tr = 
ωn 1 − ζ 2

where  = cos−1 0.456 = 1.1 rad

(π − 1.1)
ωn = 
1.2 1 − (0.456)2
= 1.912 rad/s

The characteristic equation of the given system is

s 2 + As + 2K = 0
ωn2 = 2K
ωn2 (1.912)2
K = =
2 2

K = 1.827

A = 2ζ ωn
= 2 × 0.456 × 1.912
A = 1.744

A = 1.744
K = 1.827

Example 2.19
If r (t) is the step input and c(t) is the output of the system described by a differential
equation
d 2 c(t) dc(t)
+4 + 8c(t) = 8r (t)
dt 2 dt
2.4 Time Response of a Second Order System 219

Determine the undamped natural frequency, damping ratio and damped natural fre-
quency, time for peak overshoot, and settling time.

(Anna University, May, 2007)

Solution The given differential equation is

d 2 c(t) dc(t)
+4 + 8c(t) = 8r (t)
dt 2 dt
Taking Laplace transform on both sides with zero initial conditions, we get

(s 2 + 4s + 8)C(s) = 8R(s)
C(s) 8
= 2
R(s) (s + 4s + 8)

The characteristic equation of the system is

s 2 + 4s + 8 = 0

Comparing this with the characteristic equation of a standard second order system
(s 2 + 2ζ ωn s + ωn2 = 0) we get
√
ωn = 8 = 2.8284 rad/s.
2ζ ωn = 4
2
ζ = = 0.707
2.8284
The damped frequency of oscillation is given by

ωd = ωn 1 − ζ 2

= 2.8284 1 − (0.707)2 = 2 rad/s.

The time for peak overshoot is given by

π π
tp = = s.
ωd 2

The percentage peak overshoot is given by

√−ζ π √−0.707π
%M p = e 1−ζ 2 × 100 = e 1−(0.707)2 × 100 = 4.32%

The settling time for 2% error tolerance is given by

220 2 Time Response Analysis

3.91
ts =
ζ ωn
3.91
ts = = 1.955 sec
0.707 × 2.8284

ωn = 2.8284 rad/s; ζ = 0.707; ωd = 2 rad/s

π
t p = s; M p = 4.32%; ts = 1.955 s
2

Example 2.20

A second order system is represented by the T.F.

C(s) 1
=
R(s) (J s 2 + Fs + K )

A step input of 10 N.m is applied to a system and test results are:

1. maximum overshoot = 6%,
2. time at peak overshoot = 1 s. and
3. the steady state value of the output is 0.5 radians.
Determine the value of J , F and K .

(Anna University, May, 2010)

Solution
C(s) 1 1
= =
R(s) J s + Fs + K
2
J s + FJ s +
2 K
J
10
R(s) =
s
10
C(s) =
s J s 2 + FJ s + KJ

Using final value theorem,

10
C(∞) = lim sC(s) = lim
s→0 s→0 J s 2 + FJ s + K
J
10
= = 0.5
K
K = 20 N.m/rad.
2.4 Time Response of a Second Order System 221

The percentage overshoot is given by

√−ζ π
e 1−ζ 2 = 0.06
−ζ π
 = −2.8134
1 − ζ2
ζ = 0.667
π
tp =  =1s
ωn 1 − ζ 2
π π
ωn =  =
tp 1 − ζ 2 1 − (0.667)2
= 4.2166 rad/s.

The characteristic equation of the given system is

F K
s2 + s+ =0
J J
√
Here ωn = K /J and 2ζ ωn = F/J

K 20
J = =
ωn2 (4.2166)2
= 1.125 Kg-m2
F = 2ζ ωn J
= 2 × 0.667 × 4.2166 × 1.125
= 6.33 N.m/rad/s

J = 1.125 Kg-m2
F = 6.33 N.m/rad/s
K = 20 N.m/rad

Example 2.21

A certain unity feedback control system has the open loop T.F. given by

s+a
G(s) =
(s + b)2
222 2 Time Response Analysis

The system is to have a natural frequency of oscillation ωn = 11 rad/s. and the

damping factor ζ = 0.5. Find the values of a and b and also the closed loop T.F. Is
it possible to find time domain specifications using Eqs. (2.27) to (2.41).

Solution
(s + a)
G(s) =
(s + b)2

The closed loop T.F. is obtained from

C(s) G(s)
=
R(s) 1 + G(s)
(s + a)
=
s2 + (1 + 2b)s + (a + b2 )

The characteristic equation of the above equation is

s 2 + (1 + 2b)s + (a + b2 ) = 0

For the given ζ = 0.5 and ωn = 11 rad/s,

1 + 2b = 2ζ ωn
= 2 × 0.5 × 11
b=5
a + b2 = ωn2
a + 25 = 121
a = 96

a = 96
b =5
C(s) (s + 96)
=
R(s) (s 2 + 11s + 121)

Using Eqs. (2.27)–(2.41), the time domain specifications cannot be determined. This
is because in the closed loop T.F. there is a zero at s = −96, which will give different
expressions for the time response and hence different time domain specifications.

Example 2.22
Consider a certain second order system shown in Fig. 2.20.
2.4 Time Response of a Second Order System 223

Fig. 2.20 Second order

system for Example 2.22 R(s) a C(s)
s(s b)

(a) Evaluate the time constant, % peak overshoot, settling time, peak time and the
rise time of the closed loop system when subjected to unit step input for the value
of a = 16 and b = 4 and find the closed loop poles.
(b) Evaluate the values of a and b so that the real part of the poles remains constant
and the imaginary part is increased 2 times over that of part (a).
(c) Evaluate the values of a and b so that the imaginary part remains constant and
the real part is increased 2 times over that of part (a).
(d) Evaluate a and b if the poles are moved along the radial line from the origin by
2 times over that of part (a). In all the above cases find %M p and tr .

Solution From Fig. 2.20 the closed loop T.F. is obtained as

C(s) a
T (s) = = 2
R(s) s + bs + a

Substituting a = 16 and b = 4, we get

16
T (s) =
s2 + 4s + 16

The T.F. of the prototype second order system is

ωn2
T (s) =
(s 2 + 2ζ ωn s + ωn2 )

By comparison we get ωn = 4; ζ = 0.5


ωd = ωn 1 − ζ 2

= 4 1 − 0.52 = 3.464
224 2 Time Response Analysis

(a)

1 1
T = = = 0.5 s
ζ ωn 0.5 × 4
−ζ π −0.5π
%M p =  × 100 = √ × 100
e 1−ζ 2 e 1 − 0.52
= 16.3%
ts = 3.91T = 3.91 × 0.5
= 1.955 s
π π
tp = =
ωd 3.464
= 0.907 s
φ = cos−1 ζ = cos−1 0.5
π
= rad.
3
(π − φ) π − (π/3)
tr = =
ωd 3.464
= 0.6046

Real part of the pole is

ζ ωn = 0.5 × 4 = 2

The imaginary part of the pole is

ωd = 3.464

The poles of the closed loop system is

s12 = −ζ ωn ± jωd
= −2 ± j3.464

The poles are located as shown in Fig. 2.21.

T = 0.5 s
%M p = 16.3
ts = 1.955 s
t p = 0.907
tr = 0.6046
ζ ωn = 2 (Real part)
ωd = 3.464 (Imaginary part)
2.4 Time Response of a Second Order System 225

(b) Real Part Remains Same

ζ ωn = 2

Imaginary Part is Increased by 2 Times

ωd = 2 × 3.464 = 6.928

The pole locations are shown in Fig. 2.22.


ωn 1 − ζ 2 = 6.928
ζ ωn = 2

Dividing one by the other we get


1 − ζ2 6.928
= = 3.464
ζ 2

Solving for ζ we get

ζ = 0.2773
2 2
ωn = =
ζ 0.2773
ωn = 7.211
a = ωn2 = 7.2112 = 52
b = 2ζ ωn
b = 2 × 0.2773 × 7.211 = 4
√−ζ π
%M p = e 1−ζ 2 × 100
√−0.2773π
= e 1−0.27732 × 100 = 40%
φ = cos−1 ζ = cos−1 0.2773
= 1.29 rad
(π − 1.29)
tr = = 0.2673 s
6.928

a = 52
b =4
%M p = 40%
tr = 0.2673 s

(c) Imaginary Part remains the Same.

226 2 Time Response Analysis

j
2 j3.464

s-plane
n 4

d
60º
0
n

2 j3.464

Fig. 2.21 Pole location of a second order system

ωd = 3.464

Real Part is Increased Two Times.

ζ ωn = 2 × 2 = 4

The pole–zero location is shown in Fig. 2.23.


ωn 1 − ζ 2 = ωd = 3.464
ζ ωn = 4

Dividing one by the other we get


1 − ζ2 3.464
= = 0.866
ζ 4

Solving for ζ and ωn we get

ζ = 0.756
ωn = 5.29
a = ωn2 = 5.292 = 28
b = 2ζ ωn
= 2 × 0.756 × 5.29 = 8
−ζ π
%M P = e √ × 100
1 − 0.7562
= 2.66%
2.4 Time Response of a Second Order System 227

φ = cos−1 ζ = cos−1 0.756

= 0.714 rad.
(π − φ)
tr =
ωd
π − 0.714
= = 0.7 s
3.464

a = 28
b =8
%M P = 2.66%
tr = 0.7 s

(d) The poles are moved along a radial line originating from the origin by 2 times
as in (a).

ωn = 4 × 2 = 8 rad/s.

ζ remains the same as ζ = 0.5. The pole locations are shown in Fig. 2.24.

ωd = ωn 1 − ζ 2
√
= 8 1 − 0.25 = 6.928
π
φ = cos−1 0.5 = rad.
3
ζ ωn = 0.5 × 8 = 4
b = 2ζ ωn = 2 × 4 = 8
a = ωn2 = 64

√−ζ π
%M p = e 1−ζ 2 × 100
√−0.5π
=e × 100 = 16.3%
1−0.25

π −φ
tr =
ωd
π − (π/3)
= = 0.302 s
6.928

a = 64
b =8
%M p = 16.3%
tr = 0.302 s
228 2 Time Response Analysis

Fig. 2.22 Real part remains j

constant. Imaginary part is 2 j6.928
increased by 2 times
n 7.211

s-plane
d 6.928
74º
0
n 2

2 j6.928

Fig. 2.23 Imaginary part is j

kept fixed. Real part is 4 j3.464
increased by 2 times
n 5.29

s-plane
d 3.464
41º
0
n 4

4 j3.464

Fig. 2.24 The pole is moved j

along the radial line by 2 4 j6.928
times
n 8

s-plane
d 6.928
60º
0
n 4

4 j6.928
2.5 Steady State Error 229

(a) (b)
c(t) c(t)

ess ess

Output
Output Step input Ramp input

t t

(c)
c(t)

ess

Parabolic input

Output

Fig. 2.25 Steady state error for a Step input; b Velocity input; and c Acceleration input

2.5 Steady State Error

Control systems are used to control physical variables. How accurately the system
performs is very important. The variables under control should be as close as to the
desired value. To be able to measure and predict accuracy in control system, a standard
measure of performance is widely used. This standard measure of performance is
steady state error. The difference between the input (the desired response) and the
output (the actual response) is referred to as the error. This error may be a function
of time. This is called transient error. The expression in Eq. (2.27) which is the peak
overshoot is the transient error. This error when t tends to ∞ is called the steady
state error and is denoted by ess . The system under test is simulated with standard
inputs such as step, ramp and acceleration. The steady state error of certain control
system for these signals is shown in Fig. 2.25.
The steady state error in control systems mainly come from non-linearities of
some of the components connected in the systems. The non-linearities include the
backlash in gear or a motor that will not move unless the input voltage exceeds a
threshold. However, here we study about the steady state errors which are due to
system configuration and the type of input that is applied.
230 2 Time Response Analysis

R(s) E(s) C(s)

G(s)

Fig. 2.26 Block diagram showing error function E(s) for a unity feedback system

2.5.1 Type of the System

It can be established that the steady state errors are dependent upon the type of the
system. The type of the system is decided by the number of pure integrators that are
present in the open loop T.F. G(s). As explained in Chap. 1, consider the following
open loop T.F:

K (1 + sz 1 )(1 + sz 2 ) × · · ·
G(s) = (2.42)
s n (1 + sp1 )(1 + sp2 ) × · · ·

In Eq. (2.42) the value of n in the denominator, or equivalently, the number of pure
integrations in the denominator of G(s) gives the Type of the system. Thus, a system
with n = 0 is a Type 0 system, with n = 1, it is Type 1 system and so on. We will
establish that the steady state errors are functions of the input and system Type.

2.5.2 Steady State Error Using Static Error Constants

Consider the system with forward path T.F. G(s) connected with unity feedback as
shown in Fig. 2.26. In Fig. 2.26, E(s) is the error which is the difference between the
input and the output and can be written as

E(s) = R(s) − C(s)

= R(s) − E(s)G(s)
E(s)[1 + G(s)] = R(s)

R(s)
E(s) = (2.43)
1 + G(s)

From Eq. (2.43) it is evident that the error E(s) is a function of system configuration
G(s) and the input R(s). By taking inverse Laplace transform of E(s) one can get
e(t), the error which is expressed as a function of time t. The steady state error ess is
obtained by substituting t = ∞ in e(t). However, this procedure is little cumbersome.
2.5 Steady State Error 231

Fig. 2.27 Step input signal r(t)

r (t)

Instead, using Laplace final value theorem the steady state error is obtained from E(s)
as

ess = lim s E(s) (2.44)

s→0

Using Eq. (2.44), the steady state errors are calculated for the step, ramp and parabolic
inputs as explained below. It is to be noted that final value theorem as applied to Eq.
(2.44) is valid iff the closed loop system is stable and the transfer function G(s) is
proper.

2.5.2.1 Steady State Error Due to Step Input-The Position Error

Constant

Consider the step input signal shown in Fig. 2.27 which is described as
r (t) = Ru(t) or

R t ≥0
r (t) = (2.45)
0 t <0

The Laplace transform of r (t) is given by

R
R(s) = (2.46)
s
Substituting Eq. (2.46) in (2.43) we get

R
E(s) =
s + sG(s)

The steady state error is obtained by using Eq. (2.44) in the above equation.

R R
ess = lim s =
s→0 s + sG(s) lims→0 G(s) + 1
232 2 Time Response Analysis

R
ess = (2.47)
1 + Kp

where

K p = lim G(s) (2.48)

s→0

K p is called the position error constant. Now consider the system with the following
T.F.
K (1 + sz 1 )(1 + sz 2 ) · · ·
G(s) =
s n (1 + sp1 )(1 + sp2 ) · · ·

For n = 0, the system is Type 0 system. K p for Type 0 system is calculated using
Eq. (2.48).

K p = lim G(s)
s→0
K (1 + sz 1 )(1 + sz 2 ) · · ·
= lim =K
s→0 (1 + sp1 )(1 + sp2 ) · · ·

The steady state error is obtained using Eq. (2.47)

R
ess = (2.49)
1+K

For Type 1 system

K (1 + sz 1 )(1 + sz 2 ) · · ·
K p = lim =∞
s→0s(1 + sp1 )(1 + sp2 ) · · ·
R
ess = =0
(1 + ∞)

For Type 2 and Type 3 systems the steady state error will similarly be zero.
Thus, when step input is applied, for Type 0 system the steady state error
is finite and is given by Eq. (2.49). For systems of Type 1 and above the steady
state error is zero.

Example 2.23
Consider the closed loop system with the following forward path T.F.

200(s + 2)(s + 5)
G(s) =
(s + 4)(2s + 6)
2.5 Steady State Error 233

A step input of height 12 size is applied. Find the steady state error.

Solution
200(s + 2)(s + 5)
G(s) =
(s + 4)(2s + 6)
200 × 2 × 5 1000
K p = lim G(s) = =
s→0 4×6 12
R 12
ess = =
1 + Kp 1 + (1000/12)

144
ess =
1012

Example 2.24

A certain unity negative feedback control system has the following forward path T.F.

K
G(s) =
s2 + 2s + 10

Determine the value of K to achieve a steady state error ess ≤ 0.5 to a step input of
size 2.

Solution
K
G(s) =
s 2 + 2s + 10
K
K p = lim G(s) =
s→0 10
2 20
ess = =
1 + Kp 10 + K
20
0.5 =
10 + K

K = 30
234 2 Time Response Analysis

Fig. 2.28 Ramp or velocity r(t)

input signal

2.5.2.2 Steady State Error Due to Velocity (Ramp) Input-The Velocity

Error Constant

Consider the velocity or ramp input signal shown in Fig. 2.28. The slope of the ramp
is R. The signal is described as r (t) = Rtu(t) or

Rt t ≥ 0
r (t) =
0 t <0

The Laplace transform of the above signal is

R
R(s) = (2.50)
s2
R
E(s) =
s2 + s 2 G H (s)
R
ess = lim s E(s) = lim
s→0 s→0 sG(s)

R
ess = (2.51)
Kv

where

K v = lim sG(s) (2.52)

s→0

K v is called the velocity error constant and has the unit s−1 .
Consider the system with the following T.F.

K (1 + sz 1 )(1 + sz 2 ) · · ·
G(s) =
s n (1 + sp1 )(1 + sp2 ) · · ·

For Type 0 system,

2.5 Steady State Error 235

s K (1 + sz 1 )(1 + sz 2 ) · · ·
K v = lim =0
s→0 (1 + sp1 )(1 + sp2 ) · · ·
R
ess = =∞
Kv

For Type 1 system,

s K (1 + sz 1 )(1 + sz 2 ) · · ·
K v = lim =K
s→0 s(1 + sp1 )(1 + sp2 ) · · ·
R
ess =
K
For Type 2 system,

s K (1 + sz 1 )(1 + sz 2 ) · · ·
K v = lim
s→0 s 2 (1 + sp1 )(1 + sp2 ) · · ·
=∞
R
ess = =0
∞
Thus for
Type 0 system, ess = ∞
R
Type 1 system, ess =
Kv
Type 2 system, ess = 0

For Type 2 and higher Type system ess = 0.

Example 2.25
A negative feedback control system has the following loop T.F.

10(2s + 3)(s − 4)
G(s) =
s(3s + 2)(4s + 5)

The system is subjected to ramp input of slope 2 rad/s. Find the steady state error.

Solution
10(2s + 3)(s − 4)
G(s) =
s(3s + 2)(4s + 5)
−10 × 3 × 4
K v = lim sG(s) = = −12
s→0 (2)(5)
R −2
ess = =
Kv 12
236 2 Time Response Analysis

Fig. 2.29 Parabolic or r(t)

acceleration input signal

1 2
2 Rt

1
ess = − rad.
6

The negative sign indicates that the output is higher than the input at steady state.

Example 2.26
Consider a unity negative feedback system with the following open loop T.F.

K
G(s) =
s(s + 1)(s + 4)

The steady state error ess ≤ 2 rad for a velocity input of 2 rad/s. Find K v and hence
K.

Solution
K
G(s) =
s(s + 1)(s + 4)
sK K
K v = lim sG(s) = lim =
s→0 s(s + 1)(s + 4)
s→0 4
R 2×4 8
ess = = =
Kv K K
8
2=
K

K =4
2.5 Steady State Error 237

2.5.2.3 Steady State Error Due to Parabolic Input-The Acceleration

Error Constant

Consider the parabolic (acceleration) input signal shown in Fig. 2.29. The signal is
described as r (t) = 21 Rt 2 u(t) or

1
Rt 2 t ≥0
r (t) = 2
0 t <0

The Laplace transform of the above signal is

R
R(s) = (2.53)
s3
R
E(s) =
s3 + s 3 G(s)
R
ess = lim s E(s) = lim
s→0 s→0 s 2 + s 2 G(s)
R
=
lims→0 s 2 G(s)

R
ess = (2.54)
Ka

where

K a = lim s 2 G(s) (2.55)

s→0

K a is called acceleration or parabolic error constant and has the unit of sec−2 . Con-
sider the system with the following T.F.

K (1 + sz 1 )(1 + sz 2 ) · · ·
G(s) =
s n (1 + sp1 )(1 + sp2 ) · · ·

For Type 0 system,

K a = lim s 2 G(s)
s→0
s 2 K (1 + sz 1 ) · · ·
= lim =0
s→0 (1 + sp1 ) · · ·
R
ess = =∞
0
238 2 Time Response Analysis

For Type 1 system,

K a = lim s 2 G(s)
s→0
s 2 K (1 + sz 1 ) · · ·
= lim =0
s→0 s(1 + sp1 ) · · ·
ess = ∞

For Type 2 system,

K a = lim s 2 G(s)
s→0
s 2 K (1 + sz 1 ) · · ·
= lim =K
s→0 s 2 (1 + sp1 ) · · ·
R
ess =
K
Thus, for parabolic input,

Type 0 system, ess = ∞

Type 1 system, ess = ∞
R
Type 2 system, ess =
Ka
For Type 3 or higher Type system, ess = 0

Example 2.27
A certain control system has the following T.F.

1(s + 2)
G(s) =
s 2 (2s + 3)(s + 5)

The input applied to the system is

3 2
r (t) = t u(t)
2
Find the steady state error.
2.5 Steady State Error 239

Solution Given
3 2
r (t) =t
2
3
R(s) = 3
s
s 2 (s + 2) 2
K a = lim s 2 G(s) = lim =
s→0 s→0 s 2 (2s + 3)(s + 5) 15
R
ess =
Ka

where R = 3
3 × 15
ess =
2

45
ess =
2

By using the static error constants K p , K v and K a , the steady state error can be
obtained as described above. When these signals are simultaneously applied to the
system, the steady state error can be derived for the individual signals and using
superposition theorem, the total steady state error can be determined. Further, for
any other signals, in a similar way, the steady state error can be obtained from R(s)
and G(s) using final value theorem. It is to be noted here that the procedure described
above is valid under the following conditions:
(1) The method is applicable to system configuration shown in Fig. 2.26 where the
feedback is unity. For non-unity feedback system it is not valid.
(2) The error analysis is determined using final value theorem. Hence, it is important
to check whether s E(s) has any pole on the jω axis or in the right half s-plane.
In such cases, the system is unstable and the use of final value theorem is not
valid.
(3) The error constants are defined with respect to the forward path T.F. G(s) with
unity feedback. If the system is non-unity feedback, then it is necessary to convert
it to an equivalent system whose forward path T.F. is G e (s) and its corresponding
feedback path has unity feedback. For this error constants are found out.
The summary of error analysis showing the error relations among the error constants,
the Types of systems and the inputs are given in Table 2.1.

Example 2.28
A unity feedback system is characterised by the open loop T.F. as

1
G(s) =
s(1 + 0.5s)(1 + 0.2s)
240 2 Time Response Analysis

Table 2.1 Summary of the steady state errors due to step, ramp and parabolic inputs for unity
feedback systems
Type of Error constants Steady state error
system
Kp Kv Ka Step input Ramp input Parabolic
input
     
R R R
1 + Kp Kv Ka
R
0 K 0 0 ∞ ∞
1+K
R
1 ∞ K 0 0 ∞
K
R
2 ∞ ∞ K 0 0
K
3 ∞ ∞ ∞ 0 0 0

Determine the steady state error for unit step, unit ramp and unit acceleration.

Solution
1
G(s) =
s(1 + 0.5s)(1 + 0.2s)
1
K p = lim G(s) = lim
s→0 s→0 s(1 + 0.5s)(1 + 0.2s)
Kp = ∞
s
K v = lim sG(s) = lim =1
s→0 s(1 + 0.5s)(1 + 0.2s)
s→0

s2
K a = lim s 2 G(s) = lim =0
s→0 s→0 s(1 + 0.5s)(1 + 0.2s)

The steady state error for unit step input is

1 1
ess = = =0
1 + Kp 1+∞

The steady state error for unit velocity input is

1 1
ess = = =1
Kv 1

The steady state error for unit acceleration input is

1 1
ess = = =∞
Ka 0
2.5 Steady State Error 241

The steady state error when all the above three inputs are simultaneously applied is
given as

ess = 0 + 1 + ∞ = ∞

ess = ∞

Example 2.29
Find the steady state error for a system with open loop T.F.

10(s + 2)(s + 3)
G(s) =
s(s + 1)(s + 5)(s + 4)
H (s) = 1

when the input is r (t) = 3 + t + t 2 .

Solution
10(s + 2)(s + 3)
K p = lim =∞
s→0 s(s + 1)(s + 5)(s + 4)
s10(s + 2)(s + 3)
Kv = lim =3
s→0 s(s + 1)(s + 5)(s + 4)

s 2 10(s + 2)(s + 3)
Ka = lim =0
s→0 s(s + 1)(s + 5)(s + 4)
3 1 2
ess = + +
1 + Kp Kv Ka
3 1 2
= + +
1+∞ 3 0
1
= 0+ +∞
3

ess = ∞

Example 2.30
Consider a unity feedback system with a closed loop T.F.

C(s) (K s + B)
= 2
R(s) (s + As + B)
242 2 Time Response Analysis

Determine the open loop T.F. G(s). Show that the steady state error with unit ramp
input is given by ess = (A − K )/B.

(Anna University, December, 2009)

Solution
C(s)
= T (s)
R(s)
G(s)
=
1 + G(s)
T (s) + T G(s) = G(s)
T (s)
G(s) =
1 − T (s)
(K s + B)
= s + As + B
2
(K s + B)
1− 2
(s + As + B)
(K s + B)
= 2
s + As + B − K s − B
Ks + B
=
s(s + A − K )
K v = lim sG(s)
s→0
s(K s + B)
= lim
s(s + A − K )
s→0
B
=
A−K

The steady state error for unit ramp input is

1
ess =
Kv

(A − K )
ess =
B

Example 2.31
For a unity feedback system having open loop T.F. as

10(s + 2)
G(s) =
s 2 (s 2
+ 2s + 12)
2.5 Steady State Error 243

determine the Type of the system, error constant, and the steady state error for
parabolic input.

Solution
10(s + 2)
G(s) =
s 2 (s 2
+ 2s + 12)

Since there are two pure integrators (number of poles at the origin is two), the system
is Type 2. For Type 2 system, K p = ∞ and K v = ∞. For parabolic input,

s 2 10(s + 2) 5
K a = lim =
s→0 s 2 (s 2 + 2s + 12) 3
R 3
ess = =
Ka 5

Type of the system = 2

K p = ∞; K v = ∞
5 3
K a = ; ess =
3 5

Example 2.32
A certain unity feedback system when subjected to unit ramp input gives the fol-
lowing output response with zero initial conditions.

c(t) = t − 1 + e−t cos t

Find out the steady state error. Determine the forward path T.F. G(s) of the system.

Solution

c(t) = t − 1 + e−t cos t

r (t) = t
e(t) = r (t) − c(t) = t − t + 1 − e−t cos t
ess = lim e(t)
t→∞

ess = 1
244 2 Time Response Analysis

Fig. 2.30 Block diagram of

non-unity feedback system R(s) C(s)
G(s)
Ea(s)

H(s)

c(t) = t − 1 + e−t cos t

Taking Laplace transform on both sides we get

1 1 (s + 1)
C(s) = − +
s2 s (s + 1)2 + 1
(s 2 + 2s + 2) − s(s 2 + 2s + 2) + s 2 (s + 1)
=
s 2 (s 2 + 2s + 2)
2
= 2 2
s (s + 2s + 2)

For unit ramp input R(s) = 1/s 2

C(s) 2s 2 2
= T (s) = 2 2 = 2
R(s) s (s + 2s + 2) (s + 2s + 2)
2
T (s) (s 2 +2s+2)
G(s) = =
1 − T (s) 1 − s 2 +2s+2
2

2
G(s) =
s(s + 2)

2.5.3 Steady State Error for Non-unity Feedback Systems

Consider the non-unity feedback control system shown in Fig. 2.30 where G(s) is
the forward path T.F. and H (s) is the T.F. in the feedback path. Here the error is
not the difference between the input and the output. The signal at the output of the
summer is denoted as E a (s) which is the actuating signal which is different from
the error signal E(s) which was discussed in the previous section to determine the
steady state error. Error signal is different from actuating signal and therefore E a (s)
2.5 Steady State Error 245

(a)
R(s) Ea(s) C(s)
G(s)

H(s)

(b)
R(s) Ea(s) C(s)
G(s)

H(s) 1

(c)
R(s) E(s) G(s) C(s)
Ge(s) =
[1 GH(s) G(s)]

Fig. 2.31 Conversion of non-unity feedback system to unity feedback system

cannot be used in place of E(s) in estimating the steady state error. Figure 2.30 is
reduced to its equivalent as shown in Fig. 2.31.
The successive block diagram reduction shows that the equivalent forward path
T.F. G e (s) as

G(s)
G e (s) = (2.56)
[1 + G H (s) − G(s)]

For a non-unity feedback control system, Eq. (2.56) is used to determine the error
constants and the steady state error. In this case the error is

R(s)
E(s) =
1 + G e (s)
246 2 Time Response Analysis

Fig. 2.32 Block diagram for

Example 33 R(s) Ea(s) 5 C(s)
s(s 2)

1
(s 3)

Substituting for G e (s) in the above equation we get

R(s)
E(s) =
1+ G(s)
[1+G H (s)−G(s)]

R(s)[1 + G H (s) − G(s)]

E(s) = (2.57)
1 + G H (s)

Alternatively if the closed loop T.F. T (s) = C(s)/R(s) is known, G e (s) is expressed
as
T (s)
G e (s) =
1 − T (s)
E(s) = R(s)[1 − T (s)] (2.58)

Using Eq. (2.58) the static error constants can be calculated and the steady state
error determined. For any type of signal, the steady state error is calculated using
Eq. (2.58) for the non-unity feedback system. Here the error is defined as

e(t) = r (t) − c(t) (2.59)

Calculation of steady state error using

R(s)
E(s) =
1 + G H (s)

will give erroneous result. This gives steady state actuating signal.

Example 2.33
For the system shown in Fig. 2.32 find the system Type and steady state error for step
size of 2 units. Also determine the steady state value of the actuating signal ea (∞).
Assume input and output values are same.
2.5 Steady State Error 247

Solution The equivalent forward path T.F. is obtained using Eq. (2.56).

5
G(s) =
s(s + 2)
1
H (s) =
(s + 3)
5
G(s) s(s + 2)
G e (s) = =
1 + G(s)H (s) − G(s) 5 5
1+ −
s(s + 2)(s + 3) s(s + 2)
5(s + 3)
=
(s 3 + 5s 2 + s − 10)

There is no pole of G e (s) in the origin. Hence, the system is Type 0 system and
not Type 1 as seen at the first sight of Fig. 2.32. By applying appropriate stability
conditions, it can be proved that the system is stable. Hence, final value theorem is
applied to find the error constant.

5(s + 3)
K p = lim G e (s) = lim = −1.5
s→0 s→0 s 3 + 5s 2 + s − 10
R 2
ess = = = −4
1 + Kp 1 − 1.5

The negative value of ess implies that the output is larger than the input. The error
constant for the actuating signal is

5
K pa = lim G H (s) = lim =∞
s→0 s→0 s(s + 2)(s + 3)

The steady state actuating signal is

R 2
eass = = =0
1 + K pa 1+∞

Type 0 system
ess = − 4
eass = 0
248 2 Time Response Analysis

D(s)

R(s) E(s) C(s)

G1(s) G2(s)

Controller Plant

Fig. 2.33 Feedback system with disturbance

D(s)

R(s) E(s) 10 5 C(s)

(s 2) (s 3)(s 6)

Fig. 2.34 Block diagram for Example 2.34

2.5.4 Steady State Error for Disturbances

Consider the system shown in Fig. 2.33 where G 1 (s) and G 2 (s) are controller and
plant transfer functions respectively. The disturbance D(s) occurs between the plant
and the controller. From Fig. 2.33, the following equations are written:

C(s) = E(s)G 1 (s)G 2 (s) + D(s)G 2 (s)

But C(s) = R(s) − E(s)

R(s) − E(s) = E(s)G 1 (s)G 2 (s) + D(s)G 2 (s)

R(s) D(s)G 2 (s)
E(s) = − (2.60)
[1 + G 1 G 2 (s)] [1 + G 1 G 2 (s)]

Example 2.34
Consider the system shown in Fig. 2.34. Find the steady state error due to the input
r (t) = 3u(t) and the step disturbance d(t) = 2u(t).
2.5 Steady State Error 249

Solution
10
G 1 (s) =
(s + 2)
5
G 2 (s) =
(s + 3)(s + 6)

where
3 2
R(s) = ; D(s) =
s s
Equation (2.60) can be written as

E(s) = E 1 (s) + E 2 (s)

where
R(s)
E 1 (s) =
1 + G 1 G 2 (s)
D(s)G 2 (s)
E 2 (s) = −
1 + G 1 G 2 (s)
3
E 1 (s) =  
50
s 1+
(s + 2)(s + 3)(s + 6)
3(s + 2)(s + 3)(s + 6)
=
s(s 3 + 11s 2 + 36s + 86)
ess1 = lim s E 1 (s)
s→0
3×2×3×6
= = 1.26
86
−2 × 5(s + 2)
E 2 (s) =
s[s 3 + 11s 2 + 36s + 86]
ess2 = lim s E 2 (s)
s→0
−2 × 5 × 2
=
86
= −0.23
ess = ess1 + ess2
= 1.26 − 0.23

ess = 1.03
250 2 Time Response Analysis

Fig. 2.35 System R(s) C(s)

configuration to determine G(s)
the generalized error E(s)
coefficients

2.5.5 The Generalized Error Constants

The steady state errors estimated in Sect. 2.5.3 using static error constants have the
following limitations:
1. The steady state errors are estimated using static error constants K p , K v and K a
when the input signals applied are step, velocity and parabolic. Such constants
are not available for other type of signals. The steady state errors, in these cases to
be estimated from E(s) by applying final value theorem, require system stability
as the pre requisite.
2. The steady state error estimated using static error constants gives the value of
ess as finite or zero or infinite. No information is known about the accumulation
of error with respect to time.
To overcome the above two limitations, generalized error constants or dynamic error
constants are used. As stated in Sect. 2.5.3, the error is a function of the input and
the system configuration G(s) and H (s). A set of coefficients are derived which
are constants but depends upon G(s) and H (s). However, the input is treated as
a function of time. The inputs and their derivatives together with the generalized
error constants give the error as a function of time and at any time the error can be
estimated including t = ∞ which gives the steady state error ess . For establishing
dynamic error coefficients the system configuration shown in Fig. 2.35 is used.
From Fig. 2.35, the error E(s) is given as

R(s)
E(s) = (2.61)
(1 + G(s))
= R(s)F(s) (2.62)

where
1
F(s) = (2.63)
1 + G(s)

From the Laplace transform property of convolution, the following equation is writ-
ten:

e(t) = r (t) ∗ f (t) (2.64)

2.5 Steady State Error 251

where e(t), r (t) and f (t) are the inverse Laplace transform of E(s), R(s) and F(s)
respectively.
Equation (2.64) can be written as
 t
e(t) = f (τ )r (t − τ )dτ (2.65)
0

But r (t − τ ) can be expanded in Taylor’s series as

τ2 τ 3 ...
r (t − τ ) = r (t) − τ ṙ (t) + r̈ (t) − r (t) + · · · (2.66)
∠2 ∠3
where ṙ (t), r̈ (t) etc., are the first, second derivatives of r (t).
Substituting Equation (2.66) in Eq. (2.65), we get
  
t
τ2
e(t) = f (τ ) r (t) − τ ṙ (t) + r̈ (t) · · · dτ
0 ∠2
 t  t
= f (τ )r (t)dτ − τ ṙ (t) f (τ )dτ + · · ·
0 o

In the above integrations r (t), ṙ (t), r̈ (t) etc. are constants when integration is done
with respect to τ and they are taken out. Hence
 t  t 
r̈ (t) t 2
e(t) = r (t) f (τ )dτ − ṙ (t) τ f (τ )dτ + τ f (τ )dτ + (2.67)
∠2 0
 0 0

C2 Cn
= C0 r (t) + C1ṙ (t) + r̈ (t) + · · · + nṙ (2.68)
∠2 ∠n

where
 t
C0 = f (τ )dτ
0
 t
C1 = − f (τ )dτ
0
 t
C2 = τ 2 f (τ )dτ
0
 t
Cn = (−1)n τ n f (τ )dτ (2.69)
0

The constants C0 , C1 , C2 , . . . , Cn in Eq. (2.68) are called generalized or dynamic

error coefficients and are defined in Eq. (2.69). However, determination of these
constants using the integrals given in Eq. (2.69) is somewhat tedious. Hence, the
procedure is simplified as follows. Since we are interested in the steady state error,
the upper limit of integration in Eq. (2.69) becomes t = ∞. Thus, for example,
252 2 Time Response Analysis
 ∞
C0 = f (τ )dτ
0

Now consider the Laplace transform of F(s) which is given as

 ∞
F(s) = f (τ )e−sτ dτ (2.70)
0

Taking lims→0 on both sides, we get

 ∞
lim F(s) = lim f (τ )e−sτ dτ
s→0 s→0 0
 ∞
= f (τ )dτ
0
= C0
C0 = lim F(s)
s→0

Differentiating Equation (2.70), we get

 ∞
d F(s)
=− f (τ )τ e−sτ dτ
ds
 ∞
0
d F(s)
lim =− τ f (τ )dτ
s→0 ds 0
= C1
d F(s)
C1 = lim
s→0 ds

The procedure is repeated, and we get

d 2 F(s)
C2 = lim
s→0 ds 2
d n F(s)
Cn = lim
s→0 ds n

Thus, the generalized error constants are estimated from

C0 = lim F(s)
s→0
d F(s)
C1 = lim
ds
s→0
d 2 F(s)
C2 = lim
s→0 ds 2
n
d F(s)
Cn = lim (2.71)
s→0 ds n
2.5 Steady State Error 253

where
1
F(s) =
1 + G(s)

Example 2.35
The forward path T.F. of a certain unity negative feedback control system is given
as
5
G(s) =
(s + 5)

Calculate the generalized error co-efficients and the steady state error for the input
r (t) = 6 + 5t.

(Anna University, May, 2007)

Solution

r (t) = 6 + 5t
ṙ (t) = 5
r̈ (t) = 0

For the above input, since r̈ (t) is zero, the error series is

e(t) = C0 r (t) + C1ṙ (t)

1
F(s) =
1 + G(s)
1
=
1 + s+55

(s + 5)
=
(s + 10)
C0 = lim F(s)
s→0
(s + 5)
= lim
s→0 (s + 10)
= 0.5
d F(s) (s + 10) − (s + 5)
=
ds (s + 10)2
d F(s) 5
=
ds (s + 10)2
d F(s)
C1 = lim
s→0 ds
254 2 Time Response Analysis

Fig. 2.36 Block diagram for

Example 36 R(s) 2 C(s)
(s 1)(s 2)
Ea(s)

(s 3)

5
= lim
s→0 (s + 10)2
1
C1 =
20
5
e(t) = 0.5(6 + 5t) +
20
13
= + 2.5t
4
ess = lim e(t)
t→∞
13
= lim + 2.5t
t→∞ 4
=∞

ess = ∞

Example 2.36
Consider the system shown in Fig. 2.36. The forward path T.F.

2
G(s) =
(s + 1)(s + 2)

The feedback path T.F. H (s) = (s + 3). The input applied is r (t) = 2 + 3t. Deter-
mine the generalized error constants and hence the steady state error e(t) = r (t) −
c(t).

Solution
2
G(s) =
(s + 1)(s + 2)
H (s) = (s + 3)
2.5 Steady State Error 255

The system is not a unity feedback system. Hence, using Eq. (2.56), the equivalent
forward path T.F. is obtained as

G(s)
G e (s) =
[1 + G H (s) − G(s)]
2
(s + 1)(s + 2)
=  
2(s + 3) 2
1+ −
(s + 1)(s + 2) (s + 1)(s + 2)
2
G eq = 2
s + 5s + 6

The block diagram representation of the equivalence of Fig. 2.36 is shown in Fig. 2.37.
From Fig. 2.37,

1
F(s) =
1 + G e (s)
2
=
2
1+ 2
s + 5s + 6
2(s 2 + 5s + 6)
= 2
(s + 5s + 8)

The input

r (t) = 2 + 3t
ṙ (t) = 3
r̈ (t) = 0
e(t) = C0 r (t) + C1ṙ (t)

2(s 2 + 5s + 6) 3
C0 = lim =
s→0 (s + 5s + 8)
2 2
d F(s) (s 2 + 5s + 8)(4s + 10) − 2(s 2 + 5s + 6)(2s + 5)
=
ds (s 2 + 5s + 8)2
d F(s)
C1 = lim
s→0 ds
8 × 10 − 2 × 6 × 5
=
64
80 − 60 5
= =
64 16
256 2 Time Response Analysis

Fig. 2.37 Equivalent block

diagram of Fig. 2.36 R(s) E(s) 2 C(s)
Ge(s) = 2
s 5s 6

3 5
e(t) = (2 + 3t) + ×3
2 16
e(t) = 3.9375 + 4.5t
ess = lim e(t)
t→0

ess = ∞

Example 2.37
A certain unity negative feedback control system has the following forward path
T.F.
K (s + a)
G(s) =
(s + 2)(s + b)

The system is subjected

√ to step input of size 5. It is desired to have the closed loop
poles at s = −2 ± j 12. Determine the value of K , a and b if the permissible steady
state error ess = 1.

Solution

K (s + a)
G(s) =
(s + 2)(s + b)
Ka
K p = lim G(s) =
s→0 2b
R 5
ess = =
1 + Kp 1 + (K a/2b)
10b
1=
2b + K a
2b + K a = 10b
K a = 8b (a)
2.5 Steady State Error 257

For the given G(s), the closed loop T.F. is

K (s + a)
C(s) (s + 2)(s + b)
=
R(s) K (s + a)
1+
(s + 2)(s + b)
K (s + a)
= (b)
(s + 2)(s + b) + K (s + a)

The characteristics equation of Eq. (b) is

s 2 + (2 + b + K )s + (2b + K a) = 0 (c)
√ √
The desired roots are s1 = −2 + j 12 and s2 = −2 − j 12. The desired charac-
teristics equation is
√ √
(s + 2 + j 12)(s + 2 − j 12) = 0
s 2 + 4s + 16 = 0 (d)

Comparing Eqs. (c) and (d), we get

(2 + b + K ) = 4
b =2−K

and

2b + K a = 16 (e)

Substituting K a = 8b from Eq. (a), we get

2b + 8b = 16
10b = 16
b = 1.6
K = 2 − b = 2 − 1.6 = 0.4
8b
a=
K
8 × 1.6
= = 32
0.4

K = 0.4
a = 32
b = 1.6
258 2 Time Response Analysis

R(s) 25 C(s)
s(s 5)
Ea(s)

(1 as)

Fig. 2.38 A non-unity feedback system for Example 38

Example 2.38
Consider the non-unity feedback system shown in Fig. 2.38. Determine the value of
a so that the peak overshoot for step input is 5%. For this value of a determine the
type of the system and the steady state error for a velocity input of slope 4. Verify
the results using dynamic error constants.

e(t) = r (t) − c(t)

Solution
25
G(s) =
s(s + 5)
H (s) = (1 + as)
C(s) G(s)
=
R(s) 1 + G H (s)
25/(s(s + 5))
=  
25(1 + as)
1+
s(s + 5)
25
=
s + (5 + 25a)s + 25
2

The characteristic equation is

s 2 + (5 + 25a)s + 25 = 0

The required overshoot is related as

√−ζ π
e 1−ζ 2 = 0.05

Taking loge on both sides, we get

2.5 Steady State Error 259

−ζ π
 = loge 0.05 = −3
1 − ζ2

ζ = 0.69

From the characteristic equation,

√
ωn = 25 = 5
2ζ ωn = 5 + 25a
2ζ ωn − 5
a=
25

a = 0.076

Thus
25
G(s) =
s(s + 5)
H (s) = (1 + 0.076s)

Using Eq. (2.56), we get

G(s)
G e (s) =
1 + G H (s) − G(s)

25(1 + 0.076s)
G H (s) =
s(s + 5)
25
s(s+5)
G e (s) =
1+ 25(1+0.076s)
s(s+5)
− 25
s(s+5)
25
=
+ 5s + 1.9s + 25 − 25
s2
25
=
s(s + 6.9)

This is Type 1 system since there is one integrator in the open loop (forward path)
T.F. For ramp input,
260 2 Time Response Analysis

K v = lim sG e (s)
s→0
s × 25
= lim
s→0 s(s + 6.9)
25
=
6.9
Steady state error for ramp input of slope 4,

4 4 × 6.9
ess = =
Kv 25

ess = 1.1 rad/s

Verification using Dynamic Error Constants

r (t) = 4t
ṙ (t) = 4

Hence, the dynamic error constants C0 and C1 are to be determined.

1
F(s) =
1 + G e (s)
1
=
1 + s(s+6.9)
25

s(s + 6.9)
=
+ 6.9s + 25
s2
C0 = lim F(s)
s→0
(s 2 + 6.9s)
= lim =0
s→0 s 2 + 6.9s + 25

d F(s) (s 2 + 6.9s + 25)(2s + 6.9) − (s 2 + 6.9s)(2s + 6.9)

=
ds (s 2 + 6.9s + 25)2
d F(s)
C1 = lim
s→0 ds
25 × 6.9 6.9
= =
(25)2 25
e(t) = C1ṙ (t)
4 × 6.9
=
25
2.5 Steady State Error 261

d(t) = 2

E(s) 2(s 2) 3 C(s)

(s 6) s2
r(t) = 0

3 C(s)
d(t) = 2 s2

2 (s 2)
(s 6)

Fig. 2.39 a Block diagram for Example 39. b Reduced block diagram of Fig. 2.39a

ess = 1.1

Thus, the result is verified using dynamic error constants.

Example 2.39
Consider the system shown in Fig. 2.39a. Determine the steady state error for r (t) =
0 and d(t) = 2.

Solution From Fig. 2.39a when r (t) = 0, E(s) = −C(s). The block diagram is
redrawn as shown in Fig. 2.39b.
From Fig. 2.39b, C(s)/D(s) can be written as

C(s) 3/s 2 3(s + 6)

= = 2
D(s) 1 + s 2 (s+6)
3 2(s+2) [s (s + 6) + 6(s + 2)]
2
D(s) =
s
2 × 3(s + 6)
C(s) =
s[s 2 (s + 6) + 6(s + 2)]

The steady value of the output is obtained by using final value theorem
262 2 Time Response Analysis

Fig. 2.40 System (a)

representation. a With closed R(s) C(s)
loop T.F. T (s); and b Closed T(s)
loop system with forward
path T.F
(b)

R(s) E(s) C(s)

Ge(s)

6s(s + 6)
Css = lim sC(s) = lim =3
s→0 s→0 s[s 2 (s + 6) + 6(s + 2)]
ess = −Css

ess = −3

2.5.6 Steady State Error When Closed Loop T.F. is Given

Consider the system shown in Fig. 2.40a and b. From these figures, the following
equations are written:

C(s)
= T (s)
R(s)
C(s) = R(s)T (s)
E(s) = R(s) − C(s) = R(s) − R(s)T (s) (2.72)

E(s) = [1 − T (s)]R(s) (2.73)

From Eq. (2.73) steady state error can be determined using final value theorem
provided the system is stable. Alternatively G e (s) from Eq. (2.73) is obtained as

T (s)
G e (s) = (2.74)
1 − T (s)
2.5 Steady State Error 263

Using Eq. (2.74) error constants and hence ess can be determined.

Example 2.40
The closed loop T.F. of a certain unity feedback control system is given as

5
T (s) =
s 2 + 3s + 5

Find the steady state error for a ramp input of slope 5 rad/s.

Solution
5
T (s) =
+ 3s + 5
s2
(s 2 + 3s)
1 − T (s) = 2
s + 3s + 5
5 [s 2 + 3s]
E(s) = [1 − T (s)]R(s) =
s 2 [s 2 + 3s + 5]
5s 2 [s + 3]
ess = lim s E(s) = lim 2 2
s→0 s→0 s (s + 3s + 5)

ess = 3

Alternatively, from Eq. (2.74), G e (s) determined.

T (s) 5/((s 2 + 3s + 5))

G e (s) = = 2
1 − T (s) (s + 3s)/(s 2 + 3s + 5)
5
=
s(s + 3)
5 5
K v = lim sG e (s) = lim =
s→0 s→0 s + 3 3
R 5
ess = = ×3
Kv 5

ess = 3
264 2 Time Response Analysis

Fig. 2.41 Block diagram for

Example 41 R(s) K C(s)
G(s) =
(s 3)

4
H(s) =
(s 5)

Example 2.41
Consider the system shown in Fig. 2.41. Find the value of K so that the steady state
error is zero for step input. Find the type of the system. For K = 15, find ess for ramp
input. e(t) = r (t) − c(t).

Solution
K
G(s) =
(s + 3)
4
H (s) =
(s + 5)
4K
G H (s) =
(s + 3)(s + 5)
G(s)
G e (s) =
1 + G H (s) − G(s)
K /s + 3
G e (s) =
1 + (s+3)(s+5)
4K
− s+3
K

K (s + 5)
=
s2 + 8s + 15 + 4K − K s − 5K

For a step input, the position error constant is

K ×5
K p = lim G e (s) =
s→0 15 − K
R R
ess = =
1 + Kp 1 + 15−K
5K

R[15 − K ]
=
15 + 4K

For ess to be zero K = 15

K = 15
2.5 Steady State Error 265

Fig. 2.42 Block diagram for

Example 42 R(s) K C(s)
G(s) =
s(s 2)

H(s) = (1 5s)

For K = 15,

15(s + 5)
G e (s) =
s(s − 7)

This is Type 1 system,

−75
K v = lim G e (s) =
s→0 7
1
ess =
Kv

−7
ess =
75

Example 2.42
Consider the system shown in Fig. 2.42. Determine the value of K so that the steady
state error is 6, for a ramp input of slope 1. Use e(t) = r (t) − c(t).

Solution From Fig. 2.42, the closed loop T.F. is

C(s) G(s)
T (s) = =
R(s) 1 + G H (s)
K
s(s+2)
= (1+5s)
1 + Ks(s+2)
K K
= = 2
s(s + 2) + 5K s + K s + (2 + 5K )s + K

From Eq. (2.73), the error function is written as

E(s) = R(s)[1 − T (s)]

 
1 K
= 2 1− 2
s [s + (2 + 5K )s + K ]
266 2 Time Response Analysis

 
1 s 2 + (2 + 5K )s
E(s) =
s 2 [s 2 + (2 + 5K )s + K ]
ess = lim s E(s)
s→0
[s + (2 + 5K )] 2 + 5K
ess = lim = =6
s→0 [s 2 + (2 + 5K )s + K ] K
2 + 5K = 6K

K =2

Alternative Method
K
G(s) =
s(s + 2)
H (s) = (1 + 5s)
K (1 + 5s)
G H (s) =
s(s + 2)
G(s)
G e (s) =
1 + G H (s) − G(s)
K
s(s+2)
= K (1+5s)
1+ s(s+2)
− s(s+2)
K

K
=
+ 2s + 5K s
s2
K v = lim sG e (s)
s→0
K
=
2 + 5K
R
ess = where R = 1
Kv
2 + 5K
6=
K
6K − 5K = 2

K =2
2.5 Steady State Error 267

(a)

R(s) 3 2 C(s)
(s 2) (s 5)

5
s
(b)
R(s) 3 2 C(s)
(s 2) (s 7)

5
s

(c)
R(s) 6 C(s)
(s 2)(s 7)

5
s

Fig. 2.43 a Block diagram for Example 2.43. Reduced block diagram for Fig. 2.43a

Example 2.43
For the system shown in Fig. 2.43a, find the steady state error for unit step input.
e(t) = r (t) − c(t).

Solution The given block diagram is shown in Fig. 2.43a, and T (s) = C(s)/R(s)
is obtained. The successive block diagram reduction is shown in Fig. 2.43b and c.
From Fig. 2.43c, the closed loop T.F. is obtained as

C(s)
T (s) =
R(s)
6/[(s + 2)(s + 7)] 6s
= = 3
1 + [30/(s(s + 2)(s + 7))] s + 9s + 14s + 30
2

T (s) is strictly proper and stable and hence Laplace final value theorem can be
applied.
268 2 Time Response Analysis

E(s) 2 C(s)
R(s) G(s) = n
s(s 2 n)

Kt s

Fig. 2.44 Rate or tachometer feedback controller

E(s) = [1 − T (s)]R(s)
 3 
s + 9s 2 + 8s + 30 1
= 3
s + 9s 2 + 14s + 30 s

ess = lim s E(s)

s→0

ess = 1

2.6 Performance Enhancement By Using Controllers

The transient and steady state responses of a second order system for different types of
inputs were discussed in the previous sections. Formulae were derived for transient
and steady state performance specifications for typical test signals such as step,
velocity and acceleration. However, for a given system, since the damping ratio and
natural frequency of oscillations are fixed, the only option to enhance the performance
is by adjusting the open loop gain which is called the D.C. gain. By adjusting the
D.C. gain, only very little improvement can be obtained in the transient and steady
state performance. A viable option is to provide a controller either in the forward
path or in the feedback path. Each method has its own advantages and limitations.
The commonly used controllers are
1. Rate controller or tachometer feedback controller,
2. PID controller.
These controllers? performances are discussed below when they are connected to a
second order system.
2.6 Performance Enhancement By Using Controllers 269

c(t)

Kt 0
Kt 0.1
1

Kt 0.3

Fig. 2.45 Unit step response curve for different values of error rate damping

2.6.1 Rate Controller or Tachometer Feedback Controller

A second order system whose open loop T.F.

ωn2
G(s) =
s(s + 2ζ ωn )

is taken for the study of transient and steady state performances using controllers.
The second order system with rate feedback controller is shown in Fig. 2.44. Here, the
derivative of the output signal is fed back and added to the actuating signal in the miner
feedback loop. This rate feedback is also called tachometer feedback or velocity
feedback. For example, in a mechanical system, if the output is mechanical position,
it is differentiated using a tachometer which produces voltage that is proportional to
the velocity of the mechanical system. The proportionality constant K t has the units
of volts/rad/s or volts/metre/s.
From Fig. 2.44, the following equation is obtained.

C(s) ωn2
= 2 (2.75)
R(s) [s + (2ζ ωn + K t ωn2 )s + ωn2 ]

The characteristic equation of the closed loop system is

s 2 + (2ζ ωn + K t ωn2 )s + ωn2 = 0 (2.76)

From Eq. (2.76) the damping ratio of the system is

Kt
ζ =ζ+ ωn (2.77)
2
270 2 Time Response Analysis

R(s) E(s) K C(s)

s(s 2)

Kt s

Fig. 2.46 Block diagram for Example 2.44

From Eq. (2.77) it is evident that the effect of the tachometer feedback is to increase
the damping of the system by (K t /2)ωn . Since ωn is fixed for the system, by varying
K t the effective damping can be varied for a wide range. Further from Fig. 2.44 the
error function is derived as

R(s)s[s + 2ζ ωn + K t ωn2 ]
E(s) = (2.78)
s2+ (2ζ ωn + K t ωn2 )s + ωn2

From Eq. (2.78), the steady state error for step input is zero. However, for the ramp
input, the steady state error is

R
ess = [2ζ + K t ωn ] (2.79)
ωn

Without rate feedback the steady error is 2ζ R/ωn . Hence, by rate feedback the steady
state error is increased by R K t . Thus, by rate feedback, the effective damping can
be improved but it enlarges the steady state error for ramp input. The time
response curves for unit step input with ζ = 0.5 and ωn = 4 rad/s are shown for
K t = 0, 0.1 and 0.3 in Fig. 2.45.

Example 2.44
A certain second order system with rate feedback in the minor loop is shown in
Fig. 2.46. The overshoot for a step input of height 2 units should not exceed 0.2. The
steady state error for ramp input of slope 4 should be less than 0.5. Find the value of
K and K t . Compare the transient response specifications such as % overshoot, rise
time, settling time for step input and ess for ramp input without rate feedback and
with rate feedback.

Solution The required overshoot of the rate feedback system is

−ζ2 π
2e  = 0.2
1 − ζ22
2.6 Performance Enhancement By Using Controllers 271

where ζ2 is the damping ratio of the system with rate feedback.

−ζ2 π
e = 0.1
1 − ζ22

Taking loge on both sides we get

−ζ2 π
 = −2.303
1 − ζ22
ζ2 = 0.591

Let ζ1 be the damping ratio without rate feedback. Then

C(s) K
= 2
R(s) [s + 2s + K ]
√
From the above equation we get ωn = K and

1
2ζ1 ωn = 2 or ζ1 = √
K

The natural frequency ωn will remain the same for the system with and without rate
feedback. From Eq. (2.79) the steady state error for ramp input of slope 4 is

4
ess = [2ζ1 + K t ωn ]
ωn
 
4 2 √
0.5 = √ √ + Kt K
K K
4
= [2 + K t K ] (a)
K
0.125K = [2 + K t K ]

From Eq. (2.77)

Kt
ζ2 = ζ1 + ωn
2
1 Kt √
0.591 = √ + K
K 2
√
2 × 0.591 K = 2 + K t K
√
1.182 K = 2 + K t K (b)

From Eq. (a) we get

272 2 Time Response Analysis

0.125K = 2 + K t K

Substituting in Eq. (b) we get

√
1.182 K = 0.125K

K = 89.42

From Eq. (b) we get after substituting for K ,

√
1.182 89.42 = 2 + K t 89.42

K t = 0.103

1 1
ζ1 = √ = = 0.106
K 89.42

The closed loop T.F. of the rate feedback system from Eq. (2.75) is

C(s) ωn2
= 2
R(s) [s + (2ζ1 ωn + K t ωn2 )s + ωn2 ]
89.42
= 2
[s + (2 × 0.106 × 9.456 + 0.103 × 89.42)s + 89.42]
89.42
= 2 (c)
[s + 11.2s + 89.42]

The closed loop T.F. of the system without rate feedback is

C(s) 89.42
= 2
R(s) [s + 2s + 89.42]

Time Domain Specifications without Rate Feedback

The steady state error for ramp of slope 4 is
2.6 Performance Enhancement By Using Controllers 273

2ζ1 R
ess1 =
ωn
2 × 0.106 × 4
=
9.456
= 0.09
ζ1 π
For step input, overshoot, M p1 = Re− 
1 − ζ12
0.106π
M p1 = 2e− √
1 − 0.1062
= 1.43

The rise time is calculated from

(π − φ)
tr1 =
ωd

where φ = cos−1 ζ1 = cos−1 0.106 = 1.46 rad


ωd = ωn 1 − ζ12

= 9.456 1 − (0.106)2 = 9.4
(π − 1.46)
tr1 = = 0.179 s
9.4
The settling time is calculated from (for 2% error tolerance).

3.91
ts1 =
ζ ωn
3.91
=
0.106 × 9.456
ts1 = 3.9 s

Time Domain Specifications with Rate Feedback Control

ess2 = 0.5

The peak overshoot,

M p2 = 0.2

The rise time is calculated from

274 2 Time Response Analysis

(π − φ)
tr2 =
ωd
φ = cos−1 ζ2
= cos−1 0.591
φ = 0.938 rad

ωd = ωn 1 − ζ22

= 9.456 1 − (0.591)2
= 7.628
(π − 0.938)
tr2 =
7.628
= 0.29 sec

The settling time for 2% error tolerance is given by

3.91
ts2 =
ζ2 ωn
3.91
=
0.591 × 9.456
= 0.7 s

The performance specifications obtained above with and without rate feedback are
tabulated below.

Performance specifications Without rate With rate

Feedback Feedback
Steady state error, ess 0.09 0.5
Overshoot, M p 1.43 0.2
Rise time, tr 0.179 s 0.29 s
Settling time, ts 3.9 s 0.7 s
Damping ratio, ζ 0.106 0.591
Natural frequency of oscillation, ωn 9.456 9.456

2.6.2 Proportional Plus Integral Plus Derivative (Three Term

Controllers) Controllers

A proportional + Integral + Derivative controller (PID controller) is probably the

most used feedback control and is widely used in industrial control systems. A PID
controller calculates an error value as the difference between a measured process
variable and a desired set point. The controller attempts to minimize the error by
adjusting the process control inputs. For best performance, the PID parameters used
2.6 Performance Enhancement By Using Controllers 275

(a)
Kp
+
e(t) ea(t)
Ki fdt
+
+
Kd d
dt

(b)
Kp
+
E(s) Ki Ea(s)
s +
+
Kd s

(c)
Kp
+
R(s) + E(s) Ea(s) 2
n C(s)
Ki
_ s + s(s+2 n)
+
Kd s

Fig. 2.47 Representation of the PID controller. a Time domain representation; b Frequency domain
representation; and c Connected in cascade with the prototype second order system

in the calculation must be tuned according to the nature of the operating conditions
of the system.
The PID controller calculation (algorithm) involves three parameters and is
accordingly called three term control: the proportional, the integral and deriva-
tive values, denoted by P, I, and D respectively. The proportional value determines
the reaction to the current error, the integral value determines the reaction based on
the sum of recent errors, and the derivative value determines the reaction based on
the rate at which error has been changing. The weighted sum of these three actions
is used to adjust the process via a control element. Thus, P depends on the present
error, I on the accumulation of the past error and D is the prediction of future errors
based on current rate of change.
By tuning the three constants in the PID controller algorithm, the controller can
provide the control action designed for specific process requirements. Even though,
276 2 Time Response Analysis

R(s) + E(s) Ea(s) 2

n C(s)
KKpp
_ s(s+2 n)

Fig. 2.48 Proportional controller connected in cascade with the second order system

by the use of PID controller, it is possible to achieve some or all of the transient and
steady state specifications, it does not guarantee optimal control of the system.
Some applications may require using only one or two modes to provide the appro-
priate system control. This is achieved by settling the gain of undesired control out-
puts to zero. A PID controller will be called a PI, PD, P or I controller in the absence
of the respective control actions. PI controller are fairly common, since the derivative
action is sensitive to noise whereas the absence of the integral value may prevent the
system from reaching the target value due to the control action.
While PID controllers are applicable to many control problems, and often perform
satisfactorily without even tuning according to the changes in the operating condi-
tions, they can perform poorly in some applications. The fundamental difficulty with
PID control is that it is a feedback system with constant parameters and no direct
knowledge of the process.
The PID controller is represented in the time domain as shown in Fig. 2.47a. The
frequency domain representation is shown in Fig. 2.47b. The error signal e(t) is the
input to the controller and its output is ea (t). The output of the controller is connected
to the prototype second order system which is taken up for the performance study of
the PID controller. This is shown in Fig. 2.47c. We study below the performance of
the individual controllers as well as this combination when connected in series with
the standard prototype second order system. For this, the results obtained from the
previous sections are referred to.

2.6.2.1 Proportional Controller

The proportional controller connected in cascade with the standard second order
system is shown in Fig. 2.48. From Fig. 2.48 the following equations are obtained:

C(s) K p ωn2
= 2 (2.80)
R(s) [s + 2ζ ωn s + K p ωn2 ]
E(s) s(s + 2ζ ωn )
= 2 (2.81)
R(s) [s + 2ζ ωn s + K p ωn2 ]

The characteristic equation is

s 2 + 2ζ ωn s + K p ωn2 = 0 (2.82)
2.6 Performance Enhancement By Using Controllers 277

From Eq. (2.82) it is seen that the damping factor of the system with and without
the proportional controller remains the same. However, the natural  frequency of
oscillation of the system with the controller is altered from ωn to K p ωn . The
performance study of the controller with respect to steady state error, peak overshoot,
rise time and setting is discussed and compared with and without the controller as
given below.
1. Steady State Error: The steady state error for the system without controller
(system 1) and with the controller (system 2) for step input is zero. However, for
ramp input ess1 = 2ζ R/ωn and ess2 = 2ζ R/K p ωn . The suffix one refers to the
system without compensator and suffix two refers to system with compensator.
 Overshoot: Since the damping factor is changed by the factor ζ =
2. Peak
ζ / K p the peak overshoot also changes accordingly.

ζ1 π
M p1 = e− √ R
1 − ζ1
−ζ1 π
√
− Kp
M p2 = e  R
1 − ζ12 /K p

The time at which these overshoot occur differs as given below.

π
t p1 = 
ωn 1 − ζ12
π
t p2 =  
K p ωn 1 − ζ22

Thus, for K p > 1, t p2 < t p1 . The system two has faster transient response.
3. Rise Time: The rise time of system 1 is

(π − φ)
tr 1 = 
ωn 1 − ζ 2

The rise time of system 2 is

(π − φ)
tr 2 =  
K p ωn 1 − ζ 2

Thus, tr 2 < tr 1 and the system with the controller is faster. Since rise time is
inversely proportional to the bandwidth, the bandwidth of the system when con-
troller is incorporated is increased as the proportional controller gain K p is
increased.
278 2 Time Response Analysis

4. Settling Time: The settling time for system one is

3.91
ts1 =
ζ1 ωn1

The settling time for system two is

3.91
ts2 =
ζ2 ωn2
3.91 3.91
=  =
K p √ζK ωn1 ζ1 ωn 1
p

ts2 = ts1

The transient response curves of these two systems are shown in Fig. 2.49 for
unit step input. The T.F. of the prototype second order system taken for the study
is
16
G(s) =
s(s + 4)

The natural frequency of the original system is ωn = 4 rad/s. and ζ = 0.5. The
closed loop T.F. of the system without the controller is

C(s) 16
= 2
R(s) s + 4s + 16

The closed loop T.F. with the proportional controller is

C(s) K p 16
= 2
R(s) s + 4s + 16K p

The above T.F. is simulated in the digital computer for K p = 1, 2, 3 and 5 For
K p = 1, the T.F. is that of the prototype second order system without the con-
troller.
The following MATLAB program is used:

MATLAB Program
% · · · · · · Unit-step response · · · · · ·
num = [0 0 16];
den = [1 4 16];
t = 0 : 0.05 : 3;
c = step(num,den,t);
plot(t,c)
title(‘Unit-Step Response’)
2.6 Performance Enhancement By Using Controllers 279

Fig. 2.49 Transient response curves of a second order system for unit step input with proportional
controller (1) K p = 5; (2) K p = 3; (3) K p = 2; (4) K p = 1 (without controller)

Figure 2.49 shows the step response curve. From the step responds curve it is observed
that curve 4 represents the system without the controller. It has an overshoot of 16.3%
and occurs at t p = 0.91 s curves 1, 2 and 3 corresponds to K p = 5; K p = 3 and K p =
2 respectively. For K p = 5 the damping factor ζ = 0.2236; and %M p = 48.6% at
t p = 0.36 s, which is much faster than the system without the controller.
Summary of Proportional Controller
1. The system overshoot is changed. As K p is increased overshoot is also increased.
2. The rise time, settling time and the time at which the peak overshoot occurs
are reduced because of fast transient response. However, the settling time is
unchanged.
3. The bandwidth is increased.
4. The steady state error for step input is zero and unchanged.
 However, the steady
state error for ramp input is reduced by a factor 1/ K p .
280 2 Time Response Analysis

R(s) + E(s) Ea(s) 2 C(s)

K
Kip
n
_ s s(s+2 n)

Fig. 2.50 Second order system with integral controller

2.6.3 Integral or Reset Controller

The integral controller connected in the forward path in cascade with a standard
second order system is shown in Fig. 2.50. For Fig. 2.50 the following equations are
derived.

C(s) K i ωn2
= 3 (2.83)
R(s) [s + 2ζ ωn s 2 + K i ωn2 ]

The characteristic equation is

s 3 + 2ζ ωn s 2 + K i ωn2 = 0 (2.84)

Inspection of the characteristic Eq. (2.84) shows that the coefficient of s is zero
and hence the closed loop system will be unstable with a pure integrator connected
in cascade with the second order system and hence no further analysis is required.
However, its action will be discussed at length when combined with proportional
controller. Integral control can be used to control a second order 0 Type system
which will be stable. Similarly if an ID controller with the T.F.
 
Ki
G(s) = + Kd s
s
1
= [K d s 2 + K i ]
s
when connected in cascade with a prototype second order type 1 system, will have
the closed loop T.F. as

C(s) (K d s 2 + K i )
= 3
R(s) [s + (2ζ ωn + K d )s 2 + K i ]

From the characteristic equation it is observed that the coefficient of s is zero and
hence the system becomes unstable. Therefore, ID controller cannot be used for type
1 and order two system. However, it can be used for zero type order two system.
2.6 Performance Enhancement By Using Controllers 281

R(s) + E(s) Ea(s) 2

n C(s)
KKdsp
_ s(s+2 n)

Fig. 2.51 Second order system with derivative controller

2.6.3.1 Derivative or Rate Controller

Consider the second order system connected to the derivative controller in the forward
path as shown in Fig. 2.51. The following equations are derived for Fig. 2.51. Here
the zero of the controller cancels with the pole of the second order system.

C(s) K d ωn2
= (2.85)
R(s) [s + 2ζ ωn + K d ωn2 ]
E(s) (s + 2ζ ωn )
= (2.86)
R(s) [s + 2ζ ωn + K d ωn2 ]

The characteristic equation of the system with derivative control in the forward path
is given by [s + 2ζ ωn + K d ωn2 ].
From Eq. (2.86) it is evident that the given second order system is reduced to
a first order system and its pole is placed on the negative real axis in the complex
s-plane. The performance study of the controller with respect to steady state error,
peak overshoot, rise time and settling time is discussed and compared as given below.
(1) Steady State Error: Steady state error for system one for step input is zero and
for ramp input ess1 = 2ζ R/ωn . However, ess = (2ζ R)/(2ζ + K d ωn ) for system
with derivative controller for step input. For ramp input ess = ∞.
(2) Peak Overshoot: The peak overshoot for step input for system one is as dis-
cussed in Sect. 2.6.1. However, for system two, since it is a first order system,
the peak overshoot will be zero and the rise time will be tr 2 = 2.2T2 . The time
constants of these two systems are

1
T1 =
ζ ωn
1
T2 =
2ζ ωn + K d ωn2
1
=
(2ζ + K d ωn )ωn
T2 < T1

The rise time for system two is

282 2 Time Response Analysis

tr 2 = 2.2T2

The settling time for system two is

ts2 = 3.91T2
ts2 < Ts1

For comparison study the following numerical values are chosen. ωn = 4 rad/s;
ζ = 0.5; K d = 2.

1
T1 =
ζ ωn
1
= = 0.5
4 × 0.5
1
T2 =
2ζ ωn + K d ωn2
1
=
2 × 0.5 × 4 + 2 × (4)2
= 0.0278
√−ζ π 2
M p1 = e 1−ζ1 × 100
−0.5×π
√
= e4 1−0.52 × 100 = 16.3

π −φ
tr 1 = 
ωn 1 − ζ 2
π − cos−1 0.5
= √
4 1 − 0.52
π − π3
=
4 × 0.866
tr 1 = 0.6 s
tr 2 = 2.2T2 = 2.2 × 0.0278 = 0.061 s
ts1 = 3.91T1 = 3.91 × 0.5 = 2.0 s
ts2 = 3.92T2 = 3.91 × 0.0.278 = 0.11 s

For unit ramp input,

2.6 Performance Enhancement By Using Controllers 283

Fig. 2.52 Transient response curves of a second order system for unit step input with derivative
controller. (1) K d = 1 (without controller); (2) K d = 2; (3) K d = 3; (4) K d = 5

2ζ
ess1 =
ωn
2 × 0.5
= = 0.25
4
ess2 =∞

The performance specifications are shown in the table given below:

Steady state Peak
Performance Time Rise Settling Error for Overshoot
specifications constant T time tr time ts ramp ess Mp
System 1 (without controller) 0.5 0.6 s 2.0 s 0.25 16.3%
System 2 (with controller) 0.0278 0.061 s 0.11 s ∞ 0

The transient response curves for system 1 (without controller) and system 2 (with
derivative controller) are shown in Fig. 2.52 for unit step input for K d = 1, K d =
2, K d = 3 and K d = 5.
Summary of Derivative Controller
The second order system with derivative controller is reduced to a first order system
by pole–zero cancellation. Thus, it has no overshoot when subjected to step input.
The steady state errors due to step input is finite and for ramp input is zero. The time
284 2 Time Response Analysis

R(s) + Ea(s) 2 C(s)

E(s) + n
Kp s(s+2
_ n)
+
Ki
s

Fig. 2.53 Second order system with PI controller

constant, rise time and settling time are much less for the system with derivative
controller.

2.6.4 Proportional Plus Integral (PI) Controller

Figure 2.53 shows the PI controller connected in cascade with a second order system.
The control action provided by the controller is proportional to the error as well as
the time integral part of the error. Thus, it combines the effects of the proportional
controller discussed in Sect. 2.6.1 and integral action discussed in Sect. 2.6.2. The
transfer function of the PI controller is
Ki
G c (s) = K p +
s
s K p + Ki
= (2.87)
s

By connecting a PI controller, a pole at s = 0 and a zero at s = −(K i /K p ) are added

to the open loop T.F. The equivalent open loop T.F. now becomes

(K p s + K i ) ωn2
G e (s) = (2.88)
s s(s + 2ζ ωn )

The closed loop T.F. is obtained as

C(s) ωn2 (K i + K p s)
= 3 (2.89)
R(s) [s + 2ζ ωn s 2 + K p ωn2 s + K i ωn2 ]

From Eqs. (2.88) and (2.89) it is observed that the type of the system is increased by
one to a type 2 system. The order of the system is also increased by one to a third
order system. The increase in the type and order of the system will have the influence
on the steady state error and transient performance respectively. The increase in the
type of the system by one improves the steady state error of the original system by
order one. Thus, the second order type one system will have zero steady state error
2.6 Performance Enhancement By Using Controllers 285

for step input and finite error for ramp input. However, the second order system with
PI controller will have zero steady state error for step and ramp inputs irrespective of
K p and K i . The steady state error for parabolic input is finite which is evident from
Eq. (2.90).

E(s) s 2 (s + 2ζ ωn )
= 3 (2.90)
R(s) [s + 2ζ ωn s 2 + K p ωn2 s + K i ωn2 ]

For parabolic input

2ζ R
e(ss) = (2.91)
K i ωn

As for the stability of the system with PI controller, the order of the system is
increased from two to three. While the original second order system is always stable,
irrespective of the numerical values of ζ and ωn (ζ > 0 and ωn > 0), the system
with PI controller may become unstable if the value of K p and K i are not judiciously
chosen from view point of stability and also transient response specifications.
For retaining stability and improvement in transient response specifications, the
values of K p and K i are to be properly selected. The PI controller is basically a low
pass filter and hence the system with PI controller will have a slower rise time and
longer settling time. The values of K p and K i are chosen following the procedure
given below.
(1) Choose the range of K p and K i so that the system with PI controller is stable.
Routh–Hurwitz method of stability can be applied.
(2) Select the zero of the PI controller at s = −(K i /K p ) such that it is relatively
closer to the origin and away from the most significant poles of the system.
(3) The values of K p and K i should be as small as possible.
Since the order of the system is three and there is a zero in the closed loop T.F; it is not
easy to derive expressions for transient response specifications. However, transient
response of the original system and the system with PI controller can be simulated
and useful inferences can be established.
Again we consider a standard second order system with ζ = 0.5 and ωn = 4 rad/s.
From Eq. (2.89), the characteristic equation is obtained as

s 3 + 4s 2 + 16K p s + 16K i = 0 (2.92)

The system without the controller has the closed

 loop poles with the real part of
s = −ζ ωn = −2. Thus, we have to choose (K i /K p ) > 2 or K i > 2K p . For Eq.
(2.92) the following Routh’s array is prepared.
286 2 Time Response Analysis

s3 1 16K p
s2 1 4K i
s1 16K p − 4K i
s0 Ki

From Routh’s array, to ensure stability of the closed loop system with PI controller
K i < 4K p . Hence, the value of K i is chosen in the range 2K p < K i < 4K p .
Time response curves for unit step input are drawn and shown in Figs. 2.54 and
2.55 for the following controller parameters. [ζ = 0.5 and ωn = 4 rad/s in all cases].

I II
1 KI = 0; K p = 1 (without controller) 1 Kp = 2
2 Ki = 0.2; 2 K i = 0.2
3 Ki = 0.5 3 K i = 0.5
4 Ki =1 4 Ki = 1

Summary of PI Controller
1. Improves the damping and reduces the maximum overshoot.
2. Increases the rise time.
3. Decreases the bandwidth.
4. Filters out high frequency noise.
5. Improves phase margin and gain margin and hence the relative stability.
6. Increases the type of the system by one and ess for step and ramp inputs is zero
and for parabolic input, it is finite.

2.6.5 Proportional + Derivative (PD) Controller

From Fig. 2.56, the PD controller provides control action ea (t) which is proportional
and time derivative of the error e(t).

de(t)
ea (t) = K p e(t) + K d
dt
Taking Laplace transform on both sides of the above equation we get

E a (s) = [K p + K d s]E(s)
Ea
(s) = G c (s) = (K p + K d s) (2.93)
E
From Eq. (2.93) it is evident that the PD control is equivalent to adding a zero at s =
−(K p /K d ). Further, it has the characteristic of a high pass filter which increases the
2.6 Performance Enhancement By Using Controllers 287

Fig. 2.54 Transient response curves of a second order system for unit step input with proportional
plus integral controller with K p = 1. (1) K i = 0.2; (2) K i = 0.5; (3) K i = 1.0; (4) K i = 0 (without
controller)

bandwidth and hence reduces the rise time for step input. However, the differentiation
characteristic amplifies the noise that enters in the input. The forward path T.F. of
the combined PD controller and the second order system is

ωn2 (K p + K d s)
G(s) =
s(s + 2ζ ωn )

The closed loop T.F is

C(s) ωn2 (K p + K d s)
= 2 (2.94)
R(s) [s + (2ζ ωn + K d ωn2 )s + ωn2 K p ]

The error function is given by

E(s) s(s + 2ζ ωn )
= 2 (2.95)
R(s) [s + (2ζ ωn + K d ωn2 )s + ωn2 K p ]
288 2 Time Response Analysis

Fig. 2.55 Transient response curves of a second order system for unit step input with proportional
plus integral controller with K p = 2. (1) K i = 0.2; (2) K i = 0.5; (3) K i = 1.0; (4) K i = 0 (without
controller)

Kp +
R(s) + Ea(s) 2 C(s)
E(s) n
_ s(s+2 n)
+
Kds

Fig. 2.56 PD controller for a second order system

The steady state error for step input as seen from Eq. (2.95) is zero and for ramp
input

2ζ R
ess = (2.96)
ωn K p

The characteristic equation of the system is

s 2 + (2ζ ωn + K d ωn2 )s + ωn2 K p = 0 (2.97)

2.6 Performance Enhancement By Using Controllers 289

Fig. 2.57 Transient response curves of a second order system for unit step input with proportional
plus derivative controller with K p = 1. (1) K d = 0 (without controller); (2) K d = 0.2; (3) K d =
0.4; (4) K d = 1

Because of the introduction of the PD controller, the type and order of the system
are not changed. However, the natural frequency and damping factors are changed
and are given by

ωn1 = ωn K p (2.98)
 
Kd 1
ζ1 = ζ + ωn  (2.99)
2 Kp

For the transient response study, the second order system with ζ = 0.5 and ωn = 4
rad/s is considered. With these parameters, Eq. (2.94) is rewritten as

C(s) 16(K p + K d s)
= 2 (2.100)
R(s) [s + 4(1 + 4K d )s + 16K p ]

If we choose K p = 1, the natural frequency is unaltered and the new damping ratio
becomes

ζ = 0.5 + 2K d

For values of K d = 0, 0.2, 0.4 and 1 and for K p = 1 and K p = 2 transient response
curves for unit step input are plotted and are shown in Figs. 2.57 and 2.58 respectively.
290 2 Time Response Analysis

Fig. 2.58 Transient response curves of a second order system for unit step input with proportional
plus derivative controller with K p = 2. (1) K d = 0 (without controller); (2) K d = 0.2; (3) K d =
0.4; (4) K d = 1

Kp
+
R(s) + E(s) Ea(s) 2 C(s)
Ki + n
_ s s(s+2 n)
+
K ds

Fig. 2.59 PID controller connected in cascade with a second order system

Summary of PD Controller
1. Improves damping and reduces maximum overshoot.
2. Reduces rise time and settling time.
3. Increases bandwidth.
4. Improves phase margin and gain margin and thus the relative stability.
5. Amplifies noise at higher frequencies.
6. Steady state error for step input is zero and for ramp input it is finite.
2.6 Performance Enhancement By Using Controllers 291

2.6.6 Proportional + Integral + Derivative (PID) Controller

From Sects. 2.4 and 2.5 we observed that the PI controller could improve the relative
stability and steady state error simultaneously while PD controller could add damping
to the system but the steady state response is not affected. Further, the advantages
and disadvantages of each of these methods were summarized at the end of the
respective section. The best features of each of these controllers are taken to form
PID controller. The PID controller connected in cascade with a standard second order
system is shown in Fig. 2.59.
The PID controller provides control action ea (t) which is proportional, time inte-
grated and time derivative of the error e(t). That is

de(t)
ea (t) = K p e(t) + K i e(t)dt + K d
dt

Taking Laplace transform on both sides we get

 
Ki
E a (s) = Kp + + K d s E(s)
s
Ea
G c (s) = (s)
E 
Ki
= Kp + + Kd s
s
[K i + K p s + K d s 2 ]
= (2.101)
s
From Eq. (2.101), it is seen that the PID controller has two zeros in the LHP and a
pole at s = 0.
The forward path T.F. of the entire control system is obtained as

(K i + K p s + K d s 2 )ωn2
G(s) = (2.102)
s 2 (s + 2ζ ωn )

Equation (2.102) shows that the type of the system is increased by one from one
to two. The order of the system is increased by one from two to three. Hence, the
following information are available from Eq. (2.102):
1. The steady state errors of the system for step and velocity inputs are zero while
it is finite for parabolic input.
2. While the given second order system is always stable, for ζ > 0 and ωn > 0,
stability is not guaranteed for the system when connected with a PID controller.
3. Only judicious selection of controller parameters K p , K i and K d becomes nec-
essary not only for the system stability but also for good transient performance.
From Eq. (2.102), the closed loop T.F. is obtained as
292 2 Time Response Analysis

C(s) (K i + K p s + K d s 2 )ωn2
= 3 (2.103)
R(s) [s + (2ζ ωn + K d ωn2 )s 2 + K p ωn2 s + K i ωn2 ]

The error function is derived as

E(s) s 2 (s + 2ζ ωn )
= 3 (2.104)
R(s) [s + (2ζ ωn + K d ωn2 )s 2 + K p ωn2 s + K i ωn2 ]

From Eq. (2.104), while the ess for step and velocity inputs is zero, it is finite for
parabolic input. The ess for parabolic input is

2ζ R
ess = (2.105)
K i ωn

The characteristic equation of the system is given by

s 3 + (2ζ ωn + K d ωn2 )s 2 + K p ωn2 s + K i ωn2 = 0 (2.106)

To ensure the stability of the closed loop system, Routh–Hurwitz stability test is
performed.

s3 1 K p ωn2
s2 (2ζ ωn + K d ωn2 ) K i ωn2
s1 (2ζ ωn + K d ωn2 )K p ωn2 − K i ωn2
s0 K i ωn2

For the system to be stable the following condition is to be satisfied which is obtained
from sth row

(2ζ + K d ωn )K p ωn > K i (2.107)

For the transient response study of the second order system, with PID controller, as in
previous cases, ζ = 0.5 and ωn = 4 rad/s. are chosen. The PID controller parameters
given in Eq. (2.107) become

(1 + 4K d )4K p > K i (2.108)

The characteristic equation given in Eq. (2.106) is given as

s 3 + 4(1 + 4K d )s 2 + 16K p s + 16K i = 0 (2.109)

Selection of PID Parameters

Let us consider the PID controller consisting of PD portion connected in series with
a PI portion as shown in Fig. 2.60 where it is assumed that the proportional term in
PD controller is 1.
2.6 Performance Enhancement By Using Controllers 293

PD PI
E(s) Ea(s) E(s) Ea1(s)
(
Gc(s) = Kp + Ki + Kds
s
) 1+K
Kpd1s Kp2K+pKi2
s

Fig. 2.60 PID controller split into PD and PI controllers

From Fig. 2.60,

Ki K i2
Kp + + K d s = K p2 + K d1 K p2 s + K i2 K d1 + (2.110)
s s
Equating both sides of Eq. (2.110) we get

K p = K p2 + K d1 K i2
K d = K d1 K p2
K i = K i2 (2.111)

Based on the performance of PD controller alone, choose K d1 . Similarly, based on the

performance of PI controller alone chose K p2 and K i2 . Now using (2.111) choose K p ,
K i and K d . For simulation study, K d1 = 0.4, K p2 = 0.4 and K i2 = 0.2 are chosen.
Using (2.111) we get

K p = 0.4 + 0.4 × 0.2 = 0.48

K d = 0.4 × 0.4 = 0.16
K i = 0.2
 
0.2
G c (s) = 0.48 + + 0.16s
s
C(s) 16[0.2 + 0.48s + 0.16s 2 ]
= 3 (2.112)
R(s) [s + 6.56s 2 + 7.68s + 3.2]

The transient response curve for unit step input is plotted and is shown in Fig. 2.61.
Along with this the transient response curves with PD controller whose parameters are
K p = 1 and K d = 0.4 and PI controller whose parameters are K p = 1 and K i = 0.4
are also shown for comparison. Note that Eq. (2.108) is satisfied for the PID controller
parameters chosen.

Example 2.45

The forward path T.F. of a certain control system is given by

294 2 Time Response Analysis

Fig. 2.61 Transient response curves of a second order system for unit step input with (1) PI
controller with K p = 1; K i = 0.4; (2) PID controller with K p = 0.48; K i = 0.2; K d = 0.16; (3)
PD controller with K p = 1; K d = 0.4

4
G(s) =
s+2

An integral controller with a T.F

4
G c (s) =
s
is connected in cascade. Determine the damping ratio, natural frequency of the closed
loop system. Find the rise time, settling time and % overshoot for unit step input.
What is the steady state error for unit ramp input?

Solution The forward path T.F. of the given system is

G 1 (s) = G(s)G c (s)

16
=
s(s + 2)

The closed loop T.F is

2.6 Performance Enhancement By Using Controllers 295

C(s) G 1 (s)
=
R(s) 1 + G 1 (s)
16
= 2
(s + 2s + 16)
√
Comparing this with the standard second order system, we get ωn = 16 = 4. And
2ζ ωn = 2 or ζ = 0.25.
The rise time tr is given by

(π − φ) π − cos−1 0.25
tr =  = 
ωn 1 − ζ 2 4 1 − (0.25)2
π − 1.318
= = 0.471 s.
4 × 0.968

The settling time is given by

3.91 3.91
ts = = = 3.91 s.
ζ ωn 0.25 × 4

The % peak overshoot is given by

√−ζ π √−0.25π
%M p = e 1−ζ 2 × 100 = e 1−(0.25)2 × 100 = 44.43%

The steady state error for unit ramp is given by

1
ess =
Kv

where
s16
K v = lim sG 1 (s) = lim =8
s→0 s→0 s(s + 2)
1
ess = = 0.125
8

tr = 0.471 s.
ts = 3.91 s.
%M p = 44.43%
ess = 0.125
296 2 Time Response Analysis

Fig. 2.62 Block diagram for C(s)

R(s) 1
Example 2.46 Td s
s(s 2)(s 3)

Example 2.46

Consider a certain third order system shown in Fig. 2.62 connected in cascade with
a derivative controller. Determine the value of Td so that the closed loop system is
critically damped. For this value of Td , find the steady state error for unit step input.
Also determine the peak overshoot.

Solution
1
G(s) =
s(s + 2)(s + 3)
G(s) = Td s

The forward path T.F

G 1 (s) = G(s)G c (s)

Td
=
(s + 2)(s + 3)

The closed loop T.F. is given by

Td
C(s) G 1 (s) (s + 2)(s + 3)
= = 
R(s) 1 + G 1 (s) Td
1+
(s + 2)(s + 3)
Td
= 2
s + 5s + (6 + Td )

The characteristic equation is

s 2 + 5s + (6 + Td ) = 0

Comparing this with the characteristic equation of a standard second order system.
We get
2.6 Performance Enhancement By Using Controllers 297

ωn = (6 + Td )
2ζ ωn = 5

For the system to be critically damped ζ = 1.

5
ωn = = 2.5
 2
(6 + Td ) = 2.5
Td = 0.25
0.25
G 1 (s) =
(s + 2)(s + 3)

The position error constant K p is

K p = lim G 1 (s)
s→0
0.25 0.25
= lim =
s→0 (s + 2)(s + 3) 6
1 1
ess = =
1 + Kp 0.25
1+
6
6
= = 0.96
6.25
Since the system is critically damped, the overshoot M p = 0.

Td = 0.25
ess = 0.96
MP = 0

Example 2.47
A certain unity negative feedback system has the following forward path T.F.

36
G(s) =
s(s + 4)

The system is provided with a PD controller in the forward path. Determine the
controller parameter K p and K d so that the system damping ζ = 0.8 and natural
frequency of oscillation ωn = 9.
298 2 Time Response Analysis

Solution For the given system, the characteristic equation is

1 + G(s) =0
36
1+ 2 =0
s + 4s
s 2 + 4s + 36 =0
√
ωn = 36 = 6
2ζ ωn = 4
4
ζ = = 0.333
2×6

The new values of damping factor and natural frequency of oscillation are

ζ 1 = 0.8
ωn1 = 9

Using Eq. (2.98)


ωn1 = ωn K p

9 = 6 Kp
K p = 2.25

Using Eq. (2.99)

 
Kd 1
ζ1 = ζ + ωn 
2 Kp
 
6 1
0.8 = 0.333 + K d √
2 2.25
K d = 0.289

K p = 2.25
K d = 0.289

Example 2.48

The forward path T.F. of a certain first order system connected with unity feedback is

10
G(s) =
(s + a)
2.6 Performance Enhancement By Using Controllers 299

The system is connected in cascade with an integral controller whose T.F. is

Ki
G c (s) =
s
The permissible steady state error for unit ramp is 0.1. The required ωn = 12 rad/s.
Determine the values of K i and a.

Solution
10
G(s) =
(s + a)
Ki
G c (s) =
s
K i 10
G 1 (s) = G(s)G c (s) =
s(s + a)

For ramp input

K i 10 K i 10
K v = lim sG 1 (s) = lim =
s→0 s→0 0+a a
1
ess = = 0.1
Kv
10K i
K v = 10 or or K i = a.
a
The characteristic equation is

1 + G 1 (s) = s 2 + as + K i 10

Comparing this with the characteristic equation of a standard second order system
equation, we get

ωn = K i 10
144 = K i 10

K i = 14.4
a = 14.4

1. Time response of a system consists of two parts namely transient response and
steady state response. In transient response, the variables are expressed as a
300 2 Time Response Analysis

function of time where t is varied in the range 0 < t < ∞. The steady state
response is obtained as time t approaches infinity.
2. Transient response of first and prototype second order systems are obtained for
impulse and step inputs.
3. For step input, transient response specifications for first and prototype second
order systems are derived in terms of system parameters.
4. Steady state errors for prototype second order system are derived for step, ramp
and parabolic inputs when the system is connected with unity feedback. The
steady state errors are expressed in terms of static error constants K p , K v and
Ka .
5. For a unity feedback system, the error can be expressed as a function of time
using dynamic error constants. The steady state error is obtained by putting
t = ∞.
6. For a non-unity feedback system, estimation of steady state error with G H (s)
will give erroneous result. The non-unity feedback system has to be converted to
an equivalent unity feedback system and for this system only steady state errors
are to be determined.
7. To enhance the transient and steady state performances controllers are employed
either in the forward or in the feedback paths.
8. The controllers normally used are rate controllers PID controllers and addition
of poles and zeros.
9. A prototype second order system is considered for the study of the performance
of the controllers.
10. The rate controller is connected in the minor loop of the system. The rate con-
troller increases the damping of the system and suppresses the oscillations in the
system.
11. PID controller is the most widely used controller in industrial control systems.
This controller is connected in the forward path in cascade of the prototype
second order system.
12. The performance enhancement of P, I, D, PI, PD, ID and PID controllers is
studied by plotting transient response curves by simulation in digital computer
using MATLAB program.
13. For performance enhancement of higher order systems PID controller is the most
suited controller. By judicious choice of proportional, integral and derivative term
controller parameters, it is possible to obtain good transient response with less
overshoot, less rise time and setting time.

Short Answer Type Questions

1. What do you understand by time response of a system?
The output response, starting at some initial condition expressed as a function
of time when the input is applied, is called time response of the system. It is
divided into two parts namely transient response and steady state response.
2.6 Performance Enhancement By Using Controllers 301

2. What is transient response?

Transient response is the time response which goes from initial state to final state
as time becomes large. The variation is expressed as a function of independent
variable time t.
3. What is steady state response?
The manner in which the output behaves as time t approaches infinity is called
steady state response.
4. Define impulse, step, ramp and parabolic inputs.
(a) Impulse input:

r (t) = R t = 0
r (t) = 0 t = 0

(b) Step input:


Ru(t) t ≥ 0
r (t) =
0 t <0

(c) Ramp or velocity input:


Rtu(t) t ≥ 0
r (t) =
0 t <0

(d) Parabolic or acceleration input:


1
Rt 2 u(t) t ≥0
r (t) = 2
0 t <0

5. What are the Laplace transform of signals given in Question 4?

(a) R(s) = R (for impulse)
(b) R(s) = R/s (for step)
(c) R(s) = R/s 2 (for ramp)
(d) R(s) = R/s 3 (for parabolic)
6. How the impulse, step and ramp signals are related?

The relationship between impulse, step and ramp signals are as indicated below.

Integrate Integrate
Impulse −−−−→ Step −−−−→ Ramp
Differentiate Differentiate
Ramp −−−−−−→ Step −−−−−−→ Impulse

7. How impulse response of a system is important?

(a) If the area under the impulse response curve is finite, the system is stable.
302 2 Time Response Analysis

(b) By integrating the impulse response, step response can be obtained.

(c) If the impulse response function is known, then by taking the Laplace trans-
form, the T.F. of the system can be obtained.
8. Why systems are tested with step input?
It is easy to generate a step function. If the step response is satisfactory, the given
system is likely to give satisfactory performance to other commonly used signals.
If the step response is known, impulse response is obtained by differentiating it
and velocity response is obtained by integrating it.
9. What are the performance specifications of a first order system?
The performance specifications of first order system are
(a) Time constant T ,
(b) Rise time tr ,
(c) Setting time ts , and
(d) Time delay td .
10. Define time constant of the first order system. How is it related to the location
of the pole?
Time constant T of a first order system is defined as the time taken for the step
response to reach 63.2% of its final value for the first time. It is expressed as
T = 1/a where a is the location of the pole on the negative real axis in the
s-plane.
11. Define rise time of a first order system.
Rise time tr is defined as the time taken for the step response to go from 10% to
90% of its final value. It is expressed as tr = 2.2T .
12. Define time delay of a first order system.
Time delay td of a first order system is defined as the time taken for the step
response to reach 50% of its final value for the first time. It is expressed as
td = 0.693T .
13. Define settling time of a first order system.
The settling time ts of a first order system is defined as the time taken for the
step response to reach and stay within 2% of its final value for the first time. It
is expressed as ts = 3.91T . For 5% error tolerance ts = 3T .
14. What information we get from performance specifications of first order
system?
The time constant of a first order system is inversely proportional to the pole
located on the negative real axis of the s-plane. If the pole is moved away from
the origin, the time constant is less. The rise time, settling time and the time
delay are all directly proportional to the time constant. Hence, the time response
is faster. The time response will be slower if the pole is moved towards the origin.
15. What do you understand by prototype second order system?
For a unity feedback system, the forward path T.F. and the closed loop T.F are
respectively writing as
2.6 Performance Enhancement By Using Controllers 303

ωn2
G(s) =
s(s + 2ζ ωn )
C(s)
T (s) =
R(s)
ωn2
= 2
(s + 2ζ ωn s + ωn2 )

where ζ and ωn are system parameters. The system with the above equations is
called prototype second order system.
16. What are the parameters of a prototype second order system and how they
are differently expressed?
ζ and ωn are the only two parameters of a prototype second order system where
ζ is called the damping ratio and ωn is the natural frequencyof oscillation
expressed in rad/s. ζ ωn is called the damping factor and ωd = ωn 1 − ζ 2 where
ωd is called damped frequency of oscillation in rad/s ωn ≥ ωd .
17. What is characteristic equation? Why is it called so?
The dominator of the closed loop T.F of a system when equated to zero is called
the characteristic equation. The corresponding polynomial is called characteristic
polynomial. These polynomials contain the closed loop poles whose locations
in the s-plane decide the stability and the time response of the system. Since it
characterizes the system response, it is called characteristic equation.
18. What is the characteristic equation and characteristic polynomial of a pro-
totype second order system?
The characteristic equation and the characteristic polynomial of a prototype sec-
ond order system are respectively given below as

s 2 + 2ζ ωn s + ωn2 = 0

F(s) = s 2 + 2ζ ωn s + ωn2

19. List the time domain specifications of a prototype second order system?
The time domain specifications of a second order system are
(a) Peak overshoot,
(b) The time at which the peak overshoot occurs,
(c) Rise time,
(d) Settling time,
(e) Time delay,
(f) Time constant,
(g) The period of oscillations, and
(h) The number of oscillations during the settling time.
20. For what type of input, the time domain specifications mentioned in Ques-
tion 19 are defined?
The time domain specifications are defined for the step input.
304 2 Time Response Analysis

21. How many specifications completely determine the time response?

The percentage overshoot, rise time and settling time completely determine the
time response.
22. How maximum or peak overshoot is defined? What is the percentage max-
imum overshoot and how it is related to the damping ratio?
The maximum overshoot is defined as the difference between the maximum
value of the output Cmax at any instant and the steady value Css . Thus, M p =
Cmax − Css . The percentage maximum overshoot is the amount by which the
time response curve overshoots the steady state or final value at the peak time
expressed as a percentage of the steady state value.

Maximum overshoot, M p
%M p = × 100
Css

The % peak overshoot of a prototype second order system is

−ζ π
%M p =  × 100
e 1 − ζ2

23. What pole location characterize: (a) undamped, (b) the over damped, (c) the
under damped and (d) critically damped and (e) negatively damped?

(a) Undamped, ζ = 0; s1 , s2 = ± jω

(b) Over damped, ζ > 1; s1 , s2 = −ζ ωn ± ωn ζ 2 − 1

(c) Under damped, ζ < 1; s1 , s2 = −ζ ωn ± jωn 1 − ζ 2
(d) Critically damped, ζ = 1; s1 , s2 = −ωn

(e) Negatively damped, ζ < 0; s1 , s2 = ζ ωn ± jωn 1 − ζ 2

24. How can you justify pole–zero cancellation in the T.F.?

If the ratio C(s)/R(s) is expressed in terms of poles and zeros they can be put
into partial fraction. If any residue is negligibly small, that pole can be can-
celled with a zero which is very close to that pole. The small value of the residue
gives very insignificant contribution to the transient response and thus pole–zero
cancellation is justified.
25. What is an insignificant pole?
The pole which is far away from the imaginary axis of the s-plane corresponds to
fast decaying time response. If the real part of the magnitude of the pole is 5 to 10
times more than the real part of the dominant pole, it is called insignificant pole.
26. Name any two conditions under which the response generated by a pole can
be neglected?
The response generated by a pole can be neglected if:
(a) Pole–zero cancellation takes place; and
(b) Insignificant poles are omitted.
2.6 Performance Enhancement By Using Controllers 305

27. What are dominant poles?

Poles that are closer to the imaginary axis of the left half of the s-plane give rise
to transient response that will decay very slowly. Such poles are called dominant
poles.
28. In a system the real part of a pole generate what response?
In a system, the real part of the pole generates exponential decay if it is in LHP
and exponential rise if it is in RHP of the s-plane.
29. In a system, the imaginary part of a pole generate what response?
In a system, the imaginary part of a pole generates sinusoidal response.
30. What is the effect of the presence of repeated poles in the imaginary axis?
The presence of repeated poles in the imaginary axis will make the system unsta-
ble with unbounded output for bounded input.
31. In a system if a pole is moved with a constant imaginary part what will the
response have in common?
If a pole is moved with a constant imaginary part, as shown in Fig. 2.63, the
damped frequency of oscillation ωd is constant. The more negative poles indi-
cate higher value of damping factor ζ ωn , which makes the system transient
response more heavily damped. The damping ratio ζ is increased which results
in less % peak overshoot. The time constant T , rise time tr and the time at which
peak value occurs t p are shortened.
32. If a pole is moved with a constant real part what will the response have in
common?
Figure 2.64 shows the movement of complex poles with their real parts remain-
ing constant at ζ ωn . For upward movement of the pole, the time constant remains
the same. However, the damping ratio decreases and ωn and ωd increase. This
amounts to high-overshoot. The settling time is unaffected.
33. If a pole is moved along a radial line extending from the origin what will
the response have in common?
In Fig. 2.65, the movement of complex poles along a radial line extended from
the origin is shown. As the poles move away from the origin, the natural fre-
quency of oscillation is increased. The damping ratio ζ remains the same. This
amounts to constant peak overshoot. The time constant T , rise time tr settling
time ts are all decreased. The rate of decay is faster.
34. Consider the unity feedback system with the following forward path T.F.

16
(a) G(s) =
s(s + 4)
(s + 16)
(b) G(s) =
s(s + 3)

Determine, in each case, the characteristic equation damping factor and

natural frequency of oscillation.

C(s) G(s) 16
(a) = = 2
R(s) 1 + G(s) (s + 4s + 16)
306 2 Time Response Analysis

The characteristic equation is s 2 + 4s + 16 = 0, ωn = 4; and ζ = 0.5

C(s) G(s) (s + 16)

(b) = = 2
R(s) 1 + G(s) (s + 4s + 16)

The characteristic equation is s 2 + 4s + 16 = 0, ωn = 4; and ζ = 0.5.

35. In Question 34, the systems (a) and (b) have the same ζ and ωn . Will they
have the same time domain specifications?
Eventhough, ζ and ωn are the same for system (a) and system (b), their time
domain specifications will be totally different. The formulae derived in Sect. 2.4.4
are valid for system (a) and they are not valid for system (b). This is due to the
presence of a zero at s = −16 for system (b).
36. Consider the closed loop T.F. of the following second order systems. Find ζ ,
ωn and the nature of damping.

(8s + 4)
(a) T (s) =
+ 5s + 25
s2
(4 − 3s)
(b) T (s) = 2
s + 4s + 4
(s − 6)
(c) T (s) = 2
s + 10s + 16

(a) s 2 + 5s + 25 = 0; ωn = 5; ζ = 0.5; ζ < 1. The system is under damped.

(b) s 2 + 4s + 4 = 0; ωn = 2; ζ = 1; ζ = 1. The system is critically damped.
(c) s 2 + 10s + 16 = 0; ωn = 4; ζ = 1.25; ζ > 1. The system is over damped.

37. For a system with unity feedback,

(25 − T s)
G(s) =
s(s + 10)

Find the value of T so that the system is undamped. What is the undamped
frequency?

C(s) G(s)
=
R(s) 1 + G(s)
25 − T s
= 2
s + (10 − T )s + 25

10 − T = 0 or T = 10

The characteristic equation is

2.6 Performance Enhancement By Using Controllers 307

s 2 + 25 = 0
ωn = 5

38. For the following closed loop T.F find the value of K so that ζ = 1.

(s + 2)
T (s) =
5s 2 + 2s + K
K = 0.2

39. For the following closed loop T.F find the value of K so that ζ = 0.6.

(s + 0.6)
T (s) =
[4s 2 + 2.4s + (K + 6)]
K = 10

40. For the following closed loop T.F find the value of K so that ωd = 2 rad/s

(K s + 2)
T (s) =
5s 2 + (2K + 5)s + 25
K = 2.5

41. For the following closed loop T.F find the value of K so that the undamped
natural frequency of oscillation is 8 rad/s

(s + K )
T (s) =
+ 5s + 100
K s2
K = 1.5625

42. For the following closed loop system T.F. find the % peak overshoot.

125
T (s) =
+ 10s + 500
5s 2
%M p = 72.94%

43. For the following closed loop T.F find t p .

64
T (s) =
+ 8s + 64
s2
t p = 0.4534 s

44. For the following closed loop T.F. find the rise time tr .
308 2 Time Response Analysis

100
T (s) =
+ 10s + 100
5s 2
tr = 0.412 s

45. For the following closed loop T.F find the settling time ts for 5% error tol-
erance.
60
T (s) =
8s 2 + 32s + 72
ts = 1.5 s

46. Define steady state error.

Steady state error is the difference between the given input and the output as
t → ∞ which amounts to the natural response having decayed to zero.
47. What are the sources of steady state error?
The sources of steady state errors are
(a) System configuration,
(b) Type of input applied, and
(c) Non-linear sources such as back lash in gears or a motor that will not move
unless the input voltage exceeds the threshold value, saturation in generator etc..
48. How the steady state error is calculated for a unity feedback system if the
forward path T.F is known?
If the forward path T.F. G(s) is known, then the steady error is calculated from

R(s)
ess = lim s
s→0 1 + G(s)

where R(s) is the Laplace transform of the input.

49. For a unity feedback system, if the closed loop T.F T (s) is known how the
steady state error is calculated?
For a unity feedback system if the closed loop T.F. T (s) is known, then the steady
state error is calculated from

ess = lim s R(s)[1 − T (s)]

s→0

50. If G(s) is the forward path T.F and H (s) is the T.F of the feedback path does
the following equation give the steady state error? If not, what information
we get form this equation.

s R(s)
lim
s→0 [1 + G H (s)]

The above equation does not give the steady state error. It gives the steady state
actuating signal.
2.6 Performance Enhancement By Using Controllers 309

51. How do you calculate the steady state error for a non-unity feedback system?
The steady state error for non-unity feedback system is calculated by converting
G(s) and H (s) into G e (s) which is the equivalent forward path T.F of the unity
feedback system.

G(s)
G e (s) =
1 + G(s)H (s) − G(s)
s R(s)
ess = lim
s→0 1 + G e (s)

52. What are static error constants and how they are used to determine the
steady state error?
The position error constant K p , velocity error constant K v and acceleration error
constant K a are called static error constants. They are defined as follows:

K p = lim G(s)
s→0
K v = lim sG(s)
s→0
K a = lim s 2 G(s)
s→0

53. What is the effect of increasing the system gain upon the steady state error?
By increasing the system gain the steady state error is reduced.
54. What is the effect of increasing the system gain of a prototype second order
system upon the maximum overshoot?
If the system gain of a prototype second order system is increased, the maximum
overshoot is also increased.
55. Identify the inputs to which the static error constants are related.
The position error constant K p is defined when the input is step. The velocity
error constant K v is defined when the input is ramp and the acceleration error
constant is defined when the input is parabolic.
56. What is the disadvantage of using position error constant?
From the position error constant, the information we get about the steady state
error is either it is zero, or finite or infinite. The information about the accumu-
lation of the error is missing.
57. Define dynamic error constants.
The dynamic error constants are defined as follows:

C0 = lim F(s)
s→0
d F(s)
C1 = lim
s→0 ds
.. ..
. .
d n F(s)
Cn = lim
s→0 ds n
310 2 Time Response Analysis

where F(s) = 1/1 + G(s).

58. What is the advantage of dynamic error constants over static error con-
stants?
The dynamic error constants give error as a function of time. Thus, the accumu-
lation of error with respect to time is estimated.
59. How many integrations in the forward path should be there in order that
the steady state error is zero when the inputs are (a) step; (b) ramp; and (c)
parabolic?
(a) For step input there should be one integration.
(b) For ramp input there should be two integrations.
(c) For parabolic input there should be three integrations.
60. A certain feedback control system has the following T.F.

16
G(s) =
s(s + 4)
2
H (s) =
(s + 1)

What is the type of this system?

16(s + 1)
G e (s) =
s3 + 6s 2 − 8s + 16

Since there is no integrator in the G e (s) the system is Type 0 and not Type 1
which is erroneously obtained from

32
G H (s) =
s(s + 4)(s + 1)

61. If the following input is applied to the feedback system, what are the dynamic
error constants needed to determine the steady state error?

r (t) = t 3

The dynamic error constants C0 , C1 , C2 , and C3 need to be determined to esti-

mate the steady state error.
62. Consider the system shown in Fig. 2.66. Will the characteristic equation be
different for C(s)/R(s) and C(s)/D(s)? The characteristic equation of any
system will be same irrespective of the type of input given and the place where
it is applied. For the system shown in Fig. 2.66, the characteristic equation is
1 + G 1 G 2 H (s) = 0.
63. What are the effects of rate feedback controller on the system performance?
By rate feedback controller, the effective damping is increased. However, the
steady state error is also increased.
2.6 Performance Enhancement By Using Controllers 311

64. What are the effects of rate feedback controller on M p , tr , ts and ωn ?

Overshoot is decreased. Rise time is increased. Settling time is decreased. Nat-
ural frequency of oscillation remains unchanged.
65. What is a three term controller?
The proportional plus the integral plus the derivative controller which has three
parameters K p , K i and K d is called three term controller.
66. What are the effects of proportional controller on the prototype second
order system performance specifications?
As the proportional gain is increased, M p is increased. t p is decreased, ts is
unchanged. The bandwidth is increased. The steady state error for ramp input is
reduced by a factor 1/ K p .
67. What are the effects of derivative controller on the prototype second order
system performance specifications?
The second order system with derivative controller is reduced to a first order sys-
tem by pole–zero cancellation. There is no peak overshoot. The time constant,
rise time and settling time are less and hence the time response is faster.
68. What is the effect of I controller on the prototype second order system per-
formance?
When the integral controller is connected in cascade with a prototype second
order system, it becomes unstable. However in general, the I controller permits
the rejection of a step disturbance. It accelerates the movement of the process
set point and eliminates the residual steady state error.
69. What are the effects of PI controller on the system performance?
PI controller improves the damping and reduces the maximum overshoot.
Increases the rise time and decreases the bandwidth. Filters out high frequency
noise. Increases the type of the system by one and the ess for step and ramp
inputs is zero. For parabolic input ess is finite.
70. What are the effects of PD controller when connected with a second order
system?
The PD controller when connected in cascade with a second order system:
(a) Improves damping and reduces overshoot,
(b) Reduces tr and ts .
(c) Increases bandwidth.
(d) Improves phase margin and gain margin and thus increases the relative
stability.
(e) Amplifies noise at higher frequencies.
71. What are the effects of PID controller on the transient response of higher
order system?
By proper tuning of the PID controller parameters, the stability of the system
is improved. The maximum overshoot can be reduced. System rise time and
settling time can be reduced. The system bandwidth can be increased.
312 2 Time Response Analysis

s-plane
n
cos 1 d

0
n

Fig. 2.63 Movement of pole with constant imaginary part

s-plane
n
d
cos 1

0
n

Fig. 2.64 Movement of pole with constant real part

s-plane

cos 1

Fig. 2.65 Movement of pole along a radial line from the origin

Long Answer Type Questions

1. For the R-L circuit shown in Fig. 2.67a, the input voltage v(t) = u(t). Drive the
expression for the current i(t) and sketch the wave form. Determine the time
constant and rise time.
c(t) = 0.5(1 − e−2t )
2.6 Performance Enhancement By Using Controllers 313

D(s)

R(s) C(s)
G1(s) G2(s)

H(s)

Fig. 2.66 Block diagram for Question 62

(a) (b)
i(t)

i(t) 0.5A

L=1H
0.316A

v(t) R=2

0
T = 0.5 t

Fig. 2.67 a Electric circuit; and b current wave form

Fig. 2.68 Time response (t)

curve

0
3 t
314 2 Time Response Analysis

Time constant T = 0.5 sec; rise time tr = 1.1 s.

2. The time response of the first order system for unit step input is shown in Fig. 2.68.
The system T.F is G(s) = K /(s + a). Find K and a.

1
a= ; K =1
3
3. Consider the first order system shown in Fig. 2.69. Find the time constant, rise
time, time delay and settling time for unit step input for the closed loop system.
T = 0.25 s; tr = 0.55 s; td = 0.1733 s; ts = 0.98 s.
4. A certain non-unity feed back control system is represented as shown in Fig. 2.70.
Derive an expression for the time response of the system as a function of time
for (a) unit impulse; (b) unit step; (c) unit ramp inputs; and (d) also find T , tr , td
and ts and Css for unit step input.

(a) c(t) = 0.8e−2.4t

1
(b) c(t) = [1 − e−2.4t ]
3 
1
(c) c(t) = t − 0.139 + 0.139e−2.4t
3
1
(d) T = 0.4167sec; tr = 0.9167; td = 0.2888 s; ts = 1.63 s; Css =
3
5. A certain second order unity feedback system is described by the following
forward path T.F.
25
G(s) =
s(s + 6)

Derive an expression for the output response of the system for unit step input.
Also find tr , t p , T , M p and ts .

c(t) = 1 − 1.25e−3t sin(4t + 0.928 rad).

tr = 0.5535 sec; t p = 0.785 sec; T = 0.333 sec; ts = 1.3 sec; M p = 9.5%.

6. A certain unity feedback control system has the following open loop T.F.

6
G(s) =
s(s + 2)

Find the values for tr , t p , ts , M p and % M p for a step input of 10 units.

tr = 0.724 sec; ts = 3.91 sec; t p = 1.4 sec; M p = 2.456 units;

%M p = 24.56%.
7. For the system represented in block diagram shown in Fig. 2.71 write down the
characteristic equation for (a) K = 9.5; (b) K = 11; and (c) K = 12.
2.6 Performance Enhancement By Using Controllers 315

(a) F(s) = s 3 + 6s 2 + 11s + 63

(b) F(s) = s 3 + 6s 2 + 11s + 72
(c) F(s) = s 3 + 6s 2 + 11s + 78

8. For a certain unity feedback control system, the forward path T.F. is given by,

10(1 + K s)
G(s) =
s(s + 2)

Determine the value of K so that the damping ratio is 0.5. For this value of
K determine the maximum overshoot, time constant and rise time for unit step
input. What is the steady state error for unit ramp input with and without K ?
Also derive expression for the time response for unit step input for the above
cases.

K = 0.1162

Time domain ess for

specifications T tr tp %M p ramp input
K =0 1s 0.63 s 1.05 s 36% 0.2

K = 0.1162 0.6325 0.65 s 1.032 s 17.3% 0.2

For K = 0; c(t) = 1 − 1.054e−t sin(3t + 1.25 rad)

For K = 0.1162; c(t) = 1 − 1.02e−t sin(2.74t + 1.361 rad)

9. A unity feedback control system has the following forward path T.F.

K
G(s) =
s(1 + sT )

where K and T are constants and greater than zero. Determine the factor by
which K should be multiplied to reduce the overshoot for step input from 85 to
35%.

K 2 = 0.02658K 1

where K 1 corresponds to 85% overshoot and K 2 corresponds to 35% overshoot.

The gain has to be reduced by a factor 0.02658.
10. A certain unity feedback control system has the following forward path T.F.

K (s + b)
G(s) =
(s + a)2
316 2 Time Response Analysis

where K , a and b are constants and greater than zero. The steady state error due
to unit step input is to be limited to 0.2 with a damping ratio of 0.4 and undamped
frequency of oscillation as 6 rad/s. Find K , a and b.

K = 0.328; a = 4.472; b = 15.24

11. A certain unity feedback control system has the following forward path T.F.

25.4
G(s) =
s(s 2 + 8s + 17)

Find the dominant poles and the insignificant pole of the closed loop system.

Dominant poles, s12 = −1 ± j1.8

Insignificant pole, s3 = −5.9

12. For the rate feedback system shown in Fig. 2.72 it is desired to have the closed
loop system to be placed at s12 = −2 ± j4.

Find the values of K and K t . What is the type and order of the system? Find the
steady state error for unit ramp input.

(a) K = 23.34; K t = 0.0428

(b) Type 0 and order 2 system
(c) ess = 10.1714

13. A certain unity feedback system has the following forward path T.F.

K (s + a)
G(s) =
s(s + b)

where K , a and b are constants and greater than zero. It is required that the steady
state error for unit ramp input is less than 0.1. To improve transient response,
the closed loop system poles are to be placed at s12 = −2 ± j4. Find the values
of K , a and b.
K = 2; a = 10; b = 2

14. Consider the system shown in Fig. 2.73. (a) Find the value of K so that the
damping ratio of the closed system is 0.5. (b) Determine the values of tr , T and
ts if the system is subjected to unit step input. (c) Derive an expression for the
impulse response of the system.
2.6 Performance Enhancement By Using Controllers 317

(a) K = 0.2;
(b) tr = 0.605 s; T = 0.5 s; ts = 1.955 s;
(b) c(t) = 4.62e−2t sin 3.464t

15. A certain unity feedback system has the following forward path T.F.

40(s + 2)
G(s) =
s(s + 1)(s + 4)

Determine the type and order of the system, the static error coefficients and the
steady state error for ramp input with magnitude 4.

K p = ∞; K v = 20 s−1 ; K a = 0; ess = 0.2, Type 1 and order 3

16. A certain feedback control system has the following forward and feedback path
T.F.
20
G(s) =
(s + 1)(s 2 + 10s + 6)
5
H (s) =
(s + 3)

Find the steady state error when the system inputs are (a) r (t) = 5; (b) r (t) = 4t
and (c) r (t) = 23 t 2 .

20(s + 3)
G e (s) =
s4 + 14s 3 + 49s 2 + 34s + 58

(a) K p = 60
58
; ess=2.46 (b) K v = 0; ess = ∞ (c) K a = 0; ess = ∞.
17. For a unity feedback system, the forward path T.F. is

K
G(s) =
s 2 (s + 2)(s + 3)

Find the value of K to limit the steady state error to zero when the input to the
system is r (t) = 1 + 10t + 20t 2 .

The steady state error is zero for all values of K since K p = ∞ and also K v = ∞
which corresponds to r (t) = 1 and r (t) = 10t respectively.
18. Using the generalized error series calculate the steady state error of a unity
feedback system having the open loop T.F. as

10
G(s) =
(s + 2)
318 2 Time Response Analysis

For the following excitation: (a) r (t) = 2; (b) r (t) = 2t; (c) r (t) = 21 t 2 ; (d)
r (t) = 1 + 2t + 21 t 2 .

1 1 1
(a) C0 = ; e(t) = ; ess =
6 3 3
10 2 20
(b) C1 = ; e(t) = t + ; ess = ∞
144 3 144
1 2 10
(c) C2 = −2.5; e(t) = t + t − 1.25; ess = ∞
 12  144
1 1 10
(d) e(t) = 1 + 2t + t 2 + (2 + t) − 1.25; ess = ∞
6 2 144

19. The open loop T.F. of a unity feedback system is

s + 3.15
G(s) =
s(s + 1.5)(s + 0.5)

Find the type and order of the system. Find the static error constants and the
steady state error corresponding to them.

Type 1 order 3 system

K p = ∞; K v = 4.2; Ka = 0

1 1
ess = + 0.238 + = ∞
1+∞ 0

20. The closed loop T.F. of a unity feedback system is

20
T (s) =
s 2 + 4s + 25

Find the steady state error for unit step input.

ess = 0.2

21. Consider the system shown in Fig. 2.74. The steady state error due to a unit ramp
input is 0.2 and due to unit step disturbance is −0.5. Determine the values of K 1
and K 2 .
K 1 = 1.3333; K 2 = 125

22. Consider the system shown in Fig. 2.75. What is the type of the system? Find
the value of K to get 2% steady state error for unit step input.

Type 0 system; K = 1.0204

2.6 Performance Enhancement By Using Controllers 319

23. For the system shown in Fig. 2.76 determine the type and order. Identify the
input for which the steady state error is finite and calculate the same.

Type 0 and order 3 system. For step input only the steady state error is finite.
ess = R.
24. For the system shown in Fig. 2.77 find K 1 and K 2 so that the percentage error
for step input is 10% with a rise time of 0.2 s. What is the steady state error for
unit ramp input?

K 1 = 18.6; K 2 = 1.411; ess = 0.0866

.
25. For the system shown in Fig. 2.78 find the steady state error for unit step input?

The system is unstable. Steady state error does not exist.

26. For the system shown in Fig. 2.79, find the value of K so that the steady state
error for unit step input is zero. For this value of K what is the steady state error
for unit ramp input?
1
K = 7.5; ess =
45
for ramp input.
27. For the system shown in Fig. 2.80 find the value of K so that the steady state
error for a ramp input is less than 0.1.

K = 40

28. The forward path T.F. of a unity feedback system is

40
G(s) =
s(0.2s + 1)

Obtain the generalized error constants. Also find its steady state error when the
input is r (t) = (3 + 4t)t.

1 1
C0 = 0; C1 = ; C2 = −
40 400
1 1
e(t) = (3 + 8t) − ; ess = ∞
40 100
29. The forward path T.F. of a unity feedback system is

10
G(s) =
(s + 2)
320 2 Time Response Analysis

Find the steady state error for a step input of magnitude 5 for the following
controllers which are connected in cascade with G(s). (a) PI controller; and (b)
PD controller.
5
(a) ess = 0; (b) ess =
(1 + 5K p )

where K p is the proportional term constant of the PD controller.

30. The forward path T.F. of a unity feedback system is

5
G(s) =
s(s 2 + 3s + 5)

Find the steady state error for unit ramp input when the following controllers are
connected is cascade with G(s). (a) PI controller (b) PD controller.

1
(a) ess = 0; (b) ess =
Kp

31. A PD controller is connected in cascade with a plant as shown in Fig. 2.81. Find
the PD controller parameters K p and K d so that the closed loop system damping
ratio is 0.5, and the natural frequency of oscillation is 10 rad/s.

K p = 5.25; K d = 0.5

32. Consider the system shown in Fig. 2.82

Choose the value of the proportional controller K p so that the time constant of
the closed loop system is one fourth of its open loop value.

K p = 0.15

33. Consider the system shown in Fig. 2.83

(a) Find the values of K p and K d so that the damping ratio is 0.8 and undamped
natural frequency of oscillation is 12 rad/s for the closed loop system. (b) Find
the values of K P and K d if the closed loop poles are to be at s = −4 and s = −8.

(a) K p = 1.44; K d = 0.092

(b) K p = 0.32; K d = 0.02

34. The closed loop T.F. of a unity feedback control system is given as

K (s + 5)
T (s) =
[s 3 + 10s 2 + (3K + 5)s + 5K ]
2.6 Performance Enhancement By Using Controllers 321

2 C(s)
R(s)
(s 2)

Fig. 2.69 Block diagram for Question 3

Find K so that the steady state error for unit ramp input is 0.9.

K =2

35. A certain control system is described by the following forward and feedback
path transfer functions.

1
G(s) =
s 2 (s + 5)
H (s) = 3(s + 5)

The error function is defined as

1
e(t) = r (t) − c(t)
Kh

Find the steady state error for unit ramp input

1
ess =
75
36. The forward path T.F. of a certain unity feedback system is

K
G(s) =
(s + 2)(s + a)

K p = 5 and the required overshoot is less than 16.3%. Find K , a and the steady
state error.

The above conditions are satisfied for the following values of K and a.

(a) K = 74.64; a = 7.464; ess = 16.7%

(b) K = 5.35; a = 0.535; ess = 16.7%

37. The closed loop T.F. of a certain control system is given by

322 2 Time Response Analysis

R(s) C(s)
4
C(s) =
(s 5)

H(s) = (s 2)

Fig. 2.70 Block diagram for Question 4

R(s) C(s)
2
K
(s 1)(s 2)

3
(s 3)

Fig. 2.71 Block diagram for Question 7

Fig. 2.72 Rate feedback C(s)

R(s) K
system for Question 12
s(s 3)

Kt s

Fig. 2.73 Rate feedback

R(s) 16 C(s)
system for Question 14
s(s 0.8)

(1 Ks)

225
T (s) =
(s 2 + s + 9)(s 2 + 2s + 25)

Determine the damping ratio of the dominant characteristic roots.

1 1
(a) ζ1 = ; (b) ζ2 =
6 5
2.6 Performance Enhancement By Using Controllers 323

D(s)

R(s) K1(2s 3) K2 C(s)

(3s 2) s(s 5)

Fig. 2.74 Block diagram for Question 21

R(s) (s 5) C(s)
s2(s10)

Fig. 2.75 Block diagram for Question 22

R(s) K1 C(s)
(s2 2s 4)

K2
s

Fig. 2.76 Block diagram for Question 23

R(s) E(s) 10 1 C(s)

K1
(s 2) s

K2

Fig. 2.77 Block diagram for Question 24

324 2 Time Response Analysis

R(s) 1 C(s)
(s2 16)

4
(s 5)

Fig. 2.78 Block diagram for Question 25

R(s) K C(s)
(s 3)

3
(s 5)

Fig. 2.79 Block diagram for Question 26

R(s) K C(s)
s(s 2)

2s
K

Fig. 2.80 Block diagram for Question 27

R(s) 16 C(s)
Kp Kd s
(s2 2s 16)

Fig. 2.81 PD Controller block diagram for Question 31

2.6 Performance Enhancement By Using Controllers 325

R(s) C(s)
10
Kp
(s 0.5)

Fig. 2.82 Block diagram for Question 32

R(s) C(s)
100
Kp Kd s
s(s 10)

Fig. 2.83 Block diagram for Question 33

R(s) (s 1) C(s)
2 3
(s 2)(s 3)

Fig. 2.84 Block diagram for Question 38

s G2(s)
(s 1)

E(s) (1 2s) (1 2s) C(s)

R(s) (s 5) E1 s(s 2)
G1(s) G3(s)

Fig. 2.85 Block diagram for Question 40

38. For the system shown in Fig. 2.84, find the steady state error for unit step input.
The error is defined as e(t) = r (t) − c(t).

ess = 0.4
326 2 Time Response Analysis

G1(s) G2(s)

3 4 C(s)
R(s) E1 (s 2) E2 (s 1)

H(s)
K
(s 6)

Fig. 2.86 Block diagram for Question 41

39. For the following closed loop T.F. determine the steady state error for unit
parabolic input. The error is defined as e(t) = r (t) − c(t)

(−2s 2 + 3s + 10)
T (s) =
(s 3 + 5s 2 + 3s + 10)
ess = 0.7

40. Consider the system shown in Fig. 2.85. Determine the steady state error for the
input r (t) = 3t.

ess = 45

41. Consider the block diagram shown in Fig. 2.86. Find the value of K so that the
steady state error due to step input is zero.

For the value of K find the steady state error for unit ramp input.

e(t) = r (t) − c(t)

K =1
ess = 1.667
Chapter 3
Frequency Response Analysis

Learning Objectives
After completing this chapter, you should be able to
 Understand the concept of frequency response.
 Draw the polar plot for minimum phase, non-minimum phase and all phase
transfer functions.
 Draw the Bode plot.
 Determine the transfer function from the Bode plot.
 Establish the correlation between transient response and frequency
response.
 Define frequency domain specifications and derive expressions for these
specifications.
 Determine the closed loop frequency response from open loop frequency
response using constant M and constant N circles and Nichols chart.
 Adjust the gain to satisfy the frequency response specifications.
 Use MATLAB program for the frequency response analysis.

3.1 Introduction

In Chap. 2, the dynamic response of linear time invariant continuous time system was
studied when typical test signals such as impulse, step and ramp were applied. The
performance of many real-time control systems is judged based on the time response
of the system due to the above test signals. However, a simple and useful test signal
is a sine wave and we get valuable information when applied to a certain class of

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 327
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_3
328 3 Frequency Response Analysis

systems such as communication system where most of the signals to be processed

are either sinusoidal or composed of sinusoidal signals. When the input applied to
a system is a sine wave the output obtained from the system is also a sine wave but
with different amplitude and phase. By measuring the output amplitude and phase
of a system over a range of frequency of the input sine wave, a different dynamic
response is observed and this is called the frequency response of the system which
represents the system’s response to sinusoidal inputs at varying frequencies.
The frequency response is defined as the plot of magnitude and phase differ-
ence between the input and output sinusoids over a range of frequencies.
The frequency response method may be less intuitive than the other methods
described in the previous chapter. However, it has the following advantages and
disadvantages.
Advantages of Frequency Response Method
1. It is easy to generate sine wave and collect data from the output.
2. Experimental information can be used with ease for design purpose.
3. Raw measurements of the output amplitude and phase of the system are enough
to design a suitable feedback control. It directly quantifies system performance
and directly designs control systems.
4. In real-life situations such as modelling the transfer function no intermediate
processing of the data is required.
5. It offers the most cost-effective design method especially for systems that are
stable in open loop.
6. It is the easiest method to use for designing compensation.
7. The tests involve measurement under steady state conditions which are simpler
to analyse compared to measurements of transient responses.
8. From test results it is easy to get the control system order, gain, error constant,
resonant frequencies, etc.
9. The frequency response method can be very well used for the analysis and design
of certain non-linear systems too using describing function method.
10. Since only one frequency at a time is used, the amplitude of the sine wave can
be increased or decreased as one wishes.
11. The method is frequency selective and can therefore reject the noise in the output
in an excellent way.
Disadvantages of Frequency Response Method
1. It is not often easy to correlate the results of the frequency response characteristics
of higher order systems to that of time domain characteristics.
2. Systems like process control operate at low frequencies. It is difficult to generate
low-frequency signals and obtain necessary measurements. Measurement may
require several hours in such cases.
3. In frequency response method, the analysis is done at one frequency at a time
which requires very long test time.
4. If the spacing of the test frequencies is too far apart, it is possible to miss some
important frequency response characteristics.
3.1 Introduction 329

5. The frequency response method as considered by many researches and professors

is an old-fashioned method even though for many practical control and signal
processing engineers it is a very useful tool.

3.2 Obtaining Steady State Output to Sinusoidal Input

Consider the linear time invariant system whose transfer function is G(s). The system
is excited with a sine wave

r (t) = R sin ωt (3.1)

The system representation and input and output wave forms are shown in Fig. 3.1a, b
representing the steady state response of the system to a sinusoidal input which can
be directly obtained from G(s) by replacing s by jω.

C
( jω) = G( jω) (3.2)
R
is called the sinusoidal transfer function, which is the ratio of C( jω) to R( jω).
Thus, by knowing the T.F. G(s), the magnitude characteristic |G( jω)| and phase
characteristic ∠G( jω) completely describe the steady state frequency response of
the system. Given the poles and zeros of G(s), the frequency response at any fre-
quency ω = ω1 and other frequencies also can be evaluated graphically as illustrated
below.

(a)
R(s) Input C(s) Output
G(s)
r(t) c(t)

(b)
φ r(t) R sin t

c(t) C sin( t )
R C

Fig. 3.1 a Linear system with sine wave input and sine wave output. b Input-output wave forms
with their amplitude and phase
330 3 Frequency Response Analysis

Example 3.1
A certain control system has the following transfer function

5(s + 2)
G(s) =
(s + 1)(s + 4)

Find the frequency response at ω = 5 rad/sec.

Solution The poles and zeros of G(s) are located as shown in Fig. 3.2a. Point A
which corresponds to ω = 5 in the imaginary axis is identified. Vectors from the
poles and zeros of G(s) are drawn to the point A and their magnitude and phase
identified. The frequency response at ω = 5 is calculated as

5 × Product of directed vector distances from all the zeros of G(s)

G( j5) =
Product of directed vector distances from all the poles of G(s)
√
5 × 29∠68.2◦
= √ √
26∠78.69◦ 41∠51.34◦
= 0.8247∠−61.83◦

The ratio of the output to input magnitude is 0.8247 and the output lags behind
the input by an ∠61.83◦ at ω = 5 rad/sec. Similar procedure is adopted at other
frequencies.
The graphical method described above is somewhat tedious. The frequency
response at any frequency ω = ω1 can be obtained from the sinusoidal T.F. directly
as explained below.
The vectorial representation of G( jω) for the sinusoidal T.F. is shown in Fig. 3.2b.
From Fig. 3.2b, the magnitude and phase contributed at any frequency in general
can be determined. For any required frequency, the response can be obtained by
substituting for the given ω

5( jω + 2)
G( jω) =
( jω + 1)( jω + 4)

5 (ω2 + 4)∠tan−1 ω2
=  ω
(ω2 + 1)(ω2 + 16)∠tan−1 ω + tan−1 4
= 0.8242∠−61.83◦
|G( jω)| = 0.8247
∠G( jω) = −61.83◦

Obtaining frequency response using Fig. 3.2b is much simpler than using Fig. 3.2a.
3.3 Plotting Frequency Response 331

(a) Im

A 5

√41 9 √2
6
√2
51.34º 68.2º 78.69º 0
4 2 1 Re

(b)

Im Im

2 +4 j 2 +1 j
√ √

tan 1 tan 1
2
0 2 Re 0 1 Re

Im

2 +1
6 j
√

tan 1
4
0 4 Re

Fig. 3.2 a Sketching frequency response from pole–zero plot. b Vectorial representation of G( jω)

3.3 Plotting Frequency Response

The frequency response of a system can be plotted in several ways using the sinusoidal
transfer function G( jω). As we vary input frequency over a range, we get a vector
of frequency response data in the form of a complex vector. The complex vector can
be displayed in a variety of ways as given below.
1. Linear Frequency Response Plots
These plots include the plot magnitude of G( jω) versus ω and phase angle of G( jω)
versus ω. The horizontal axis in both plots is frequency ω with linear scaling.
332 3 Frequency Response Analysis

2. Linear Frequency Response Plots versus log ω

These plots are similar to the first case but the horizontal axes are log arithmetically
scaled (semi-log plots). This allows a larger frequency scale to be used.
3. Polar (Nyquist) Plot
This is the plot of the imaginary versus the real part of G( jω) drawn in a polar graph
with ω as the varying parameter.
4. Bode Plots
It has two separate plots one for the magnitude and another for phase angle of
G( jω). The magnitude is expressed in decibel (20 log10 |G( jω)|). By this very large
magnitude can be accommodated. The horizontal axes are logarithmic by which we
can accommodate larger frequency scale.
5. Log Magnitude versus Phase Plot (Nichols Plot)
This is the plot of the magnitude of G( jω) in dB versus the phase in degrees with
ω as the varying parameter.
For control system analysis and design, the first two methods are not used. The
widely used methods are the polar plot, Bode plot and Nichols plot. These three
methods are described below.

3.4 Polar Plot (Nyquist Plot)

The polar or otherwise called Nyquist plot of a sinusoidal transfer function G( jω)
is the plot of the magnitude of G( jω) versus the phase angle of G( jω) on polar
coordinates as ω is varied from zero to infinity. In polar coordinates the gain of the
sinusoidal T.F. G( jω) is plotted as the radial coordinate, while the phase of G( jω)
is plotted as the angular coordinate. It has a unique advantage of having the display of
both amplitude and phase angle on a single plot over a wide frequency range. It has
the property to show whether a closed loop system is stable or unstable. However, the
only disadvantage of this method is that it does not clearly indicate the contributions
made by each pole and zero. We illustrate below how a polar plot is drawn for the
given T.F.
The polar plot is one of the various graphical plots available to represent the
sinusoidal T.F. G( jω). To represent the frequency response of G( jω) in polar plot
G( jω) should be stable. Frequency response can be represented in polar, Bode and
log magnitude versus phase plots. However, all these plots need not be frequency
response plots.
Consider the T.F. shown in Example 3.1

5 (ω2 + 4) ω ω
G( jω) =  ∠tan−1 − tan−1 ω − tan−1
(ω2 + 1)(ω2 + 16) 2 4
3.4 Polar Plot (Nyquist Plot) 333

Im
(a) (b) 105º 90º 75º
120º 60º
135º 45º

90º
I 150º 30º
II 270º
Positive Imaginary ve
165º 15º

180º 2.5 Ref 180º 0

Re
180º 0 32.5º 0 0º 360º 0 0º
Negative real ve real
1.917 5
195º 4 345º
5 3 2 1
4 3 1
2 B G(j ) plot
210º 330º
III A
IV
Negative Imaginary 225º 315º
270º
240º 300º
90º
255º 285º
270º
Scale 1 cm 0.5 unit

(c)
Polar plot of G(s) (5s 10)/(s2 5s 4)
90 2.5
120 60
2

1.5
150 30
1

0.5

0
180 0

210 330

240 300

270

Fig. 3.3 a Polar plot for G(s) = 5(s + 2)/[(s + 1)(s + 4)]. b Frequency response plot in polar
graph

For the frequency 0 ≤ ω ≤ ∞, |G( jω)| and ∠G( jω) are calculated and tabulated
below.

ω 0◦ 1 2 3 4 5 ∞
φ = ∠G( jω) 0◦ −32.5◦ −45◦ −52.1◦ −57.6◦ −61.8◦ −90◦
|G( jω)| 2.5 1.917 1.414 1.14 0.959 0.8247 0

The tabulated values of φ = ∠G( jω) and |G( jω)| are transferred to polar graph
as follows. The polar coordinates are drawn as shown in Fig. 3.3a. It is divided into
four quadrants. The first quadrant lies between 0◦ and 90◦ , the second quadrant
between 90◦ and 180◦ , the third quadrant between 180◦ and 270◦ and the fourth
quadrant lies between 270◦ and 0◦ . The reference line is along the 0◦ line. Angle is
measured positive from the reference line when moved anticlockwise and if moved in
the clockwise direction, the angle is measured negative with respect to the reference.
Thus, +90◦ = −270◦ ; +180◦ = −180◦ ; +270◦ = −90◦ ; 0◦ = 360◦ . To transfer the
334 3 Frequency Response Analysis

frequency response say for ω = 1, φ = −32.5◦ and |G( jω)| = 1.917 units, draw
the radial line 0 A from the origin with φ = −32.5◦ . Choose any scale for |G( jω)|.
Measure 0B = 1.917 on 0 A. Similar procedure is followed to locate |G( jω)| in the
polar graph for other values of ω. Join these points by a smooth graph. By the side
of B put the value of ω. Put arrow in the G( jω) plot pointing in the direction of
increasing values of ω. This completes the polar plot which is manually done.
Ready-made polar graph sheets are available. Radial lines have been drawn for
each 1◦ angular space. By choosing appropriate scale factor for the magnitude, the
polar plot is drawn in polar graph as shown in Fig. 3.3b.
The polar plot can also be obtained using MATLAB program. The graph can
be plotted either in polar radial and angular coordinates or coordinates of real and
imaginary axes. The necessary programmers and the plots obtained are shown ahead.
Polar Plot in Radial and Angular Coordinates
The following MATLAB program is written to get this plot. The plot is shown in
Fig. 3.3c.

MATLAB Program
num=[0 5 10];
den=[1 5 4];
[re, im]-nyquist(num,den,w)
Z = r e + li × im
r = abs(z)
theta=angle(z)
polar(theta,r)
grid
title(‘polar plot of G(s) = (5s + 10)/(s 2 + 5s + 4)’)

3.4.1 General Shape of Polar Plot and Type and Order of the
System

As stated in the previous sections, polar plot plays an important role in the study of
linear system stability especially the Nyquist stability analysis. In the stability study,
in most of the cases, mere sketching of Nyquist plot is good enough and point by
point information may not be always necessary. In these cases, the |G( jω)| where it
cuts the negative real axis will be enough to assess the stability. The Type and order
of the system have the influence in the shape of polar plot and this is discussed below.
Polar Plot of Type 0 and Order 1 System
Consider the following T.F. whose Type is 0 and order is 1

K
G(s) = (3.3)
(1 + sT )
3.4 Polar Plot (Nyquist Plot) 335

Fig. 3.4 Polar plot of Type 0 Im

order 1 system
G(s) = K /(1 + sT )

K
0 0 Re

G(j ) plot

The sinusoidal T.F. is

K K ∠tan−1 ωT
G( jω) = = √
(1 + jωT ) 1 + ω2 T 2
K
|G( jω)| = √
1 + ω2 T 2
φ = ∠G( jω) = ∠− tan−1 ωT

For ω = 0;
G( jω) = K ∠0◦

For ω = ∞;
G( jω) = 0∠−90◦

The polar plot starts from +ve real at K , traverses through the fourth quadrant as ω
increases and terminates at the origin as ω → ∞. The polar plot is shown in Fig. 3.4.
Polar Plot of Type 0 Order 2 System
Consider the following T.F. of a Type 0, order 2 system

K
G(s) = (3.4)
(1 + sT1 )(1 + sT2 )

The sinusoidal T.F. is

K
G( jω) =
(1 + jωT1 )(1 + jωT2 )
K ∠− tan−1 ωT1 − tan−1 ωT2
= 
(1 + ω2 T12 )(1 + ω2 T22 )

For ω = 0,
G( jω) = K ∠0◦
336 3 Frequency Response Analysis

Im

K
0 0 Re

G(j ) plot

Fig. 3.5 Polar plot of Type 0 order two system G(s) = K /[(1 + sT1 )(1 + sT2 )]

For ω = ∞,
G( jω) = 0∠−180◦

The polar plot is shown in Fig. 3.5.

The polar plot starts from +ve real at |G( jω)| = K for ω = 0, traverses through
the fourth and third quadrants and terminates at the origin when ω = ∞.
Polar Plot for Type 0 and Order 3 System
Consider the following T.F. of a Type 0 order 3 system

K
G(s) = (3.5)
(1 + sT1 )(1 + sT2 )(1 + sT3 )
K
G( jω) =
(1 + jωT1 )(1 + jωT2 )(1 + jωT3 )
K
|G( jω)| = 
(1 + ω2 T1 )(1 + ω2 T22 )(1 + ω2 T32 )
2

φ = ∠G( jω) = − tan−1 ωT1 − tan−1 ωT2 − tan−1 ωT3

For ω = 0,
G( jω) = K ∠0◦

For ω = ∞;
G( jω) = 0∠−270◦
3.4 Polar Plot (Nyquist Plot) 337

Im

K
0 0 Re

G(j ) plot

K
Fig. 3.6 Polar plot of Type 0 order 3 system G(s) =
[(1 + sT1 )(1 + sT2 )(1 + sT3 )]

The polar plot is shown in Fig. 3.6. It starts from +ve real at |G( jω)| = K , for ω = 0
traverses through the fourth, third and second quadrants and terminates at the origin
when ω = ∞.
Polar Plot for Type 1 Order 2 System
Consider the following T.F. of a Type 1 order 2 system

K
G(s) = (3.6)
s(1 + sT1 )
K
G( jω) =
jω(1 + jωT1 )
K ∠−90◦ − tan−1 ωT1
= 
ω 1 + ω2 T12

For ω = 0;
G( jω) = ∞∠−90◦

For ω = ∞;
G( jω) = 0∠−180◦

The polar plot is shown in Fig. 3.7. It starts from ∞ from negative imaginary for
ω = 0, traverses through the third quadrant and terminates at the origin when ω = ∞.
Polar Plot for Type 1 Order 3 System
Consider the following T.F. of a Type 1 order 3 system
338 3 Frequency Response Analysis

Fig. 3.7 Polar plot of Type 1 Im

order 2 system
G(s) = K /[s(1 + sT1 )]

0 Re

G(j ) plot

K
G(s) = (3.7)
s(1 + sT1 )(1 + sT2 )
K
G( jω) =
jω(1 + jωT1 )(1 + jωT2 )
K
|G( jω)| = 
ω (1 + ω2 T12 )(1 + ω2 T22 )
φ = ∠G( jω) = −90◦ − tan−1 ωT1 − tan−1 ωT2

For ω = 0;
G( jω) = ∞∠−90◦

For ω = ∞,
G( jω) = 0∠−270◦

The polar plot is shown in Fig. 3.8. It starts from ∞ from negative imaginary axis,
traverses through the third quadrant, cuts the negative real axis, traverses through the
second quadrant and terminates at the origin when ω = ∞.
From Eqs. (3.8) to (3.12), the following inferences are drawn:
1. The starting portion of G( jω) plot is decided by the type of the system. Type
0 system starts from +ve real. Type 1 system starts from −ve imaginary from
infinity. Type 2 system starts from −ve real from infinitive and so on.
2. The finishing portion of the G( jω) plot is decided by the order of the system
when ω = ∞. First order system terminates in the fourth quadrant with φ =
−90◦ . The second order system terminates in the third quadrant with φ = −180◦ .
The third order system terminates in the second quadrant with φ = −270◦ .
3.4 Polar Plot (Nyquist Plot) 339

Im

0 Re

G(j ) plot

Fig. 3.8 Polar plot of Type 1 order 3 system G(s) = K /[s(1 + sT1 )(1 + sT2 )]

Im
Type 3
0

Type 2 Order 3
Order 4 0
0
Re
Order 2 Order 1 Type 0

Type 1
0

Fig. 3.9 General shape of polar plot at the start and finish

The above inferences are true iff (if and only if) there are no zeros in the T.F. and it
is a minimum phase T.F. This is illustrated in Fig. 3.9.
If the T.F. is strictly proper (numerator polynomial is greater than the denominator
polynomial), as ω tends to ∞, the polar plot converges to the origin. Further, it is to
340 3 Frequency Response Analysis

be noted that the phase angle of the poles and zeros decide the general shape of the
polar plot.
Polar Plot of a T.F. with a Zero
Consider the following T.F. whose Type is 1 and order is three with a zero.

K (1 + sT1 )
G(s) = (3.8)
s(1 + sT2 )(1 + sT3 )
K (1 + jωT1 )
G( jω) =
jω(1 + jωT2 )(1 + jωT3 )

K (1 + ω2 T12 )
|G( jω)| = 
ω (1 + ω2 T22 )(1 + ω2 T32 )
∠G( jω) = −90◦ − tan−1 ωT2 − tan−1 ωT3 + tan−1 ωT1

For ω = 0,
G( jω) = −90◦

For ω = ∞,
G( jω) = 0∠−180◦

With the addition of zero, the starting point of the polar plot is not affected.
However, as ω increases and tends to ∞, the polar plot traverses through the third
quadrant and terminates at the origin as shown in Fig. 3.10a. The finishing portion
of the polar plot is like that of a second order system. As ω tends to ∞, the phase
contribution of a zero gets nullified by the phase contribution of a pole. In general
if n is the degree of the denominator polynomial of G( jω) and m is the degree
of numerator polynomial, the order of the system is considered as (n − m) for the
purpose of deciding the finishing portion of the polar plot as ω tends to ∞. The polar
plot of Eq. (3.8) with zero is sketched as shown in Fig. 3.10a and without zero is
sketched in Fig. 3.10b. The general shape of finishing portion of the polar plot in
terms of (n − m) is shown in Fig. 3.10c.
The summary of starting and finishing portion of polar plot and magnitude and
phase angle contribution of poles and zeros are given in Table 3.1.

3.4.2 Polar Plot of Non-minimum Phase Transfer Function

Depending upon the type and order of the system, the general shape of the polar plot
can be decided and sketched for a system with a minimum phase T.F. as discussed in
Sect. 3.4.1. The inferences derived in Sect. 3.4.1 will not be valid for a system with a
non-minimum phase T.F. where at least a pole or a zero lies in RHP. The magnitude
of the T.F. of a non-minimum phase T.F. remains the same as that of the minimum
3.4 Polar Plot (Nyquist Plot) 341

(a ) (b)
Im Im

0 Re 0 Re

0 0

(c) Im

n m 3
n m 4

Re
n m 2 n m 1 order

Fig. 3.10 a Polar plot of Eq. (3.8). b Polar plot of Eq. (3.8) without zero and c General shape of
G( jω) plot as ω tends to ∞ in terms of n and m

phase T.F. The phase angle of the T.F. is totally shifted. In this section, we discuss
the sketching of polar plot for non-minimum phase poles and zeros. Consider the
following zero of a certain T.F.

G(s) = (2 − 3s)
G( jω) = (2 − j3ω)

The zero is located at s = 2/3 in RHP +ve real. The real and imaginary parts of
G( jω) are marked as shown in Fig. 3.11a.
342 3 Frequency Response Analysis

Table 3.1 Magnitude and phase angle contribution of poles and zeros of G(s)
G(s) 1
s
1
s2
(1 + sT ) (1 + sT )2 1
(1+sT )
1
(1+sT )2

∠G( jω) −90◦ −180◦ tan−1 ωT 2 tan−1 ωT − tan−1 ωT −2 tan−1 ωT


|G( jω)| 1
ω
1
ω2
(1 + ω2 T 2 ) (1 + ω2 T 2 ) √ 1 1
(1+ω2 T 2 )
1+ω2 T 2

∠G( jω) −90◦ −180◦ 0◦ 0◦ 0◦ 0◦

ω=0
∠G( jω) −90◦ −180◦ 90◦ 180◦ −90◦ −180◦
ω=∞
|G( jω)| ∞ ∞ 1 1 1 1
ω=0
|G( jω)| 0 0 ∞ ∞ 0 0
ω=∞

(a) (b) Im
Im

√4
+ 9
j3 2

2
13
tan 13 =180 tan 2
0 = tan 13 Re 2
2
2 0 Re
√4
+9 j3
2

Fig. 3.11 a Magnitude and phase representation of non-minimum phase zero. b Magnitude and
phase representation of a non-minimum zero

From Fig. 3.11a, the magnitude and phase of G( jω) are obtained as

|G( jω)| = 4 + 9ω2
3ω
∠G( jω) = − tan−1
2
The magnitude of G( jω) remains the same as that of the minimum phase T.F.
However, the phase angle is − tan−1 (3ω/2) which alters the polar plot to a completely
different quadrant.
Now consider the T.F. with the following zero which has a non-minimum phase

G(s) = −2 + 3s
G( jω) = −2 + j3ω
3.4 Polar Plot (Nyquist Plot) 343

This zero is also located at s = 2/3 in RHP +ve real. The magnitude and phase of
G( jω) are determined from Fig. 3.11b.
From Fig. 3.11b, the magnitude and phase of G( jω) are obtained as

|G( jω)| = 4 + 9ω2
3ω
∠G( jω) = 180 − tan−1
2
Here also, the magnitude of G( jω) remains the same as that of minimum phase zero
but the phase is shifted from tan−1 (3ω/2) to 180 − tan−1 (3ω/2). In a similar way,
the phase angle of non-minimum phase poles is determined.
For a system with non-minimum phase poles and zeros, Fig. 3.11a or b is
drawn as the case may be and the phase angle is expressed as a function of ω.

Example 3.2
Consider the following T.F.

5(s − 2)
G(s) =
(s + 1)(s + 4)

Sketch the polar plot.

Solution
5(s − 2)
G(s) =
(s + 1)(s + 4)
5( jω − 2)
G( jω) =
( jω + 1)( jω + 4)

5 (ω2 + 4)
G( jω) = 
(ω2 + 1)(ω2 + 16)
ω ω
∠G( jω) = 180 − tan−1 − tan−1 ω − tan−1
2 4
For ω = 0,
G( jω) = 2.5∠180◦

For ω = ∞,
G( jω) = 0∠−90◦

The polar plot starts from negative real at −2.5, traverses through the second, first
and fourth quadrants and terminates at the origin when ω = ∞. The polar plot is
sketched as in Fig. 3.12.
344 3 Frequency Response Analysis

Im

14
2.5
0 0 1 Re

Fig. 3.12 Polar plot for G(s) = 5(s − 2)/[(s + 1)(s + 4)]

Example 3.3
Sketch the polar plot for the following T.F.

5(s + 2)
G(s) =
(s + 1)(−s + 4)

Solution
5(s + 2)
G(s) =
(s + 1)(−s + 4)
5( jω + 2)
G( jω) =
( jω + 1)(− jω + 4)
√
5 ω2 + 4
|G( jω)| = 
(ω2 + 1)(ω2 + 16)
ω ω
∠G( jω) = tan−1 + tan−1 − tan−1 ω
2 4
The polar plot starts from +ve real from 2.5 for ω = 0 traverses through the fourth
quadrant since initially, tan−1 ω2 + tan−1 ω4 contributes less phase than tan−1 ω.
When ∠tan−1 ω2 + tan−1 ω4 = ∠tan−1 ω, it cuts the +ve real axis at ω = ωc and
traverses through the first quadrant. Now (tan−1 ω2 + tan−1 ω4 ) contributes more phase
than tan−1 ω. At ω = ∞, the total phase is +90◦ and the magnitude becomes zero.
The frequency ωc at which the phase angle becomes zero is calculated as follows:
3.4 Polar Plot (Nyquist Plot) 345

Im

>√2

c=√2 2.5
0 5/3 0 Re

< √2

5(s + 2)
Fig. 3.13 Polar plot for G(s) =
(s + 1)(−s + 4)

ωc ωc
tan−1 + tan−1 − tan−1 ωc = 0
2 4
ωc ωc
tan−1 + tan−1 = tan−1 ωc
2 4
Using the formula

A+B
tan−1 A + tan−1 B = tan−1
1 − AB

the above equation is written as

ωc ωc
+
tan−1 2
= tan−1 ωc
4
ωc2
1− 8
6ωc
tan−1 = tan−1 ωc
8 − ωc2

Taking tan on both sides, we get

6ωc
= ωc
8 − ωc2
√
ωc = 2

The magnitude at this frequency is 53 . The polar plot is sketched as shown in Fig. 3.13.
346 3 Frequency Response Analysis

Example 3.4
Sketch the polar plot for the following T.Fs.

(1 − 2s)
(a) G(s) =
(1 + 2s)
2(s + 5)
(b) G(s) =
(3s + 4)
(1 − 2s)(1 + 5s)
(c) G(s) =
(1 + 2s)(1 − 5s)

Solution (a)

(1 − 2s)
G(s) =
(1 + 2s)
(1 − jω2)
G( jω) =
(1 + jω2)
|G( jω)| = 1
∠G( jω) = −2 tan−1 2ω

For ω = 0,
G( jω) = 1∠0◦

For ω = ∞,
G( jω) = 1∠−180◦

The polar plot is a semi-circle of unit radius. It starts from +ve real with mag-
nitude 1, traverses through fourth and third quadrants and terminates at (−1, 0)
on the negative real axis as shown in Fig. 3.14a.
(b)

2(s + 5)
G(s) =
(3s + 4)
2( jω + 5)
G( jω) =
(3 jω + 4)
√
2 ω2 + 25
|G( jω)| = √
9ω2 + 16
ω 3ω
∠G( jω) = tan−1 − tan−1
5 4
3.4 Polar Plot (Nyquist Plot) 347

(a) Im

0
1 1 Re

(b) Im

2/3 2.5
0 Re

(c) Im

1
m=
√10
1
G(j ) plot

m= 50.78º 0
0 1 Re

Fig. 3.14 a Polar plot for G(s) = (1 − 2s)/(1 + 2s). b Polar plot for G(s) = 2(s + 5)/(3s + 4).
(1 − 2s)(1 + 5s)
c Polar plot for G(s) =
(1 + 2s)(1 − 5s)
348 3 Frequency Response Analysis

For ω = 0,
G( jω) = 2.5∠0◦

For ω = ∞;
2 ◦
G( jω) = ∠0
3

At any frequency ω, − tan−1 3ω

4
is more dominant than tan−1 ω5 except at ω = 0
where the phase contribution of each one of them is 0 and at ω = ∞ where the
phase contribution of each one of them is 90◦ . The polar plot starts from +ve
real where its magnitude is 2.5, traverses through the fourth quadrant and for
ω = ∞, it terminates at 2/3. This is shown in Fig. 3.14b.
(c)

(1 − 2s)(1 + 5s)
G(s) =
(1 + 2s)(1 − 5s)
(1 − jω2)(1 + jω5)
G( jω) =
(1 + jω2)(1 − jω5)
|G( jω)| = 1
φ = ∠G( jω) = −2 tan 2ω + 2 tan−1 5ω
φ
= tan−1 5ω − tan−1 2ω
2

tan−1 5ω, at any frequency ω dominates over tan−1 2ω except at ω = 0 they are
equal with 0◦ and at ω = ∞, they are equal at 90◦ . They polar plot starts (ω = 0)
from +ve real at 1, traverses through the first quadrant with a magnitude of 1 at
all frequencies reaches the maximum phase φm at ω = ωm , where 0 < φ < 90◦
and comes back to 0◦ when ω = ∞. The values of ωm and φm are determined
as follows
φ
= tan−1 5ω − tan−1 2ω
2  
−1 5ω − 2ω −1 −1 −1 A − B
= tan tan A − tan B = tan
1 + 10ω2 1 + AB
3ω
= tan−1
1 + 10ω2
φ
Let N = tan 2

φ
= tan−1 N
2
3ω
tan−1 N = tan−1
1 + 10ω2
3.4 Polar Plot (Nyquist Plot) 349

Taking tan on both sides, we get

3ω
N =
1 + 10ω2

Differentiating N with respect to ω and equating d N /dω = 0 we get ωm and

corresponding phase as φm .

dN (1 + 10ω2 )3 − 3ω(20ω)
=
dω (1 + 10ω2 )2
=0
1 + 10ω2 = 20ω2
1
ω= √
10
1
ωm = √
10

At ω = ωm , φ = φm

φm = 2[tan−1 5ωm − tan−1 2ωm ]

= 2[57.7◦ − 32.3◦ ]
= 50.78◦

The polar plot is sketched as shown in Fig. 3.14c.

3.4.3 Polar Plot with Transportation Lag (Time Delay)

Consider the transfer function of a first order system

K
G(s) =
(1 + sT1 )

The transfer function of the above system with a time delay (transportation lag) is
written as

K e−sT
G(s) =
(1 + sT1 )

The sinusoidal T.F. is

350 3 Frequency Response Analysis

K e− jωT
G( jω) =
(1 + jωT1 )
K
|G( jω)| = 
1 + ω2 T12
∠G( jω) = − tan−1 ωT1 − ∠ωT

With the introduction of the time delay, the magnitude of the original system is
unchanged. However, the phase angle of G( jω) is changed by −ωT .

Example 3.5
Consider the following T.F. of a certain control system with time delay. Draw the
polar plot and compare this plot with the one obtained without time delay

100e−0.2s
G(s) =
(2 + 3s)(2s + 5)

Solution

100e−0.2s
G(s) =
(2 + 3s)(2s + 5)
100e−0.2 jω
G( jω) =
(2 + j3ω)( j2ω + 5)
100
|G( jω)| = 
(4 + 9ω2 )(25 + 4ω2 )
3 2
∠G( jω) = − tan−1 ω − tan−1 ω − 0.2 × 57.3ω
2 5
= − tan−1 1.5ω − tan−1 0.4ω − 11.46ω

The magnitude and phase of G( jω) are tabulated below for different values of ω.

ω 0 1 2 3 4 5 ∞
|G( jω)| 10 5.15 2.47 1.39 0.87 0.6 0
∠G( jω) 0◦ −78.1◦ −110.3◦ −127.66◦ −138.5◦ −146◦ −180◦
without
delay
∠11.46ω 0◦ −11.46◦ −22.92◦ −34.38◦ −45.84◦ −57.3◦ 0
∠G( jω) 0◦ −89.56◦ −133.22◦ −162.04◦ −184.34◦ −203.3◦ −180◦
with delay

The polar plots of G( jω) with and without delay are shown in Fig. 3.15.
3.4 Polar Plot (Nyquist Plot) 351

Im

10
O 0 Re

G(j ) Plot without delay

G(j ) Plot with delay

Fig. 3.15 Polar plot with and without delay

3.4.4 Polar Plot of Prototype Second Order System

Consider the following T.F. of a prototype second order system

ωn2
G(s) = (3.9)
(s 2 + 2ζ ωn s + ωn2 )
ωn2
G( jω) =
−ω2 + j2ζ ωn ω + ωn2
1
= ω2
(1 − ωn2
) + j2ζ ωωn
1
|G( jω)| =  (3.10)
ω2 2
+ 4ζ 2 ωω2
2
(1 − ωn2
)
n

2ζ ωωn
∠G( jω) = − tan−1   (3.11)
ω2
1− ωn2

For ω = 0,
G( jω) = 1∠0◦

For ω = ∞,
G( jω) = 0∠−180◦

At ω = ωn ,
352 3 Frequency Response Analysis

Im

1
0 0 Re

0.8
0.6
n
0.4

n
0.2

Fig. 3.16 Polar plot of a prototype second order system

∠G( jω) = −90◦

The polar plot starts from +ve real for ω = 0 from 1, traverses through the fourth and
third quadrants and ends at the origin when ω = ∞. The phase angle of G( jω) is
−90◦ when ω = ωn . For ζ = 0.2, 0.4, 0.6 and 0.8, polar plots are sketched and shown
in Fig. 3.16. For lower values of ζ , the polar plot is larger in size. As ω is increased,
the size shrinks and the polar plot more or less takes the shape of a semi-circle.

3.5 Correlation Between Transient Response and

Frequency Response

In Chap. 2, expressions were derived for transient response specifications such as

overshoot, rise time, settling time and so on in terms of damping ratio ζ and undamped
natural frequency ωn of a prototype second order system. Similar expressions in
the frequency domain can be derived for prototype second order systems. These
expressions form the specifications in the frequency domain. Thus, if specifications
in the frequency domain are known, the damping ratio and undamped frequency of
oscillation can be determined. Once these parameters are known, the time domain
specifications are determined. Unfortunately, this correlation exists only for a second
order system and it is not valid for a higher order system. However, the higher order
3.5 Correlation Between Transient Response and Frequency Response 353

2
R(s) n C(s)
G(s) =
s(s+2 n)

Fig. 3.17 Prototype second order system

systems are reduced to lower second order system and one-to-one correlation can be
approximately established.
Consider the following T.F. of a prototype second order system shown in Fig. 3.17

C(s) ωn2
= M(s) = 2 (3.12)
R(s) s + 2ζ ωn s + ωn2

The sinusoidal T.F. of Eq. (3.12) is

ωn2
M( jω) =
−ω2 + j2ζ ωωn + ωn2
1
=  
1 − ω2 + j2ζ ωωn
ω2
n

Let u = ω/ωn

1
M( ju) =
(1 − u2) + j2ζ u
1∠− tan−1 (1−u
2ζ u
2)
=  (3.13)
(1 − u ) + 4ζ 2 u 2
2 2

The maximum value of M is obtained by differentiating M( ju) with respect to u

and equating d M( ju)/du = 0. The corresponding peak value is called the resonant
peak denoted by Mr . The frequency at which the resonant peak occurs is called
Resonant Frequency denoted by ωr

d M( ju) [2(−2u)(1 − u 2 ) + 8ζ 2 u]
= 3 =0
du 2[(1 − u 2 )2 + 4ζ 2 u 2 ] 2
4u(1 − u 2 ) = 8ζ 2 u

u = 1 − 2ζ 2

Substituting for u, we get the resonant frequency as

354 3 Frequency Response Analysis


ωr = ωn 1 − 2ζ 2 (3.14)

Substituting u = 1 − 2ζ 2 in Eq. (3.14), the resonant peak is obtained as

1
Mr = 
(1−1 + 2ζ 2 )2 + 4ζ 2 (1 − 2ζ 2 )

1
Mr =  (3.15)
2ζ 1 − ζ 2

2ζ ωωn
∠M( jω) = − tan−1 ω2
(3.16)
1− ωn2

Equation (3.15) is valid for 0 ≤ ζ ≤ 0.707. For ζ > 0.707 there won’t be any res-
onance. For ζ > 0.707, the resonant frequency ωr = 0 and Mr = 1. As seen from
Eq. (3.15), the resonant peak is a function of the damping ratio. If the resonant peak
is higher, the system is loosely damped. If the resonant peak is smaller, the system
is heavily damped. Thus, in the frequency domain specifications, resonant peak and
resonant frequency are known, then it is possible to get the damping ratio ζ and
undamped frequency ωn . From this, the time domain specifications can be obtained.
Thus, a correlation is obtained between time domain and frequency domain which
is true only for a second order system. More frequency response specifications are
obtained and discussed in the later part of the chapter. The linear frequency response
plots of Eqs. (3.15) and (3.16) are shown in Fig. 3.18a and b, respectively.

3.6 Bandwidth of a Second Order System

The frequency response of a certain second order system is as shown in Fig. 3.19.
As established in Sect. 3.5, correlation between the frequency response and time
response can be established by another frequency response specification called
BANDWIDTH. The bandwidth ωb of the closed loop frequency response is defined
as the frequency at which the magnitude response curve |M( jω)| does not fall below
0.707 of its value at zero frequency. In terms of decibel (dB), it is defined as the fre-
quency at which 20log10 |M( jω)| does not fall below −3 dB from its value at zero
frequency. For a prototype second order system the expression for the bandwidth is
derived as given below. From Eq. (3.12) we write
3.6 Bandwidth of a Second Order System 355

Fig. 3.18 a Magnitude versus normalized frequency. b Phase angle versus normalized frequency
of a second order closed loop system

ωn2
M( jω) =
−ω2 + j2ζ ωωn + ωn2
ωn2
|M( jω)| = 
(ωn2 − ω2 )2 + 4ζ 2 ω2 ωn2

Let ω√ √ at ω = 0) is reduced to 0.707

b be the frequency at which the D.C. gain (gain
or 1/ 2. Substituting ω = ωb and |M( jωb )| = 1/ 2 in the above equation we get
356 3 Frequency Response Analysis

Im

1
M(j )

0.707

Bandwidth, b

0
b

Fig. 3.19 Frequency response and bandwidth of a second order system

1 ωn2
√ =
2 (ωn2 − ωb2 )2 + 4ζ 2 ωb2 ωn2

Solving the above equation, we get

 
Bandwidth = ωb = ωn 1 − 2ζ 2 + (4ζ 4 − 4ζ 2 + 2) (3.17)

ωb /ωn versus ζ is plotted and shown in Fig. 3.20. The bandwidth plays an impor-
tant role in the frequency response characteristic. Large bandwidth means the system
will pass a wide range of frequency of input signal. In the time domain it amounts
to smaller rise time and hence fastness of transient response. Bandwidth is inversely
proportional to the rise time. A flat frequency response characteristic gives higher
bandwidth which requires high-cost equipment. However, higher bandwidth allows
noise to pass through the system. Further, certain systems require low bandwidth and
hence high frequency noise is attenuated.

3.6.1 Cut off Frequency and Cut Off Rate

√
In Fig. 3.19, the frequency ωb at which the magnitude curve has 1/ 2 times its D.C.
value is called the cut off frequency. The slope of the magnitude curve at cut off
frequency is called cut off rate. d M( jω)/dω is called cut off rate.
3.6 Bandwidth of a Second Order System 357

1.554

1.2
n
BW/

0.8

0.4

0
0.2 0.4 0.6 0.8 1.0 1.2
Damping ratio,

Fig. 3.20 Bandwidth/ωn versus damping ratio of a second order system

Example 3.6
The open loop T.F.s of certain control systems are given. In each case, determine the
resonant peak, resonant frequency and the bandwidth of the closed loop system.

64
(a) G(s) =
s(s + 8)
16
(b) G(s) =
s(s + 8)

Solution
(a)

64
G(s) =
s(s + 8)
G(s)
M(s) =
1 + G(s)
64
= 2
(s + 8s + 64)

Comparing the above equation with the standard second order system, we get
358 3 Frequency Response Analysis

ωn2 64
= 2
(s 2 + 2ζ ωn s + ωn2 ) s + 8s + 64
ωn = 8
ζ = 0.5

Using Eq. (3.15), we get

1
Mr = 
2ζ 1 − ζ 2
1
= 
2 × 0.5 1 − (0.5)2
= 1.1547

Using Eq. (3.14), we get


ωr = ωn 1 − 2ζ 2

= 8 1 − 2(0.5)2
= 5.6569 rad/sec

Using Eq. (3.17), we get

 
ωb = ωn 1 − 2ζ 2 + 4ζ 4 − 4ζ 2 + 2
 
= 8 1 − 2(0.5)2 + 4(0.5)4 − 4(0.5)2 + 2
√
= 8 1 − 0.5 + 1.118
= 10.176 rad/sec

Mr = 1.1547
ωr = 5.6569 rad/sec
ωb = 10.176 rad/sec

(b)

16
G(s) =
s(s + 8)
16
M(s) = 2
s + 8s + 16

Comparing this with the standard second order system, we get ωn = 4 and ζ =
1. Since ζ > 0.707, Mr = 1 and ωr = 0. The bandwidth is calculated using
Eq. (3.17)
3.6 Bandwidth of a Second Order System 359

R(s) K C(s)
G(s) =
s(s+b)

Fig. 3.21 Transfer function, for Example 3.7


√
ωb = 4 1 − 2 + 2
= 2.5737 rad/sec

Mr = 1
ωr = 0
ωb = 2.5737 rad/sec

Example 3.7
For the system shown in Fig. 3.21 find the value of K and b to satisfy the following
frequency domain specifications. Mr = 1.2; ωr = 14 rad/sec. For these values of K
and b, calculate the settling time and bandwidth.

(Anna University, May, 2011)

Solution From Eq. (3.15), we have

1
Mr = 
2ζ 1 − ζ 2
1
1.2 = 
2ζ 1 − ζ 2

Squaring and cross multiplying, we get

ζ 2 (1 − ζ 2 ) = 0.1736
ζ − ζ + 0.1736 = 0
4 2

Solving the above equation, we get

ζ = 0.88 or ζ = 0.473

Since resonance occurs for ζ < 0.707, the value of ζ = 0.473 alone is considered.
360 3 Frequency Response Analysis

From Eq. (3.14), we have


ωr = ωn 1 − 2ζ 2
ωr
ωn = 
1 − 2ζ 2
14
= √ = 18.834
1 − 2 × 0.4732

From Fig. 3.21, we get

C(s) K
= 2 (a)
R(s) (s + bs + K )

For ζ = 0.473 and ωn = 18.834

C(s) ωn2
= 2
R(s) (s + 2ζ ωn s + ωn2 )
354.72
= 2 (b)
s + 17.817s + 354.72

Comparing Eqs. (a) and (b), we get

K = 354.72
b = 17.817

The settling time for 2% error tolerance is

3.91
ts =
ζ ωn
3.91
= = 0.439 sec
0.473 × 18.834

From Eq. (3.17), the bandwidth is

 
ωb = ωn 1 − 2ζ 2 + (4ζ 4 − 4ζ 2 + 2)
 
= 18.834 1 − 2(0.473)2 + 4(0.473)4 − 4(0.473)2 + 2
√
= 18.834 1 − 0.447 + 1.3
= 25.673 rad/sec
3.6 Bandwidth of a Second Order System 361

K = 354.72
b = 17.817
ωb = 25.673

Example 3.8
The magnitude of frequency response of an √under-damped second order
√ system is
5 at a frequency of 0 rad/sec. and peaks to 5/3 at a frequency of 5 2 rad/sec.
Determine the transfer function.

Solution

5
Mr =
3
√
ωr = 5 2

1 5
Mr =  =
2ζ 1 − ζ2 3

Squaring and cross multiplying, we get

ζ 2 (1 − ζ 2 ) = 0.15
ζ 4 − ζ 2 + 0.15 = 0

Solving the above equation, we get

ζ = 0.429 or ζ = 0.9

Since ζ = 0.9 is greater than 0.707, only the value of ζ = 0.429 is considered
√
ωr 5 2
ωn =  =
1 − 2ζ 2 1 − 2 × (0.429)2
= 8.894

The closed loop T.F. is written as

K ωn2
M(s) =
s 2 + 2ζ ωn s + ωn2
K 79.1
= 2
s + 7.639s + 79.1
M(0) = K
362 3 Frequency Response Analysis

But it is given that M(0) = 5 and K = 5

5 × 79.1
M(s) =
s 2 + 7.639s + 79.1

395.5
M(s) =
(s 2 + 7.639s + 79.1)

Example 3.9
A certain second order closed loop system when subjected to step input exhibits a
peak overshoot of 16.3% and the rise time is observed as 0.2 sec. Find the values of
resonant peak Mr , resonant frequency ωr and the bandwidth.

Solution The peak overshoot is given by

−ζ π
Mp = e  = 0.163
1 − ζ2

Taking ln on both sides of the equation, we get

−ζ π
 = −1.814
1 − ζ2

Solving the above equation, we get

ζ = 0.5
π
φ = cos−1 ζ = rad
3
(π − φ)
tr = 
ωn 1 − ζ 2
(π − φ) π − (π/3)
ωn =  = 
tr 1 − ζ 2 0.2 1 − (0.5)2
= 12.1 rad/sec

Resonant peak is given by

1 1
Mr =  = 
2ζ 1 − ζ 2 2 × 0.5 1 − (0.5)2
= 1.155

The resonant frequency is given by

3.6 Bandwidth of a Second Order System 363
 
ωr = ωn 1 − 2ζ 2 = 12.1 1 − 2(0.5)2
= 8.556 rad/sec

The bandwidth is given by

 
ωb = ωn 1 − 2ζ 2 + 4ζ 4 − 4ζ 2 + 2
 
= 12.1 1 − 2(0.5)2 + 4(0.5)4 − 4(0.5)2 + 2
√
= 12.1 1 − 0.5 + 1.25
ωb = 14.82

Mr = 1.155
ωr = 8.556
ωb = 14.82

Example 3.10
A certain unity feedback control system when subjected to sinusoidal input has a
resonant peak of 1.2 and the corresponding frequency is 8 rad/sec. Determine the
bandwidth of the closed loop system. Also determine the % overshoot, rise time and
settling time when subjected to step input.

Solution
1
Mr = 
2ζ 1 − ζ 2
1 1
4ζ 2 (1 − ζ 2 ) = 2 =
Mr (1.2)2
ζ 2 (1 − ζ 2 ) = 0.1736
ζ − ζ + 0.1736 = 0
4 2

Solving for ζ , we get

ζ 2 = 0.7764 or ζ 2 = 0.2236
ζ = 0.882 or ζ = 0.473

Resonance occurs only for ζ < 0.707. Hence, ζ = 0.473 is considered

364 3 Frequency Response Analysis

ωr = ωn 1 − 2ζ 2
ωr 8
ωn =  =√
1 − 2ζ 2 1 − 2 × 0.2236
= 10.76 rad/sec

ωd = ωn 1 − ζ 2
√
= 10.76 1 − 0.2236 = 9.481
φ = cos−1 ζ = cos−1 0.473
= 61.77◦ or 1.078 rad.
ζ ωn = 0.473 × 10.76 = 5.1
 
ωb = ωn 1 − 2ζ 2 + 4ζ 4 − 4ζ 2 + 2
 
= 10.76 1 − 2 × 0.2236 + 4(0.2236)2 − 4(0.2236) + 2
= 14 rad/sec

−ζ π −0.473π
%M p = e  × 100 = e √ × 100 = 18.52%
1−ζ 2 1 − 0.2236
(π − φ) (π −1.078)
tr = = = 0.2177 sec
ωd 9.481
3.91 3.91
ts = = = 0.767 sec
ζ ωn 5.1

ωb = 14 rad/sec; %M p = 18.52%
tr = 0.2177 sec; ts = 0.767 sec

3.7 Bode Plots

In Sect. 3.4, the display of frequency response in polar graph was explained for
the frequency range 0 ≤ ω ≤ ∞. The polar plot has the main drawback in that
information near the origin could not be obtained with clarity since the magnitude and
frequency both vary over many powers of 10 and they are got compressed. Further,
the method does not clearly indicate the contribution made by each pole and zero in
the frequency response plot which will be very useful in the design. To overcome
these difficulties, Hendrik Wade Bode, who worked in Bells Telephone Laboratories,
U.S.A. in the year 1938, developed his asymptotic phase and magnitude plots what
is now popularly known as Bode plots.
These plots are the most widely used means of displaying and communicating
frequency response information. Bode plot is very commonly used to communicate
3.7 Bode Plots 365

frequency response specifications in stereo equipment such as amplifiers, speakers,

microphones and headsets. Further, these plots give useful information for designing
a variety of control systems. Stability criteria can be interpreted on Bode plots, and
furthermore, there are numerous design techniques available which are based on
Bode plots. Bode plot contains two separate plots and they are called magnitude and
phase plots. In magnitude plot the gain |G( jω)| is expressed in decibel (dB) and
displayed in the vertical axis. 20log10 |G( jω)| is called a decibel which is written in
the abbreviated form as dB. The horizontal axis represents frequency on a log scale.
In the second plot which is called the phase plot ∠G( jω) is plotted in the vertical
axis while frequency ω is represented in the horizontal axis which is graduated on a
log scale.
Advantages of Bode Plots
1. Bode plots are sketched by approximating the magnitude and phase with straight
lines which are easily sketched in Bode plot.
2. Since the horizontal axis is on log scale, it takes care of logarithmic value of
frequency ω. Thus, a wide range of frequency study is possible with clarity at all
frequencies.
3. The vertical axis represents the magnitude of G( jω) which is expressed in log-
arithmic form. Very large amplitude of |G( jω)| can be represented in this way.
4. The magnitude of G( jω) is expressed in dB. The multiplications and divisions in
|G( jω)| become addition and subtraction in Bode plot display which can be eas-
ily manipulated. Similarly, phase angles of zeros and poles of G( jω) are added
and subtracted from each other algebraically.
5. Frequency response specifications such as phase margin, gain margin, phase
crossover and gain crossover frequencies are more easily determined from the
Bode plot than the polar plot.
6. In the design using Bode plots, the effects of adding controllers and their param-
eters can be easily visualized than on the polar plot.
Disadvantages of Bode Plot
1. If the Bode plots are manually drawn without the use of computers, the plots
become approximate. Unless otherwise, necessary corrections are made in appro-
priate places in the Bode plots, which is very tedious to accomplish, the results
obtained deviate from the actual ones.
2. Absolute and relative stability of the system are obtained from Bode plots provided
the system under study is of minimum phase. For non-minimum phase transfer
function, the Bode plot cannot be used to assess the absolute and relative stability.
366 3 Frequency Response Analysis

A Historical Note: Hendrik Wade Bode (1905–1982)

Hendrik Wade Bode was an American engineer,

researcher, inventor, author and scientists of Dutch
ancestry. Bode was born in Madian, Wisconsin, USA
in the year 1905. His father was a professor of educa-
tion at the University of Illinois at Urbana. He received
his B.A. degree in 1924 at age 19 his M.A. degree in
1926 both in mathematics and his Ph.D. in physics
from Columbia University in the year 1935. He joined
Bell Laboratories in New York City as a design engi-
neer of electronic filters and equalizers. During World
War II, Dr. W.H. Bode joined the armed services of
USA. During this period, he developed automatic anti-
aircraft gun control system whereby radar information was used to locate the enemy
aircraft. The information was feedback to anti-aircraft artillery to shoot down the
enemy’s aircrafts. He also developed robot weapons which brought down the NaZe
V-1 flying bombs, the weapon which terrorized London city during World War II. His
research work impacted many other engineering disciplines such as computer, robot
and mobile phone. In academic circle, worldwide he is widely known for developing
the asymptotic magnitude and phase plots that bear his name, the Bode Plot. He was
married to Barbara Bode. They had two children: Dr. Katharine Bode Darlington
and Mrs. Anne Hathway Bode Aarnes. Hendrik Wade Bode died on 21 June 1982 at
the age of 76, at his home in Cambridge, Massachusetts.

3.7.1 Bode Asymptotic Magnitude Plot

Consider the following transfer function

K (1 + sT1 )(1 + sT2 ) · · · (1 + sTm )

G(s) = (3.18)
s k (1 + sTa )(1 + sTb ) · · · (1 + sTn )

Equation (3.18) is said to be written in Time Constant Form. The sinusoidal T.F.
of the above equation is written as

K (1 + jωT1 )(1 + jωT2 ) · · · (1 + jωTm )

G( jω) = (3.19)
( jω)k (1 + jωTa )(1 + jωTb ) · · · (1 + jωTn )

K (1 + ω2 T12 )(1 + ω2 T22 ) · · · (1 + ω2 Tm2 )
|G( jω)| = (3.20)
ωk (1 + ω2 Ta2 )(1 + ω2 Tb2 ) · · · (1 + ω2 Tn2 )
∠G( jω) = −k × 90◦ + tan−1 ωT1 + tan−1 ωT2 + · · · + tan−1 ωTm
−(tan−1 ωTa + tan−1 ωTb + · · · + tan−1 ωTn ) (3.21)
3.7 Bode Plots 367

Bode magnitude plot is drawn for 20log10 |G( jω)| versus log10 ω and phase plot is
drawn for ∠G( jω) versus log10 ω. We discuss below drawing of magnitude plot first
followed by the phase plot. 20 log |G( jω)| is called a decibel which is written in
the abbreviated form as dB. Now taking log10 on both sides of Eq. (3.20) we get
(hereafter log10 is simply written as log)
 
dB = 20 log K + 20 log (1 + ω2 T12 ) + · · · + 20 log (1 + ω2 Tm2 )
 
−20 log (1 + ω2 Ta2 ) − · · · − 20 log (1 + ω2 Tn2 ) (3.22)

In Eq. (3.22), each term on the right-hand side is plotted separately and added or
subtracted accordingly and the total magnitude plot in dB is obtained. The curve is
sketched on a semi-log graph. dB is represented in the vertical axis, and frequency
ω is represented in the horizontal axis which is marked in logarithmic scale.
In Eq. (3.19), G( jω) can contain just five simple factors which are given below
1. G( jω) = K (a constant factor)
2. G( jω) = K ( jω)± p (poles or zeros at the origin of order p)
3. G( jω) = (1 + jωT )±q (poles or zeros at s = −(1/T ) of order q)
4. G( jω) = [ωn2 /−ω2 + j2ζ ωn ω + ωn2 ]±r (complex poles and zeros of order r )
5. G( jω) = e− jωT d (Time delay)
We discuss below how Bode magnitude plots are sketched for the above five factors.
(1)

G(s) = K
G( jω) = K
|G( jω)| = K
dB = 20 log K (3.23)

The magnitude of G( jω) is independent of frequency and hence dB is constant

at all frequency. For K = 1, dB = 20 log 1 = 0. For K > 1, dB is +ve and for
K < 1, dB is −ve. The Bode magnitude plot is shown in Fig. 3.22.
(2)

G(s) = K s (3.24)
G( jω) = j K ω
|G( jω)| = K ω
dB = 20 log K ω
1
At ω = , dB = 20 log 1 = 0
K
10
At ω = , dB = 20 log 10 = 20
K
368 3 Frequency Response Analysis

Fig. 3.22 Bode plot of

G( jω) = K

20
dB K 10

0
log
dB K 1

20
K 0.1

0.1
At ω = , dB = 20 log 1 = −20
K
This is a straight line. The Bode plot is shown in Fig. 3.23a.
Figure 3.23a is drawn using the following steps:
1. Find ωl , the frequency at which the magnitude of G( jω) = 1. In the above
case |G( jω)| = K ω and hence ωl = 1/K . At this frequency the dB is
20 log 1 = 0. This corresponds to point A in Fig. 3.23a.
2. Choose ω = 10 ωl = 10/K . At this point, the dB is 20 log 10K /K = 20.
This corresponds to point B.
3. Join points A and B and extend it as much as you want. The straight line
drawn corresponds to Bode plot for G( jω) = j K ω.
4. The above straight line is called +20dB/decade line. This is the slope of
the straight line. The meaning of decade is the frequency range which is
multiplied or divided by the factor 10. If the frequency 1 is multiplied by 10,
the frequency range from 1 to 10 is a decade. The frequency range from 10
to 100 is a decade. The frequency range from 100 to 1000 is a decade and
so on. The frequency range from 1 to 100 is a two decade. The frequency
range from 1 to 1000 is a 3 decade and so on.
The meaning of the slope is to be well understood. For every decade of
frequency rise, the increase in dB is 20. Thus, for the frequency range from
1 to 10, the increase in dB is 20, from 10 to 100, the increase in dB is 20
and so on. For the frequency range from 1 to 100, the increase is dB 40.
There are two decades in the frequency range from 1 to 100. The straight line
simply follows the law y = mx, where m is the slope which is determined
by the ratio of the vertical height to the base. x is the logarithmic frequency
difference. That is x = log ω2 /ω1 . y is the increase in the decibel.
5. The straight line in Fig. 3.23a can also be expressed as 6 dB/octave straight
line. Octave means, the frequency is multiplied or divided by a factor 2.
Thus, at ω = 2ωl in Fig. 3.23a dB = 20 log 2ωl /ωl = 6. Thus, 20 dB/dec
3.7 Bode Plots 369

(a) (b)

dB One decade

20dB/dec. line y mx
20dB
B
One octave
y
6dB
0 A
10 log
dB 0.1 10 l K
K 2 1 log 2
20dB l
1 x log 2
l K 1

(c) (d)
dB
p 20dB

p 20dB/dec
40dB 40dB/dec

log
0 0
log
10 l 10 l

1 1
l K l p
√K
dB

(e) (f)
dB dB
20dB/dec

10 log 10 log
l 0 l
0

K p
p 20dB

l l √K
20dB p 20dB/dec
dB
dB

Fig. 3.23 a Bode plot of G(s) = K s. c Bode plot for G(s) = K s 2 , and d G(s) = K s m . e Bode
K K
plot for G(s) = , and f G(s) = p
s s
370 3 Frequency Response Analysis

has the same meaning as 6 dB/octave. In this book, we follow 20 dB/dec.

The frequency ωl is calculated from K ωl = 1.
Now consider the following T.F.

G(s) = K s 2
|G( jω)| = K ω2
K ωl2 = 1
1
ωl = √
K
√
At ω = 10ωl , dB is 20 log K (10/ K )2 = 40 dB. Thus, the straight line has a
slope of +40 dB/dec slope as shown in Fig. 3.23c. In general for G(s) = K s p ,
the Bode plot is a straight line with a slope of p × 20 dB/dec slope and ωl = p √1 K .
Now consider the following T.F.

K
G(s) =
s
K
|G( jω)| =
ω
|G( jω)| = 1 at ω = K or ωl = K . At this frequency the dB = 20 log 1 = 0. At
ω = 10ωl = 10K
K
dB = 20 log = −20
10K
Thus, the slope m here is −20 dB/dec√ and the Bode plot is a straight line. In
general, for G(s) = K /s p , ωl = p K and the slope is − p × 20 dB/dec. These
plots are shown in Fig. 3.23e and f, respectively.
It is to be noted that as the order of integration or differentiation increases, the
slope of the straight line plot also increases and plot becomes much steeper.
(3)

G(s) = (1 + sT ) (3.25)
G( jω) = 1 + jωT

|G( jω)| = (1 + ω2 T 2 )

Now consider the T.F. shown above. In the Bode plots drawn for the transfer func-
tions in cases 1 and 2, there was absolutely no approximation incorporated in the
Bode plot. However, for quicker sketching of Bode plot, now some approxima-
tions are made. These approximations can be improved, if necessary by adding
corrections in appropriate places and totally nullified by computer programmed
plots using MATLAB. For the T.F. described in case 3 above for ωT <<< 1,
3.7 Bode Plots 371

|G( jω)| = 1
dB = 20 log 1 = 0

For ωT >>> 1,

|G( jω)| = ωT

The above plot is similar to case 2. It is a straight line with +20 dB/dec
slope and ωl = 1/T . Thus, an approximate plot is drawn with 0 dB upto
ω = ωl = 1/T . Another approximate plot is drawn with +20 dB/dec slope for
the frequency range ωl ≤ ω < ∞. These approximate plots are called Asymp-
totes. The points at which these two asymptotes meet in the horizontal axis and
the corresponding frequency is called Corner frequency ωc . Thus, ωl is now
replaced by the conventional symbol ωc when asymptotic plots are drawn.
The corner frequency of a Bode plot is therefore defined as the frequency
at which the two asymptotes (approximate plots) meet.
The Bode plot for G(s) = (1 + sT ) is shown in Fig. 3.24a.
In the asymptotic plot, by introducing necessary corrections at strategic points,
a Bode plot which is reasonably closer to the actual Bode plot can be drawn.
These points are chosen at frequencies one octave above the corner frequency
and one octave below the corner frequency.
Correction at the Corner Frequency

|G( jω)| = 1 + ω2 T 2
√
dB|ω= T1 = 20 log 2 = 3 dB

Correction at One Octave Above the Corner Frequency

Actual dB at ω = 2/T is
√
dB = 20 log 1+4=7

Asymptotic dB at ω = 2/T is

dB = 20 log 2 = 6

The correction to be made is (7 − 6) = 1 dB.

Correction at One Octave Below the Corner Frequency
Actual dB at ω = 1/2T is

1
dB = 20 log 1 + =1
4

Asymptotic dB at ω = 1/2T is
372 3 Frequency Response Analysis

(a)

20 dB/dec.
20 dB
Low frequency asymptote High frequency
asymptote

dB 1dB Asymptote plot

Bode plot after
error correction
3dB 2 c
1dB
1 10 log
c c
c T
2
Corner frequency
(b)

q 20dB/dec
slope straight line
q 20dB

log
0
1 10
c c
T

(c) (d)
dB dB
1 1
c T 10 log c T
c 10 c
0 0
log

20dB
dB q 20dB/dec
dB stright line
20dB/dec
q 20dB

stright line

Fig. 3.24 a Bode plot for G(s) = (1 + sT ). b Bode plot for G(s) = (1 + sT )q . Bode plot for c
1 1
G(s) = , and d G(s) =
(1 + sT ) (1 + sT )q
3.7 Bode Plots 373

dB = 0

The correction to be made is 1 − 0 = 1 dB.

The complete Bode plot is shown in Fig. 3.24a. Since the Bode plot is very
extensively used for the design of control system compensators, where the Bode
plot has to be drawn according to the poles and zeros alterations, the asymptotic
Bode plot is invariably used. However, the final design is verified by simulation
study of actual Bode plot obtained with the help of MATLAB. We discuss below
the asymptotic Bode plot of other factors.
Now consider the following T.F.

G(s) = (1 + sT )q
G( jω) = (1 + jωT )q

|G( jω)| = [ (1 + ω2 T 2 )]q
dB = 10 × q log(1 + ω2 T 2 )

This is the straight line with ωc = 1/T and slope q × 20 dB/dec as shown in
Fig. 3.34b.
The Bode plot for G(s) = 1/(1 + sT ) and G(s) = 1/(1 + sT )q are the mir-
ror images of Fig. 3.24a and b, respectively, and they are shown in Fig. 3.24c,
d. It is to be noted that the corner frequency for G(s) = (1 + sT ) and for
G(s) = (1 + sT )q are the same at ω = 1/T .
(4) Complex quadratic factors

ωn2
G(s) = (3.26)
s 2 + 2ζ ωn s + ωn2
ωn2
G( jω) =
−ω2 + j2ζ ωωn + ωn2
ωn2
|G( jω)| = 
(ωn2 − ω2 )2 + 4ζ 2 ω2 ωn2
1
=   2
1 − ωω2 + 4ζ 2 ωω2
2 2

n n

For at ω <<< ωn ,

|G( jω)| = 1
dB = 0

For ω >>> ωn ,
374 3 Frequency Response Analysis

ωn2
|G( jω)| =
ω2
ωn2
dB = 20 log
ω2
= −40dB/dec with ωc = ωn

At ω = ωn

1
|G( jω)| =
2ζ
dB = −20 log 2ζ
For ζ = 0.5, dB = 0
ζ > 0.5, dB = −ve
ζ < 0.5, dB = +ve

For the quadratic factor, the Bode asymptotic plot is drawn with −40 dB/dec
straight line with corner frequency ωc = ωn . This is represented in Fig. 3.25a.
The variation at ω = ωn with respect to ζ is shown in Fig. 3.25b.
(5) Time delay

G(s) = e−sT d
|G( jω)| = e− jωTd
|G( jω)| = 1

The magnitude plot is a zero dB line at all frequencies.

3.7.2 Bode Asymptotic Phase Angle Plots

For the five factors for which the Bode magnitude plot was discussed, we give below
the asymptotic phase angle plots (Table 3.2).
(1)

G(s) = K
G( jω) = K

The phase angle is zero at all frequencies.

3.7 Bode Plots 375

(a)
dB

c n 10 c
0
log

dB

40dB
40dB/dec

(b)

Fig. 3.25 Bode plot for G(s) = ωn2 /s 2 + 2ζ ωn s + ωn2 . a Asymptotic and b Effect of ζ at ωc

Table 3.2 Summary of Bode magnitude plot

K K
T.F. G(s) K Ks K s2 (1 + sT )
s s2
√ 1 1 1
Corner − K K √
frequency, K K T
ωc
Slope, 0 −20 −40 20 40 20
dB/decade
1 1 ωn2
T.F. G(s) (1 + sT )2
(1 + sT ) (1 + sT )2 (s 2 + 2ζ ωn s + ωn2 )
1 1 1
Corner ωn
frequency, T T T
ωc
Slope, 40 −20 −40 −40
dB/decade
Note Phase plots are normally drawn for the actual phase angle which is calculated at each ω. The
asymptotic phase plot is usually avoided
376 3 Frequency Response Analysis

(2)

G(s) = K s
G( jω) = j K ω
∠G( jω) = 90◦
G(s) = K s 2
G( jω) = −K ω2
∠G( jω) = 180◦

In general for

G(s) = K s p
∠G( jω) = ∠ p × 90◦

For G(s) = K /s p

∠G( jω) = ∠− p × 90◦

The phase angle plots are shown in Fig. 3.26.

(3)

G(s) = (1 + sT )
G( jω) = 1 + jωT
∠G( jω) = tan−1 ωT

The corner frequency is ωc = 1/T . At ωc , φ = tan−1 1 = 45◦ . At ω = 0, φ = 0;

At ω = ∞, φ = 90◦ . For G(s) = (1 + sT )q these angles will be multiplied by
the factor q. Similarly, for G(s) = 1/(1 + sT ), the phase angle is the negative
of (1 + sT ). The phase angle plots are shown in Fig. 3.27.
(4) Quadratic factor

ωn2
G(s) =
s2 + 2ζ ωn s + ωn2
ωn2
G( jω) =
(ωn2 − ω2 ) + j2ζ ωωn

2ζ ωωn
φ = − tan−1
ωn2 − ω2

For ω = 0,
φ=0
3.7 Bode Plots 377

(a) (b)

log
0
90º

90º

0
log
K
Fig. 3.26 Bode phase plots. a G(s) = K s and b G(s) =
s

(a) (b)

1
c T log
90º 90º

45º 45º

0 90º
1
c T
1
Fig. 3.27 Bode phase angle plots. a G(s) = (1 + sT ) and b G(s) =
(1 + sT )

At corner frequency

ω = ωn
φ = −90◦

For ω → ∞
φ → −180◦

The phase plots for different values of ζ are shown in Fig. 3.28.
378 3 Frequency Response Analysis

ωn2
Fig. 3.28 Phase angle plots for G(s) =
(s 2 + 2ζ ωn s + ωn2 )

Fig. 3.29 Phase angle plots

of G(s) = e−sTd log
0

90º

180º

270º

360º

(5) Time delay

G(s) = e−sT d
G( jω) = e− jωTd
G( jω) = ∠−ωTd in rad
= ∠−57.3ωTd in degree

This is a straight line with a slope Td . The phase angle plot is shown in Fig. 3.29.

3.8 Step by Step Procedure to Draw Bode Magnitude Plot

1. The poles and zeros of the given T.F. is written in the time constant form as given
in Eq. (3.18).
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 379

2. Identify the low frequency term in the transfer function. This may be K , K s, K s 2 ,
K /s, K /s 2 , etc. The Bode plot starts at low frequency from any of these terms.
If it contains pure constant, the low frequency response starts with 20 log K .
Otherwise find ωl where K s or K /s, etc. magnitude becomes 1 or its decibel is
zero.
3. Identify the corner frequencies of each pole and zero and write in the ascending
order. Locate these corner frequencies in the log ω scale in the horizontal axis.
Draw vertical dotted lines passing through these corner frequencies. Also locate
ωl in the log ω axis.
4. Start the Bode plot for the low frequency part of the T.F. which will pass through
ωl had there been no other pole or zeros to interfere. (If there is no pure integrator
or differentiator, the low frequency response is due to K only and a horizontal
line is drawn).
5. When a pole or a zero interferes with the low frequency magnitude response, stop
and calculate the dB at that point. If a zero interferes, the slope is changed by
+20 dB/dec. If two zero interfere the slope is changed by +40 dB/dec and so on.
On the other hand if a pole interferes, the slope of the straight line is changed
by −20 dB/dec. When two poles interfere, the slope is changed by −40 dB/dec
and so on. Thus, with the new slope the magnitude plot is extended up to the
next corner frequency where poles or zeros interfere. At that point, the Bode
plot is to be stopped and the dB is calculated. Depending upon whether the zero
interferes or the pole interferes, the new slope of the magnitude plot is decided.
The procedure is repeated till the Bode plot crosses past all the corner frequencies
and the dB either goes to ∞, 0 finite or −∞ depending upon G(s).

Example 3.11
Consider the following T.F.

3000
G(s) =
(s + 2)(1 + 10
s
)(s + 50)

Sketch the Bode magnitude plot. What is the gain crossover frequency? Also deter-
mine the position error constant K p .

Solution
(1)
3000
G(s) =
(s + 2)(1 + 10
s
)(s + 50)

The given T.F. is written in time constant form as

380 3 Frequency Response Analysis

30
G(s) =
(1 + s
2
)(1 + (s/10))(1 + (s/50))

The factors of G(s) written in ascending order of corner frequency are

G 1 (s) = 30
1
G 2 (s) =
(1 + 21 s)
1
G 3 (s) =
(1 + 101
s)
1
G 4 (s) =
(1 + 501
s)

The range of corner frequencies and the corresponding slopes of the straight
lines are given in the table below. The corner frequencies are ωc = 2, 10 and 50.

Corner frequency range ω≤2 2 ≤ ω ≤ 10 10 ≤ ω ≤ 50 ω ≥ 50

Slope in dB/dec 0 −20 −40 −60

(2) The corner frequencies are identified as 2, 10 and 50 and marked in Fig. 3.30a.
(3) The low frequency factor of the T.F. is 30. Taking decibel for this we get

dB = 20 log 30
= 29.5

A horizontal line ab starting from 29.5 dB is drawn.

(4) At point b, a pole starts and changes the slope from 0 to −20 dB/dec. Line bc is
drawn to represent this. The height h in the triangle bgc is obtained as

(higher frequency)
|h| = 20 log
(lower frequency)
10
= 20 log
2
= 14

The point c is located at (29.5 − 14) = 15.5 dB.

(5) At point c, where the corner frequency is 10, a pole interferes and changes the
slope from −20 dB/dec to −40 dB/dec. Straight line cde is drawn. The decibel
at e is calculated as in step 4.

50
|h| = 40 log
10
= 28

The decibel at e = 15.5 − 28 = −12.4.

3.8 Step by Step Procedure to Draw Bode Magnitude Plot 381

The frequency where the Bode magnitude plot crosses the 0 dB line is called
gain crossover frequency. The gain crossover frequency is calculated as
ωgc
40 log = 15.5
10
ωgc = 24.4

(6) At point e, another pole interferes and changes the slope from −40 dB/dec to
−60 dB/dec. Any frequency ω > 50 is chosen. We choose ω = 70. At this fre-
quency

70
|h| = 60 log
50
= 8.8

The point f is chosen with a dB of −12.4 − 8.8 = −21.2. The straight line
connecting e and f is drawn and extended. This completes the Bode magnitude
plot.
(7) The position error constant is calculated from the zero dB slope line. From
Fig. 3.30a.

20 log K p = 29.5
K p = 30

Figure 3.30a is transferred to semi-log graph and is shown in Fig. 3.30b.

K p = 30
ωgc = 24.4 rad/sec

How to Draw Bode Plot in Semi-log Graph

1. In semi-log graph in the vertical axis, the decibel is plotted by choosing appro-
priate scale factor.
2. For the given transfer function, the lower and upper frequencies are determined
so that the Bode plot is drawn with clarity at all frequencies. The lower frequency
is chosen as 0.01 or 0.1, or 1 or 10, etc. depending upon the requirement. The
frequency range 0.01 to 0.1 is called one decade. Similarly, 0.1 to 1.0 is another
decade.
3. In the decade 0.01 to 0.1, the frequencies are marked for each division as 0.01,
0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09 and 0.1. For the next decade the
frequencies are marked as 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 and 1. Like this
the frequencies in each decade are marked.
4. Since the frequency axis has already been converted into logarithmic scale, one
can choose the division directly for the given ω.
Now we transfer Fig. 3.30a to semi-log sheet and is shown in Fig. 3.30b.
382 3 Frequency Response Analysis

(a)

dB

0 dB/dec
a b 20dB/dec. line
29.5
h
15.5 c
g 40dB/dec. line
20 log Kp
gc 24.4
Pole
0
2 10 d 50 70 log
Pole Pole
12.4 e

60dB/dec. line

dB

21.2 f

(b)
10 2 10 1 100 101 102 103 104 105 106

50
Decade Decade Decade Decade
1 2 3 4
40

a 0 dB/dec b
29.5
dB 20dB/dec
20
15.5 c

10 40dB/dec

gc 24.4
0
2 10 d 50 70 log

10
12.4 e

f
21.2
dB 60dB/dec. line

30

40

3000
Fig. 3.30 a Bode magnitude plot for G(s) = . b Bode magnitude plot for
(s + 2)(1 + 10
s
)(s + 50)
3000
G(s) = drawn in semi-log graph
(s + 2)(1 + 10
s
)(s + 50)
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 383

Example 3.12
Consider the following transfer function

15(1 + 10s
)
G(s) =
s(1 + 4 )(1 + 20
s s
)

Sketch the Bode magnitude plot. Find the velocity error constant. What is the gain
crossover frequency?

Solution
(1) The given transfer function is in standard time constant form. The corner fre-
quencies are 4, 10 and 20. The frequency ωl where 15/ω = 1 or dB of 15/ω is
0 is obtained as ωl = 15. For a type 1 system this is K v . Hence, K v = 15 sec−1 .
The factors of G(s) in ascending order of corner frequencies are written as given
below.
15
G 1 (s) =
s
1
G 2 (s) =
(1 + 41 s)
1
G 3 (s) = 1 + s
10
1
G 4 (s) =
(1 + 20
1
s)

The range of corner frequencies and the corresponding slope of the straight lines
are given in the following table:

Corner frequency range ω≤4 4 ≤ ω ≤ 10 10 ≤ ω ≤ 20 ω > 20

Slope in dB/dec −20 −40 −20 −40

2. For ω ≤ 4, the magnitude contribution comes only from 15/ω. Hence

15
dB = 20 log
ω
This is a straight line with slope −20 dB/dec with ωl = 15. At ω = ωl = 15,
dB = 0. At ω = 4,
15
dB|ω=4 = 20 log = 11.5
4
and point b is located. The points at ωl and b are joined by a straight line and
extended beyond ω < 4. Only line ab is drawn, and the remaining portion is
drawn by dotted line as shown in Fig. 3.31.
384 3 Frequency Response Analysis

a
20dB/dec
b
11.5 40 dB/dec
gc 7.8 Zero
dB
Pole Pole
l Kv
0
4 c 10 15 20 30 log
4.4 20dB/dec
d
10.4
e
17.4 f

dB 40dB/dec

1
15(1+ 10s )
Fig. 3.31 Bode magnitude plot for G(s) = s(1+ 4s )(1+ 20
s
)

3. At point b, the pole 1/ 1 + 41 s changes the slope from −20 dB/dec to −40 dB/dec
and the straight line goes with this slope till ω = 10. At ω = 10, the dB is calcu-
lated as
10
dB|ω=10 = 11.5 − 40 log = −4.4
4
Line bd is drawn. The Bode plot crosses the 0 dB line. The gain crossover fre-
quency is obtained from
ωgc
40 log = 11.5
4
ωgc = 7.8

4. At point d, the zero 1 + 10

1
s interferes and changes the slope from −40 dB/dec
to −20 dB/dec and goes upto ω = 20. The decibel at ω = 20 is calculated as

20
dB|ω=20 = −4.4 − 20 log = −10.4
10
Thus, point e is fixed and the straight line de is drawn.
5. At ω = 20, the pole 1/(1 + 20 1
s) interferes and changes the slope from −20
dB/dec to −40 dB/dec. Choose any frequency ω which is greater than 20; ω = 30
is chosen. At ω = 30, the dB is calculated as

30
dB = −10.4 − 40 log = −17.4
20
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 385

The point f is fixed at −17.4 dB and the straight line e f is drawn and extended
beyond ω > 30. This completes the Bode magnitude plot

K v = 15 sec−1
ωgc = 7.8 rad/sec

Example 3.13
Consider the following T.F.

4(s + 5)(s + 10)

G(s) =
s 2 (s + 20)

Sketch the Bode magnitude plot. Find the acceleration error constant and the gain
crossover frequency.

Solution
1.
4(s + 5)(s + 10)
G(s) =
s 2 (s + 20)
10 1 + 15 s 1 + 1
s
= 10
s 2 (1 + 20
s
)

The factors of G(s) are written in the ascending order of corner frequency as

10
G 1 (s) =
s2
1
G 2 (s) = 1 + s
5
1
G 3 (s) = 1 + s
10
1
G 4 (s) =
1 + 20
1
s
386 3 Frequency Response Analysis

a
− 40dB/dec.
+ dB
ωl = ωgc = Ka = 10
0
10 5 20 30 log ω

− dB
− 20dB/dec.
−8
b

− 14 d − 20dB/dec.
c 0dB/dec.

− 17.5 e

4(s + 5)(s + 10)

Fig. 3.32 Bode magnitude plot for G(s) =
s 2 (s + 20)

The range of corner frequencies and the corresponding slopes of the straight line
√ in the following table. The corner frequencies are ωc = 5, 10 and
plots are shown
20 and ωl = 10.

Corner frequency range ω≤5 5 ≤ ω ≤ 10 10 ≤ ω ≤ 20 ω ≥ 20

Slope in dB/dec −40 −20 0 −20

The corner frequencies and ωl are fixed in the semi-log graph as shown in √
Fig. 3.32.
2. The low frequency factor is 10/ω2 . By equating this to 1 we get ωl = 10. For
ω ≤ 5, the magnitude contribution is due to 10/ω2 only. At ω = 5, dB is

10
dB|ω=5 = 20 log
ω2 ω=5
10
= 20 log
25
= −8
√
Thus, point b is fixed.√A straight line joining ωl = 10 and point b is drawn and
extended below ω < 10 as shown in Fig. 3.32.
3. At point b, the zero (s + 5) changes the slope from −40 dB/dec to −20 dB/dec.
The straight line goes up to ω = 10. At ω = 10, the dB is
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 387

10
dB|ω=10 = −8 − 20 log
5
= −14

Thus, point c is fixed. The straight bc is drawn which has a slope of −20 dB/dec.
4. At ω = 10, another zero (s + 10) comes and alters the slope from −20 dB/dec
to 0 dB/dec. This is a horizontal line upto ω = 20 without any change in dB. At
ω = 20, the point d is fixed. Horizontal straight line cd is drawn.
5. At ω = 20, the pole changes the slope from 0 dB/dec to −20 dB/dec. The Bode
plot for ω > 20 is extended as follows. Choose ω = 30. At this frequency, the dB
is
30
dB|ω=30 = −14 − 20 log
20
= −17.5

The point e is fixed at −17.5 dB. The straight line joining the points de is extended
beyond ω > 20. Thus, the Bode magnitude plot becomes√complete.
6. As seen from Fig. 3.32, the dB becomes
√ zero at ω = 10. This is then ωgc .
Further, for a type 2 system ωl = K a . Hence, K a = 10 sec−2

K a = 10 sec−2
√
ωgc = 10 rad/sec

Example 3.14
Consider the following T.F.

10s 2 (s + 10)
G(s) =
(s + 20)(s + 5)(s + 30)

Sketch the Bode magnitude plot. What is the gain crossover frequency?

Solution
1.

10s 2 (s + 10)
G(s) =
(s + 5)(s + 20)(s + 30)

The given T.F. is rewritten in time constant form as

1 2 (1 + 10s
)
G(s) = s
30 (1 + 5 )(1 + 20 )(1 +
s s s
30
)
388 3 Frequency Response Analysis
√
The corner frequencies are 5, 10, 20 and 30 with ωl = 30. The factors of G(s)
are written as follows with the ascending order of corner frequencies

1 2
G 1 (s) = s
30
1
G 2 (s) =
1 + 15 s

1
G 3 (s) = 1 + s
10
1
G 4 (s) =
(1 + 20
s
)
1
G 5 (s) =
(1 + 30
1
s)

Range of corner frequencies ω≤5 5 ≤ ω ≤ 10 10 ≤ ω ≤ 20 20 ≤ ω ≤ 30 ω ≥ 30

Slope in dB/dec +40 +20 +40 +20 0

2. The low frequency response √is due to s 2 /30 or ω2 /30 in sinusoidal T.F. By equat-
ing ω /30 = 1 we get ωl = 30 = 5.5. For the factor ω2 /30, the Bode plot is a
2

straight line with +40 dB/dec slope and ωl = 5.5. A pole interferes at ω = 5 and
changes the slope. Hence, the dB at ω = 5 is calculated as

ω2
dB|ω=5 = 20 log
30 ω=5
25
= 20 log
30
= −1.6

The two points with ω = 5 and dB = −1.6 and ω = 5.5 and dB = 0 are joined
by a straight line. Only the response for ω ≤ 5 is sketched. Straight line ab is
drawn.
3. At ω = 5 the pole 1/(1 + 15 s) changes the slope from +40 dB/dec to +20 dB/dec
in the frequency range 5 ≤ ω ≤ 10. At ω = 10, the dB is calculated

10
dB|ω=10 = −1.6 + 20 log
5
= 4.4

In this frequency range the magnitude plot crosses the 0 dB line. The gain
crossover frequency is calculated from
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 389

ωgc
20 log = 1.6
5
ωgc = 6

Straight line bc is drawn.

4. At ω = 10, the zero 1 + 10 1
s interferes and changes the slope from +20 dB/dec
to +40 dB/dec. A straight line in the frequency range 10 ≤ ω ≤ 20 is to be drawn.
The dB at ω = 20 is calculated from
20
dB|ω=20 = 4.4 + 40 log
10
= 16.4

The straight line cd is drawn.

5. At ω = 20, the pole 1/(1 + 201
s) interferes and changes the slope from +40 dB/
dec to +20 dB/dec. A straight line in the frequency range 20 ≤ ω ≤ 30 is to be
drawn. The dB at ω = 30 is obtained from
30
dB|ω=30 = 16.4 + 20 log
20
= 20

The straight line de is drawn.

6. At ω = 30, the pole 1/(1 + 301
s) interferes and changes the slope from +20 dB/
dec to 0 dB/dec for ω ≥ 30. Hence, the horizontal line e f is drawn and extended
beyond ω > 30. The complete Bode magnitude plot is shown in Fig. 3.33

ωgc = 6 rad/sec

Example 3.15
Consider the following transfer function

s(s + 10)
G(s) =
(s 2 + 2s + 9)

Sketch the Bode magnitude plot. Find the gain crossover frequency if any?

Solution
1.
s(s + 10)
G(s) =
(s 2 + 2s + 9)
390 3 Frequency Response Analysis

e 0dB/dec.
20 f

+ 20dB/dec.
16.4 d

+ dB + 40dB/dec.

4.4 c
ωl = 5.5
+ 20dB/dec.
0
5 10 20 30 log ω
ωgc = 6
− 1.6 b

− dB + 40dB/dec.
a

10s 2 (s + 10)
Fig. 3.33 Bode magnitude plot for G(s) =
(s + 20)(s + 5)(s + 30)

The given T.F. is rewritten in time constant form of poles zeros and the standard
prototype second order system as given below.

10 s 1 + 10 1
s 9
G(s) =
9 (s 2 + 2s + 9)

The following factors of G(s) are written in the ascending order of corner fre-
quency

10
G 1 (s) = s
9
9
G 2 (s) =
+ 2s + 9
s2
1
G 3 (s) = 1 + s
10
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 391

The corner frequencies are 3 and 10 with ωl = 10 9

= 0.9. The range of corner
frequencies and the corresponding slopes of the straight lines are shown in the
following table:

Range of corner frequency ω≤3 3 ≤ ω ≤ 10 ω ≥ 10

Slope in dB/dec +20 −20 0

2. The low frequency response is due to the factor G 1 ( jω) = 10

9
ω. This is a straight
line with ωl = 10
9
for the frequency range ω ≤ 3. At ω = 3, the dB is

10
dB|ω=3 = 20 log 3 = 10.5
9
Straight line ab is drawn which passes through 0 dB at ωl = 9/10. Straight line
ab is drawn.
3. At ω = 3, the quadratic factor (9/(s 2 + 2s + 9)) interferes and changes the slope
from +20 dB/dec to −20 dB/dec. The straight line is drawn for the frequency
range 3 ≤ ω ≤ 10. The dB at ω = 10 is calculated as

10
dB|ω=10 = 10.5 − 20 log =0
3
The straight line bc is drawn.
4. At ω = 10, the zero 1 + 10 1
s interferes and changes the slope from −20 dB/dec
to 0 dB/dec in the frequency range ω ≥ 10. A horizontal line cd is drawn and the
Bode plot is completed as shown in Fig. 3.34.

Example 3.16
Consider the following T.F.

30(s 2 + 4s + 16)
G(s) =
8(s + 1)(s + 20)

Sketch the Bode magnitude plot. Find the gain crossover frequency.

Solution
1.

30(s 2 + 4s + 16)
G(s) =
8(1 + s)(s + 20)

The T.F. is rewritten in time constant and standard second order forms as given
below
392 3 Frequency Response Analysis

b
10.5
20dB/dec

dB
c 0 dB/dec log
0
3 10 d
l 0.9

dB
20dB/dec
a
s(s + 10)
Fig. 3.34 Bode magnitude plot for G(s) =
(s 2 + 2s + 9)

30 × 16 (s 2 + 4s + 16)
G(s) =
8(1 + s) 1 + 20
1
s 20 16
3 (s 2 + 4s + 16)
=
(1 + s)(1 + 1
20
s) 16

The factor of G(s) in the ascending order of frequency are written as follows:

G 1 (s) = 3
1
G 2 (s) =
1+s
(s 2 + 4s + 16)
G 3 (s) =
16
1
G 4 (s) =
(1 + 201
s)

The corner frequencies are 1, 4 and 20. The corner frequency range and the slope
of the Bode magnitude are given in the following table.

Corner frequency range ω≤1 1≤ω≤4 4 ≤ ω ≤ 20 ω ≥ 20

Slope in dB/dec 0 −20 20 0

2. The low frequency response is due to G 1 (s) = 3. The magnitude plot is a hori-
zontal line with 0 dB/dec with dB = 20 log 3 = 9.5 upto ω ≤ 1. Line ab is drawn.
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 393

3. At ω = 1, the pole 1/1 + s starts which changes the slope from 0 dB/dec to
−20 dB/dec in the frequency range 1 ≤ ω ≤ 4. The dB at ω = 4 is calculated as

4
dB = 9.5 − 20 log = −2.5
1
The straight bc is drawn. It cuts 0 dB line at ωgc1 . This frequency is calculated
from
ωgc1
20 log = 9.5
1
ωgc1 = 3

4. At ω = 4, the quadratic factor (s 2 + 4s + 16)/16 starts and changes the slope

from −20 dB/dec to +20 dB/dec in the frequency range +4 ≤ ω ≤ 20. The dB
at point d is obtained form

20
dB|ω=4 = −2.5 + 20 log = 11.5
4
Straight line cd is drawn. This line crossover the 0 dB at ωgc2 which is calculated
from
ωgc2
20 log = 2.5
4
ωgc2 = 5.3

5. At ω = 20, the pole 1/(1 + 20

1
s) starts which changes the slope from +20 dB/dec
to 0 dB/dec. Hence, for ω ≥ 20, the dB is constant at 11.5. The complete Bode
magnitude plot is shown in Fig. 3.35. The T.F. has two gain crossover frequencies
at ωgc1 = 3 and ωgc2 = 5.3.

Example 3.17
Consider the following T.F.

100(s + 2)
G(s) =
s(s + 5)(s + 10)

Sketch the asymptotic graph and hence plot the actual Bode plot. Also sketch the
Bode phase plot in semi-log graph.
394 3 Frequency Response Analysis

0dB/dec
11.5
d e
0dB/dec b
9.5
a
20dB/dec 20dB/dec
dB gc1 3
log
0
1 4 20
gc2 5.3
2.5
c

30(s 2 + 4s + 16)
Fig. 3.35 Bode magnitude plot for G(s) =
8(s + 1)(s + 20)

Solution
1.

100(s + 2) 4(1 + 2s )
G(s) = =
s(s + 5)(s + 10) s(1 + 5s )(1 + s
10
)

The corner frequencies are 2, 5 and 10 with ωl = 4. The factors of G(s) are
written in the ascending order of corner (break) frequency as follows:

4 1
G 1 (s) = ; G 2 (s) = 1 + s
s 2
1 1
G 3 (s) = ; G 4 (s) =
(1 + 15 s) 1 + 10
1
s

The slopes of the straight lines are given in the following table:

Frequency range ω≤2 2≤ω≤5 5 ≤ ω ≤ 10 ω ≥ 10

Slope in dB/dec −20 0 −20 −40

2. The low frequency response is solely due to 4/ω with ωl = 4 and it is a straight
line with −20 dB/dec slope. At ω = 2
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 395

4
dB|ω=2 = +20 log
ω ω=2
4
= +20 log
2
= +6

Straight line ab is drawn which passes through 6 dB at ω = 2 and 0 dB at ω = 4.

3. At ω = 2, the zero 1 + 21 s starts changing the resultant slope as 0 dB/dec.
Horizontal straight line bc is drawn.
4. At ω = 5, the pole 1/(1 + 15 s) starts and changes the slope from 0 dB/dec to
−20 dB/dec in the frequency range 5 ≤ ω ≤ 10. At ω = 10, the dB is

10
dB|ω=10 = 6 − 20 log =0
5
Straight line cd is drawn.
5. At ω = 10, the pole 1/ 1 + 10
1
s starts making the change in the slope as
−40 dB/dec. For ω ≥ 10. Any frequency say ω = 20 is chosen. At ω = 20, the
dB is
20
dB|ω=20 = −40 log = −12
10
The straight line de is drawn and extended. The Bode magnitude plot is drawn as
shown in Fig. 3.36a.
6. To draw the phase angle plot, the sinusoidal T.F. is

100( jω + 2)
G( jω) =
jω( jω + 5)( jω + 10)
ω ω ω
φ = ∠G( jω) = −90 + tan−1 − tan−1 − tan−1
2 10 5
For ω = 0;
φ = −90◦

For ω = ∞;
φ = −180◦

In between, few values of ω are chosen and the phase angle is calculated and
tabulated below

ω 0 1 2 5 8 12 20 30 ∞
φ −90◦ −80◦ −78◦ −94◦ −111◦ −127◦ −145◦ −156◦ −180◦

The data from the above table is transferred to the semi-log sheet where the vertical
axis represents the phase angle while the horizontal axis represents the frequency
396 3 Frequency Response Analysis

(a) a
−20dB/dec.
0dB/dec.
6 c
b
−20dB/dec.
+dB

0 d
2 4 5 10 20 30 log ω

−dB
−40dB/dec.

−12 e

(b)
φ

0
2 5 10 20 30 log ω

−90
−φ

−180

(c) Bode diagram

40
20
Magnitude (dB)

0
−20
−40
−60

−80
−45

−90
Phase (deg)

−135

−180
10−1 100 101 102 103
Frequency (rad/sec)

Fig. 3.36 a Bode magnitude plot. b Bode phase plot. c Bode magnitude and phase plots (actual)
100(s + 2)
drawn using MATLAB for the T.F. G(s) =
s(s + 5)(s + 10)
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 397

in logarithmic scale. The phase angle plot is shown in Fig. 3.36b. Usually both
magnitude and phase plots are drawn one below the other to get information
easily. The plot of Fig. 3.36 is obtained (actual magnitude and phase angle plots)
using the following MATLAB program.

Bode Diagram
Transfer Function
100s + 200
s3 + 15s 2 + 50s
Program
num=[0 0 100 200];
den=Conv([1 0], Conv([1 5],[1 10]));
sys=tf(num,den)
bode (sys)
grid

The actual Bode magnitude and phase plots are shown in Fig. 3.36c. The error
between the actual plot and the asymptotic plots is obviously seen. Since the
access to MATLAB is easy to get nowadays, control system design problems are
invariably carried out with actual Bode plots.

3.8.1 Transfer Function from Bode Plot

In Sect. 3.7, for the given T.F. the Bode magnitude as well as the phase plots were
drawn. By the reverse process, that is given the Bode plot, the transfer function can
be obtained. If the magnitude plot alone is given, then it is possible to find the T.F.
only if the T.F. is of minimum phase. However, if both magnitude and phase plots
are given, then it is possible to find the T.F. of non-minimum system also. We discuss
below the method to find the minimum phase T.F. when the Bode magnitude plot is
given. The step-by-step procedure is given below
1. Identify the low frequency response of the system in the Bode plot. This may
correspond to the factors K , K s, K s 2 , K /s or K /s 2 in the system T.F.
2. Wherever there is a change of slope, identify that with a corner frequency T1 , T2 ,
T3 , etc..
3. A change of −20 dB/dec slope at ωc = T1 , there is a pole 1/(1 + T11 s) and
−40 dB/dec slope at ωc = T2 , there is a pole 1/(1 + T12 s)2 and so on. Simi-
larly for +20 dB/dec change of slope at corner frequency ωc = T3 , there is a zero
(1 + T13 s). For +40 dB/dec change of slope at corner frequency ωc = T4 , there
are two zeros (1 + T14 s )2 and so on.
4. From the Bode plot, information about the decibel values and corner frequencies
are obtained and K , T1 , T2 , T3 , T4 are known and hence the T.F.
398 3 Frequency Response Analysis

a 0dB/dec. b
32
+dB −40dB/dec.

ωgc= 6.31

0
T1 T2 log ω

−12
c
−20dB/dec.
−dB
d

Fig. 3.37 Bode plot, for Example 3.18

Example 3.18
From the Bode magnitude plot shown in Fig. 3.37, determine the T.F.

Solution
1. Straight line ab which has 0 dB/dec. slope represents the low frequency response.
The factor for this portion of the T.F. is

G 1 (s) = K
dB = 20 log K = 32 for ω ≤ T1
or K = 39.8
G 1 (s) = 39.8

2. There are changes in the slopes at points b and c. At these points, the corner
frequencies are identified as T1 and T2 .
3. At point b, the T.F. G 2 (s) = 1/(1 + T11 s)2 starts. For this

ωgc
40 log = 32
T1
6.31
40 log = 32
T1
T1 = 1
1
G 1 (s) =
(1 + s)2
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 399

dB/dec.
c

b
24
dB dB/dec.

dB/dec.
63.3
0
T1 T2 log 
 1
dB 4

a d

Fig. 3.38 Bode plot, for Example 3.19

4. At point c zero with the T.F. G 2 (s) = (1 + 1

T2
s) starts. T2 is determined from

T2
−40 log = −12
ωgc
T2
−40 log = −12
6.31
T2 = 12.6
G 2 (s) = (1 + 0.0794s)

5.

G(s) = G 1 (s)G 2 (s)G 3 (s)

39.8(1 + 0.0794s)
G(s) =
(1 + s)2

Example 3.19
For the Bode plot shown in Fig. 3.38, determine the T.F.

Solution
1. There are two corner frequencies in Fig. 3.38. These frequencies are identified as
T1 and T2 .
2. The low frequency response is contributed by the straight line ab. The T.F. of line
ab is
400 3 Frequency Response Analysis

G 1 (s) = K s 2

3. At the corner frequency T1 the slope changes from +40 dB/dec to +20 dB/dec.
A pole interferes with the following T.F.

1
G 2 (s) =
(1 + T11 s)

4. At the corner frequency T2 the slope changes from +20 dB/dec to −20 dB/dec.
Hence, two repeated poles start from T2 as the corner frequency. The T.F. of these
poles is

1
G 3 (s) =
(1 + T12 s)2

5. Now K , T1 and T2 are to be determined. From straight line ab, the decibel con-
tribution is solely due to G 1 (s). At ω = 41

|G 1 ( jω)|ω= 41 = 1
K ω2 |ω= 41 = 1
K = 16
G 1 (s) = 16s 2

6. Also for the straight line ab, the following equation is written

40 log 4T1 = 24
T1 = 1
1
G 2 (s) =
(1 + s)

7. At point c, the equation for the dB is written as

T2
dB|ω=T2 = 24 + 20 log
T1

dB at ω = T2 can also be written as

63.3
dB|ω=T2 = 20 log
T2

Equating these equations after substituting T1 = 1, we get

3.8 Step by Step Procedure to Draw Bode Magnitude Plot 401

− 20dB/dec.
c 0dB/dec. d
20
a
+20dB/dec.

14
b
+dB

0
T1 T2 10 20 200 log ω
−dB
e

Fig. 3.39 Bode magnitude plot, for Example 3.20

63.3
24 + 20 log T2 = 20 log
T2
63.3
20 log = 24
T22
T2 = 2
1
G 3 (s) =
(1 + 0.5s)2

8.

G(s) = G 1 (s)G 2 (s)G 3 (s)

16s 2
G(s) =
(1 + s)(1 + 0.5s)2

Example 3.20
For the Bode plot shown in Fig. 3.39, determine the T.F.

Solution
1. From Fig. 3.39, the straight line ab forms the low frequency response solely. This
transfer function is
402 3 Frequency Response Analysis

K
G 1 (s) =
s
2. At points b, c and d there are changes in the slope of the straight lines. At these
points, the corner frequencies are identified as T1 , T2 and 20. The values of K , T1
and T2 are to be determined.
3. At corner frequency ω = T1 , the change of slope is from −20 dB/dec to
+20 dB/dec. There should be two repeated zeros at this corner frequency. Hence,
the T.F.s of these zeros are
2
1
G 2 (s) = 1 + s
T1

4. At corner frequency ω = T2 , there is a change of slope from +20 dB/dec to zero

dB/dec. There should be a pole starting from this corner frequency. The T.F. of
this pole is

1
G 3 (s) =  
1 + T12 s

5. At corner frequency ω = 20, the 0 dB/dec slope is changed to a negative slope. The
frequency range between ω = 20 and ω = 200 is one decade. In this frequency
range, the fall in dB is 20 dB. Hence, the slope of the straight line de is −20 dB/dec.
Therefore, at corner frequency ω = 20, a pole starts. The T.F. of this straight line
is
1
G 4 (s) =
(1 + 20
1
s)
1
=
(1 + 0.05s)

6. The straight line ab has the magnitude

K
|G 1 ( jω)| =
ω
K
=1
ω ω=10
K = 10
10
G 1 (s) =
s

7. At corner frequency ω = T1 , the dB due to |G 1 ( jω)| is 14. Hence, we write the

following equation
3.8 Step by Step Procedure to Draw Bode Magnitude Plot 403

K
20 log = 14
ω ω=T1
10
20 log = 14
T1
T1 = 2
2
1
G 2 (s) = 1 + s
T1
= (1 + 0.5s)2

8. For the frequency range T1 ≤ ω ≤ T2 , the following equation is written

T2
20 log =6
T1
T2
20 log =6
2
T2 = 4
1 1
G 3 (s) = =
1 + 4s1 (1 + 0.25s)
G(s) = G 1 (s)G 2 (s)G 3 (s)G 4 (s)

10(1 + 0.5s)2
G(s) =
s(1 + 0.25s)(1 + 0.05s)

Example 3.21
The approximate Bode magnitude plot of a minimum phase system is shown in
Fig. 3.40. Determine the T.F. of the system.

Solution
1. The slopes of the low frequency and high frequency responses are given as
0 dB/dec. This means the poles and zeros are equal in number. Further, the corner
frequencies where the change of slope occurs are given as 0.1, 10, 100 with a
frequency range of 2 decades and one decade. The constant term and the slope of
the straight lines are to be determined to get to T.F.
2. For the frequency range ω ≤ 0.1, the dB is constant with 0 dB/dec. Hence

G 1 (s) = K

But 20 log K = 20
404 3 Frequency Response Analysis

d
160 e
c
140

+dB b
20 a

0
0.1 10 100 log ω

Fig. 3.40 Bode magnitude plot, for Example 3.21

K = 10

3. For the frequency range 0.1 ≤ ω ≤ 10, there are two decades. The rise in dB is
(140 − 20) = 120 dB. Hence, the slope of the straight line is +60 dB/dec. Hence,
three repeated zeros with a corner frequency of ωc = 0.1 start. The T.F. of these
zeros is

G 2 (s) = (1 + 10s)3

4. For the frequency range 10 ≤ ω ≤ 100, the change of frequency is one decade
and rise in dB is (160 − 140) = 20 dB. The slope of the straight line is therefore
20 dB/dec. At ωc = 10, the slope changes from +60 dB/dec to +20 dB/dec. This
is possible if two repeated poles start from ωc = 10. The T.F. of these poles is

1
G 3 (s) =
(1 + 0.1s)2

5. For the frequency ω ≥ 100, the slope of the straight line is 0. This is possible if
a pole starts from the corner frequency ωc = 100. The T.F. of the corresponding
pole is

1
G 4 (s) =
(1 + 0.01s)

6.

G(s) = G 1 (s)G 2 (s)G 3 (s)G 4 (s)

3.8 Step by Step Procedure to Draw Bode Magnitude Plot 405

0dB/octave
6dB/octave
c d
36
+dB − 6dB/octave
b

12 dB/octave e
a
0
2 4 12 log ω

Fig. 3.41 Bode asymptotic magnitude plot, for Example 3.22

10(1 + 10s)3
G(s) =
(1 + 0.1s)2 (1 + 0.01s)

Example 3.22
Consider the Bode magnitude asymptotic plot of a certain minimum phase T.F. shown
in Fig. 3.41. Determine the T.F.

Solution
1. While drawing the Bode plot or getting the T.F. from Bode plot the following
points are to be noted:
(a) 20 dB/dec = 6 dB/octave. It is easier to handle 20 dB/dec. So 6 dB/octave
slopes are converted into 20 dB/dec slopes when calculations are made.
(b) Each pole will contribute −20 dB/dec slope starting at its corner frequency.
Similarly, each zero will contribute +20 dB/dec slope starting at its corner
frequency. The constant term will have 0 dB/dec slope.
2. For the frequency range ω ≤ 2, the low frequency response starts and it is the
straight line ab which has a slope of 12 dB/oct or 40 dB/dec. Corresponding to
this, the T.F. is

G 1 (s) = K s 2

3. For the frequency range 2 ≤ ω ≤ 4, which is one octave, the dB is 6 dB/oct or

20 dB/dec. The slope has changed from 40 dB/dec to 20 dB/dec. A pole with a
corner frequency ωc = 2 starts. The transfer function of this pole is
406 3 Frequency Response Analysis

1
G 2 (s) =
1 + 21 s
1
=
(1 + 0.5s)

4. For the frequency range 4 ≤ ω ≤ 12, the slope remains constant. This is possible
if a pole with a corner frequency ωc = 4 starts. The T.F. of this pole is

1
G 3 (s) =
1 + 41 s
1
=
(1 + 0.25s)

5. For the frequency range ω ≥ 12, the slope of the straight line is −6 dB/oct or
−20 dB/dec. The change from 0dB/dec to −20 dB/dec is by a pole with ωc = 12.
The T.F. of this pole is

1
G 4 (s) =
(1 + 12
1
s)
1
=
(1 + 0.0833s)

6. For the frequency range ωc = 2 and ωc = 4, it is an octave. The slop is 6 dB/oct.

The point c is at 36 dB. Hence, point b is at dB = 36 − 6 = 30 dB.
7. At ω = 2, the dB contribution is solely by K ω2 . Hence

dB|ω=2 = 20 log K ω2 |ω=2 = 30

log 4K = 1.5
K = 7.9

8.

G(s) = G 1 (s)G 2 (s)G 3 (s)G 4 (s)

7.9 s 2
G(s) =
(1 + 0.5s)(1 + 0.25s)(1 + 0.0833s)

Note: The order of the numerator polynomial is 2 which will contribute

a total slope of 40 dB/dec. The order of the denominator polynomial is 3
which will contribute a total slope of −60 dB/dec. The net slope of these
two polynomials is (40 dB/dec−60 dB/dec) = −20 dB/dec = −6 dB/oct as ω tends
to ∞ as seen from Fig. 3.41.
3.9 Log Magnitude Versus Phase Plot (Nichols Plot) 407

3.9 Log Magnitude Versus Phase Plot (Nichols Plot)

The frequency response characteristics of G( jω) were displayed in Bode diagrams in

two separate parts as magnitude plot and phase plot. In the log magnitude versus phase
plot which is also known as Nichols plot the information about the magnitude and
phase of G( jω) are combined together and displayed in one graph. The horizontal
axis represents the phase of G( jω) while the vertical axis represents the magnitude
of G( jω) expressed in dB. The function G( jω) is expressed in its magnitude and
phase angle as a function of frequency ω. For various frequencies 20 log |G( jω)|
and ∠G( jω) are calculated and plotted. Thus, the frequency response of G( jω)
is plotted by preparing 20 log |G( jω)| and φ = ∠G( jω) for different values of ω.
Alternatively, if Bode plot is available, it can be easily transferred to a log magnitude-
phase plot with ω as the variable parameter.
Advantages of Log Magnitude-Phase Plot
1. Information about the magnitude of G( jω) and phase of G( jω) along with ω
as variable parameter is available in a single graph. The plot is easily drawn with
few easy computing which is not so in Bode plot.
2. Relative stability such as phase margin and gain margin can be quickly determined
and compensation can be worked out easily.
3. Frequency domain specifications such as resonant peak, resonant frequency, gain
crossover and phase crossover frequencies and the bandwidth of the closed loop
system can be easily determined. Some of these things are not represented in
Bode plot.
4. Closed loop frequency response can be determined from open loop frequency
response by superimposing on the Nichols chart.
5. When the loop gain of G( jω) is changed, the frequency response plot is simply
shifted up and down vertically. Similarly when a constant phase is added, the
frequency response plot is simply shifted horizontally without any distortion. By
these two important properties, the amount of gain to be adjusted or the amount
of phase to be added is determined to achieve the required frequency domain
specification.
6. The effect of time delay is to simply add −57.3Td◦ and shift the G( jω) plot
horizontally. Thus, the effect of time delay is easily studied.
Disadvantage of Log-Magnitude-Phase Plot
1. In the Bode plot, the effect of individual poles and zeros are clearly known which
will help modify the plot according to the desired response. However, when Bode
plot is transferred to log magnitude-phase plot this important information is not
carried over.
408 3 Frequency Response Analysis

Example 3.23
Consider the following T.F.

20
G(s) =
s(s + 1)(s + 10)

Draw (a) Bode asymptotic magnitude and actual phase plots. (b) Log magnitude
phase plot.

Solution
(a) Bode Plot

20
G(s) =
s(s + 1)(s + 10)
2
=
s(1 + s) 1 + 10
1
s

The corner frequencies are 1 and 10. ωl = 2. The factors of G(s) are written in
the ascending order of the corner frequencies as follows:

2
G 1 (s) =
s
1
G 2 (s) =
(1 + s)
1
G 3 (s) =
1 + 101
s

The range of corner frequencies and the slope of the corresponding straight lines
are given below

Range of corner frequency ω≤1 1 ≤ ω ≤ 10 ω ≥ 10

Slope in dB/dec −20 −40 −60

ω
φ = ∠G( jω) = −90◦ − tan−1 ω − tan−1
10
For 0 ≤ ω ≤ ∞, the phase angle φ is calculated and shown in the table below

ω 0 0.5 1 1.5 2 3 5 10 ∞
φ −90◦ −119.5◦ −140.7◦ −154.8◦ −164.7◦ −178◦ −195◦ −220◦ −270◦

At ω = 1
3.9 Log Magnitude Versus Phase Plot (Nichols Plot) 409

2
dB|ω=1 = 20 log =6
ω
At ω = 10

dB|ω=10 = 6 − 40 = −34

At ω = 15 (say)

15
dB|ω=15 = −34 − 60 log = −44.6
10
The Bode magnitude and phase angle plots are shown in Fig. 3.42.
(b) Log Magnitude-Phase Plot

20
G(s) =
s(s + 1)(s + 10)
20
|G( jω)| = 
ω (ω + 1)(ω2 + 100)
2

At any ω, the decibel is

20
20 log |G( jω)| = 20 log 
ω (1 + ω2 )(100 + ω2 )
ω
φ = −90◦ − tan−1 ω − tan−1
10
dB and phase for different values of ω are calculated and tabulated below

ω 0.5 1 1.5 2 3 5 10
dB = 20 log 11.1 3 −2.7 −7.18 −13.9 −23.2 −37.1
|G( jω)|
φ = ∠G( jω) −119◦ −140.7◦ −154.8◦ −165◦ −178◦ −195◦ −220◦

The values of dB and φ for different values of ω are transferred to the log
magnitude-phase graph as shown in Fig. 3.43 with ω as the varying parameter.

3.10 Frequency Domain Specifications

In Chap. 2, expressions for time domain specifications such as peak overshoot M p ,

time at which peak overshoot occurs t p , rise time tr , time delay td , settling time ts ,
etc. were derived when the system was subjected to step input for a prototype second
order system. For the same second order system, expressions for the resonant peak,
resonant frequency and bandwidth were derived in the frequency domain when the
410 3 Frequency Response Analysis

a
+ dB − 20dB/dec.
6 b
ωl
0
1 2 10 15 log ω

− 40dB/dec.
− dB

− 34 c

− 60dB/dec.
− 44.6
Asymptotic
G(jω) plot
d
+φ
0
log ω
−φ

− 90º
Actual phase plot
− 180º

− 270º

20
Fig. 3.42 Bode magnitude and phase plots for the T.F. G(s) =
s(s + 1)(s + 10)

input applied is a sinusoidal. These expressions were derived in terms of the sec-
ond order system parameters damping ratio ζ and undamped natural frequency of
oscillation ωn . Thus, a correlation between time domain specifications and some
frequency domain specifications was established. By knowing the specifications in
one domain, the specifications in the other domain are obtained. However, for higher
order systems such correlation does not exist. Further, the three frequency domain
specifications mentioned above are not adequate enough to study the performance
of the linear system which need not be a second order system. These specifications
are useful to study the system relative stability, fastness of time response, etc. They
are useful to design the right type of compensators which ensure the desired perfor-
mance. The following frequency domain specifications are defined and the method
to evaluate them for the linear system of any order is discussed. These frequency
domain specifications are
3.10 Frequency Domain Specifications 411

20
Fig. 3.43 Log magnitude-phase plot for G(s) =
s(s + 1)(s + 10)

1. Resonant peak, Mr
2. Resonant frequency, ωr
3. Bandwidth (BW), ωb
4. Phase margin, φm
5. Gain margin, G m
6. Phase crossover frequency, ω pc
7. Gain crossover frequency, ωgc
The above specifications are determined using the following frequency response
plots:
1. Polar plot (Nyquist plot)
2. Bode plot
3. Log magnitude-phase plot (Nichols plot).
412 3 Frequency Response Analysis

3.10.1 Phase Margin and Gain Margin via Nyquist (Polar)

Diagram

For the analysis and design of linear system, system stability is classified as absolute
stability and relative stability. Even though we give exclusive study of the topic
on stability in Chap. 4, we take relative stability as performance specification in
frequency domain and discuss it below.
The term absolute stability refers to whether the given system is stable or unstable
with the answer yes or no. If the answer is yes, how far it is closer to the stability
limit. This is called relative stability. Quantitatively the relative stability is measured,
in the frequency domain by means of gain margin (G.M.) and phase margin (P.M.)
and these quantities are symbolically represented by G m and φm , respectively.
Consider the Nyquist plot of G( jω) shown in Fig. 3.44a in polar coordinates for
a minimum phase transfer function. From open loop frequency response plot shown
in Fig. 3.44a the closed loop system is stable as it does not enclose (−1, 0) point.
(Detailed stability study is given in Chap. 4). The frequency response plot where it
cuts the negative real axis is called phase crossover and the corresponding frequency
ω pc is called the phase crossover frequency. The phase crossover is denoted by d
where d is calculated from

d = |G( jω)||ω=ω pc

The gain margin which is the measure of relative stability is defined as

1
Gm = (3.27)
d

The gain margin is expressed in decibel dB as

G m = −20 log d (3.28)

From Fig. 3.44, we give the definition for the following:

1. Phase crossover is defined as the intersection of G( jω) plot at the negative real
axis.
2. Phase crossover frequency ω pc is defined as the frequency at the phase crossover
where the phase angle of G( jω) is 180◦ .
3. (a) Gain margin is defined as the reciprocal of |G( jω)| at the phase crossover
frequency ω pc .
(b) Gain margin is defined as the factor (gain) by which the open loop system
gain can be increased before the closed loop system becomes unstable.
(c) Gain margin is defined as the amount of gain in decibel (dB) that can be added
to the open loop system before the closed loop system becomes unstable.
3.10 Frequency Domain Specifications 413

(a)
Im

d
Phase
crossover
ω=
(−1,0) ωpc 0 Re
Phase
crossover Gm 1 1
frequency d
G(jω) plot Gain margin
Gm 1 is positive
d
= Gain margin ω=0

(b) (c)
Im Im

d d

ω= ω=
0 Re (−1,0) 0 Re
(−1,0)

ωpc ωpc 1
Gm 1 <1 Gm 1
d d
Gain margin G(jω) plot Gain margin
G(jω) plot
is negative is critical
ω=0 ω=0

Fig. 3.44 a Definition of gain margin in polar coordinates. b Negative gain margin and c Critical
gain margin

From Fig. 3.44, the following observations are made with respect to gain margin:
1. If the G( jω) plot does not intersect the negative real axis (system of order two
and less)

Gm = ∞

2. If the G( jω) plot passes through (−1, 0) point

G m = 1 or G m = 0 dB

The system is oscillatory now. (Fig. 3.44c).

3. If the G( jω) plot intersects the negative real axis between 0 and (−1, 0) points
414 3 Frequency Response Analysis

G m > 1 or G m > 0 dB (Fig. 3.44a)

Here, the G m in decibel is +ve and the G m > 1. The gain margin is said to
be positive. For the closed loop system to be stable the gain margin should be
positive.
4. If the G( jω) plot has the crossover to the left of (−1, 0) point

G m < 1 or G m < 0 dB (Fig. 3.44b)

Here, the G m in decibel is negative and G m < 1 and the gain margin is said to be
negative. If the gain margin is negative, the system is unstable.
Higher the gain margin, higher is the relative stability. If the gain margin is
negative, the closed loop system is unstable. If the gain margin is positive, the
closed loop system is stable.
Phase Margin: in Polar (Nyquist) Plot
If the gain margin alone is positive, that does not guarantee the stability of the closed
loop system. As stated earlier, another quantitative measurement of relative stability
is the phase margin (P.M) denoted by φm .
Consider Fig. 3.45a where G( jω) is plotted in the G( jω) plane. The unit circle
with the origin of the G( jω) plane is drawn. The (−1, 0) is also marked. For a
minimum phase T.F. the G( jω) plot may have positive gain margin. Due to phase
shift near (−1, 0) point the G( jω) plot may be very closer but due to low gain at
phase crossover, the gain margin may be high. High gain margin implies high relative
stability. But low phase margin implies low relative stability. The effect of phase shift
is studied below via phase margin. We give below a few definitions.
1. Gain crossover: The intersection point of G( jω) plot on the unit circle
(|G( jω)| = 1) is called gain crossover.
2. Gain crossover frequency ωgc : The frequency at the gain crossover is called
the gain crossover frequency denoted by ωgc . It is the frequency at which the gain
of
G( jω)|ω=ωgc = 1

3. Phase Margin: Phase margin is defined as the amount of phase angle through
which the G( jω) plot should be rotated about the origin so that the gain crossover
passes through (−1, 0) point. Phase margin is also defined as 180◦ + the phase
angle of G( jω) at unity gain

φm = −180◦ + ∠G( jω)|ω=ωgc (3.29)

It is to be noted that while phase margin is measured at the gain crossover, the
gain margin is measured at the phase crossover. In other words to calculate phase
margin, gain of G( jω) is known which is unity. To calculate gain margin, phase
angle of G( jω) is known which is 180◦ .
3.10 Frequency Domain Specifications 415

(a)
Im
j

Unit cricle
ωgc = Gain cross-over ωpc
frequency
φm = Positive phase margin ω=
(−1, 0) +1 Re
φm
in Gain
e marg
Phas cross-over
φm
ωgc
−j Positive phase
margin
G(jω) plot
ω=0

(b) (c)

Im Im

ωgc j j
Unit circle
G(jω) Unit circle
φm
ω= −1 ω=
−1 1 Re 1 Re
ωpc ωpc ωgc
G(jω)
−j plot −j
G(jω) plot
ω=0 φm = 0
φm is negative ω=0

Fig. 3.45 a Definition of phase margin in polar coordinates. b Negative phase margin in polar
coordinate. c Zero phase margin in polar coordinate

In Fig. 3.45a the G( jω) plot of a third order type 1 system is shown. The G( jω)
plot cuts the unit circle in the third quadrant. The intersection point is called gain
crossover. This point is connected by a radial line. The angle between this radial line
and the negative real axis gives the phase margin. If the gain crossover lies in the
third quadrant of the complex G( jω) plane the phase margin is positive. For the
system with positive phase margin, an additional phase can be included in the system
before the system becomes unstable.
416 3 Frequency Response Analysis

Now consider Fig. 3.45b where the gain crossover is identified in the second
quadrant. The radial line from the origin to the gain crossover is drawn. The phase
margin is measured between this straight line and the negative real axis. Now the
system is said to have negative phase margin. With negative phase margin, the system
is unstable.
Consider Fig. 3.45c where the gain crossover is identified with (−1, 0) point. At
this point, the phase margin is zero. Further, it has zero dB gain margin. Systems
which have zero phase margin and zero dB gain margin are said to be oscillatory.
Here ωgc = ω pc .
Important points to remember with regard to phase margin and gain margin.
1. Gain margin is calculated at phase crossover. To calculate this the phase angle
of G( jω) is known which is 180◦ . If G m > 1, the gain margin is positive. If
G m < 1, the gain margin is negative. For the system to be stable it is necessary
that the gain margin is positive. But this is not sufficient. Further, higher the gain
margin, higher is the relative stability.
2. The phase margin is calculated at gain crossover. To calculate this the magnitude
of G( jω) is known which is one. If ∠G( jω) > 180◦ the phase margin is positive.
If ∠G( jω) < 180◦ the phase margin is negative. For the system to be stable it
is necessary (but not sufficient) that the phase margin is positive. The higher the
phase margin, the higher is the relative stability.
3. For the linear system with minimum phase transfer function, it is necessary and
sufficient that both phase margin and gain margin are positive. However, if any
one of them is negative, the system is unstable.
4. If the G( jω) passes through (−1, 0) point, the gain crossover and phase crossover
lie at the same point at (−1, 0). Now the phase crossover and gain crossover
frequencies are the same (ω pc = ωgc ). The system is neither stable nor unstable.
It is said to be oscillatory.
5. For a stable system ω pc > ωgc . For an unstable system ω pc < ωgc .
6. The phase margin which is between 30◦ and 60◦ and gain margin which is greater
than +6 dB give satisfactory performance of the closed loop system.
7. In the Bode diagram if the slope of log magnitude curve is −40 dB/dec at the
gain crossover frequency a phase margin between 30◦ and 60◦ is obtained. The
slope of −20 dB/dec at the gain crossover frequency is more desirable. However,
−60 dB/dec slope magnitude curve at the gain crossover frequency the system is
likely to be unstable.
8. The damping ratio ζ of the system is roughly given by the following relation with
the phase margin
π
ζ = φm (3.30)
360
3.10 Frequency Domain Specifications 417

The following examples illustrate the method of calculating the phase margin and
gain margin using polar plot. Both analytical and graphical methods are described.

Example 3.24
A certain unity feedback system has the following forward path T.F.

1
G(s) =
s(s + 1)2

Find the gain margin and phase margin using Nyquist plot. What are phase and gain
crossover frequencies?

(Anna University, 2007)

Solution
1.
1
G(s) =
s(s + 1)2

The sinusoidal T.F. is obtained by replacing s by jω

1
G( jω) =
jω( jω + 1)2
1
= ∠−90◦ − 2 tan−1 ω
ω(1 + ω2 )
1
|G( jω)| =
ω(1 + ω2 )
∠G( jω) = −90 − 2 tan−1 ω

Here, it is not necessary to compute |G( jω)| and ∠G( jω) over a range of values
of ω. The given T.F. is Type 0 order 3 system. It starts from the 3rd quadrant when
ω = 0, crosses over the negative real axis at ω = ω pc and terminates at the origin
when ω = ∞. Only 2 selective points will be enough to find G m , φm , ω pc and
ωgc . This is explained in step 2, Fig. 3.46 is referred to.
2. To find the gain margin: To find the gain margin, phase angle ∠G( jω) = 180◦ .
The corresponding frequency is the phase crossover frequency ω pc

−90◦ − 2 tan−1 ω pc = −180◦

tan−1 ω pc = 45◦
ω pc = 1 rad/sec

The phase crossover or d is obtained by substituting ω pc in |G( jω)|

418 3 Frequency Response Analysis

Im

ver
e cr osso d = 0.5
Phas
ω=
0 1 Re
(−1,0) φm= 20°
160°
ωpc = 1.0 Unit circle

ωgc = 0.68
−j
G(jω) plot
Gain crossover
ω=0

Fig. 3.46 Gain margin and phase margin, for Example 3.24

d = |G( jω)|ω=ω pc
1
=
ω pc (1 + ω pc2 )
1
= = 0.5
1(1 + 1)
1 1
Gain margin, G m = =
d 0.5
= 2 (positive)

3. To find the phase margin: To find the phase margin |G( jω| = 1. The corre-
sponding frequency is the gain crossover frequency. A unit circle with origin as
the centre is drawn and the gain crossover is identified. ωgc is calculated from
|G( jω)| = 1

1
=1
ωgc (1 + ωgc
2 )

Solving the above equation by trial and error, we get ωgc = 0.68. At the gain
crossover, the phase angle is

∠G( jω)|ω=ωgc = −90◦ − 2 tan−1 ωgc

= −90◦ − 2 tan−1 0.68 = −160◦
3.10 Frequency Domain Specifications 419

Phase margin is

φm = 180◦ − 160◦
= 20◦ (positive)

G m = 2 (positive)
φm = 20◦ (positive)
ω pc = 1 rad/sec
ωgc = 0.68 rad/sec

Both phase margin and gain margin are positive and ω pc > ωgc . The system is
stable.

Example 3.25
A certain unity feedback control system has the following forward path T.F.

K e−0.2s
G(s) =
s(s + 2)(s + 8)

Determine K so that (1) G m = 6 dB; (2) φm = 45◦ . Use Nyquist plot in polar coor-
dinate.

(Anna University, 2009)

Solution

K e−0.2s
G(s) =
s(s + 2)(s + 8)
K e− j0.2ω
G( jω) =
jω( jω + 2)( jω + 8)
K
|G( jω)| =  (a)
ω (ω + 4)(ω2 + 64)
2
ω ω
∠G( jω) = −90◦ − tan−1 − tan−1 − 0.2 × 57.3◦ ω (b)
2 8
1. To Find K for the given Gain Margin (Analytical Method)
Method 1
The T.F. of the given system is Type 1 order 3 with a transportation lag. The G( jω)
is sketched as shown in Fig. 3.47a. The gain margin is given as 6 dB. From this d is
calculated as
420 3 Frequency Response Analysis

(a)
Im

ω=
0 Re

ωpc

G(jω) plot

ω=0

(b) Im
d=0.0176

0 Re
45º
1.5
ωpc=2.26 ωgc=1
G(jω) plot
with K=1
A
ω = 0.5

Fig. 3.47 a G( jω) plot, for Example 3.25. b G( jω) plot, for Example 3.25 with K = 1

1
20 log =6
d
d = 0.5

Thus, the phase crossover is identified as (−0.5, 0). The phase crossover frequency
is calculated from the following equation
3.10 Frequency Domain Specifications 421

ω pc ω pc
−90◦ − tan−1 − tan−1 − 0.2 × 57.3ω pc = 180◦
2 8
ω ω pc
tan−1 + tan−1 = 90◦ − 0.2 × 57.3ω pc
pc
2 8
ω pc ω pc
+
tan−1 2 ω2 8 = 90◦ − 0.2 × 57.3ω pc
1 − 16pc
10ω pc
tan−1 = 90◦ − 0.2 × 57.3ω pc
16 − ω2pc

Taking tan on both sides, we get

10ω pc
= tan(90 − 0.2 × 57.3ω pc )
16 − ω2pc
= cot(0.2 × 57.3ω pc )
1
=
tan(0.2 × 57.3ω pc )
10ω pc tan(0.2 × 57.3ω pc ) = 16 − ω2pc
ω2pc + 10ω pc tan(0.2 × 57.3ω pc ) − 16 = 0

Solving for ω pc by trial and error, we get

ω pc = 2.26 rad/sec

At the phase crossover, |G( jω pc )| = 0.5

K
 = 0.5
ω pc (ω2pc+4 )(ω2pc + 64)
K
√ = 0.5
2.26 9.1 × 69.1
K = 28.35

Method 2: To Find K for the given G m by Graphical Method

Let us assume K = 1. For different values of ω, |G( jω)| and ∠G( jω) are calculated
and tabulated as given below

ω 0 0.5 1.0 1.5 2.0 2.26 3 ∞

|G( jω)| −∞ 0.121 0.055 0.033 0.022 0.0176 0.0108 0
∠G( jω) −90◦ −113.3◦ −135◦ −154.7◦ −172◦ −180◦ −201◦ 0◦

The above data in the table is transferred to draw Nyquist plot in the polar coordinates.
This is shown in Fig. 3.47b.
From Fig. 3.47b, the values of d = 0.0176 is obtained. For the required gain
margin of 6 dB, the value of d = 0.5. Hence, the value of necessary to achieve this
422 3 Frequency Response Analysis

is

K (0.0176) = 0.5

K = 28.4

The above value which is obtained by graphical method is closer to the values as
obtained by analytical method.
2. To Find K for the Required Phase Margin
Method 1: Graphical Method
In Fig. 3.47b, with φm = 45◦ radial line O A is drawn. It cuts the G( jω) plot and
the corresponding frequency is obtained as 1 rad/sec which is the gain crossover
frequency. At gain crossover |G( jωgc )| = 1

K
 =1
ωgc (ωgc
2 + 4)(ωgc
2 + 64)

K
√ =1
1 5 × 65

K = 18

Method 2: Analytical Method

The required phase margin is +45◦ . Hence

∠G( jω)|ω=ωgc = −180◦ + 45◦ = −135◦

ωgc ωgc
−90◦ − tan−1 − tan−1 − 0.2 × 57.3ωgc = −135◦
2 8
ωgc ωgc
tan−1 + tan−1 = 11.46ωgc + 45◦
2 8
ωgc ωgc
+
tan−1 2 ω28 = 11.46ωgc + 45◦
1 − 16gc
10ωgc
tan−1 = 11.46ωgc + 45◦
16 − ωgc2

Solving by trial and error, we get ωgc = 1

K
 =1
ωgc (ωgc
2 + 4)(ωgc
2 + 64)

√
K = 1 5 × 65 = 18

K = 18
3.10 Frequency Domain Specifications 423

Example 3.26
The forward path T.F. of a unity feedback control system is

(as + 1)
G(s) =
s2
Find the value of a so that the phase margin is π/4.

Solution
(as + 1)
G(s) =
s2
jaω + 1
G( jω) =
(−ω2 )

(1 + a 2 ω2 )
G( jω) = ∠180 + tan−1 aω
ω2
|G( jω)|ω=ωgc = 1

1 + a 2 ωgc
2
=1
ωgc
2

a 2 ωgc
2
+ 1 = ωgc
4
(a)

∠G( jω) = 180 + tan−1 aω

π
∠G( jω)|ω=ωgc = 180◦ +
4
−1 π
tan aωgc =
4
Taking tan on both sides

aωgc = 1 (b)

Substituting Eq. (b) in Eq. (a), we get

ωgc
4
= 1+1=2
ωgc = 1.1892
1 1
a= =
ωgc 1.1892

a = 0.841
424 3 Frequency Response Analysis

3.10.2 Phase Margin and Gain Margin via Bode Plots

The Bode log magnitude and phase plots of a certain third order minimum phase T.F.
are shown in Fig. 3.48. To find the phase margin, the gain crossover is known which
has 0 dB and the corresponding frequency is the gain crossover frequency ωgc . From
ωgc if we come down straight, the phase angle plot is crossed. The phase margin is
obtained as −180+ phase angle at ωgc which is 22◦ in Fig. 3.48a. The phase margin
is said to be +ve and we can add more phases before the system becomes unstable.
To obtain the gain margin, the phase crossover is identified. This corresponds to 180◦
phase shift and the frequency at this point is phase crossover frequency ω pc . From ω pc
go straight up to the log-magnitude plot and identify the dB at this frequency. This
gives the gain margin. In Fig. 3.48a corresponding to ω pc , the dB = −12. The gain
margin is therefore 12 dB and is positive. In Fig. 3.48a both gain margin and phase
margin are positive. Further ω pc > ωgc . Hence, the closed loop system is stable. On
the other hand observe the Bode magnitude and phase plots given in Fig. 3.48b. Here
corresponding to ωgc , the phase angle is −210◦ . This is −30◦ above −180◦ . The
system has a phase margin of −30◦ and is unstable. To make the system stable a
phase angle greater than +30◦ is to be added. Similarly, corresponding to ω pc , the
log-magnitude has +ve dB of 18. The gain margin is −18 dB negative and the closed
loop system is unstable. To make the system stable, the system gain has to be reduced
at least by a factor which corresponds to 18 dB.
Summary of Phase Margin and Gain Margin from Bode Plots
1. For a minimum phase linear system, the gain margin is positive if the magnitude
of G( jω) at the phase crossover is negative in dB. This means the gain margin
is measured below the 0 dB axis. If the gain margin is measured above the 0 dB
axis, the gain margin is negative.
2. For a minimum phase linear system, the phase margin is positive if the phase
angle of G( jω) is greater than −180◦ at the gain crossover. This means the
phase margin is measured positive above the −180◦ axis. If the phase margin is
measured below the −180◦ axis, the phase margin is negative.
3. For the system to be stable both the phase margin and gain margin should be
positive and ω pc > ωgc .
4. If ω pc = ωgc , the phase margin is 0◦ and the gain margin is 0 dB. The system
executes sustained oscillations.
5. Increasing or decreasing the system T.F. gain K , the Bode magnitude plot is
shifted upwards or downwards while the phase plot is unaltered.

Example 3.27
A certain unity negative feedback system has the following open loop T.F.

80
G(s) =
s(s + 2)(s + 20)
3.10 Frequency Domain Specifications 425

(a)
− 20dB/dec.
− 40dB/dec.
+dB
ωgc
0
log ω
Gain margin (+ve) 12dB
−dB
−12
− 60dB/dec.

ωpc > ωgc

+φ
0 ωpc log ω

−90°
−158°
−φ Phase margin + φm = 22°
−180°

−270°

(b) − 20dB/dec.
18 − 40dB/dec.
+dB Gain margin
(−ve) −18 dB ωgc
0
log ω

−dB
− 60dB/dec.

ωpc < ωgc

+φ
ωpc log ω

−90°
−φ
−180°
φm (−ve)=−30° Phase margin
−210°
−270°

Fig. 3.48 Phase margin and gain margin from Bode plots. a Stable system (both φm and G m are
positive). b Unstable system (both φm and G m are negative)
426 3 Frequency Response Analysis

Using Bode plot determine the gain margin, phase margin, phase crossover frequency
and gain crossover frequency.

Solution
80
G(s) =
s(s + 2)(s + 20)
2
=
s(1 + 2 )(1 + 20
s 1
s)

The corner frequencies are ωc = 2, 20 and ωl = 2

2 1 1
G 1 (s) = ; G 2 (s) = ; G 3 (s) =
s (1 + 2s ) (1 + 20
1
s)

1. The low frequency T.F. starts with a corner frequency of 2. At ω = 1, dB|ω=1 =

20 log ω2 = 20 log 2 = 6. At ω = 1, the dB = 6 and ω = 2, dB = 0. A straight
line ab is drawn. This represents −20 dB/dec slope.
2. At ω = 2, G 2 (s) starts. For the frequency range 2 ≤ ω ≤ 20 the slope of the
straight line is −40 dB/dec. This frequency range is one decade and hence point
c is at −40 dB. Straight line bc is drawn.
3. Choose any frequency ω > 20. G 3 (s) starts from ω = 20 and the total slope of
the straight line now becomes −60 dB/dec. At ω = 30, the dB is

30
dB|ω=30 = −40 − 60 log = −50.6
20
Straight line cd is drawn and extended. The Bode magnitude plot is completed
and is shown in Fig. 3.49.
4. To draw the phase plot, ωgc and ω pc are necessary. At ω = 2, the dB = 0. Hence,
ωgc = 2. The expression for the phase angle is

ω ω
φ = −90◦ − tan−1 − tan−1
2 20
φ|ω=2 = −140.7◦

Phase margin φm = 180 − 140.7◦ = 39.3◦

φm = +39.3◦

5. To find ω pc , the phase angle is known which is 180◦

3.10 Frequency Domain Specifications 427

a
+dB
− 20dB/dec.
6
ωgc = 2
ω = 20 ω = 30
0
ω=1 b log ω
− 40dB/dec. Gm= 20dB
−20

−40 c

−dB

−50.6
− 60dB/dec.
d

φ
ωpc = 6.325
0
log ω

−90°
−140.7°
φm = 39.3°
−180°
−φ
−270°

Fig. 3.49 Bode plot, for Example 3.27

ω pc ω pc
−90 − tan−1 − tan−1 = 180◦
2 20
ω pc ω
+ pc
tan−1 2 ω220 = 90◦
1 − 40pc

Taking tan on both sides, we get

428 3 Frequency Response Analysis
 
−1 22ω pc
tan tan = tan 90◦
40 − ω p22
=∞
22ω pc
=∞
40 − ω2pc

This is possible if 40 − ω p2 c = 0
√
ω pc = 40 = 6.325

6. At this frequency, the dB is

6.325
dB|ω pc = −40 log = −20
2

G m = 20 dB (+ve); φm = 39.3◦ (+ve)

ω pc = 6.325 rad/sec; ωgc = 2 rad/sec
ω pc > ωgc

The system is stable.

Example 3.28
The open loop T.F. of a certain unity feedback control system is given by

K
G(s) =
s(1 + 0.5s)(1 + 0.2s)

Using Bode asymptotic plot find K so that

(a) Gain margin is 6 dB.
(b) Phase margin is 25◦ .

(Anna University, December, 2010)

Solution
K
G(s) =
s(1 + 0.5s)(1 + 0.2s)
K
G( jω) =
jω(1 + 0.5 jω)(1 + 0.2 jω)
K
|G( jω)| = 
ω (1 + 0.25ω2 )(1 + 0.04ω2 )
3.10 Frequency Domain Specifications 429

a − 20dB/dec. ωgc = 1.928

+dB
ω =1 ω =2 ω =5 ω =10
0
20 log K = 5.7 log ω
−5.7
−6 b 20 log K = (14 − 6) = 8
up by 5.7 dB B − 40dB/dec.
up by 8 dB
−14
A
−21.9 c
− 60dB/dec.
−dB

−40 G(jω) plot for K = 1

d

+φ ωpc = 10

−φ

−90°
−155°
φm = + 25° D
−180°

−270°

Fig. 3.50 Phase margin and gain margin, for Example 3.28

∠G( jω) = −90 − tan−1 0.5ω − tan−1 0.2ω

The corner frequencies are 2 and 5 and ωl = K . Let us put K = 1 and sketch the
Bode magnitude plot as shown in Fig. 3.50 as abcd. To draw the phase plot, the
following table is prepared.
√
ω 0 1 1.928 10 5 ∞
∠G( jω) −90◦ −128◦ −155◦ −180◦ −203◦ −270◦

The phase plot is also sketched in Fig. 3.50.

(a) To Determine K Value for the Desired G m = 6 dB √
From Fig. 3.50, the phase crossover frequency is obtained as ω pc = 10. This
is obtained as follows:
430 3 Frequency Response Analysis

−90 − tan−1 0.5 × ω pc − tan−1 0.2ω pc = 180◦

0.5ω pc + 0.2ω pc
tan−1 = 90◦
1 − 0.1ω2pc

Taking tan on both sides, we get

0.7ω pc
= tan 90◦ = ∞
1 − 0.1ω pc2

This gives
1 − 0.1ω p2 c = 0

or √
ω pc = 10

This point is identified as point D in phase plot. If we move vertically up the

log-magnitude plot is intercepted at point A. At point A, the dB is
√
10
dB|ω=√10 = −6 − 40 log
2
= −14

The gain margin occurs at phase crossover. The required gain margin is 6 dB. In
other words, the Bode magnitude plot should be shifted vertically upwards so
that point A is located at −6 dB. The required upward shift is 8 dB. That is

20 log K = 8

K = 2.51

(b) To Determine K Value for the Required Phase Margin of φm = 25◦

The phase margin is calculated at gain crossover which is calculated as follows:

φm = 180◦ + ∠G( jω)|ω=ωgc

25◦ = 180◦ + ∠G( jω)|ω=ωgc
∠G( jωgc ) = −155◦
−90 − tan−1 0.5ωgc − tan−1 0.2ωgc = −155◦
0.7ωgc
tan−1 = 65◦
1 − 0.1ωgc
2

0.7ωgc = (1 − 0.1ωgc
2
)(2.1445)
2
0.21445ωgc + 0.7ωgc − 2.1445 = 0
ωgc = 1.928
3.10 Frequency Domain Specifications 431

Corresponding to 1.928, the dB should be zero. For K = 1

dB|ω=1.928 = 20 log 1.928 = 5.7

This corresponds to point B in the log-magnitude plot. The point B should be

shifted upwards by adding +5.7 dB so that it is located at zero dB. The required
gain for the upward shift of 5.7 dB is

20 log K = 5.7

K = 1.928

Example 3.29
Consider the following open loop T.F. of a certain closed loop system

20
G(s) =
s(1 + 0.5s)(1 + 0.05s)

Sketch the Bode asymptotic log-magnitude and phase plots and determine the φm ,
G m , ω pc and ωgc .

Solution
20
G(s) =
s(1 + 0.5s)(1 + 0.05s)
20∠−90◦ − tan−1 0.5ω − tan−1 0.05ω
G( jω) = 
ω (1 + 0.25ω2 )(1 + 0.0025ω2 )

The asymptotic Bode log-magnitude plot is sketched as straight line segments abcd
shown in Fig. 3.51. The gain crossover frequency is obtained as
ωgc
40 log = 20
2
ωgc = 6.32

At this frequency, the phase angle of G( jω) is

φ = −90 − tan−1 0.5ωgc − tan−1 0.05ωgc

= −90◦ − tan−1 0.5 × 6.32 − tan−1 0.05 × 6.32
= −180◦
432 3 Frequency Response Analysis

− 20dB/dec.
a
+dB ωpc = ωgc = 6.32
20 b ω = 20

ω=2 ω = 20 ω = 30 log ω
− 40dB/dec.

−20 c
G(jω) plot
−dB
− 60dB/dec.
−30.6 d

φ ωgc = 6.32

−90°
−φ
−180°
Phase plot

−270°

Fig. 3.51 Bode plots, for Example 3.29

Since φ = −180◦ at ωgc , the phase margin is 0. To find the gain margin, the phase
crossover is determined. For this ∠G( jω) = −180◦ . Corresponding to −180◦ phase,
the phase crossover frequency is ω pc = 6.32. Corresponding to this frequency dB = 0

G m = 0 dB
φm = 0 ◦
ωgc = 6.32 rad/sec
ω pc = 6.32 rad/sec

The system is oscillatory.

3.10.3 Phase Margin and Gain Margin Using MATLAB

Phase margin and gain margin play an important role in the study of relative sta-
bility. The examples illustrated above make use of asymptotic Bode magnitude plot
and therefore the results obtained are not accurate. Standard MATLAB programs
are available to draw the Bode log-magnitude and phase plots very accurately and
quickly. The symbols used in the MATLAB program are
3.10 Frequency Domain Specifications 433

G m = Gain margin
pm = Phase margin
ωcp = Phase crossover frequency
ωcg = Gain crossover frequency

The following example is solved using MATLAB program to obtain the above fre-
quency domain specifications.

Example 3.30
The open loop T.F. of a certain unity feedback control system is given by

10(s + 2)
G(s) =
s(s + 6)(s 2 + 4s + 9)

Determine, using MATLAB program, gain margin in dB, phase margin, phase
crossover frequency and gain crossover frequency.

Solution
10(s + 2)
G(s) =
s(s + 6)(s 2 + 4s + 9)
(10s + 20)
= 2
(s + 6s + 0)(s 2 + 4s + 9)

The MATLAB program is given below

Bode plots
To find Phase margin and Gain margin
num=[0 0 0 10 20];
den=conv([1 6 0],[1 4 9]);
sys=tf(num,den);
w=logspace(−1, 2, 100);
bode (sys,W)
[G m , pm , ωcp , ωcg ]=margin(sys);
G m dB = 20∗ log 10(G m );
[G m dB, pm , ωcp , ωcg ]

ans= 22.5146 87.4196 4.3310 0.3768

From Fig. 3.52, the following results are obtained

434 3 Frequency Response Analysis

Bode Diagram
20
ωgc = 0.3768
0
Gm = 22.5146
−20

−40
Magnitude (dB)

−60

−80

−100

−120
−90

−135 φm = 87.4196°

ωpc = 4.3310
−180
Phase (deg)

−225

−270
10−1 100 101 102
Frequency (rad/sec)

Fig. 3.52 Bode log-magnitude and phase plots obtained by MATLAB program to determine phase
margin, gain margin, phase crossover and gain crossover frequencies

G m = + 22.5146 dB
φm = + 87.4196◦
ωgc = 0.3768 rad/sec
ω pc = 4.3310 rad/sec

Phase margin and gain margin are +ve. ω pc > ωgc . The system is stable.
3.10 Frequency Domain Specifications 435

35
ω1 ω1<ω2<ω3
30 B
25 ω2 ω1

Magnitude in dB
20 ωpc<ωgc A
+dB G(jω)plot
15 Phase ω3
crossover
10 ωpc ω2 ωpc>ωgc

5 Gm=−10dB ωgc ω4 ωgc G(jω)plot

Gain Crossover
0
−5 Critical ω3 Gm=+15dB
ω5 point
−10 ωpc
−15dB
−dB −15 ω4 +φm
Unstable =+80˚
−20
system Stable system
−25 ω5

−30 −φm
=−20˚
−35
−360˚ −270˚ −200˚ −180˚ −100˚ −90˚ 0˚
Phase in degress

Fig. 3.53 Phase margin and gain margin is log magnitude-phase plot of G( jω)

3.10.4 Phase Margin and Gain Margin via

Log-Magnitude-Phase Plot

The magnitude-phase plot has several advantages over the Bode plots and Nyquist
plot as discussed earlier. The relative stability parameters such as the phase margin,
gain margin, phase crossover frequency and gain crossover frequency can be easily
and quickly determined with high degree of accuracy. The log-magnitude-phase plots
of G( jω) of two different systems A and B are shown in Fig. 3.53. The following
points are to be noted in magnitude-phase plot:
1. The G( jω) plot is drawn with frequency as the varying parameter. The arrow in
G( jω) plot indicates that the value of ω increases in the direction of the arrow.
2. The gain crossover is where the G( jω) plot intersects the 0 dB axis. The corre-
sponding frequency is gain crossover frequency ωgc .
3. The phase crossover is where the G( jω) plot intersects the −180◦ axis. The
corresponding frequency is phase crossover frequency ω pc .
4. The critical point is the intersection of 0 dB horizontal axis and −180◦ vertical
axis.
436 3 Frequency Response Analysis

5. The phase margin is the horizontal distance in degrees measured from the gain
crossover to the critical point. If the gain crossover is to the right of the critical
point, the phase margin is measured as positive. If it is to the left of the critical
point, the phase margin is measured as negative.
6. The gain margin is the vertical distance in dB measured from the phase crossover
to the critical point. If the phase crossover is below 0 dB horizontal axis, it is
measured as positive. If the phase crossover is above the 0 dB horizontal axis, it
is measured as negative.
It is to be noted that in the horizontal axis the phase of open loop T.F. ∠G( jω)
is represented. Similarly, in the vertical axis 20 log |G( jω)| is represented.
Now consider the log magnitude-phase plot of system A shown in Fig. 3.53. The
gain crossover is at ωgc and it is to the right of critical point. The phase margin as
indicated in Fig. 3.53 is 80◦ and is positive. For the same system, the phase crossover
is at ω pc . It is below the 0 dB horizontal line which is at −15 dB. The gain margin
is therefore +15 dB positive. Here ω pc > ωgc . The system is stable.
Consider the log magnitude-phase plot of system B shown in Fig. 3.53. The gain
crossover is at ωgc and it is to the left of the critical point. The phase margin as
indicated in Fig. 3.53 is −20◦ and it is negative. For the same system, the phase
crossover is at ω pc . It is above 0 dB horizontal line at +10 dB. The gain margin is
−10 dB and is negative. ω pc < ωgc . Both phase margin and gain margin are negative
and the system is unstable.

Example 3.31
A certain closed loop system has the following open loop T.F.

0.75(1 + 0.1s)
G(s) =
s(1 + 0.5s)(1 + 0.2s)

Draw the log-magnitude-phase plot and determine the phase margin, gain margin,
phase crossover frequency and gain crossover frequency. Comment on the stability
of the system.

Solution
0.75(1 + 0.1s)
G(s) =
s(1 + 0.5s)(1 + 0.2s)
0.75(1 + 0.1 jω)
G( jω) =
jω(1 + j0.5ω)(1 + j0.2ω)
√
0.75 1 + 0.01ω2
|G( jω)| = 
ω (1 + 0.25ω2 )(1 + 0.04ω2 )
∠G( jω) = φ = tan−1 0.1ω − 90 − tan−1 0.5ω − tan−1 0.2ω
3.10 Frequency Domain Specifications 437

20
ω = 0.5 +dB
ωgc = 0.7
0

Gm = +30dB G(jω)plot
Open-loop gain (dB)

Magnitude in dB of G(jω)
−20
ωpc = 5.77
−40 −dB
ω =7
φm = 67°
−60
ω = 10

−80

−100
−225 −180 −135 −90
Open-Loop phase (deg)

Fig. 3.54 Log magnitude-phase plot, for Example 3.31

For different values of ω in the range 0 ≤ ω ≤ ∞, |G( jω)|, 20 log |G( jω)| and φ
are calculated as shown in the table given below

ω 0 0.5 0.7 5.77 7 10

φ −90◦ −107◦ −113◦ −180◦ −183◦ −187◦
|G( jω)| ∞ 1.45 1 0.032 0.021 0.0092
dB = 20 log |G( jω)| ∞ 3.23 0 −30 −33.6 −40.6

The data from the above table is transferred to the log magnitude-phase plot, and
it is shown in Fig. 3.54. The values of G m , φm , ωgc and ω pc are given below.

G m = + 30 dB
φm = + 67◦
ωgc = 0.7 rad/sec
ω pc = 5.77 rad/sec

Both phase margin and gain margin are positive and ω pc > ωgc . Hence, the system
is stable.
438 3 Frequency Response Analysis

Example 3.32
The open loop T.F. of a certain feedback control system is

K s2
G(s) =
[(1 + 0.2s)(1 + 0.02s)]

Determine the value of K for the gain crossover frequency to be 5 rad/sec. Use log
magnitude-phase plot.

Solution

K s2
G(s) =
(1 + 0.2s)(1 + 0.02s)

Let us put K = 1

250 s 2
G(s) =
(s + 5)(s + 50)
250ω2
|G( jω)| = 
(ω2 + 25)(ω2 + 2500)
φ = ∠G( jω)
ω ω
= 180◦ − tan−1 − tan−1
5 50
The following table is prepared to draw G( jω) plot in the log magnitude-phase plot:

ω 1 3 5 7 10
φ −192.5◦ −214.4◦ −230◦ −242.5◦ −254.7◦
|G( jω)| 0.98 7.7 17.59 28.2 43.85
dB = 20 log |G( jω)| −0.172 17.7 24.9 29 32.84

The log magnitude-phase plot is shown in Fig. 3.55. For ωgc = 5 rad/sec, the log
magnitude plot should be shifted vertically down so that at ωgc = 5, the dB is zero.
This means the gain has to be reduced by a factor of 24.9 dB. This is related in terms
of K as

20 log K = −24.9

K = 0.05688

If the same problem is solved using Bode asymptotic plot, the answer obtained will
be slightly different because Bode plot is drawn with some approximation. However,
in Bode plot obtained using MATLAB will give the same value.
3.10 Frequency Domain Specifications 439

Nochols chart
15
ω = 10
30
ω =7
25 ω =5
24.9
Open-loop gain (dB)

20

20 log G(jω)
+dB ω =3
15
G(jω) plot
24.9 dB
10

0
−dB ω =1
−5
−360° −300° −240° −180° −120° −60° 0°
Open-loop phase (degree)

Fig. 3.55 Log magnitude-phase plot, for Example 3.32

Example 3.33
The forward path transfer function of a certain unity feedback control system is given
by
K
G(s) =
s(1 + 0.2s)(1 + 0.05s)

Sketch the log-magnitude phase plot and determine the value of K for
(a) Gain margin is 25 dB.
(b) Phase margin is 65◦ .

Solution
K
G(s) =
s(1 + 0.2s)(1 + 0.05s)

Put K = 1
1
G(s) =
s(1 + 0.2s)(1 + 0.05s)
100
=
s(s + 5)(s + 20)
100
G( jω) =
jω( jω + 5)( jω + 20)
440 3 Frequency Response Analysis

100
|G( jω)| = 
ω (ω2 + 25)(ω2 + 400)

∠G( jω) = −90 − tan−1 0.2ω − tan−1 0.05ω

For different values of ω, |G( jω)| and 20 log |G( jω)| and ∠G( jω) = φ are calcu-
lated as shown in the table given below

ω 0.5 0.98 4 8 10 15
φ = ∠G( jω) −97◦ −104◦ −140◦ −170◦ −180◦ −198◦
|G( jω)| 1.99 1 0.1914 0.0615 0.04 0.01687
dB = 20 log |G( jω)| 6 0 −14.36 −24.22 −28 −35.5

The data from the table is transferred, and the log magnitude-phase plot is drawn
as shown in Fig. 3.56. For K = 1, from Fig. 3.56, the phase margin is +76◦ and gain
margin is +28 dB. The gain crossover frequency is ωgc = 0.98 rad/sec and the phase
crossover frequency is 10 rad/sec
(a) Determination of K for the required gain margin +25 dB.
The required gain margin is 25 dB. The actual gain margin is 28 dB. To achieve
this the G( jω) plot is to be shifted vertically up by 3 dB by increasing the gain.
Under this condition

Nichols chart
20
ωgc = 0.98 5 dB up

0
A
Gm = 25
Open-Loop gain (dB)

Gm = 28
−20
ωpc = 10
−40 3 dB up φm = 65°

−60
φm = 76°
−80

−100
−270 −225 −180 −135 −115° −90
Open-loop phase (deg) −104°

Fig. 3.56 Log magnitude-phase plot, for Example 3.33

3.10 Frequency Domain Specifications 441

20 log K = 3

K = 1.4125

The phase crossover frequency remains the same at ω pc = 10 rad/sec

(b) Determination of K for the required phase margin +65◦ .
Corresponding to 65◦ phase margin point A is identified in the G( jω) plot.
This point is down by −5 dB from 0 dB horizontal axis. To make point A to be
on the horizontal axis G( jω) plot should be shifted vertically up by +5 dB by
increasing the gain. Under this condition

20 log K = 5

K = 1.778

By the vertical shift, the gain crossover frequency is changed and the new gain
crossover frequency is roughly estimated as ωgc = 2 rad/sec

Example 3.34
A certain unity feedback control system has the following open loop T.F. with trans-
portation lag.
0.1e−T s
G(s) =
s(1 + 0.1s)(1 + 0.5s)

Find the value of T so that the phase margin is +25◦ . Use log magnitude-phase plot.

Solution

0.10e−T s
G(s) =
s(1 + 0.1s)(1 + 0.5s)
0.10e− jωT
G( jω) =
jω(1 + 0.1 jω)(1 + 0.5 jω)
2(e− jωT )
=
jω( jω + 10)( jω + 2)
2
|G( jω)| = 
ω (ω2 + 100)(ω2 + 4)
∠G( jω) = −90◦ − tan−1 0.1ω − tan−1 0.5ω − 57.3T ω◦
442 3 Frequency Response Analysis

Let T be zero

G( jω) = −90◦ − tan−1 0.1ω − tan−1 0.5ω

ω 0.05 0.1 1 3 4.472 8 10

φ = ∠G( jω) −92◦ −93.5◦ −122.26◦ 163◦ −180◦ −205◦ −214◦
|G( jω)| 2 1 0.089 0.0177 0.00833 0.00237 0.00139
dB = 20 log 6 0 −21 −35 −41.6 −52.5 −57
|G( jω)|

From Fig. 3.57, the following frequency domain specifications are observed.

G m = +41.6 dB
φm = +86.5◦
ωgc = 0.1
ω pc = 4.472

The required phase margin is +25◦ . The balance of −(86.5◦ − 25◦ ) = −61.5◦ is to
be contributed by the transportation large at ω = ωgc = 0.1

Nichols chart
0
ωgc = 0.1

−20 Gm = 41.6 dB
Open-loop gain (dB)

−40 φm = 61.5°
25°
ωpc = 4.472
−60

−80

φm = 86.5°
−100

−120
−270 −225 −180 −135 −90
Open-loop phase (deg)

Fig. 3.57 Log magnitude-phase plot, for Example 3.34

3.10 Frequency Domain Specifications 443

−57.3ωgc T = −61.5
5.73T = 61.5

T = 10.73

It is to be noted that when phase angle is added the G( jω) plot is horizontally shifted
without affecting the dB of |G( jω)|. The gain crossover frequency ωgc moves in
the same horizontal 0 dB axis hence it is unchanged. However, the phase crossover
frequency is changed during horizontal shift.

3.10.4.1 MATLAB Programme to Determine Phase Margin and Gain

Margin from Log Magnitude-Phase (Nichols) Plot

MATLAB toolbox is very extensively used in solving control system problems espe-
cially in frequency response plots which otherwise involves tedious calculations. The
frequency response plots can be quickly and very accurately drawn using MATLAB.
The several advantages of log magnitude-phase (Nichols) plot were listed. We illus-
trate below a few examples to draw exact Nichols plot and determine phase margin,
gain margin, phase crossover and gain crossover frequencies.
The same program which was used to determine these parameters from Bode plot
can be used except the command bode (sys, W) is replaced by Nichols (sys, W).

Example 3.35
Consider the closed loop system described in Example 3.30 which has the open loop
T.F.
10(s + 2)
G(s) =
s(s + 6)(s 2 + 4s + 9)

Using MATLAB program draw the Nichols plot and determine gain margin, phase
margin, phase crossover frequency and gain crossover frequency. Comment on the
stability of the closed loop system.

Solution The MATLAB program for the given T.F.

10(s + 2)
G(s) =
s(s + 6)(s 2 + 4s + 9)

is given below. The program is exactly same as for the Bode diagram except that bode
(sys, W) is replaced by Nichols (sys, W). The log magnitude phase plot is shown in
Fig. 3.58. From the print out and Fig. 3.58, the following results are obtained
444 3 Frequency Response Analysis

Fig. 3.58 To determine G m , Nichols chart

φm , ω pc and ωgc from 20
Nichols plot for G(s) =
10(s + 2)/[s(s + 6)(s 2 + 4s + 9)] φm = +87.4196°

0
ωgc = 0.3768 rad/sec
Gm = +22.5146dB
−20
ωpc = 4.331 rad /sec

Open-Loop Gain (dB)

−40

G(jω)plot
−60

−80

−100

−120
−270 −225 −180 −135 −90
Open-loop phase (deg)

G m = 22.5146 dB
φm = 87.4196◦
ω pc = 4.3310 rad/sec
ωgc = 0.3768 rad/sec

Since both gain margin and phase margin are +ve and ω pc > ωgc , the system is
STABLE. The same result is obtained using Bode plot in Example 3.30.
3.10 Frequency Domain Specifications 445

MATLAB Program
Nichols plot
To find Phase margin and Gain margin
num=[0 0 0 10 20];
den=conv([1 6 0],[1 4 9]);
sys=tf(num,den);
w=logspace(−1, 2, 100);
Nichols (sys,W)
[G m , pm , ωcp , ωcg ]=margin(sys);
G m dB = 20∗ log 10(G m );
[G m dB, pm , ωcp , ωcg ]

ans=
22.5146 87.4196 4.3310 0.3768

Example 3.36
Consider the following open loop T.F. of a certain closed loop system.

(s + 4)(s + 5)
G(s) =
s(s + 1)(s + 2)(s + 30)

Write a MATLAB program for the frequency response plot of the open loop system
in the Nichols plot and determine G m , φm , ω pc and ωgc . Determine the stability of the
closed loop system. How the above specifications are changed if the gain is changed
by a factor of 10.

Solution
(s + 4)(s + 5)
G(s) =
s(s + 1)(s + 2)(s + 3)
(s 2 + 9s + 20)
= 2
(s + s + 0)(s 2 + 5s + 6)

The MATLAB program is written as given below and the G( jω) plot in log
magnitude-phase is shown in Fig. 3.59. The following results are given in the print
out and marked in Fig. 3.59
446 3 Frequency Response Analysis

Fig. 3.59 To determine G m , Nichols chart

φm , ω pc and ωgc . by Nichols 40
plot for G(s) =
(s + 4)(s + 5)
s(s + 1)(s + 2)(s + 3) φm = +7.8555°
20

G(jω)plot
0

Open-Loop Gain (dB)

Gm = +3.5685 dB
ωgc = 1.4878 rad/sec
ωpc = 1.8061 rad /sec

−20

−40

−60

−80
−225 −180 −180 −90
Open-loop phase (deg)

G m = + 3.5685 dB
φm = + 7.8555◦
ω pc = 1.8061 rad/sec
ωgc = 1.4878 rad/sec

Both gain margin and phase margin are positive. ω pc > ωgc . The closed loop system
is stable.
3.10 Frequency Domain Specifications 447

MATLAB Program
Nichols plot
To find Phase margin and Gain margin
num=[0 0 1 9 20];
den=conv([1 1 0],[1 5 6]);
sys=tf(num,den);
w=logspace(−1, 2, 100);
Nichols (sys,W)
[G m , pm , ωcp , ωcg ]=margin(sys);
G m dB = 20∗ log 10(G m );
[G m dB, pm , ωcp , ωcg ]
ngrid

ans=
3.5685 7.8555 1.8061 1.4878

G m , φm , ω pc and ωgc and stability when the system gain is increased

10(s + 4)(s + 5)
G(s) =
s(s + 1)(s + 2)(s + 3)
(10s 2 + 90s + 200)
= 2
(s + s + 0)(s 2 + 5s + 6)

The MATLAB program is written as given below and the G( jω) plot in log
magnitude-phase is shown in Fig. 3.60. The following results are given in the print
out and marked in Fig. 3.60.

G m = − 16.4315 dB
φm = − 18.8153◦
ω pc = 1.8061 rad/sec
ωgc = 3.9713 rad/sec

Both gain margin and phase margin are negative and ω pc < ωgc . The closed loop
system is unstable. An increase in gain of 10 is equivalent to increase in dB by +20 dB.
Fig. 3.59 is vertically shifted up by +20 dB. The gain crossover is unaffected and the
dB at this point is −3.5685 + 20 = +16.4315. Hence, G m = −16.4315 dB as seen
is Fig. 3.60.
448 3 Frequency Response Analysis

Nichols chart
60

40

G(jω)plot
20
ωpc = 1.8061 rad /sec
Open-Loop Gain (dB)

Gm =
−16.4315dB

0
ωgc = 3.9713 rad /sec

φm φm = −18.8153°
−20

−40

−60
−225 −180 −135 −90
Open-loop phase (deg)

10(s + 4)(s + 5)
Fig. 3.60 To determine G m , φm , ω pc and ωgc . by Nichols plot for G(s) =
s(s + 1)(s + 2)(s + 3)
3.10 Frequency Domain Specifications 449

MATLAB Program
Nichols plot
To find Phase margin and Gain margin
num=[0 0 10 90 200];
den=conv([1 1 0],[1 5 6]);
sys=tf(num,den);
w=logspace(−1, 2, 100);
Nichols (sys,W)
[G m , pm , ωcp , ωcg ]=margin(sys);
G m dB = 20∗ log 10(G m );
[G m dB, pm , ωcp , ωcg ]
Warning: The closed-loop system is unstable.
ans= −16.4315 −18.8153 1.8061 3.9713

3.11 Determining Closed Loop Frequency Response from

Open Loop Frequency Response (Unity Feedback
System)

Consider the block diagram of a certain unity feedback control system with the
forward path T.F. as shown in Fig. 3.61a. The closed loop T.F. is obtained as

C(s) G(s)
= M(s) = (3.31)
R(s) 1 + G(s)

The sinusoidal T.F. of Eq. (3.31) is

G( jω)
M( jω) = (3.32)
1 + G( jω)

Instead of using Eq. (3.32) which is complex in nature to evaluate the closed loop
frequency response M( jω), it can be easily obtained by graphical means. Further
earlier methods developed to determine resonant peak Mr , resonant frequency ωr
and the BW are restricted to second order system only. The information that we get
from Bode and log magnitude-phase plots is about the phase margin and gain margin
which is used to determine the absolute and relative stability. From the graphical
method we discuss below, we get information about Mr , ωr and BW for higher order
systems in addition to getting information about the closed loop frequency response
which is obtained very easily.
Consider the frequency response G( jω) of the open loop system shown in
Fig. 3.61b. For any frequency ω1 , point B in G( jω) can be located. Draw vec-
tors O B from the origin and AB from (−1, 0) point in the G( jω) complex plane.
450 3 Frequency Response Analysis

(b)
Im

(a)
C(s) (−1,0) ω=
G(s)
R(s) A θ2 0 Re
(θ1−θ2) θ1

B ω1
G(jω) plot

ω=0

Fig. 3.61 a Block diagram of a unity feedback control system. b G( jω) plot in the polar coordinate

The vectors in Fig. 3.61b represent the following

−→
OA = (−1, 0)
−→
OB = G( jω)
−→ −→ −→
AB = O A + AB = 1 + G( jω)
−→
OB |G( jω)|
|M( jω)| = −→ = (3.33)
AB |1 + G( jω)|
−→ −→
∠M( jω) = ∠ O B − ∠ AB
= −θ1 + θ2
= −(θ1 − θ2 )
= −∠AB O (3.34)

From Eq. (3.33) by calculating the ratio of O B to AB, the closed loop magnitude
M( jω)|ω=ω1 is obtained. Similarly, from Eq. (3.34) the negative value of AB O gives
∠M( jω)|ω=ω1 . The procedure is repeated for other values of ω and for each frequency
the magnitude and phase of the closed loop system are obtained. However, the above
procedure is still further simplified. A set of circles each indicating a particular value
of closed loop magnitude and closed loop phase angle separately are available. These
sets of circles are called constant M circles and constant N circles respectively. The
G( jω) plot is simply superimposed on the set of constant M circles. The points
of intersections of G( jω) plot with the constant M circles give the closed loop
magnitude at different values of frequency. Similarly, by superimposing G( jω)
plot on the set of constant N circles, the closed loop phase angles are obtained for
different values of frequency. Once |M( jω)|, ∠M( jω) and ω are known, closed
loop frequency response can be drawn either in polar coordinate, Bode diagram or
3.11 Determining Closed Loop Frequency Response from Open … 451

B (x,y)

y G(jω) plot

A θ2 θ1
x
−1, 0 0 ω= x

(1+x)

ω =0

Fig. 3.62 G( jω) plot in the complex x-y plane

log magnitude-phase plot. We give below, how the magnitude loci from constant M
circles and the phase angle loci from constant N circles are obtained.

3.11.1 Constant M Circles

Consider the G( jω) plot drawn in the complex x − y plane as shown in Fig. 3.62.
The symbols used in Fig. 3.62 have the same meaning as those used in Fig. 3.61b.
From Fig. 3.62, at point B, the closed loop magnitude is obtained from Eq. (3.33) as
−→
| O B|
M( jω) = −→
| AB|

x 2 + y2
M= 
(1 + x)2 + y 2

Squaring and cross multiplying, we get

M 2 [(1 + x)2 + y 2 ] = x 2 + y 2

Rearranging the terms, we get (for M < 1)

(1 − M 2 )x 2 + (1 − M 2 )y 2 − 2M 2 x = M 2 (3.35)

For M > 1, the above equation becomes

452 3 Frequency Response Analysis

(M 2 − 1)x 2 + (M 2 − 1)y 2 − 2M 2 x = M 2 (3.36)

 2
M2
By dividing Eq. (3.35) by (1 − M 2 ) and adding (1−M 2 )
on both sides we get

 2  2
2M 2 x M2 M2 M2
x 2 + y2 + + = +
(1 − M ) 2 (1 − M )2 (1 − M )
2 (1 − M 2 )
 2  2
M2 M
x− + y 2
= (3.37)
(1 − M 2 ) (1 − M 2 )

Equation (3.37) represents a circle with centre at centre,

M2
x= ; y=0 (3.38)
(1 − M 2 )

and
M
radius, r = for M < 1 (3.39)
(1 − M 2 )

For M > 1, from Eq. (3.36), on similar line described above, we get
 2  
M2 M
x− +y =2
(3.40)
(M 2 − 1) (M − 1)
2

The centre and radius of the circle for any particular value of M is

M2
Center, x = ; y=0
(M 2 − 1)
M
Radius, r = (3.41)
(M 2 − 1)

For M = 1, Eq. (3.35) is written as

1
x =− (3.42)
2

This is the vertical straight line passing through x = − 21 . Using Eqs. (3.38), (3.41)
and (3.42), for any value of M chosen, the centre and radius using the above equations
are drawn. For M = ∞, the circle converges to a point at (−1, 0). These circles are
known as Constant M circles, and they are shown in Fig. 3.63. Ready-made constant
M circles charts are also available.
3.11 Determining Closed Loop Frequency Response from Open … 453

Im G
G(jω) plane

M =1.2 M=0.833

1.3 0.766
1.5 0.67
2.0 0.5
M=0
3
1 Re G
M= 0.33

M=1

−1
2

Fig. 3.63 Constant M circles

3.11.2 Constant N Circles

Similar to constant M circles, a set of circles are available to determine the closed loop
phase angle from open loop frequency response. These circles are called constant N
circles where tan ∠M( jω) = tan−1 N . Referring to Fig. 3.62

∠O B
∠M( jω) = = (θ1 − θ2 )
∠AB

tan θ1 = y
x
and tan θ2 = y
1+x
. Let

∠M( jω) = θ
θ = (θ1 − θ2 )
y y
tan−1 N = tan−1 − tan−1
x 1+x
y
−1 x
y
− 1+x
= tan y2
1 + x(1+x)
y
= tan−1 2
x + y2 + x

Taking tan on both sides, we get

454 3 Frequency Response Analysis

y
N=
x 2 + y2 + x
y
x 2 + x + y2 − =0
N

Adding 1
4
+ 1
4N 2
on both sides, we get

y 1 1 1 1
x 2 + x + y2 − + + = +
N 4 4N 2 4 4N 2
 2
2 2
1 1 1 1
x+ + y− = + (3.43)
2 2N 4 4N 2

Equation
 (3.43) represents the circle with centre at (−1/2, 1/2N ) and radius
1
4
+ 4N 2
1

−1 1
Center, x = − and y =
2 2N


1 1
Radius, r = + (3.44)
4 4N 2

For any chosen value of N , the centre and radius of the circles are obtained and
the circle is drawn. Unlike M circles, the constant N circles are not marked with
closed loop phase N . Instead N is converted into degrees. For example, for N = 1,
x = −12
, y = 21 ; r = √12 . The closed loop phase angle of the circle is not marked with
N = 1 but θ = tan−1 N = tan−1 1 = 45◦ . Ready-made constant phase angle circles
are available for use. The constant N circles are shown in Fig. 3.64.

3.11.3 The Nichols Chart

In Sects. 3.11.1 and 3.11.2, obtaining the closed loop frequency response from open
loop frequency response using constant M and constant N circles was discussed.
The data has to be obtained twice, once from constant M circles and the other time
from the constant N circles. The laborious task is very much simplified by combining
constant M circles and constant N circles in one chart called Nichols Chart which
was introduced by Nathaniel B. Nichols (1914–1997). The Nichols chart is derived
from constant M and N circles and represented in log magnitude-phase coordinates.
3.11 Determining Closed Loop Frequency Response from Open … 455

Fig. 3.64 Constant N Im G

circles
φ = 15˚ (−165˚)
G(jω) plane

30˚ (−150˚)

(−165˚) 45˚
60˚

90˚

−1 0 Re G
−90˚

−60˚
−45˚

−30˚

φ = −15˚

Advantages of Nichols Chart

1. The Nichols chart gives both the gain and phase characteristics of the closed loop
system simultaneously at one instant.
2. The phase margin, gain margin, phase crossover and gain crossover, the critical
point, the resonant peak, resonant frequency and bandwidth which are the fre-
quency domain specifications can be easily and quickly determined from the open
loop frequency response plot G( jω).
3. Using Nichols chart, the closed loop frequency response can be easily obtained
from the open loop frequency response.
4. The required gain of the system can be obtained for the desired phase margin,
gain margin, bandwidth and resonant peak.
456 3 Frequency Response Analysis

A Historical Note: Natheniel B. Nichols (1914–1997)

N.B. Nichols was born in 1914 in Michigan, USA. He

received his B.S. degree from Central Michigan Uni-
versity in 1936 and M.S. degree in physics in 1937 from
the University of Michigan. He joined Taylor Instru-
ments in Rochester in 1937 as a process control engi-
neer and worked with John Ziegler. These two intellec-
tuals worked together at Taylor and their job included
determining means for tuning PID controllers and suc-
cessfully developed Ziegler–Nichols tuning parame-
ters which were rough approximations to optimal set-
tings for open loop T.F. to ensure prescribed behaviour in closed loop PID realizations.
After working for several years at Taylor, Nichols turned his attention to technical
innovations of war weapon design. It was the beginning of World War II then. His
innovative ideas included the then emerging field of control of radar and ground anti-
aircraft gun and angle tracking systems. Nichols developed a graphical design aid to
design this system. This graphical aid now we call as Nichols chart forms a standard
chapter in automatic control theory. Nichols contributions to the concepts, design
and performance of electric drive systems became very popular worldwide at that
time and it was considered several orders of magnitude greater than the introduction
of the famous Ward Leonard drives. In 1987, Nichols visited China to participate in
technical conference. On that occasion every control engineer in China wanted to
meet Dr. Nichols, talk with him and sit next to him for breakfast, lunch and dinner.
Everyone wanted to have a personal conversation with this great engineer who had
made feedback control theory understandable. His influence on the control field has
been continuous since 1942 and for every contribution he was honoured by U.S.
government. On April 17, 1997, Nick died after a prolonged illness.

3.11.4 Nichols Chart from Constant M and Constant N

Circles

The Nichols chart is a tool for the designer to read off closed loop gain and phase
directly from the plot of the open loop frequency response. It is derived from constant
M and constant N circles. When constant M circles are transferred to log magnitude-
phase plot, then it becomes constant dB loci. Similarly when constant N circles are
transferred to log magnitude-phase plot, they become constant phase loci. Consider
Fig. 3.65, where a constant M circle is shown. Let us consider the circle with M = 1.2.
The centre is
3.11 Determining Closed Loop Frequency Response from Open … 457

−230˚

−210˚ Im
g2

b1 a1
−190˚

5.2

2.2
g2
c1
−180˚ e
d 0
bn φ=0 Re
130˚
g2 c
−170˚ a g
1
b
M =1.2 circle g2
−150˚ −130˚ −1
2

Fig. 3.65 Constant M-circles in G( jω) plane

M2 1.44
c= = = 3.2727
M2 − 1 0.44

and radius
M 1.2
r = = = 2.7272
M2 −1 1.44 − 1

The corresponding dB to M = 1.2 is 20 log 1.2 = 1.58 dB.

φ −230◦ −210◦ −190◦ −180◦ −170◦ −150◦ −130◦

g1 2.2 1.0 0.8 0.81 0.8 1.0 2.2
20 log g1 6.85 0.0 −1.94 −1.95 −1.94 0.0 6.85
g2 − 5.2 6.2 6.21 6.2 5.2 −
20 log g2 − 14.32 15.85 15.86 15.85 14.32 −

Radial lines from the origin are drawn to the constant M circle. Consider the
line oa 1 . The phase angle is −230◦ and the distance g1 = 2.2 which corresponds
to 20 log 2.2 = 6.85 dB. Locate the point a 1 in the log magnitude-phase coordinate
in Fig. 3.66. Draw another radial line ob1 . The angle this radial line makes with
the reference axis is φ = −210◦ . It intersects the constants M circle at g1 = 1 and
g2 = 5.2. Corresponding to g1 , the dB = 20 log 1 = 0 and corresponding to g2 = 5.2,
the dB = 20 log 5.2 = 14.32. These points are fixed at b1 and b11 respectively in
Fig. 3.66. Thus, for different radial lines, φ and dB are calculated from this constant
M circle and transferred to log magnitude-phase coordinate. The complete constant
dB locus for M = 1.2 is shown in Fig. 3.66 as 1.58 dB curve. By similar procedure
458 3 Frequency Response Analysis

20

15 c1 d c
b1 b
10 +dB

5 a1 a
(0dB,−180) Critical point
0
−5 b11 e −dB
c11
−10 M =1.58 Constant dB locus

−225˚ −210˚ −195˚ −180˚ −165˚ −150˚ −135˚

Phase angle φ

Fig. 3.66 Constant dB locus for M = 1.2 or dB = 1.58

other constant M circles are transferred to the log magnitude phase coordinate. A set
of constant dB lines are now available for use.
Now consider any constant N circle. The same procedure as was done for trans-
ferring the constant M circles, constant N circles is transferred to the log magnitude-
phase coordinate. These loci are called constant phase angle loci.
The constant dB loci together with constant phase angle loci is called Nichols
chart which is shown in Fig. 3.67. Nichols charts are readily available and can be
used to solve control system problems very easily and quickly. From Fig. 3.67, it is
seen that the Nichols chart is symmetric about the −180◦ vertical axis. The point
(0 dB, −180◦ ) is called the critical point. The G( jω) plot which is tangential to any
one of constant dB lines gives the resonant peak Mr and the corresponding frequency
is the resonant frequency ωr . The bandwidth of the closed loop system is obtained at
the intersection point of G( jω) plot with M = −3 dB locus. If the gain of the open
loop system is small, the G( jω) will not be tangent to any of the constant dB loci
and for such cases the resonant peak does not exist for the closed loop system.

Example 3.37
A certain unity feedback control system has the following open loop transfer function

15
G(s) =
(s + 1)(s + 2)(s + 3)

Draw the G( jω) plot in log magnitude-phase plot and determine resonant peak
Mr , resonant frequency ωr and BW of the closed loop system. Also determine the
phase margin φm , gain margin G m , phase crossover frequency ω pc and gain crossover
frequency ωgc .
3.11 Determining Closed Loop Frequency Response from Open … 459

Fig. 3.67 Nichols chart derived from constant M and constant N circles

Solution
15
G(s) =
(s + 1)(s + 2)(s + 3)
15
G( jω) =
( jω + 1)( jω + 2)( jω + 3)
15
|G( jω)| = 
(ω + 1)(ω2 + 4)(ω2 + 9)
2
ω ω
φ = ∠G( jω) = − tan−1 ω − tan−1 − tan−1
2 3
460 3 Frequency Response Analysis
 
dB = 20 log 15 − 20 log (ω2 + 1) − 20 log (ω2 + 4)

−20 log (ω2 + 9)

For different values of ω, φ and dB are calculated and tabulated below

ω 0.5 1.5 2.5 3.5 11 40

∠G( jω) −50◦ −120◦ −159◦ −183◦ −239◦ −261◦
dB = 20 log |G( jω)| 6.6 −0.07 −7.0 −13.1 −39.5 −72.6

The frequency response data in the table is transferred to log magnitude phase
Nichols plot and is shown in Fig. 3.68. The following results are obtained from
Fig. 3.68.
1. The phase margin and gain crossover frequency are obtained where the G( jω)
plot intersects 0 dB horizontal axis. This intersection is at point B in Fig. 3.68.
φm = 180◦ − 119◦ = 61◦ and ωgc = 1.49 rad/sec
2. The gain margin and phase crossover frequency ω pc are obtained where the
G( jω) plot intersects −180◦ vertical axis at point D. G m = 12 dB and
ω pc = 3.32 rad/sec

Fig. 3.68 Nichols plot for G(s) = 15

(s+1)(s+2)(s+3)
3.11 Determining Closed Loop Frequency Response from Open … 461

3. The bandwidth of the closed loop system is obtained from the intersection of the
G( jω) plot with −3 dB constant phase angle locus. This intersection is at point
C in Fig. 3.68. This dB is different from the dB marked in the vertical axis. The
frequency at this point is BW which is denoted by ωb = 2.46 rad/sec
4. The resonant peak Mr and resonant frequency ωr are obtained from the inter-
secting point of G( jω) plot with a constant dB locus which is tangential to it.
From Fig. 3.68, at point B, the G( jω) plot is just tangential to 0 dB constant dB
locus. Hence, Mr = 0 dB or Mr = 1. The corresponding frequency is the resonant
frequency ωr = 1.49 rad/sec

Resonant peak, Mr = 1
Resonant frequency, ωr = 1.49 rad/sec
Bandwidth, BW ωb = 2.46 rad/sec
Phase margin, φm = +61◦
Gain crossover frequency, ωgc = 1.49 rad/sec
Gain margin, G m = +12 dB
Phase crossover frequency, ω p c = 3.32 rad/sec

ω pc > ωgc ; φm and G m are +ve. The closed loop system is stable.

Example 3.38
Consider the following open loop T.F. of a certain closed loop system

K (1 + 0.1s)
G(s) =
s(1 + 0.5s)(1 + 0.2s)

Find the value of K so that the resonant peak of the closed loop system is 6 dB.

Solution
K (1 + 0.1s)
G(s) =
s(1 + 0.5s)(1 + 0.2s)
K (s + 10)
=
s(s + 2)(s + 5)
K ( jω + 10)
G( jω) =
jω( jω + 2)( jω + 5)

K (ω2 + 100)
|G( jω)| = 
ω (ω2 + 4)(ω2 + 25)
φ = tan−1 0.1ω − 90◦ − tan−1 0.5ω − tan−1 0.2ω
462 3 Frequency Response Analysis

Let K = 1

(ω2 + 100)
|G( jω)| = 
ω (ω2 + 4)(ω2 + 25)

For different values of ω, dB = 20 log |G( jω)| and φ are calculated and are tabulated
below

ω 0.5 1.0 1.5 6 30

φ −104◦ −122◦ −135◦ −181◦ −184.5◦
2 dB = 20 log |G( jω)| 5.7 −1.1 −5.73 −28.1 −88

The frequency response plot of G( jω) is represented in Nichols chart as shown

in Fig. 3.69.
From Fig. 3.69, the G( jω) plot is not tangential to 6 dB locus. The G( jω) plot is
vertically shifted upwards. At point A it is just tangential to 6 dB constant magnitude
locus. The amount of the shift of point B which is vertically below to point A is
measured as 11 dB +ve. The required gain is calculated from

(1 + 0.1s)
Fig. 3.69 Frequency response in Nichols chart for G(s) =
s(1 + 0.5s)(1 + 0.2s)
3.11 Determining Closed Loop Frequency Response from Open … 463

20 log K = 11

K = 3.55

Note: If the vertical shift is downwards, the dB is measured negative.

For example, for a downward shift of 11 dB, the value of K may be calculated as

20 log K = −11
1
K = = 0.2817
3.55

Example 3.39
A certain unity feedback control system has the following open loop T.F.

10 K
G(s) =
(s + 1)(s + 2)(s + 3)

It is desired to have a resonant peak Mr = 6 dB. or Mr = 2. Determine the value of

K . For this value of K , find φm , G m , ω pc , ωgc . Is the closed loop system stable?

Solution
10 K
G(s) =
(s + 1)(s + 2)(s + 3)

Put K = 1
10
|G( jω)| = 
(ω2 + 1)(ω2 + 4)(ω2 + 9)
ω ω
∠G( jω) = φ = − tan−1 ω − tan−1 − tan−1
2 3
For different values of ω, φ, |G( jω)| and dB = 20 log |G( jω)| are calculated and
tabulated below

ω 1 2 3 4 25
φ −90◦ −142◦ −173◦ −193◦ −254◦
|G( jω)| 1 0.4385 0.2067 0.108 6.33 × 10−4
dB = 20 log |G( jω)| 0 −7.16 −13.69 −19.3 −64
464 3 Frequency Response Analysis

10K
Fig. 3.70 Determine of K for Mr = 6 dB G(s) =
(s + 1)(s + 2)(s + 3)

The above data is transferred to Nichols chart. This is shown in Fig. 3.70 as curve
A. The graph A is vertically shifted up until any point in the graph is just tangential
to the 6 dB constant dB locus. The curve A when shifted vertically up is drawn as
curve B. The amount of vertical shift is observed as 8.7. Since the shift is upwards

20 log K = 8.7

K = 2.72

The new T.F. of the system now becomes

27.2
G(s) =
(s + 1)(s + 2)(s + 3)

The following MATLAB program may be used to get graph A and graph B.
3.11 Determining Closed Loop Frequency Response from Open … 465

MATLAB Program
Nichols plot
Gain Adjustment for desired Mr
G(s) = 10K /(s + 1)(s + 2)(s + 3)
num=[0 0 0 10]
den=conv([1 3 2],[1 3]);
sys=tf(num,den);
[G m , pm , ωcp , ωcg ]=margin(sys);
[G m dB, pm , ωcp , ωcg ]
num=[0 0 0 27.2]
den1=conv([1 3 2],[1 3]);
sys1=tf(num1,den1);
[G m1 , pm1 , ωcp1 , ωcg1 ]=margin(sys1);
[G m dB1, pm1 , ωcp1 , ωcg1 ]
Nichols (sys,sys1)

ans=
15.5630 90.0000 3.3166 1.0000

ans=
6.8717 29.6873 3.3166 2.2222

The answer for G m , φm , ω pc , ωgc for K = 1 and K = 2.72 is given at the end of
the program. In both cases, φm and G m are +ve and ω pc > ωgc . Hence, the closed
loop system is stable. Further, Mr = 6 dB or Mr = 2. Resonant frequency is obtained
as ωr = 2.5 rad/sec

Example 3.40
Consider the system shown in Example 3.39 where

10 K
G(s) =
(s + 1)(s + 2)(s + 3)

Find the value of K so that the gain margin is +30 dB. Use Nichols plot.

Solution
10 K
G(s) =
(s + 1)(s + 2)(s + 3)

Put K = 1;
466 3 Frequency Response Analysis

10
|G( jω)| = 
(ω2 + 1)(ω2 + 4)(ω2 + 9)
ω ω
∠G( jω) = − tan−1 ω − tan−1 − tan−1
2 3
The frequency response data prepared in Example 3.39 is used and is plotted in
Nichols chart as shown in Fig. 3.71. Graph A represents G( jω) plot for K = 1. with
ω pc = 3.32 and gain = −15.5 dB. The curve is shifted down so that it reaches point
B where the gain is −30 dB. Hence, the decrease in dB is −(30 − 15.5) = −14.5 dB.
The required gain is calculated from

20 log K = −14.5
K = 0.188

The T.F. of the system becomes

10K
Fig. 3.71 Gain Adjustment for the desired G m = 30 dB; G(s) =
(s + 1)(s + 2)(s + 3)
3.11 Determining Closed Loop Frequency Response from Open … 467

1.88
G(s) =
(s + 1)(s + 2)(s + 3)

Example 3.41
Consider the Example 3.39 where

10 K
G(s) =
(s + 1)(s + 2)(s + 3)

Determine the value of K so that phase margin of the system is 45◦ .

Solution
10 K
G(s) =
(s + 1)(s + 2)(s + 3)

For K = 1, the data prepared in Example 3.39 is used to plot G( jω) in Nichols chart
and is shown in Fig. 3.72 as curve A. The phase margin is now found as 90◦ . The
point A is vertically shifted up. Now the G m is decreased. φm is also decreased. ωgc
is also decreased. But ω pc is unchanged. The vertical shift is made until the phase
margin is 45◦ as seen from graph B. From point A to point B, the upward shift is
(15.6 − 9.64) = 5.96. The corresponding value of K is calculated from

20 log K = 5.96

K =2

The T.F. becomes

20
G(s) =
(s + 1)(s + 2)(s + 3)

G m = +9.64; ωgc = 1.82; ω pc = 3.33.

Example 3.42
Consider the system shown below, where

K (s + 0.5)
G(s) =
(s + 1)(s + 2)(s + 3)
468 3 Frequency Response Analysis

10K
Fig. 3.72 Gain adjustment to get the desired φm = 45◦ ; G(s) =
(s + 1)(s + 2)(s + 3)

It is desired to have a BW= 11 rad/sec by adjusting the gain K . Determine the value
of K . Under this condition find φm and G m .

Solution
K (s + 0.5)
G(s) =
(s + 1)(s + 2)(s + 3)

Putting K = 1;

(s + 0.5)
G(s) =
(s + 1)(s + 2)(s + 3)
3.11 Determining Closed Loop Frequency Response from Open … 469

K (s + 0.5)
Fig. 3.73 Gain adjustment for the desired BW= 11 rad/sec G(s) =
(1 + s)(s + 2)(s + 3)

ω 1 3 5 8 11 15
φ = ∠G( jω) −26.6◦ −92.32◦ −121.6◦ −142◦ −151.8◦ −159.2◦
|G( jω)| 0.118 0.062 0.0313 0.01422 7.82 × 10−3 4.31 × 10−3
dB −18.56 −24 −30 −37 −42 −47.3

The data for G( jω) plot is prepared and used to plot G( jω) in Nichols chart and
is shown in Fig. 3.73 as curve A. The curve does not intersect −3 dB constant dB
plot. Hence, the curve A is shifted vertically upwards until −3 dB locus is intersected.
Further it is shifted such that at the point of intersection the value of ω on G( jω)
curve is 11 rad/sec. This corresponds to the required BW. At point B of the shifted
curve ω = 11. The amount of vertical shift of point A to point B in Fig. 3.73 is
obtained as 35.15 dB. The value of K is calculated from

20 log K = 35.15

K = 57
470 3 Frequency Response Analysis

57(s + 0.5)
G(s) =
(s + 1)(s + 2)(s + 3)

From Fig. 3.73, we obtain

φm = + 42.3◦
Gm = + ∞
ωgc = 7.15 rad/sec
ωb = 11 rad/sec
K = 57

Example 3.43
A certain unity feedback control and system has the following open loop T.F.

e−10Td s
G(s) =
s(s + 2)(s + 8)

Determine the value of Td so that the phase margin is 45◦ . What is the gain crossover
frequency?

Solution

e−10Td s
G(s) =
s(s + 2)(s + 8)

Let Td = 0

1
G(s) =
s(s + 2)(s + 8)
1
|G( jω)| = 
ω (ω + 4)(ω2 + 64)
2
ω ω
∠G( jω) = φ = −90◦ − tan−1 − tan−1
2 8
3.11 Determining Closed Loop Frequency Response from Open … 471

Fig. 3.74 Determination of transportation lag for the required phase margin φm = 45◦ ; G(s) =
e−10Td s /s(s + 2)(s + 8)

The following table is prepared.

ω 0.06 0.5 1 2 4 10
∠G( jω) −92◦ −108◦ −124◦ −149◦ −180◦ −220◦
|G( jω)| 1 0.121 0.0555 0.0214 6.25 × 10−3 7.66 × 10−4
dB 0 −18.3 −25.1 −33.4 −44 −62.3

The G( jω) plot without time delay is shown in Fig. 3.74. It is noted that φm = 88◦
at ωgc = 0.06 rad/sec. The required phase margin is 45◦ . Hence, the G( jω) plot is
horizontally shifted towards left by an ∠ − 43◦ which makes the φm = 45◦ with time
delay. The magnitude contribution by time delay is 0 dB at all frequency. Hence, at
ω = ωgc = 0.06, the phase of −43◦ is contributed by e−Td s . In other words

−10Td ωgc × 57.3 = −43

10Td × 0.06 × 57.3 = 43

Td = 1.25 sec
472 3 Frequency Response Analysis

SUMMARY

1. The frequency response is the plot of magnitude and phase difference between
the input and output sinusoids over a range of frequencies.
2. It has several advantages and a few disadvantages too in comparison with time
domain approach.
3. When sinusoidal input is applied to a linear system, the output is also sinusoidal
with a different amplitude and a phase difference at steady state.
4. The ratio of the output to input sinusoid is called magnitude or amplitude ratio and
the phase difference is called phase angle which are functions of frequency. The
magnitude and phase angle of the given linear system is obtained by replacing s
by jω in the T.F. provided the given linear system is stable.
5. The plot of magnitude ratio and phase angle with frequency as the variable
parameter is called frequency response plots.
6. The frequency response can be represented in graph which can be used for the
analysis and design. The very extensively used methods include (a) Polar plot,
(b) Bode plot and (c) Log magnitude phase (Nichols) plot. Each of these methods
has its own merits and demerits.
7. A correlation between time domain and frequency domain can be established
for a second order system. Similar to time domain specifications which are
expressed in terms of damping ratio and natural frequency of oscillations, in
the frequency domain, expressions for resonant peak Mr , resonant frequency ωr
and the bandwidth ωb are derived. Thus, there exists a one-to-one relationship
between time domain and frequency domain specifications. However, such a
relationship is valid only for a second order system.
8. In the frequency domain, other important specifications such as gain margin,
phase margin, gain crossover frequency and phase crossover frequency are
defined which are applicable to higher order systems also.
9. Gain margin and phase margin are the important measures of relative stability
of linear system. The higher the phase margin and gain margin are the higher is
the degree of stability.
10. For the linear system to be stable, both phase margin and gain margin should be
positive. If any one of them is negative then the system is unstable. Further the
condition ω pc ≥ ωgc should be satisfied.
11. Frequency response methods not only give information about the stability of the
closed loop system but they also give information about the transient response of
the system via phase margin, gain margin, bandwidth resonant peak and resonant
frequency.
12. Obtaining frequency response of an open loop system by analytical methods
as well as by experimental method is easy. However, due to non-availability
of closed loop poles, obtaining closed loop frequency response is somewhat
difficult if not impossible.
3.11 Determining Closed Loop Frequency Response from Open … 473

13. Closed loop frequency response can be obtained from open loop frequency
response by using graphical aids such as constant M and constant N circles
and Nichols chart.
14. With the availability of computers and the MATLAB software today, frequency
response plots described in this chapter which are drawn manually can be
obtained very quickly and accurately. Side by side the required MATLAB pro-
gram and the plots obtained are given.

EXERCISE

Short Answer Type Questions

1. Define frequency response.

The frequency response is the steady state response of the system to a sinusoidal
input whose frequency is varied over a certain range.
2. What are the two components of frequency response?
The two components of frequency response are the magnitude component and
phase component.
3. What is a sinusoidal transfer function?
In the transfer function of a stable system if s is replaced by jω, then G(s)
becomes G( jω). G( jω) is called the sinusoidal T.F.
4. How magnitude and phase components of frequency response is obtained
for the given T.F.?
In the T.F. G(s), if s is replaced by jω, then the sinusoidal T.F. is obtained as
G( jω) which is a complex quantity. |G( jω)| which is the ratio of the output
amplitude to input amplitude gives the magnitude component. ∠G jω gives the
phase shift of the output from the input and this represents the phase component
of the frequency response.
5. Mention any three major advantage of frequency response method.
The advantages of frequency response methods are as follows:
(a) It is easy to generate sine wave and collect data to draw frequency response
which can be used for design purpose.
(b) The frequency response tests involve measurement under steady state con-
ditions which are simpler to analyse.
(c) It can be applied to a certain class of non-linear system also.
6. Mention any two disadvantages of frequency response method.
The disadvantages of frequency response methods are as follows:
(a) It is not often easy to correlate the results of the frequency response char-
acteristics of higher order system (order greater than two) to that of time
domain characteristics.
(b) In frequency response method, the analysis is done at one frequency at a
time which requires very long test time.
474 3 Frequency Response Analysis

Fig. 3.75 T.F. for Question 7

1
G(s)=
x(t)= 5 sin (2t+20º) (s+4)

7. For the system shown in Fig. 3.75 find the output.

1
G( jω)|ω=2 =
j2 + 4
1
= ∠−63.4◦
4.472
y(t) = 1.118 sin(2t − 43.4◦ )

8. What are the magnitude and phase angle components of a prototype second
order system at ω = ωn ?

The magnitude component |G( jω)| = 1/2ζ

The phase angle component ∠G( jω) = ∠−90◦

9. What are the methods available to represent the frequency response in

graph?
The methods available to represent the frequency response in graph are as fol-
lows:
(a) Linear frequency response plots vs. ω
(b) Linear frequency response plots vs. log ω
(c) Polar (Nyquist) plot
(d) Bode plot
(e) Log magnitude vs. phase (Nichols) plot.
10. What is a polar (Nyquist) plot?
Polar plot is the plot of the imaginary versus the real part of the complex sinu-
soidal T.F. G( jω) drawn in a polar graph with ω as the varying parameter.
11. What are the advantages and disadvantages of polar plot representation of
frequency response?
The advantage of polar plot representation of frequency response is that both
amplitude and phase angle are displayed on a single plot over a wide frequency
range. It has the property to show clearly, from the open loop frequency response
plot whether the closed loop system is stable or unstable. The famous Nyquist
stability criterion has been developed using polar plot only. The only disadvan-
tage of polar plot is that it does not clearly indicate the contributions made by
each pole and zero of the T.F.
12. What part of the T.F. influences the starting portion of a polar plot?
The Type of the system decides the starting portion of the polar plot. For example,
3.11 Determining Closed Loop Frequency Response from Open … 475

the polar plot of a type o system starts from +ve real axis and a Type 1 system
starts from − j∞.
13. What part of the T.F. influences the ending part of a polar plot?
The difference in the number of poles and zeros of the T.F. decides the ending
portion of the polar plot. For example, if there are 4 poles and one zero, the polar
plot ends in the second quadrant at the origin with a phase angle of −(4 − 1)π =
−270◦ .
14. For the following T.Fs. determine the magnitude and phase angle of the
sinusoidal T.F.
K (s + 10)
(a) G(s) =
s(s + 5)
K (s + 10)
(b) G(s) =
s 2 (s + 5)
K (s − 10)
(c) G(s) =
s(−s + 5)
−K (s + 10)
(d) G(s) =
s 2 (s + 5)
K (s + 10)
(e) G(s) =
s 2 + 4s + 16)

Ans:

K (ω2 + 100) ω ω
(a) G( jω) =  ∠tan−1 − 90◦ − tan−1
ω (ω + 25)
2 10 5

K (ω + 100)
2 ω ω
(b) G( jω) =  ∠tan−1 − 180◦ − tan−1
ω (ω + 25)
2 2 10 5

K (ω + 100)
2 ω ω
(c) G( jω) =  ∠90◦ − tan−1 + tan−1
ω (ω + 25)
2 10 5

K (ω2 + 100) ω ω
(d) G( jω) =  ∠tan−1 − tan−1
ω (ω + 25)
2 2 10 5

K (ω2 + 100) ω 4ω
(e) G( jω) = ∠tan−1 − tan−1
(16 − ω ) + 16ω
2 2 2 10 (16 − ω2 )

15. For the following T.F.s Sketch the polar plots.

(a) G(s) = K (s + 10)

(b) G(s) = K (s − 10)
(c) G(s) = K (−s + 10)
K
(d) G(s) =
(s + 10)
476 3 Frequency Response Analysis

K
(e) G(s) =
(s − 10)
K
(f) G(s) =
(−s + 10)
K (10 − s)
(g) G(s) =
(s − 5)

K (10 − s)
(h) G(s) =
s(s − 5)
K
(i) G(s) =
s(s 2 + 4s + 16)
K (s − 2)
(j) G(s) =
(s + 2)

16. What are Bode plots or Bode diagrams?

The log-magnitude and phase frequency response curves plotted as a function
of logarithmic frequency are called Bode plots or Bode diagrams.
17. Mention any three advantages of Bode plots.
In Bode magnitude and phase plots, the horizontal axis is on log scale. Since the
horizontal axis represents frequency, a wide range of frequency can be included
with clarity at all the frequencies. The vertical axis of the magnitude plot rep-
resents magnitude of G( jω) in dB. This enables to provide large amplitude
of G( jω) being represented. Multiplication and division in the sinusoidal T.F.
become addition and subtraction when magnitude is expressed in dB. This pro-
vides clear information about the magnitude contribution of each pole and zero
at any frequency range.
18. What are the disadvantages of Bode plots?
Bode plots are sketched using straight line approximation which introduces error.
The estimation of relative stability using phase margin and gain margin is appli-
cable for minimum phase T.F. and not for non-minimum phase T.F.
19. What are asymptotes in Bode magnitude plot?
Bode magnitude plot is sketched by approximating the magnitude with straight
line segments. These approximate straight line segments are called the asymp-
totes.
20. What is break frequency or corner frequency in Bode plot?
The break frequency or corner frequency which is denoted by ωc is the frequency
at which the low frequency asymptote and high frequency asymptote meet.
21. What do you understand by octave and decade?
If the frequency is multiplied by a factor two, this frequency range is called an
octave. If the frequency is multiplied by a factor ten, this frequency range is
called a decade. The frequency range 2 to 4 is an octave. The frequency range
from 2 to 20 is a decade.
22. What do you understand by dB/oct and dB/dec?
The slope of the straight line of the magnitude plot is expressed as dB/oct or
3.11 Determining Closed Loop Frequency Response from Open … 477

dB/dec. dB/oct represents the rise or fall in dB in the frequency range of one
octave. Similarly, dB/dec represents the rise or fall in dB in the frequency range
of one decade.
23. How dB/oct and dB/dec are related?
6 dB/oct is same as 20 dB/dec.
24. Each pole and each zero of a system contributes how much of a slope to the
Bode magnitude plot?
Each pole contributes −20 dB/dec and each zero contributes +20 dB/dec. slope
to the slope of Bode magnitude plot.
25. What is the slope of the magnitude Bode plot at high frequency for a system
with four poles and two zeros?
The slope of the Bode magnitude plot at high frequency for a system with 4 poles
and 2 zeros is −40 dB/dec.
26. The Bode magnitude plot of a certain system is shown below with frequency
in logarithmic scale. Find the rise in dB between ω1 and ω2 .
Between ω1 = 5 and ω2 = 80, there are 4 octaves. Hence, the rise in dB between
ω1 and ω2 = 4 × 12 = 48 dB. Alternatively, 12 dB/oct = 40 dB/dec. Hence, the
rise in dB = 40 log 80
5
= 48 dB.
27. What is the error introduced at ωc by asymptotic plot for a pole.
The error introduced at the corner frequency ωc for a pole is −3 dB.
28. How the error is corrected in Bode magnitude plot for a pole with a corner
frequency ωc .
The error is corrected as follows. At corner frequency ωc , −3 dB is located. At
one octave above and below corner frequency −1 dB points are located below
the asymptotes. By joining these points by a smooth curve, the error is corrected
to the minimum value.
29. A certain system has the following T.F.

G(s) = K (s + 4)3

What is the corner frequency and how much error is to be added to the
magnitude plot?
The corner frequency is 4 rad/sec. The error is 9 dB at ω = 4 and +9 dB is to be
added to the asymptotic plot at this frequency.
30. What is the effect of time delay e−sT d in the Bode diagram?
The sinusoidal T.F. of the time delay is

e− jωT d = 1∠−57.3ωT d

The magnitude is 1 and the dB is zero. Hence, the magnitude plot of the system
with time delay is not affected. However, a phase of −57.3ωT d degrees is to be
added at each frequency in the phase angle plot of Bode diagram.
31. What is log magnitude-phase (Nichols) plot?
The log magnitude-phase plot of G( jω) is a plot of G( jω) magnitude in dB
478 3 Frequency Response Analysis

versus its phase in degrees with ω as a parameter on the curve. The Nichols plot
can be derived from Bode diagram very easily.
32. Of the methods namely Nyquist, Bode and Nichols which one is more suit-
able for stability analysis?
Nyquist or polar plot is the most suitable one for stability study. The stability
analysis of system with non-minimum phase T.F. has to be very carefully anal-
ysed using Bode and Nichols plots. However, in Nyquist plot both minimum
phase and non-minimum phase T.F.s are treated in a similar way.
33. Bode diagram is more suitable for what applications?
Bode plot is an ideal one for the determination of phase margin, gain margin,
phase crossover and gain crossover frequencies of minimum phase systems.
It gives ω pc and ωgc very explicitly. It is also widely used in the design of
compensator where frequent adjustment of poles and zeros of the compensators
is required.
34. Nichols plot is more suitable for what applications?
Nichols plot is extensively used to plot the closed loop frequency response from
open loop frequency response using Nichols chart. It is a very ideal method to
apply for the determination of gain for the desired resonant peak and BW of the
closed loop system and also to meet the required phase margin and gain margin
of the closed loop system.
35. Define bandwidth.
The bandwidth BW is the frequency ωb at which |M( jω)| drops to 70.7 per cent
of, or 3 dB down from its zero frequency value.
36. What are frequency domain specifications?
The frequency domain specification are as follows:
(a) Resonant peak,
(b) Resonant frequency,
(c) Bandwidth,
(d) Phase margin,
(e) Gain margin,
(f) Phase crossover frequency, and
(g) Gain crossover frequency.
37. What are constant M circles?
The constant M circles are the family of circles drawn on the G( jω) plane.
These circles represent the
 magnitude of the  closed loop system. The centre of
the circle is given by c = 0, M 2 /(M 2 − 1) and the radius is r = M/(M 2 − 1).
38. How the constant M loci are placed in the G( jω) plane?
The constant M circles are symmetrical with respect to M = 1 line and the
real axis. For M = 1, the circle becomes a vertical straight line passing through
(−1/2, 0) point. The circles to the left of M = 1 correspond to M > 1 and circles
to the right of M = 1 correspond to M < 1. At (−1, 0) point M = ∞.
3.11 Determining Closed Loop Frequency Response from Open … 479

39. What are the applications of the constant M circles?

(a) The points of intersection of G( jω) plot when superimposed on constant M
loci give the closed loop magnitude at the corresponding frequency on the
G( jω) curve. Thus, if the open loop frequency response is superimposed
on the constant M loci, the closed loop magnitudes at different frequencies
can be obtained.
(b) The resonant peak of the closed loop system is obtained where the G( jω)
plot is tangential to the constant M circle. The corresponding frequency on
the G( jω) gives the resonant frequency ωr .
(c) When the G( jω) plot intersects M = 0.707 constant M circle, the corre-
sponding frequency on the G( jω) curve gives the BW of the closed loop
system.
40. What are constant N circles?
The constant N circles are the family of circles drawn on the G( jω) plane.
These circles represent the phase angle of the closed loop
 system. The centre of
the circle is given by c = [−1/2, 1/2N ] and radius r = (1 + N 2 )/4N 2 , where
N = tan φ. Here φ is the closed loop phase angle.
41. How constant N circles are used to get closed loop phase if open loop fre-
quency response is known?
The G( jω) plot is super imposed on the loci of constant N circles. The points
of intersections of G( jω) with these loci give the closed loop phase angle at the
corresponding frequency on the G( jω) curve.
42. How closed loop frequency response is obtained if the open loop frequency
response is known using constant M and constant N circles?
The G( jω) plot which represents the open loop frequency response is superim-
posed on the constant M-loci and constant N loci separately. From the points
of intersection of G( jω) on constant M loci closed loop magnitude at different
frequency values on the G( jω) curve are obtained and tabulated. Similarly, for
the same frequency values, the points of intersection when G( jω) is superim-
posed on constant N circles are obtained. These values give the closed loop
system phase. Knowing closed loop system magnitude and phase for different
frequencies, the closed loop frequency response is drawn.
43. Can we use constant M and constant N circles for non-unity feedback sys-
tem also?
Constant M and constant N circles cannot be used to get the closed loop fre-
quency response of a non-unity feedback system. However, it can be used with
small modification.
44. The G( jω) plot of certain closed loop system passes through the left of
(−1, 0) point. Is it possible to get closed loop magnitude from constant N
circles?
For a system with minimum phase T.F. if G( jω) passes to the left of (−1, 0)
point, the closed loop system becomes unstable and the output becomes
unbounded. Under this condition, the constant M circles no longer have any
meaning.
480 3 Frequency Response Analysis

45. What are the disadvantages of using constant M and constant N circles?
To get the closed loop frequency response from open loop frequency response,
the procedure has to be followed twice, once using the constant M loci and at
another time using the consent N loci. This is the main disadvantage of using
constant M and N loci separately.
46. What is a Nichols Chart?
Log magnitude versus phase plot is called Nichols plot with frequency as the
variable parameter. A set of radial lines drawn from the origin of G( jω) plane
intersect a constant M circle. The points of intersection of these radial lines
are converted into dB. The phase angles of these radial lines are also measured
and transferred to the log-magnitude and phase plane. The loci of these circles
become constant dB loci. Similar procedure is adopted to the constant N loci
and transferred as constant phase angle loci which together form Nichols Chart
(see Sect. 3.11.4).
47. What are the advantages of Nichols Chart?
(a) The Nichols chart gives both the gain and phase characteristics of the closed
loop system simultaneously at one instant.
(b) The frequency response specifications can be easily and quickly determined
from Nichols chart in addition to obtaining closed loop frequency response
from open loop frequency response.
48. What is a critical point in the Nichols chart? Why is it called so?
(0 dB, −180◦ ) point is called the critical point in the Nichols chart. It is called so
because, for a minimum phase system, if the G( jω) plot passes to the right of
it, the closed loop system is stable. On the other hand if the G( jω) plot passes
through the left of the critical point, the system is unstable.
49. Define gain margin in the polar coordinate.
Gain margin is defined as the amount by which the loop gain is increased before
the closed loop system becomes unstable. It is the reciprocal of the gain of
G( jω) plot when its phase angle is −180◦ .
50. Define phase margin in the polar coordinate?
Phase margin is defined as the phase angle in degrees through which the G( jω)
plot be rotated about the origin before the closed loop system becomes unstable.
It is 180◦ + phase angle of G( jω) when |G( jω)| = 1.
51. What is gain crossover and gain crossover frequency in the polar coordi-
nate?
The gain crossover is defined as a point on the G( jω) plane at which |G( jω)| =
1. The corresponding frequency on the G( jω) curve gives the gain crossover
frequency denoted by ωgc .
52. What is phase crossover and phase crossover frequency in the polar coor-
dinate?
The phase crossover is defined as the point on the G( jω) plane at which the plot
intersects the negative real axis. The corresponding frequency on the G( jω)
plot is called phase crossover frequency denoted by ω pc .
3.11 Determining Closed Loop Frequency Response from Open … 481

53. For the estimation of gain margin what is known in G( jω)?

When gain margin is to be estimated, the phase angle of G( jω) is known which
is −180◦ . By this ω pc is calculated.
54. For the estimation of phase margin what is known in G( jω)?
When phase margin is to be estimated, the magnitude of G( jω) is known which
is unity. By this ωgc is calculated.
55. How relative stability is defined in the frequency domain?
In the frequency domain, the relative stability is defined by how close the Nyquist
plot of G( jω) is to the (−1, 0) point.
56. What are the measures of relative stability in the frequency domain?
In the frequency domain, relative stability is measured in terms of phase margin
and gain margin. For the system to be stable both phase margin and gain margin
should be positive. Further ω pc > ωgc . The higher the phase margin and gain
margins are the higher is the relative stability.
57. If ω pc = ωgc , what is the status of stability of the closed loop system?
If ω pc = ωgc , the G( jω) plot passes through the critical point (−1, 0) in the
G( jω) plane and the system is said to be critically stable or oscillatory.
58. Under what condition the positive gain margin is not an indication of closed
loop system stability?
For a system with non-minimum phase T.F. positive gain margin may not ensure
closed loop system stability.
59. Define gain margin in Bode plot?
In Bode plot, the gain margin is defined as the amount of gain in dB that can be
added to the loop before the closed loop system becomes unstable.
60. Define phase margin in Bode plot.
In Bode plot, the phase margin is defined as the amount of the phase angle that
can be added to the loop before the closed loop system becomes unstable.
61. How to identify positive and negative gain margins in polar plot?
If the gain margin is greater than one, the gain margin is +ve and ω pc > ωgc . If
the gain margin is less than one, G m is negative and ω pc < ωgc . If G m = 1, the
system is oscillatory and ω pc = ωgc .
62. How do you identify positive and negative phase margin in polar plot?
If at unity gain of G( jω), the phase angle of G( jω) is less than −180◦ , the
phase margin is positive and now ω pc > ωgc . If at unity gain of G( jω), the phase
angle of G( jω) is greater than −180◦ , the phase margin is negative and now
ω pc < ωgc . The phase margin is zero when ω pc = ωgc and the G( jω) plot passes
through (−1, 0) point and the closed loop system has oscillatory behaviour.
63. What is the corner frequency of a prototype second order system in Bode
plot? What is the error introduced at ωc by approximation?
The natural frequency of oscillation ωn of the prototype second order system
is the corner frequency in the Bode plot. The error introduced at ωc by the
asymptote is given by −20 log 2ζ .
64. For a second order system, how the phase margin is related with the damping
factor?
For a second order system, the phase margin is related with damping factor as
482 3 Frequency Response Analysis

π
ζ = φm
360◦
65. What is the effect of bandwidth on transient response?
A large bandwidth allows high frequency signals more easily to pass through.
This corresponds to faster time response in the time domain. If the BW is small
the time response will be slow. The rise time and BW are inversely proportional
to each other. However, large bandwidth increases noise and component cost.
66. How static error constants are determined from Bode diagrams?
From Bode magnitude plot the static error constants are determined. These con-
stants are determined from low frequency response part of Bode plot. The posi-
tion error constant K p exits only for a Type 0 system. For Type 0 system the
frequency Bode magnitude is 20 log K = K p . The velocity error constant exits
for a Type 1 system. This is obtained from ω where Kω cuts the frequency axis.
That is K v = ω where ω = K . The acceleration error constant exits for Type 2
√ from ω where K /ω cuts the frequency axis. That is
2
system. This is obtained
K a = ω where ω = K .
2

67. How to identify positive and negative gain margins from Bode plots?
The gain margin is measured positive if the magnitude of the G( jω) plot at the
phase crossover is negative in dB. The gain margin is measured +ve below the
0 dB axis. Here ω pc > ωgc . The gain margin is measured −ve if it is measured
above the 0 dB axis at the phase crossover. Here ω pc < ωgc for a minimum phase
T.F.
68. How to identify the positive and negative phase margins from Bode plots?
The phase margin is measured positive if the phase angle of G( jω) plot at the
gain crossover is greater than −180◦ . The phase margin is measured +ve above
the −180◦ axis. Here ω pc > ωgc . The phase margin is measured −ve if it is
measured below the −180◦ axis at the gain crossover. Here ω pc < ωgc .
69. What is phase margin and gain margin when ω pc = ωgc ?
When ω pc = ωgc , the phase margin is 0◦ and the gain margin is 0 dB. The system
is now oscillatory.
70. What is the effect of increasing and decreasing the value of open loop system
gain on the Bode diagrams?
When the gain of open loop system is increased, the Bode magnitude plot is
vertically shifted up. When the gain of the open loop system is decreased, the
Bode magnitude plot is vertically shifted down. In both the cases, the phase plot
is unaffected.
71. What is the effect of incorporating a transportation lag on the Bode dia-
gram? How the relative stability is affected?
By incorporating a transportation lag in the system, the magnitude plot will not
be affected. However, the phase angle plot is shifted downwards very rapidly
as ω increases. ω pc is shifted towards left which decreases the gain margin. ωgc
is unaffected. Since the phase curve is shifted down, the phase margin is also
decreased.
3.11 Determining Closed Loop Frequency Response from Open … 483

72. How do you identify positive and negative gain margin in Nichols plot?
If the dB of G( jω) at the critical point (0 dB, −180◦ ) is negative, the gain margin
has +ve dB. That is the gain margin is +ve at the phase crossover and measured
below the 0 dB axis. Here ω pc > ωgc . The gain margin is measured negative (dB
is +ve) at the phase crossover and is above the 0 dB axis. Here ω pc < ωgc .
73. How do you identify positive and negative phase margin in Nichols plot?
In the Nichols plot, the phase margin is measured positive at the gain crossover
if it is to the right of the critical point (0 dB, −180◦ ). That is the G( jω) plot is
to the right of −180◦ vertical axis. Here ω pc > ωgc . If the gain crossover is to
the left of the critical point, the phase margin is negative. Here ω pc < ωgc .
74. What will be the nature of Nichols plot when ω pc = ωgc ?
For ω pc = ωgc , the G( jω) plot will pass through the critical point (0 dB, −180◦ ).
The system will have oscillatory behaviour.
75. How do you find BW from Nichols plot?
The Nichols plot of G( jω) is superimposed on the Nichols chart. The point of
intersection of G( jω) plot with the constant −3 dB locus is found out. The value
of ω on the G( jω) curve at this intersection point gives the BW of the closed
loop system.
76. How the gain of the system is determined for the desired gain margin using
Nichols chart?
The G( jω) plot with K = 1 is superimposed on the Nichols chart. ω pc is iden-
tified on the −180◦ vertical axis. The G( jω) plot is shifted up or down so that
at the phase crossover the required gain margin is obtained. For upward shift
the required gain is calculated from 20 log K = + ve dB and for downward shift
from 20 log K = −ve dB.
77. How do you calculate the gain of the system for the desired phase margin
using Nichols chart?
The G( jω) plot with K = 1 is superimposed on the Nichols chart. The G( jω)
plot is shifted up or down so that at gain crossover the required phase margin is
obtained. For upward shift, the gain is calculated from 20 log K = + dB and for
downward shift, 20 log K = − dB.
78. Sketch the polar plot for the following T.F. for 0 ≤ ω ≤ ∞.

10
(a) G(s) =
(s 2 + 4)
(b) G(s) = 10−0.1s
484 3 Frequency Response Analysis

79. For the following T.F.s. Sketch the Bode magnitude plot and determine K p ,
K v and K a respectively.

12
(a) G(s) =
(s + 5)
120
(b) G(s) =
s(s + 5)
200
(c) G(s) = 2
s (s + 5)

80. What is the damping ratio when the percentage overshoot is 100%? What
is the corresponding resonant peak?
For 100% overshoot, the damping ratio ζ = 0. The corresponding resonant peak
Mr = ∞.
81. The log magnitude-phase plot of a certain open loop T.F. gave the following
data.

G( jω)|ω=5 = −18 dB∠−180◦

G( jω)|ω=2 = 0 dB∠−140◦

Find φm , G m , ω pc and ωgc and the stability of the closed loop system.

φm = +40◦
G m = +18 dB
ω pc = 5 rad/sec
ωgc = 2 rad/sec

The closed loop system is stable.

82. What are the recommended values of phase margin and gain margin of a
system for satisfactory performance?
For satisfactory performance of a system, the recommended value of phase mar-
gin is between 30◦ and 60◦ and the gain margin is greater than 6 dB.
83. What is the desirable slope of the log magnitude plot at the gain crossover
frequency for the system to be stable?
−20 dB/dec slope characteristic of the log magnitude plot at the gain crossover
frequency is the desirable slope for the system to be stable (Figs. 3.76, 3.77, 3.78
and 3.79).

Long Answer Type Questions

1. For the following T.F. sketch the polar plot. What is the phase crossover
frequency and gain margin?
3.11 Determining Closed Loop Frequency Response from Open … 485

(a) (b) Im
Im
ω=
ω=

ω=0 ω=0
0 10 K Re −10 K 0 Re

G(s) = K(s+10) G(s) = K(s−10)

(c) (d) Im
Im

10 K
ω= K/10
0 ω=0 Re
0 ω=0 Re

K
ω= G(s) =
(s+10)
G(s) = K(10 −s)

(e) Im (f)
Im

−K/10 ω=
ω= ω=0
ω=0 0 Re
0 K/10 Re

K K
G(s) = G(s) =
(s−10) (10 −s)

(g) Im
(h) ω=0
Im
−2K −K
ω= Re
ω=0 0

ω= Re
K (10−s)
G(s) =
(s−5) K (10−s)
G(s) =
s(s−5)

(i) (j)
−K/64 Im Im

ω=
0 Re
ω=0 ω=
ω=4 −K 0 K Re

K
G(s) =
s(s2+4s+16) K (s−2)
G(s) =
ω=0 (s+2)

Fig. 3.76 Polar plot for different T.F.s of Question 15

486 3 Frequency Response Analysis

Fig. 3.77 Sample Bode

magnitude plot for Question 12 dB/oct
26
48 dB

1 5 2 80

(a) Im (b) Im

ω>2 ω=0 ω<2

ω=∞ − 10 10 ω =
2.5 Re 0 Re
ω=0
10

10
G(s) =
(s2+4) G(s) = 10e −0.1s

Fig. 3.78 Polar plot of Question 78

10
G(s) =
s(2s + 1)(s + 1)

Ans: ω pc = 0.707 rad/sec; G m = −0.15.

2. For the following T.F. sketch the polar plot and find phase margin, gain
margin, ω pc and ωgc .

20(s + 6)(s + 8)
G(s) =
(s − 4)(s − 7)

Ans: G m = −0.0394; ω pc = 6.06 rad/sec; φm and ωgc do not exist since unit
circle does not intersect the G( jω) plot. The system is unstable.
3. For the following T.F. sketch the polar plot and obtain G m , φm , ω pc & ωgc .

100
G(s) =
s(s + 2)2

Ans: G m = −0.16; or −15.9 dB; φm = −40.6◦ ; ω pc = 2 rad/sec; ωgc = 4.35

rad/sec.
3.11 Determining Closed Loop Frequency Response from Open … 487

(a) (b)

dB dB 20dB/dec

7.6 13.6
20 log Kp
log
24
5 12 5 11 log
dB dB
Kv 24
20dB/dec 40dB/dec
Kp 2.4

(c)
dB 40dB/dec

4.08

√40
0 log
5 5.85

dB √Ka √40
60dB/dec

Ka 40

Fig. 3.79 Bode plots of Question 79

4. For the following T.F. sketch the polar plot and obtain φm , G m , ω pc and ωgc .

10s 2
G(s) =
(s + 1)(s + 2)(s + 3)

Ans: φm = −90◦ ; ωgc = 1 rad/sec; G m = ∞; ω pc = 0.

5. Consider the following open loop T.F. of a certain unity feedback system.

20e−0.1Td s
G(s) =
s(s + 2)(s + 4)

Find φm , G m , ω pc and ωgc of the system without transportation lag. With

delay included what is the critical value of Td ?
Ans: Without delay: G m = +7.6 dB; φm = +25.64◦ ; ω pc = 2.82 rad/sec; ωgc =
1.73 rad/sec. Critical value of Td = 25.65/(5.73 × 1.73) = 2.59 sec.
488 3 Frequency Response Analysis

6. The forward path T.F. of a certain unity feedback system is given by.

200
G(s) =
(s + 2)3

By drawing the polar find φm , G m , ω pc and ωgc . Is the system stable? If

unstable, by what value the gain is to be reduced. With 90% of this value
what are the new φm , G m , ω pc and ωgc .
Ans: φm = −30◦ ; G m = −9.84 dB; ω pc = 3.46; ωgc = 5.5. The gain should be
< 64 for the system to be stable. For 90% of this value is K = 57.6. At this
value G m = 0.886 dB; φm = +3.6◦ ; ω pc = 3.46 rad/sec and ωgc = 3.3 rad/sec.
Both φm and G m are +ve ω pc > ωgc system is stable now.
7. For the forward path T.F. draw the polar plot and find φm , G m , ω pc and ωgc .
Is the system stable?

(5s 2 + 2s + 10)
G(s) =
(2s 2 − 3s − 2)

Ans: φm = −13.3◦ ; G m = 3.52 dB; ω pc = 1.16; rad/sec; ωgc = 1.82 rad/sec.

Since φm is negative the system is unstable.
8. For the following T.F. draw the asymptotic Bode magnitude and phase plots.
(a) Find φm , G m , ω pc , ωgc and stability of the closed loop system for (a)
K = 10. (b) Find K for φm = 60◦ and (c) Find K for G m = 40 dB.

K
G(s) =
s(s + 1)(s + 10)

Ans: (a) G m = 20.82 dB; φm = 47.4◦ ; ω pc = 3.16 rad/sec; ω pc = 0.784 rad/sec;

closed loop system is stable. (b) K = 4.47 and (c) K = 1.1 (Results obtained
from actual Bode plots).
9. Consider the following T.F.

12e−0.1Td s
G(s) =
s 1 + 21 s

Using Bode plot determine Td so that the phase margin φm = 45◦ .

Ans: Td = 0.8132 sec. (−0.1Td × 4.7 × 57.3 = −21.9) (Results obtained from
actual Bode plots).
3.11 Determining Closed Loop Frequency Response from Open … 489

10. A certain control system has the following T.F.

K e−0.2s
G(s) =
s(s + 2)(s + 8)

Using Bode plots determine the value of K if.

(a) φm = 45◦
(b) G m = 6 dB

Ans:
(a) For required φm = 45◦ , the vertical upward shift required is 25 dB. Hence,
20 log K = 25 or K = 17.8.
(b) For required G m = 6 dB, the required vertical upward shift is (31 − 6) =
25 dB. Hence, 20 log K = 25 or K = 17.8. Thus, with K = 17.8 the required
φm of 45◦ and the required G m of 6 dB are achieved. (Results obtained from
actual Bode plots).
11. Consider the following open loop T.F.

80
G(s) =
s(s + 2)(s + 20)

Find φm , G m , ω pc and ωgc using Bode diagrams. Is the closed loop system
stable? By what factor the gain is to be increased so that the gain crossover
frequency ωgc = 0.2 rad/sec? What is the new φm ?
Ans: G m = 20.8 dB; φm = 47.4◦ ; ω pc = 6.32 rad/sec; ωgc = 1.57 rad/sec. Both
G m and φm are positive. ω pc > ωgc . The closed loop system is stable. The gain
has to be increased by a factor of 0.1 so that ωgc is at 0.2 rad/sec. The new phase
margin φm = 84◦ . (Results obtained from actual Bode plots).
12. Consider the following T.F.

K (1 + 0.2s)
G(s) =
(1 + s)(s 2 + 16s + 80)
490 3 Frequency Response Analysis

For K = 1000 draw Bode diagram and find φm , G m , ω pc and ωgc and the
stability of the closed loop system. What is the value of K to have ωgc =
1 rad/sec and what is the corresponding φm and G m ?
Ans: For K = 1000; φm = 51.8◦ ; G m = ∞; ωgc = 12.5 rad/sec. For ωgc = 1;
K = 80; φm = 135◦ . (Results obtained from actual Bode plots).
13. Consider the following T.F.

K
G(s) =
s(s + 1)(1 + 0.5s)

Find the value of K so that ωgc = 0.5 rad/sec. Use Bode diagrams.
Ans: K = 0.5762, and φ= 50◦ . (Results obtained from actual Bode plots).
14. Consider the following T.F.

K
G(s) =
s(s + 10)

Determine the value of K using Bode diagrams so that φm = 60◦ . What is

the gain margin for this value of K ?
Ans: For upward shift 20log K = 36.7. K = 68.5 for φm = 60◦ . G m = ∞ for
all values of K . (Results obtained from actual Bode plots).
15. Consider the following open loop T.F. of a closed loop system.

15
G(s) =
s(s + 1)(s 2 + 4s + 9)

Sketch the log magnitude-phase plot. Determine φm , G m , ω pc and ωgc . Also

find the BW, ωr , Mr and the stability of the closed loop system.
Ans: G m = 11.4 dB; φm = 90◦ ; ω pc = 3.6 rad/sec; ωgc = 1.34 rad/sec; ωb =
2.23 rad/sec. Mr does not exist; closed loop system is stable.
16. Using Nichols chart determine the closed loop frequency response of the
unity feedback system whose open loop frequency response is (Figs. 3.80,
3.81, 3.82, 3.83, 3.84, 3.85, 3.86, 3.87, 3.88, 3.89, 3.90, 3.91, 3.92, 3.93, 3.94
and 3.95)
3.11 Determining Closed Loop Frequency Response from Open … 491
G(s)=10/(2s+1)(s+1)
2
−6.667

0
ω=
ω= 1
2
−2

ω
Im

−4

−6

−8

ω=0
−10
−20 −15 −10 −5 0
Real

10
Fig. 3.80 Polar plot for G(s) =
s(2s + 1)(s + 1)

G(s) 20(s 6)(s 8)/(s 4)(s 7)

40

30

20

10

pc=6.06
Im

34.3
0
25.41 20 0

10

20

30
30 20 10 0 10 20 30 40
Real

Fig. 3.81 Polar plot for G(s) = 20(s + 6)(s + 8)/(s − 4)(s − 7)
492 3 Frequency Response Analysis

G(s) 100/s(s 2)2

6.25
0
2

10
Im

15

20

25

30
25 20 15 10 5 0
Real

Fig. 3.82 Polar plot for G(s) = 100/s(s + 2)2

G(s) 10s2/(s 1)(s 2)(s 3)

1.5

1 gc 1

m 90°

0.5
Im

√11
0
0

1.833

0.5

1
0.5 0 0.5 1 1.5 2
Real

10s 2
Fig. 3.83 Polar plot for G(s) =
(s + 1)(s + 2)(s + 3)
3.11 Determining Closed Loop Frequency Response from Open … 493
G(s) 20/s(s 2)(s 4)
1

1
Im

0
5
1.5 1 0.5 0 0.5
Real

20e−0.1Td s
Fig. 3.84 Polar plot for G(s) =
s(s + 2)(s + 4)

G(s) 200/(s 2)3

5

0
0
25

5
Im

10

15

20
10 5 0 5 10 15 20 25 30
Real

Fig. 3.85 Polar plot for G(s) = 200/(s + 2)3

494 3 Frequency Response Analysis
G(s) (5s2 2s 10)/ (2s2 3s 2)
1.5

0.5
Im

0 5
0 2.5
pc 1.16

0.5

1
6 5 4 3 2 1 0 1 2 3
Real

Fig. 3.86 Polar plot for G(s) = (5s 2 + 2s + 10)/(2s 2 − 3s − 2)

Bode Diagram
Gm= 20.8 dB (at 3.16 rad/sec), m= 47.4 deg (at 0.784 rad/sec)

G(s) 10/s(s 1)(s 10)

50
7 dB pc=3.16
0
Magnitude (dB)

20.8 dB
19.2 dB
40
50

100

150
90
120
135 m=47.4
Phase (deg)

180
gc=0.784
225

270
10 2 10 1 100 101 102 103
Frequency (rad/sec)

10
Fig. 3.87 Polar plot for G(s) =
s(s + 1)(s + 10)
3.11 Determining Closed Loop Frequency Response from Open … 495

Bode Diagram
G m= dB (at rad/sec), m= 23.1 deg (at 4.7 rad/sec)
G(s) = 12/s(1 0.5s)
60

40
Magnitude (dB)

20
4.7
0
20

40
60
90
Phase (deg)

135
21.9º
156.9
m=23.1
180
10 1 100 101 102
Frequency (rad/sec)

12
Fig. 3.88 Bode diagram for G(s) =
s(s + 0.5s)

Bode Diagram
Gm= 44.1 dB (at 4 rad/sec), Pm= 87.8 deg (at 0.0625 rad/sec)
G(s) = 1/s(s+2)(s+8)
50

0
25 dB
31 dB
50
Magnitude (dB)

100

150
200
90

135
m=45°
Phase (deg)

180
Phase plot with delay
225 Phase plot without delay

270
10 2 10 1 100 101 102 103
Frequency (rad/sec)

e−0.2s
Fig. 3.89 Bode diagram for G(s) =
s(s + 2)(s + 8)
496 3 Frequency Response Analysis

Bode Diagram
Gm= 20.8 dB (at 6.32 rad/sec), Pm= 47.4 deg (at 1.57 rad/sec)
G(s) 80/s(s 2)(s+20)
50
20
20 dB 1.57 6.32
Magnitude (dB) 0
=0.2 Gm= 20.8 dB

50

100

150
90
96
135 84°
Phase (deg)

m
m 47.4°
180

225

270
10 1 gc=0.2 100 101 102 103
Frequency (rad/sec)

80
Fig. 3.90 Bode diagram for G(s) =
s(s + 2)(s + 20)

Bode Diagram
Gm= dB (at rad/sec), Pm= 51.8 deg (at 12.5 rad/sec)
G(s) 1000(1 0.2s)/(s 1)(s2 16s 80)
40

20
22 dB
Magnitude (dB)

12.5
0

20
40

60
80
0
gc 1
45
Phase (deg)

90
m 135°
135
m 51.8°
180
10 2 10 1 100 101 102 103
Frequency (rad/sec)

1000(1 + 0.2s)
Fig. 3.91 Bode diagram for G(s) =
(s + 1)(s 2 + 16s + 80)
3.11 Determining Closed Loop Frequency Response from Open … 497

Bode Diagram
Gm= 9.54 dB (at 1.41 rad/sec), Pm= 32.6 deg (at 0.749 rad/sec)
G(s) 1/s(s 1)/(1 0.5s)
50

4.8 dB
0
Magnitude (dB)

Gm= 9.54 dB

50

100

pc 1.41
150
90 0.5
gc
130
32.6°
Phase (deg)

m
m 50°
180

225 gc 0.744

270
10 2 10 1 100 101 102
Frequency (rad/sec)

1
Fig. 3.92 Bode diagram for G(s) =
s(s + 1)(1 + 0.5s)

Bode Diagram
G m= dB (at rad/sec), Pm= 89.4 deg (at 0.1 rad/sec)
G(s) 1/s(s 10)
50
Mangnitude (dB)

0
36.7 dB
36.7
50

100

150
90
gc 5.77

120º
Phase (deg)

135 m 60°

180
10 2 10 1 100 101 102 103
Frequency (rad/sec)

1
Fig. 3.93 Bode diagram for G(s) =
s(s + 10)
498 3 Frequency Response Analysis

15
Fig. 3.94 Bode diagram for G(s) =
s(s + 1)(s 2 + 4s + 9)
3.11 Determining Closed Loop Frequency Response from Open … 499

Fig. 3.95 Closed loop frequency response from open loop frequency response in Nichols chart
500 3 Frequency Response Analysis

200(s + 1)
G(s) =
s(s + 10)2

G( jω) Plot:

ω 1 2 3 5 10 15 20 25 30 40
∠G( jω) −56◦ −49◦ −52◦ −64◦ −96◦ −117◦ −130◦ −138◦ −145◦ −154◦
|G( jω)| 2.8 2.15 1.93 1.63 1.0 0.616 0.4 0.276 0.2 0.116
20 log 8.9 6.7 5.73 4.25 0 −4.2 −8.0 −11.2 −14 −18
|G( jω)| dB

M( jω):

ω 1 2 3 5 10 15 20 25 30 40
∠M( jω) −14◦ −15◦ −18◦ −22◦ −37◦ −70◦ −90◦ −110◦ −120◦ −130◦
|M( jω)| dB −2 −3 −2.8 −3 −3 −4 −6 −9 −12 −14
Chapter 4
Stability Analysis of Linear Control
System

Learning Objectives

After completing this chapter, you should be able to:

 Define linear system stability in time and frequency domains.
 Understand different methods available for stability study.
 Establish the general stability conditions for linear system.
 Apply Routh–Hurwitz stability criterion for the linear system.
 Give definitions of Nyquist stability criterion via Cauchy’s principle of the
argument.
 Apply Nyquist stability criterion for the stability analysis of linear system.

4.1 Introduction

In Chaps. 2 and 3, time domain and frequency domain specifications were defined
as applied to the linear system. A control system, when designed, has to satisfy
some of these specifications required by the individual. In these chapters, one other
important specification, namely system stability was not discussed in detail. If this
important specification is not satisfied, the system is of no use. The time and frequency
domain specifications discussed earlier have no meaning for an unstable system.
Physically, for an unstable system, the time response grows without bound and causes
severe damages not only to the system itself, but also to the environment and human
life. Stability studies have been carried out both in the time domain and frequency
domain. Routh–Hurwitz methods, root locus technique proposed by W. R. Evans and
Liapunov stability criterion are the time domain approaches developed in the past
for the stability studies. While Routh–Hurwitz and root locus methods are applicable
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 501
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_4
502 4 Stability Analysis of Linear Control System

only for a linear system, Liapunov method is applicable for both linear as well as non-
linear system. The methods proposed by Nyquist and Popov belong to the frequency
domain approach. While Nyquist stability criterion which is based on Cauchy’s
principle of the argument is applicable only to a linear system, Popov’s method can
be extended to non-linear systems with certain constraints. In the time domain, the
method developed by Routh–Hurwitz and Evan’s root locus is extensively used, while
Nyquist stability criterion is more popular in the frequency domain. In this chapter, we
deal with the application of Routh–Hurwitz stability criterion and Nyquist stability
criterion. A separate chapter is devoted to the study of root locus technique. We
discuss below the concept of stability and various terminologies associated with it.

4.2 The Concept of Stability

Stable systems are defined as the systems which will return to a stable equilibrium
position after some small disturbance. On the other hand, systems which are given
small initial disturbance, move off position and do not return to stable equilibrium
position are termed as unstable systems. Consider the simple pendulum shown in
Fig. 4.1a. When small disturbance is given to the pendulum, it oscillates horizontally
and comes to the same vertical equilibrium position, which is stable. Now consider
the inverted simple pendulum shown in Fig. 4.1b. When the pendulum is disturbed
it falls to the horizontal. This is an example of an unstable system.
There are many definitions for stability depending upon the kind of system. We
restrict our definitions to linear time invariant continuous time system. They are given
below.
Linear system stability is defined in terms of the system time response. The time
response consists of two parts. One is due to the initial condition and the other is
the response due to the input. The total response = zero state response + zero input
response.

(a) (b)

Stable Unstable

Fig. 4.1 a Simple pendulum and b Inverted simple pendulum

4.2 The Concept of Stability 503

1. Zero input response: The zero input response is due to the initial conditions
only. All the inputs are zero.
2. Zero state response: The zero state response is the response of the system due
to the input only. All the initial conditions are zero.
3. Zero input stability: It refers to the stability condition due to zero input response
which is due to the initial conditions only and all the inputs are zero. This stability
also depends on the roots of the characteristic equation.
A linear time invariant system is said to be zero input stable if the zero input response
of the system with finite initial conditions reaches zero as time t tends to infinity.

4.2.1 Asymptotic (Internal) Stability

A linear time invariant system is said to be in its equilibrium state if it remains in a

particular state indefinitely in the absence of external input. For a linear time invariant
system, zero state is the equilibrium state. When the system is given a small distur-
bance, the zero initial state is changed to non-zero state (initial conditions). These
initial conditions will generate signals consisting of characteristic modes (roots) in
the system. Let to itself if every mode in the system created by non-zero initial
conditions returns back to zero as time t tends to infinity, the system is said to be
asymptotically stable. On the other hand, even if any one of the modes grows with
time and does not come back to zero state, the system is said to be asymptotically
unstable. Asymptotic stability is also called internal stability or stability in the sense
of Liapunov.

4.2.2 Marginal (Neutral) Stability

When a linear time invariant system is disturbed, the equilibrium is shifted from
zero state to non-zero state. Some characteristic modes of the system neither decay
to zero state nor grow indefinitely but remains constant, while other characteristic
modes decay to zero state. These systems are said to be in borderline case. Such
systems are said to be marginally stable with neutral equilibrium.

4.2.3 Bounded Input Bounded Output (BIBO) (External)

Stability

The bounded input bounded output (BIBO) stability is defined with respect to the
total response which is the sum of the zero input and zero state responses.
504 4 Stability Analysis of Linear Control System

A linear time invariant system is said to be bounded input bounded output (BIBO)
stable if every bounded input yields a bounded output. BIBO stability is also called
as external stability.

4.2.4 Relative Stability

System stability is classified as absolute stability and relative stability. Absolute

stability refers to the condition whether the given system is stable or not. If the
system is stable, the closeness of stability is measured by relative stability. In the
frequency domain, as discussed in Chap. 3, the relative stability is measured by phase
margin and gain margin. In the time domain, the relative stability is measured by the
closeness of the real part of the characteristic roots to the jω axis. If the real part of
the characteristic root is closer to the jω axis, the relative stability is less and vise
versa. Consider the following T.F. of a certain system.

K (s + 1)
G(s) = (4.1)
(s + 5)(s + 2 + j3)(s + 2 − j3)

The poles and zero are located in Fig. 4.2. From Fig. 4.2, the relative stability unit of
the system is 2 units. This is the nearest distance of the pole of G(s) to the jω axis.
Zero location does not affect the stability or the relative stability of the system.

4.2.5 BIBO Stability via Impulse Response Function

Consider the system with T.F. G(s) represented as in Fig. 4.3. g(t) is the impulse
response function, while r (t) is the input and c(t) is the external output. The external
output c(t) is obtained using the following convolution integral

Fig. 4.2 Pole–zero locations j

of Eq. (4.1)
2 j3

2 units

5 1

2 j3
4.2 The Concept of Stability 505

Fig. 4.3 Block diagram

representation of a system R(s) C(s)
G(s)
with T.F.G(s)

c(t) = g(t) ∗ r (t)

 ∞
= g(τ )r (t − τ )dτ
0 ∞ 
 

|c(t)| =  g(τ )r (t − τ )dτ  (4.2)
0

Equation (4.2) can be rewritten using the following property

   ∞
 ∞ 
 g(τ )r (t − τ )dτ  ≤ |g(τ )||r (t − τ )|dτ (4.3)

0 0
 ∞
|c(t)| ≤ |g(τ )||r (t − τ )|dτ (4.4)
0

For the input r (t) being bounded, it can be expressed as |r (t)| < A. Now Eq. (4.4)
is written as
 ∞
|c(t)| ≤ A |g(τ )|dτ (4.5)
0

For the output c(t) to be bounded, it can be expressed as

c(t) ≤ B < ∞ (4.6)

Substituting Equation (4.6) in Eq. (4.5), we get

 ∞
B
|g(τ )dτ | ≤ <∞ (4.7)
0 A

Equation (4.7) gives the area of the impulse response curve. For the system to be
BIBO stable, the area of the impulse response curve should be finite in the time
interval 0 < t < ∞.
The above requirement is shown in Fig. 4.4a and b.

4.2.6 BIBO Stability and the Characteristic Roots

It is not always possible to test the system with impulse input and measure the
output with respect to time to asses the BIBO stability requirement. It can be easily
506 4 Stability Analysis of Linear Control System

(a) (b)

c(t) c(t) Infinite area

Finite area

t t
Stable Unstable

Fig. 4.4 Impulse response curve with a Finite area; and b Infinite area

determined by knowing the characteristic roots of the system. Consider the system
shown in Fig. 4.3. By definition of Laplace transform, G(s) can be written as
 ∞
G(s) = g(t)e−st dt (4.8)
0

Making use of the property stated in Eq. (4.5), we get

 ∞
|G(s)| ≤ |g(t)||e−st |dt (4.9)
 0
∞
≤ |g(t)||e−σ t |dt (4.10)
0

where σ is the real part of s = σ + jω and |e jωt | = 1. G(s) = ∞ at s-poles of G(s).

Thus, Eq. (4.10) is written as
 ∞
∞≤ |(g(t))||e−σ t |dt (4.11)
0

For one or more number of poles in RHP of the s-plane or in the jω axis, σ ≥ 0 and

|e−σ t | ≤ M or e−σ t = 1 (4.12)

Substituting Equation (4.12) in Eq. (4.11), we get

 ∞
|g(t)|Mdt ≥ ∞
0
 ∞
g(t)dt ≥ ∞ (4.13)
0
4.2 The Concept of Stability 507

(a) ( b)

r(t) = 1 C(s) C(s)

1 1
G(s) = −
s G(s) =−
r(t) = sin at (s2+a2)
1
R s =−
s

Fig. 4.5 System poles at a Origin and b jω-axis

Equation (4.13) implies that the system is BIBO unstable when atleast one pole lies
in RHP or on the jω axis. It is to be noted that the stability condition (4.13) does
not depend upon the input applied. A linear time invariant continuous time system
is said to be BIBO stable if all the roots are located in the left half s-plane. A system
is unstable if it is not BIBO stable.
Consider the system whose pole is at the origin of the s-plane. The input is the
unit step which is bounded. The output

1
c(t) = L −1 =t
s2
The output is unbounded and hence it is BIBO unstable.
Consider the system shown in Fig. 4.5b. The poles of G(s) are located at s = ± ja.
The input r (t) = sin at is bounded, whereas the output

1
c(t) = L −1 = t sin at
(s + a 2 )2

is unbounded.
The above two examples illustrate that for BIBO stability, there should not be
even a single root in RHP, and further the characteristic roots should not be at the
origin as well as in the jω axis. All the characteristic roots should be in LHP for the
system to be BIBO stable.

Note: When a pole at the origin is intentionally introduced the system is

said to be stable.

4.2.7 Relationship Between BIBO and Asymptotic Stability

The impulse response of a system is nothing but the linear combination of system
characteristic modes. If the characteristic root λ is in LHP, the corresponding mode
e−λt is absolutely integrable. On the other hand, if the characteristic root is either
in RHP or in the imaginary axis, the characteristic mode is not integrable. The
first statement implies that if the system is asymptotically stable, then it is BIBO
stable. According to the second statement, marginally stable system or asymptotically
508 4 Stability Analysis of Linear Control System

Fig. 4.6 Two systems r(t) c(t)

connected in cascade h1(t) h2(t)

S1 S2

unstable system is BIBO unstable. The converse is that BIBO stability does not always
imply asymptotic stability.
Consider two systems S1 and S2 with impulse response functions h 1 (t) = e−2t u(t)
and h 2 (t) = e−3t u(t) cascade. The characteristic roots are λ1 = −2
 ∞ connected in ∞
and λ2 = −3 and 0 e−2t dt and 0 e−3t dt are absolutely integrable. The system
is asymptotically stable. From Fig. 4.6, H1 (s) = 1/(s + 2) and H2 (s) = 1/(s + 3)
and
1
H (s) = H1 (s)H2 (s) =
(s + 2)(s + 3)

The overall impulse response of the system is

h(t) = (e−2t − e−3t )u(t)

The area of the impulse response curve is finite and the system is BIBO stable. The
system is asymptotically and BIBO stable.
Now consider Fig. 4.6 which has h 1 (t) = (δ(t) − 4e−2t ) and h 2 (t) = e2t . The
characteristic
 ∞ +2t root of system S2 is λ = 2 and the corresponding characteristic mode
0 e dt and is not integrable. Hence, system S2 is asymptotically unstable with
the characteristic root 2 in RHP. However,
 
4 (s − 2)
H1 (s) = 1 − =
s+2 s+2

and
1
H2 (s) =
s−2

The overall impulse response function is

H (s) = H1 (s)H2 (s)

(s − 2) 1
=
s+2 s−2
1
=
(s + 2)

The output of the system is c(t) = e−2t u(t) which is finite, and therefore, it is BIBO
stable. In the above example, BIBO stable system does not imply that it is asymp-
totically stable. This is due to pole–zero cancellation in the system. With pole–zero
cancellation, the system is not controllable or observable. Thus, if the system is both
4.2 The Concept of Stability 509

controllable and observable, BIBO stability for a linear time invariant continuous
system guarantees asymptotic stability.
Locations of the characteristic root and zero input response are shown in Fig. 4.7.
The characteristic roots location and system stability is shown in Fig. 4.8.
Figure 4.7a shows the characteristic root in LHP only. The system is asymptot-
ically stable. For such pole location, the impulse response will be exponentially
decaying and hence the system is BIBO stable.
Figure 4.7b shows the presence of a single root at the origin and no root in RHP.
The system is marginally stable. The impulse response curve is a step in nature and
the area under the curve is not finite as t tends to ∞. Hence, the system is BIBO
unstable.
Figure 4.7c shows complex conjugate roots in LHP and no root in RHP and the
jω axis. Hence, the system is asymptotically stable. The impulse response curve is
decaying with oscillations. However, the area under the curve is finite and hence the
system is BIBO stable.
Figure 4.7d shows the presence of complex conjugate roots in the jω axis. The
system is marginally stable. The impulse response curve has sustained oscillations
and hence the area under this curve is not finite. Hence, the system is BIBO unstable.
Figure 4.7e shows one characteristic root in RHP. The system is, therefore, asymp-
totically unstable. The corresponding impulse (zero input) response curve is expo-
nentially increasing, and therefore, BIBO unstable.
Figure 4.7f shows two characteristic roots placed at the origin of the s-plane.
Hence, the system is asymptotically unstable. The corresponding impulse response
curve is linearly increasing with time and hence it is unbounded. The system is BIBO
unstable.
Figure 4.7g shows two complex roots are placed in the RHP of the s-plane. Hence,
the system is asymptotically unstable. The impulse response curve is exponentially
increasing with oscillations. Hence, the system is BIBO unstable.
Figure 4.7h shows that there are two pairs of repeated roots in the jω axis.
Hence, the output is asymptotically unstable. The impulse response curve is lin-
early increasing with respect to time with unbounded oscillations. Hence, the system
is BIBO unstable.
In Fig. 4.8, the stable region, unstable region and the marginally stable region in
the s-plane are sketched and shown. Characteristic roots in LHP indicate asymptotic
stability, the roots in RHP region indicates instability and the presence of single root
in the jω axis indicates marginal stability. Presence of repeated roots in the jω axis
indicates instability.
510 4 Stability Analysis of Linear Control System

Characteristic Root Location Impulse Response Plot

(a) j

0 0 t

(b) j

0 0 t

(c) j

0 t
0

(d) j

0 t

Fig. 4.7 Locations of characteristic roots and corresponding zero input response
4.2 The Concept of Stability 511

Characteristic Root Location Impulse Response Plot

(e) j

0 0 t

(f) j

0 0 t

(g) j

0 t
0

(h) j

0 t
0

Fig. 4.7 (continued)

512 4 Stability Analysis of Linear Control System

Fig. 4.8 Characteristic root j

location in the complex
s-plane and system stability.

LHP Stable RHP Unstable

Region Region

LHP RHP

Unstable for
Marginally stable multiple roots
for simple root
j axis

To Summarize

1. A linear time invariant continuous time system is asymptotically stable if all the
characteristic roots are in LHP.
2. A linear time invariant continuous time system is unstable either at least one
characteristic root is in the RHP or there are repeated roots in the jω axis.
3. A linear time invariant continuous time system is marginally stable if there are
no roots in the RHP and there are some unrepeated roots in the jω axis.
4. An asymptotically stable system is also BIBO stable. An asymptotically unstable
system or a marginally stable system is BIBO unstable.
5. BIBO stable system need not be asymptotically stable. This may occur when there
is pole–zero cancellation in the system.
6. If the system is both controllable and observable, BIBO stability guarantees
asymptotic stability also.

Example 4.1
Investigate the asymptotic and BIBO stability of the following linear time invariant
continuous time system with the following characteristic polynomial:
4.2 The Concept of Stability 513

(1) F(λ) = (λ + 2)(λ + 2 + j3)(λ + 2 − j3)

(2) F(λ) = (λ − 2)(λ + 2 + j3)(λ + 2 − j3)
(3) F(λ) = (λ + 1)(λ2 + 9)
(4) F(λ) = (λ + 1)(λ2 + 9)2
(5) F(λ) = λ(λ + 3)
(6) F(λ) = λ2 (λ + 3)

Solution
(1)
F(λ) = (λ + 2)(λ + 2 + j3)(λ + 2 − j3)

The characteristic roots are

λ1 = −2
λ2 = −2 + j3
λ3 = −2 − j3

All the characteristic roots are in LHP and hence the system is asymptotically
stable.
The characteristic modes corresponding to the above characteristic roots are as
given below
 ∞
For λ1 , e−2t dt

0
∞
For λ2 , e−(2− j3)t dt

0
∞
For λ3 , e−(2+ j3)t dt
0

The above characteristic modes are absolutely integrable and hence the system
is BIBO stable.

Asymptotically and BIBO stable

(2)
F(λ) = (λ − 2)(λ + 2 + j3)(λ + 2 − j3)

The characteristic roots are

514 4 Stability Analysis of Linear Control System

λ1 = 2
λ2 = −2 − j3
λ3 = −2 + j3

Since λ1 = 2 is in RHP, the system is Asymptotically unstable.

Corresponding to λ1 = 2, the characteristic mode is
 ∞
e2t dt
0

and it is not integrable and the system is BIBO unstable.

Asymptotically and BIBO unstable

(3)
F(λ) = (λ + 1)(λ2 + 9)

The characteristic roots are

λ1 = −1
λ2 = + j3
λ3 = − j3

There are unrepeated roots in the jω axis. The system is marginally stable but
BIBO unstable.

Marginally stable and BIBO unstable

(4)
F(λ) = (λ + 1)(λ2 + 9)2

The characteristic roots are

λ1 = −1
λ2 = + j3
λ3 = + j3
λ4 = − j3
λ5 = − j3

The roots λ2 and λ3 and the roots λ4 and λ5 are repeated in the jω axis. Hence,
it is asymptotically and hence BIBO unstable.
4.2 The Concept of Stability 515

Asymptotically and BIBO unstable

(5)
F(λ) = λ(λ + 3)

The characteristic roots are

λ1 = 0
λ2 = −3

Since there is a single root in the jω axis, the system is marginally stable but
BIBO unstable.

Marginally stable and BIBO unstable

(6)
F(λ) = λ2 (λ + 3)

The characteristic roots are

λ1 = 0
λ2 = 0
λ3 = −3

Since there are repeated roots in the jω axis, the system is asymptotically unsta-
ble and hence BIBO unstable.

Asymptotically and BIBO unstable

Example 4.2
A system consists of two subsystems connected in cascade with impulse responses as
h 1 (t) and h 2 (t). Find whether the overall system is asymptotically and BIBO stable
(Fig. 4.9).

r(t) c(t)
h1(t) h2(t)

Fig. 4.9 Subsystems connected in cascade for Example 4.2

516 4 Stability Analysis of Linear Control System

1. h 1 (t) = (δ(t) − 2e−4t )u(t)

h 2 (t) = e4t u(t)
2. h 1 (t) = (δ(t) − 8e−4t )u(t)
h 2 (t) = e4t u(t)

Solution
1.

h 1 (t) = (δ(t) − 2e−4t )u(t)

 
2
H1 (s) = 1 −
(s + 4)
(s + 2)
=
s+4
h 2 (t) = e u(t)
4t

1
H2 (s) =
(s − 4)

The overall impulse response function is

H (s) = H1 (s)H2 (s)

(s + 2)
=
(s + 4)(s − 4)

The characteristic roots are

λ1 = −4
λ2 = 4

Since λ2 is in RHP, the system is asymptotically unstable. When it is asymptot-

ically unstable, it is also BIBO unstable.

Asymptotically and BIBO unstable

2.

h 1 (t) = (δ(t) − 8e−4t )u(t)

h 2 (t) = e4t u(t)
4.2 The Concept of Stability 517

8
H1 (s) = 1 −
(s + 4)
(s − 4)
=
(s + 4)
1
H2 (s) =
(s − 4)

The overall impulse response function is

H (s) = H1 (s)H2 (s)

(s − 4)
=
(s + 4)(s − 4)

The characteristic roots are

λ1 = −4 and λ2 = 4

Since λ2 is in RHP, the system is internally or asymptotically unstable.

If we measure the output c(t), it is due to h(t) = L −1 H (s)

1
h(t) = L −1
s+4
= e−4t u(t)

The overall impulse response function is integrable and finite. Hence, the system
is BIBO stable or externally stable. This is due to pole–zero cancellation.

Asymptotically or Internally unstable.

BIBO or Externally stable.

4.3 Routh–Hurwitz Criterion

Routh–Hurwitz criterion is a time domain approach to determine the number of

roots of the characteristic polynomial with constant real coefficients with respect to
the LHP, RHP and the jω axis of the s-plane. The method requires two steps. For
the given characteristic equation, using the coefficients, the Routh table is prepared.
From the Routh table, the information about how many roots of the characteristic
polynomial are in the LHP, RHP and the jω axis is interpreted. Even though this
information can be easily obtained using modern calculators and computers, and
also determine the exact location of the closed loop system poles, the power of this
method lies in design rather than analysis. The Routh–Hurwitz criterion is applied
to determine the following:
518 4 Stability Analysis of Linear Control System

1. Absolute stability of the system.

2. Relative stability of the system in the time domain.
3. The gain of the system can be increased before the closed loop system becomes
unstable.
4. To factorize the given polynomial in some special cases and locate the roots in
the s-plane.
5. To find the number of LHP, RHP and jω axis roots.
6. Stability of the system is determined without solving the characteristic equation.
It is a straightforward method, and therefore, invariably used in the design of
compensators.
Disadvantages of Routh’s Method

1. If the system is stable, Routh’s criterion does not tell how stable it is.
2. If the system is unstable, the method does not suggest how to improve the stability
of the system.
3. In the preparation of Routh’s array, when the complete row contains zero coef-
ficients, and there is no sign change in the first column, the hasty conclusion
that the system is stable is erroneous. In such cases, careful evaluation of roots
in the jω axis is very much essential. Presence of simple roots in the jω axis
indicates that the system is neither stable nor unstable. It is called marginally
stable. However, if repeated roots are identified in the jω axis, the system is no
more marginally stable but it is unstable. Routh–Hurwitz criteria will not reveal
this form of instability which is a major disadvantage of this method.
4. Routh–Hurwitz criterion is applicable for the characteristic polynomial whose
coefficients are real. For complex coefficient polynomial, it is not applicable.
Further, if the characteristic equation is not algebraic, such as containing expo-
nential functions or sinusoidal functions of s, the Routh–Hurwitz criterion cannot
be applied.
5. jω axis is considered as the stability boundary when the Routh–Hurwitz crite-
rion is applied. The method cannot be applied to any other stability boundaries
in a complex plane. Thus, for a discrete time system, where the unit circle is
the stability boundary, Routh–Hurwitz criterion cannot be applied. By linear
transformation, the unit circle has to be transformed to jω axis and then Routh
criterion is applied.
4.3 Routh–Hurwitz Criterion 519

A Historical Note: Edward John Routh (1831–1907)

Edward John Routh was born on 20 January, 1831 in Que-

bec, Canada. He was an English mathematician. Routh
came to England aged eleven and entered University Col-
lege, London, in 1847, and chose a career in mathematics.
He obtained his B.A.; and M.A.; degrees in mathemat-
ics in 1849 and 1853, respectively. On graduation, Routh
took up work as a private mathematics teacher in Cam-
bridge for students preparing for Tripos examination. He
published a textbook titled “Dynamics of a System of
Rigid Bodies”.
In addition to teaching and writing, he also contributed
original research work such as the Routh–Hurwitz the-
orem. Routh’s stability criterion, Routh’s array, Routhian, Routh’s theorem and
Routh’s Algorithm. J.J. Thomson and George Darwin are his notable students. Routh
married Hilda in 1864, and the couple had five sons and a daughter. One of his sons,
John Routh served in the Indian Civil Service. Routh died on 7 June 1907, at the age
of 76.
Adolf Hurwitz (1859–1919)

Adolf Hurwitz was a German mathematician. His father,

was in the manufacturing business, and his mother died
when he was just three years old. He entered the Univer-
sity of Munich in 1877 and then migrated to the Univer-
sity of Berlin. In 1881, he obtained his doctoral degree
from the University of Leipzig. In 1884, Hurwitz mar-
ried Ida Sumuel, the daughter of a professor of medicine.
They had three children. Hurwitz spent most of his time in
Zurich, Switzerland. He suffered from severe migraines
when he was a student in Munich. In 1905, one of his kid-
neys became diseased and the other one was removed. He
died on 18 November 1919, at the age of 60.

4.3.1 Routh–Hurwitz Criterion to

Determine Absolute Stability

Routh–Hurwitz stability criterion was developed independently by Routh in 1892

and Hurwitz in 1895. The combined theory suggests the analytical method of finding
the number roots of the characteristic polynomial in the LHP, RHP and the jω-axis
without actually solving for them.
Consider the system shown in Fig. 4.10. The closed loop T.F. of the above system
is given by
520 4 Stability Analysis of Linear Control System

C(s) G(s)
= (4.14)
R(s) 1 + G H (s)

For a strictly proper T.F., Eq. (4.14) in polynomial form is as given below

C(s) b0 s m + b1 s m−1 + · · · + bm−1 s + bm

= (4.15)
R(s) a0 s n + a1 s n−1 + · · · + an−1 s + an

where n ≥ m and a  s and b s are constants and real. The characteristic equation of
(4.15) is

a0 s n + a1 s n−1 + · · · + an−1 s + an = 0 (4.16)

The roots of the characteristic Eq. (4.16) should all be in the LHP for the system to
be stable. For this, the following conditions are necessary before proceeding further
to investigate the absolute stability using Routh–Hurwitz stability criterion.
1. The coefficient a0 , a1 , . . . , an−1 , an must be positive.
2. All the coefficients in the characteristic equation must be present. If any one
coefficient is zero, there will be a root or roots in the imaginary axis or in the
RHP, which means the system will be unstable.
Formation of Routh’s Array
If the above necessary conditions are satisfied, we can proceed further to establish
the sufficient condition required for the closed loop system to be stable. For this,
Routh’s array is prepared as explained below.
The coefficients of Eq. (4.16) are arranged in rows and columns with even powers
of s in one row and odd power of s in the next row. If the highest degree of the
polynomial is even, then the even row is started first followed by the odd row. If the
highest degree of the polynomial is odd, the odd row components are started first
followed by even power components. By this, the first two rows of the Routh array
are formed, from which subsequent rows are formed as explained below.

Fig. 4.10 Control system

with forward and feedback R(s) C(s)
transfer functions G(s)

H(s)
4.3 Routh–Hurwitz Criterion 521

sn a0 a2 a4 ···
s n−1
a1 a3 a5 ···
s n−2
b1 b2 b3 ···
s n−3
c1 c2 c3 ···
.. .. ..
. . .
s2 d1 d2
1
s e1
0
s f1
For an nth degree characteristic polynomial, there will be (n + 1) Routh’s rows.
The first two rows are prepared from the coefficients of the given characteristic
equation. The coefficients of the remaining (n − 1) rows are evaluated as follows:

a1 a2 − a0 a3
b1 =
a1
a1 a4 − a0 a5
b2 =
a1
a1 a6 − a0 a7
b3 =
a1
..
.

b1 a3 − a1 b2
c1 =
b1
b1 a5 − a1 b3
c2 =
b1
b1 a7 − a1 b4
c3 = (4.17)
b1
..
.

The process is continued until the s 0 th row is completed. The complete array of
coefficients looks like a triangle. After preparing the Routh’s array following the
procedure described above, the Routh’s stability criterion is applied to Eq. (4.16).
The Routh–Hurwitz criterion states that the number of roots of the characteristic
equation with positive real parts is equal to the number of changes in sign of the first
column of the Routh array.
In assessing the sign changes in the first column array, suitable modification in
the calculation procedure should be followed, especially in the following cases:
1. Zero only in the first column.
2. Zeros in one complete row.
522 4 Stability Analysis of Linear Control System

3. Zeros in one complete row with repeated roots in the jω axis.

If case 3 is not properly evaluated, false indication of stability condition will be
arrived at.

Note: For simplifying the numerical calculation, the entire row may be
multiplied or divided by a positive number by which the stability condition
is unaffected.

Example 4.3
For the following characteristic polynomial, determine whether the closed loop sys-
tem is stable. Use Routh–Hurwitz stability criterion

(1) F(s) = s 5 + 4s 4 − 8s 3 + 11s 2 + 6s + 4

(2) F(s) = s 6 + 7s 5 + 4s 3 + 2s 2 + 10s + 6

Solution
(1)

F(s) = s 5 + 4s 4 − 8s 3 + 11s 2 + 6s + 4

In the above characteristic polynomial, −8 is one coefficient with negative sign.

This indicates that there is at least one root of the characteristic equation in RHP
and hence the system is unstable. There is no need to prepare Routh’s array.
(2)

F(s) = s 6 + 7s 5 + 4s 3 + 2s 2 + 10s + 6

In the above characteristic polynomial, the coefficient of s 4 is zero. Hence, all

the roots will not be in LHP. Some roots may be in the jω axis or in RHP. Hence,
the system is not stable.
In the above two cases, the necessary condition for Routh–Hurwitz stability
criterion is not satisfied and the system is unstable. There is no need to prepare
the Routh’s array.
4.3 Routh–Hurwitz Criterion 523

Example 4.4
Determine the stability of the system determined by

F(s) = s 4 + 6s 3 + 23s 2 + 40s + 50

Solution

F(s) = s 4 + 6s 3 + 23s 2 + 40s + 50

s4 1 23 50

s3 6 40

s2 16.33 50

s1 21.63

s0 50
Since there is no sign change in the first column of Routh’s array, no root is in
RHP. Hence, the system is stable.

Example 4.5
Find the number of roots in this equation

s 6 + 2s 5 + 8s 4 + 8s 3 + 20s + 10s + 16 = 0

with positive real part, zero real part and negative real part.

Solution

F(s) = s 6 + 2s 5 + 8s 4 + 8s 3 + 20s 2 + 10s + 16

The following Routh’s array is formed.

524 4 Stability Analysis of Linear Control System

s6 1 8 20 16
5
s 2 8 10
4
s 4 15 16
3
s 0.5 2
s2 −1 16
s1 10
0
s 16

There are two sign changes from s 3 th row to s 2 th row and from s 2 th row to s 1 th
row as shown in the Routh’s table. The polynomial has 2 RHP roots.
To find the LHP roots the polynomial F(−s) is obtained and the Routh’s array is
formed. Here the number of sign changes will give the number of LHP roots

F(−s) = s 6 − 2s 5 + 8s 4 − 8s 3 + 20s 2 − 10s + 16

s6 1 8 20 16
s 5
−2 −8 −10
4
s 4 15 16
s 3
−0.5 −2
s 2
−1 16
s 1
−10
s0 16
As seen from Routh’s array, there are four sign changes in the first column. Hence,
there are 4 LHP roots. There is no root in the jω axis.

RHP root 1
LHP root 4
jω axis root Nil

4.3.2 Routh–Hurwitz Criterion: Special Cases

When the Routh array is prepared, sometimes we will have the first column of any
row becoming zero or the entire row becoming zero. When these cases occur, we say
the Routh array is broken and it cannot be completed. These are called special cases
and they are discussed ahead.
4.3 Routh–Hurwitz Criterion 525

First Column Becoming Zero

When the first column becomes zero, division by the zero becomes infinity and the
Routh’s array cannot be completed. This phenomenon can be avoided by following
any one of the methods described below:
1. By forming the reciprocal of the polynomial.
2. A small value epsilon ε is assigned to replace the zero.
3. The given polynomial is multiplied by a factor (s + a) where a > 0 and the
Routh’s array is formed for the new polynomial.
Method one is computationally easier than the epsilon method. However, when the
reciprocal of a polynomial is formed, for the new polynomial also the first column
becoming zero may repeat and the method fails. The third method increases the
degree of the polynomial by one and an additional row is required in the Routh’s
array. The special case is illustrated in the following examples.

Example 4.6
Examine the stability by Routh’s criterion for

s 5 + s 4 + 2s 3 + 2s 2 + 3s + 15 = 0

(Anna University, May, 2009)

Solution

F(s) = s 5 + s 4 + 2s 3 + 2s 2 + 3s + 15

The following Routh’s array is formed.

s5 1 2 3
4
s 1 2 15
s 3
0 −12

The Routh array is broken with the first column of s 3 th row becoming zero. This
is solved using the method given below.
Method 1: By Forming Reciprocal of the Polynomial
Substitute s = 1/x and form F(x)

1 1 2 2 3
F(x) = 5
+ 4 + 3 + 2 + + 15 = 0
x x x x x
15x 5 + 3x 4 + 2x 3 + 2x 2 + x + 1 = 0

The Routh array is formed for the above polynomial as given below.
526 4 Stability Analysis of Linear Control System

x5 15 2 1
4
x 3 2 1
x3 −2 −1 (After dividing by 4)
2
x 0.5 1
x1 3
0
x 1
There are two sign changes in the first column of Routh’s array. The system is
unstable.

Method 2: Epsilon Method

The first column of s 3 th row becomes zero. Zero is replaced by a small value ε+ and
the Routh’s array is completed. After completion, ε is put to zero and the number of
sign changes in the first column is counted.
With ε With ε = 0
s5 1 2 3 1 2 3
s4 1 2 15 1 2 15
s3 ε+ −12 0+ −12
2ε + 12
s2 15 +∞ 15
ε
−15ε2 − 24ε − 144
s1 −12
2ε + 12
s0 15 15
There are two sign changes in the first column of Routh’s array and hence there
are 2 RHP roots. The system is unstable.

Method 3: Multiplying by a Factor (s + 1)

The given polynomial is multiplied by any factor and the new polynomial is obtained.
For the new polynomial, Routh’s array is formed. Arbitrarily, we choose the factor
(s + 1). The new polynomial is

F(s) = (s + 1)(s 5 + s 4 + 2s 3 + 2s 2 + 3s + 15)

= s 6 + 2s 5 + 3s 4 + 4s 3 + 5s 2 + 18s + 15

For the above polynomial, the following Routh’s array is formed.

4.3 Routh–Hurwitz Criterion 527

s6 1 3 5 15
5
s 2 4 18
s 4
1 −4 15
s 3
12 −12
or 1 −1 After dividing by 12
s 2
−3 15
1
s 4
0
s 15
There are two sign changes in the first column of the Routh’s array. There are two
RHP roots. The system is unstable.

Example 4.7
Consider the following characteristic polynomial

F(s) = 2s 4 + s 3 + 2s 2 + s + 2

Find whether the system is stable.

Solution

2s 4 + s 3 + 2s 2 + s + 2 = 0

The following Routh’s array is formed

s4 2 2 2
3
s 1 1
2
s 0 2
The Routh array is broken. We substitute s = 1/x and get the following charac-
teristic polynomial is x

2x 4 + x 3 + 2x 2 + x + 2 = 0

Both original polynomial and the polynomial of the reciprocal are the same; the
method fails. We follow the method by multiplying by the factor (s + 2)

F(s) = (s + 2)(2s 4 + s 3 + 2s 2 + s + 2)
= 2s 5 + 5s 4 + 4s 3 + 5s 2 + 4s + 4

The Routh’s array is formed below.

528 4 Stability Analysis of Linear Control System

s5 2 4 4
4
s 5 5 4
3
s 2 2.4
s 2
−1 4
1
s 10.4
1
s 4
There are two sign changes and there are two roots in RHP. The system is unstable.
By Epsilon Method
With ε With ε = 0
4
s 1 1 1 (After dividing by 2) 1 1 1
3
s 1 1 1 1
+
s 2
ε 1 0 1
ε−1
s 1
ε
−∞
0
s 1 1
There are 2 sign changes and there are 2RHP roots. The system is unstable.
Routh Array with Entire Row Being Zero
The second special case of Routh array is when the entire row becomes zero which
is due to even polynomial which is a factor of the original polynomial. This should
be handled in a different way from the case of a zero only in the first column of a
row. Let us consider a sixth degree polynomial to demonstrate how to construct and
interpret the Routh array when an entire row of zero is present

a0 s 6 + a1 s 5 + a2 s 4 + a3 s 3 + a4 s 2 + a5 s + a6 = 0 (4.18)

The following Routh array is formed. We assume that s 3 th row has zeros in all the
columns.
s6 a0 a2 a4 a6
s 5 a1 a3 a5
... ................................................
s4 b1 b2 b3
s3 0 0 A(s) = b1 s 4 + b2 s 2 + b3
or 4b1 2b2 d A(s)/ds = 4b1 s 3 + 2b2 s
s2 c1 c2
s1 d1
0
s e1
To complete the Routh’s array, we use the following procedure. Draw a horizontal
dotted line just above the row which has zeros in the entire row. The equation of the
4.3 Routh–Hurwitz Criterion 529

row which is just below the dotted line is called auxiliary equation. For the example
cited above, the auxiliary equation is

A(s) = b1 s 4 + b2 s 2 + b3 (4.19)

The auxiliary equation is differentiated with respect to s and the following equation
is obtained
d A(s)
= 4b1 s 3 + 2b2 s (4.20)
ds

The elements of the s 3 th row are taken from Eq. (4.20) and the Routh array is
completed as shown. Now the interpretation of Routh table is very important. An
entire row of zeros in the Routh array is possible when purely even or purely odd
polynomial is a factor of the original polynomial.
Equation (4.19) is a factor of the original polynomial. This is an even polynomial.
If this polynomial divides the original polynomial, the reminder is also a factor. The
even polynomial of Eq. (4.19) has the following properties:
1. The roots of the even polynomial are symmetric about the origin and real.
2. Some of the roots of the even polynomial could be on the jω-axis as simple
roots are repeated roots.
3. The roots are quadrantal.
The root locations are shown in Fig. 4.11.
(a) Real and symmetrical about the origin
(a) Simple jω axis roots symmetrical about the origin
(c) Repeated jω axis roots symmetrical about the origin
(d) Quadrantal and symmetrical roots about the origin.
Routh Array Interpretation
1. Routh array from the row of the even polynomial (in the above example s 4 th
row) down to the s 0 th row is completed and examined.
2. If there are any sign changes, from s 4 th row to s 0 th row the number of sign
changes give the RHP roots of the even polynomial. By symmetry, the same
number of roots are in LHP. The remaining roots are in the jω axis symmetrically
placed with respect to the origin. If there is no sign change, then RHP as well as
LHP roots are zero for the even polynomial.
3. From step two, the number of roots in the jω axis roots are known. These roots
may be simple (Fig. 4.11b) or repeated (Fig. 4.11c). For simple roots, the system
is marginally stable. For repeated roots, the system is unstable.
4. For repeated roots in the jω axis, the auxiliary Eq. (4.19) is of the form A(s) =
(s 2 + a 2 ).
5. Routh array is now interpreted above the horizontal dotted line to evaluate the
remaining roots. The number of sign changes give the number of RHP roots.
(The sign change is evaluated from s 6 th row to s 5 th row).
530 4 Stability Analysis of Linear Control System

(a) (b)
j j

0 0

(c) (d)
j j

0
0

Fig. 4.11 Root locations of an even polynomial

6. The sum of the roots found in steps 2 and 5 together give the LHP, RHP and jω
axis roots. The following examples illustrate the above procedure.

Example 4.8
For the characteristic polynomial given below find positive real, negative real and
imaginary roots. Is the system stable? Factorize the polynomial and locate the roots
in the s-plane

s 6 + 2s 5 + 8s 4 + 12s 3 + 20s 2 + 16s + 16 = 0

(Anna University, April, 2010)

Solution

F(s) = s 6 + 2s 5 + 8s 4 + 12s 3 + 20s 2 + 16s + 16

The Routh’s array is formed as given below

4.3 Routh–Hurwitz Criterion 531

Other polynomial

s6 1 8 20 16
s5 2 12 16

s4 2 12 16 Auxiliary polynomial

s3 0 0
or 1 3 Even polynomial

s2 6 16

s1 1/3 A(s)=2s2+12s2+16
dA(s)
s0 16 =8s3+24s
ds
=(s3+3) (After dividing by 8)
Routh array interpretation
1. The Routh array is first inspected from s 4 th row down to s 0 th row. It is seen that
there is no sign change in the first column. This implies that there is no RHP
root of the even polynomial.
2. If there is no RHP root for the even polynomial, by symmetry (see Fig. 4.11d),
there should not be any LHP root also for the even polynomial.
3. All the 4 roots of the even polynomial are in the jω axis.
4. Now we inspect the other polynomial. This starts from s 6 th row down to s 5 th
row. Since there is no sign change, there is no RHP root. There are 2 LHP roots.
5. In step 3, it is evaluated that there are 4 roots in the jω-axis. In step 4, it is known
that there is no RHP root and two roots are in LHP.
6. We divide the characteristic polynomial by the auxiliary polynomial as given
below.
s 2 + 2s + 2

s + 6s + 8 s 6 + 2s 5 + 8s 4 + 12s 3 + 20s 2 + 16s + 16
4 2

s6 + 6s 4 + 8s 2
2s 5 + 2s 4 + 12s 3 + 12s 2 + 16s + 16
2s 5 + 12s 3 + 16s
2s 4
+ 12s 2 + 16
2s 4 + 12s 2 + 16

s 6 + 2s 5 + 8s 4 + 12s 3 + 20s 2 + 16s + 16 = (s 4 + 6s 2 + 8)(s 2 + 2s + 2)

s 2 + 2s + 2 = (s + 1 + j)(s + 1 − j)

(s 4 + 6s 2 + 8) = (s 2 + 4)(s 2 + 2)
√ √
= (s + j2)(s − j2)(s + j 2)(s − j 2).
532 4 Stability Analysis of Linear Control System

Fig. 4.12 Locations of poles j

of F(s) LHP RHP
j2
1 j j√2

1 j j√2
j2

The poles of F(s) are located as shown in Fig. 4.12.

From Fig. 4.12, the following conclusions are drawn.
1. There is no pole in RHP.
2. There are 4 poles in the jω axis. They are distinct and not repeated. This implies
the system is marginally stable.
RHP roots = 0
LHP roots = 2
jω-axis roots = 4
System is marginally stable

It is erroneous to come to the conclusion that the system is stable since there is no
sign change in the first column of Routh’s array.

Example 4.9
Consider the following characteristic polynomial

s 7 + 9s 6 + 24s 5 + 24s 4 + 24s 3 + 24s 2 + 23s + 15 = 0

By using Routh–Hurwitz stability criterion, determine the stability of the system.

Also, determine the number of roots in RHP, LHP and jω axis.

(Anna University, December, 2011)

Solution

F(s) = s 7 + 9s 6 + 24s 5 + 24s 4 + 24s 3 + 24s 2 + 23s + 15

The following Routh’s array is formed.

4.3 Routh–Hurwitz Criterion 533

s7 1 24 24 23
s6 9 24 24 15
Other polynomial
192 192 192
s5 9 9 9
or 1 1 1

s4 1 1 1

s3 0 0
or 2 1 Even polynomial

s2 1 2

s1 3 A(s)=s4+s2+1
dA(s)
s0 2 =2(2s3+s)
ds

1. The even polynomial starts from s 4 th row and comes down to s 0 th row. The
number of sign changes are two. Hence, there are 2 RHP roots. By symmetry,
there will be 2 LHP roots for the even polynomial. There is no jω axis roots.
2. By inspection of other polynomial which starts from s 7 th row down to s 5 th row,
there is no sign change. All the three roots are in LHP.
3. The overall root positions are

RHP = 2
LHP = 5
jω-axis = Nil
System stability = unstable

Example 4.10
Find the number of roots in the equation

s 6 + 4s 5 + 3s 4 − 16s 2 − 64s − 48 = 0

with positive, negative and zero real parts. Determine the stability of the system.

(Anna University, December, 2011)

Solution The characteristic polynomial is

F(s) = s 6 + 4s 5 + 3s 4 − 16s 2 − 64s − 48

534 4 Stability Analysis of Linear Control System

By inspection of F(s), it is found that the system is unstable because of negative

sign in the coefficient −16, −64 and −48. The Routh’s array is formed below.
s6 1 3 16 48 Other polynomial

s5 1 0 16 (After dividing by 4)

s4 1 0 16 (After dividing by 3)
s3 0 0 Even polynomial
or
s3 4 0

s2 16
A(s) s4 16
s1 64
dA(s)
4s3
ds
s0 16

By inspection of even polynomial of Routh’s array, it is found that there is one

sign change. Hence, there is 1 RHP root and 1 LHP root and 2 jω axis roots. In the
other polynomial, there is no sign change in the first column. Both the roots are in
LHP. The roots are distributed as

RHP = 1
LHP = 3
jω-axis = 2

The system is unstable.

Example 4.11
Consider the following characteristic polynomial

F(s) = s 6 + 4s 5 + 11s 4 + 32s 3 + 40s 2 + 64s + 48

Using Routh–Hurwitz stability criterion, determine the stability of the system.

Solution

F(s) = s 6 + 4s 5 + 11s 4 + 32s 3 + 40s 2 + 64s + 48

The following Routh’s array is prepared.

Inspection of first column of Routh’s array for even polynomial shows that there
are no sign changes. This shows that there is no RHP and LHP roots for the even
polynomial. All the 4 roots are in the jω-axis. Inspection of first column of other
polynomial shows that there is no sign change. This shows that the remaining 2 roots
are in LHP. To determine the stability of the system, it is necessary to investigate
the locations of the jω-axis poles. For the even polynomial, auxiliary equation is a
factor
4.3 Routh–Hurwitz Criterion 535

A(s) = (s 4 + 8s 2 + 16) = (s 2 + 4)2

s6 1 11 40 48 Other polynomial

s5 4 32 64

s4 1 8 16 (After dividing by 3)

s3 0 0 Even polynomial

or 1 4
A(s) s4 8s2 16
s2 4 16 dA(s)
4(s3 4s)
ds
s1 0
A(s) 4s2 16
or 8 dA(s)
8s
ds
s0 16

For the above equation, the 4 poles are in the jω-axis (s 2 + 4)2 = 0

s1 = + j2 s3 = − j2
s2 = + j2 s4 = − j2
LHP roots = 2; RHP roots = 0; jω − axis = 4

The poles at s = + j2 and s = − j2 are repeated. The impulse response for these
poles will be t sin 2t which is unbounded. The system is unstable.
It is erroneous to conclude that the system is stable since there is no sign change
in the first column of the overall Routh array.

System is unstable

4.3.3 Parameters Estimation via Routh–Hurwitz Stability

Criterion

Any one parameter which is varied to obtain desired system performance can be
estimated by applying Routh–Hurwitz stability criterion. This is illustrated in the
following example.
536 4 Stability Analysis of Linear Control System

Example 4.12
Apply Routh–Hurwitz criterion to determine the value of K for the stability of a
control system whose open loop T.F. is given by

5
G H (s) =
s(s 2 + 2K s + 4 + K )

(Anna University, December, 2010)

Solution
5
G H (s) =
s(s 2 + 2K s + 4 + K )

The characteristic equation is

F(s) = 1 + G H (s) = 0
= s 3 + 2K s 2 + (4 + K )s + 5

The following Routh’s array is formed.

s3 1 (K + 4)
2
s 2K 5
2K + 8K − 5
2
s1
2K
s0 5
For the system to be stable, there should not be any sign change in the first column
of Routh’s array. Thus, we get the following conditions:
1. K >0
2. 2K 2 + 8K − 5 > 0
3. (K + 4.55)(K − 0.55) > 0
4. K > −4.55
5. K > 0.55
If the condition K > 0.55 is satisfied, K > −4.55 and K > 0 conditions are satisfied.
Hence, for the system to be stable.

K > 0.55

Example 4.13
The open loop T.F. of a unity feedback control system is given by
4.3 Routh–Hurwitz Criterion 537

K
G(s) =
(s + 2)(s + 4)(s 2 + 6s + 25)

By applying the Routh criterion, discuss the stability of the closed loop system as a
function of K . Determine the values of K which will cause sustained oscillations in
the closed loop system. What are the corresponding oscillation frequencies?

(Anna University, May, 2009)

Solution
K
G(s) =
(s + 2)(s + 4)(s 2 + 6s + 25)

The characteristic equation is

1 + G(s) = 0
(s + 2)(s + 4)(s + 6s + 25) + K = 0
2

F(s) = s 4 + 12s 3 + 69s 2 + 198s + (200 + K )

The following Routh’s array is formed.

s4 1 69 (200 + K )
3
s 12 198
s 2
52.5 (200 + K )
s 1
7995 − 12K
s 0
200 + K
For the system to be stable, the following conditions are to be satisfied:

(1) 7995 − 12K > 0

K < 666.25
(2) 200 + K > 0
K > −200

The system is stable for the value of K in the following range

−200 < K < 666.25

At K = 666.25 the system oscillates. The oscillating frequency is obtained from the
auxiliary equation which is given by

52.5s 2 + 200 + K = 0

Substituting s = jω in the above equation, we get

538 4 Stability Analysis of Linear Control System

Fig. 4.13 System R(s) + K(s+1)

representation for Example G(s) = 3 2 C(s)
− (s +as +2s+1)
4.14.

−52.5ω2 + 200 + 666.25 = 0

ω = 4.06 rad/sec

Example 4.14
A certain unity feedback system is represented in Fig. 4.13. Find the +ve value of K
and a so that the system oscillates at a frequency of 2 rad/s.

Solution
K (s + 1)
G(s) =
s 3 + as 2 + 2s + 1

The characteristic equation is

1 + G(s) = 0
s 3 + as 2 + (K + 2)s + (K + 1) = 0

The following Routh’s array is formed.

s3 1 (K + 2)
s 2
a (K + 1)
s 1
a(K + 2) − (K + 1)
s 0
(K + 1)
For the system to oscillate

a(K + 2) = (K + 1)
(K + 1)
a=
(K + 2)

Under this condition, the auxiliary equation is

as 2 + (K + 1) = 0
4.3 Routh–Hurwitz Criterion 539

Substituting s = jω in the above equation, we get

(K + 1)
ω2 =
a
(K + 1)(K + 2)
=
(K + 1)
= K +2

For ω = 2 rad/sec
K =2
a = 0.75

Example 4.15
Consider the following characteristic polynomial of a certain control system

F(s) = s 4 + 2s 3 + 6s 2 + K s + 10

Determine the real value of K so that the system is stable.

Solution

F(s) = s 4 + 2s 3 + 6s 2 + K s + 10

The following Routh’s array is formed.

s4 1 6 10
3
s 2 K
s 2
6 − 0.5K 10
K (6 − 0.5K ) − 20
s1
6 − 0.5K
s0 10
For the system to be stable, the following conditions are to be satisfied

1. 6 − 0.5K > 0
K < 12
2. K (6 − 0.5K ) − 20 > 0
−0.5K 2 + 6K − 20 > 0

Solving for the quadratic equation for K , we get

540 4 Stability Analysis of Linear Control System
√
−6 ± 36 − 40
K =
(−1)
= 6 ± j2

Since K is complex the inequalities cannot be satisfied for any real value of K . For
any value K < 12 in condition (1) would not satisfy condition (2). Thus, the system
has RHP roots for all values of K and it is always unstable.

Example 4.16
The characteristic equation of a feedback control system is

s 4 + 3s 3 + 12s 2 + 16s + K = 0

Determine the range of gain K so that the system is stable.

Solution

s 4 + 3s 3 + 12s 2 + 16s + K = 0

For the above characteristic equation, the following Routh’s array is formed.
s4 1 12 K
3
s 3 16
2
s 20 3K
s 1
320 − 9K
0
s K
From the Routh’s array, the following conditions are to be satisfied for the system
to be stable

(1) 320 − 9K > 0

320
K <
9
K < 35.556
(2) K >0

The range of K for the system to be stable is

0 < K < 35.556

4.3 Routh–Hurwitz Criterion 541

Example 4.17
The open loop T.F. of a certain unity feedback system is given by

K (s + 10)
G(s) =
s3
Determine the value of K so that the closed loop system is stable. If the closed loop
system is unstable for all values of K , determine the RHP and LHP roots.

Solution
K (s + 10)
G(s) =
s3
The characteristic equation is given by

1 + G(s) = 0
s + K s + 10K = 0
3

Since the coefficient of s 2 is missing in the Routh’s array, there will be a zero in the
first column. The characteristic polynomial is multiplied by the factor (s + 1) and
the new characteristic equation is obtained as

(s + 1)(s 3 + K s + 10K ) = 0
s + s 3 + K s 2 + 11K s + 10K = 0
4

The following Routh’s array is formed.

s4 1 K 10K
3
s 1 11K
s 2
−K K
s 1
11K + 1
0
s K
For all positive values of K , the system is unstable. There are two sign changes
in the Routh’s array. There are 2 RHP roots and one LHP root for the closed loop
system.
542 4 Stability Analysis of Linear Control System

4.3.4 Relative Stability Using Routh–Hurwitz Stability

Criterion

In most of the applications, Routh–Hurwitz stability criterion is applied to know the

immediate answer whether the system is stable or not. This information is useful
for the system design. Once the system is designed with absolute stability, it can be
tested how far it is closer to absolute stability which is termed as relative stability. To
find the relative stability, the s-plane axis is shifted to the desired value and Routh–
Hurwitz criterion in the new complex s-plane is applied. If the jω axis of the s-plane
is to be shifted to the left by σ where σ is the desired relative stability factor, then
we substitute s = (λ − σ ) in the characteristic equation and form the characteristic
polynomial F(λ). The Routh–Hurwitz test is carried out for F(λ). The number of
changes of sign in the first column of the new Routh’s array give the number of
roots located to the right of the vertical line passing through s = −σ . The following
example illustrates the above procedure.

Example 4.18
Consider the following characteristic polynomial

F(s) = s 3 + 12s 2 + 47s + 60

Test whether the system has a relative stability of 2 units.

Solution

F(s) = s 3 + 12s 2 + 47s + 60

The desired relative stability unit is σ = −2. Substitute s = (λ − 2) in the original

equation.

s 3 = (λ − 2)3
= λ3 − 6λ2 + 12λ − 8
12s 2 = 12(λ − 2)2
= 12λ2 − 48λ + 48
47s = 47(λ − 2)
= 47λ − 94
F(λ) = λ3 + 6λ2 + 11λ + 6

The following Routh’s array is formed.

4.3 Routh–Hurwitz Criterion 543

λ3 1 11
λ 2
6 6
λ 1
10
λ 0
6
There is no sign change in the first column of Routh’s array. This means all the
roots of F(s) are to the left of the vertical line passing through (−2, 0) point. The
system has a relative stability unit of more than 2.

4.3.5 Routh’s Test for System with Transportation Lag

In the process control system, the effect of transportation lag is important. The trans-
fer function of the transportation lag is given as e−sT . The effect of transportation lag
cannot be studied by Routh’s criterion. e−sT is expressed in the approximated alge-
braic form. There are few methods available in the literature. We use the following
approximation here

e−sT = (1 − sT )

Example 4.19
A certain unity feedback control system has the following open loop T.F.

10e−sT
G(s) =
(s + 1)(s 2 + 2s + 4)

Determine the value of T , so that the closed loop system is stable.

Solution
10e−sT
G(s) =
(s + 1)(s 2 + 2s + 4)

The characteristic equation is

(s + 1)(s 2 + 2s + 4) + 10e−sT = 0
s 3 + 3s 2 + 6s + 4 + 10(1 − sT ) = 0
s 3 + 3s 2 + (6 − 10T )s + 14 = 0

The following Routh’s array is formed.

544 4 Stability Analysis of Linear Control System

s3 1 6 − 10T
2
s 3 14
s 3
4 − 30T
0
s 1
For the system to be stable 4 − 30T > 0 or T < 0.1333

T < 0.1333

4.4 Nyquist Stability Criterion

Nyquist stability criterion was developed by H. Nyquist, in 1932, at Bell Telephone

Laboratories, USA. It determines the closed loop system stability if the open loop
frequency response and open loop poles are known H.S. Black, at Bell Laboratory
was developing the negative feedback amplifier for use in telephone repeater circuits
using vacuum tubes. When the amplifier became an oscillator, the reason was not
understood by any one at that time. Nyquist was asked to study the problem, and as
a result, he developed the famous stability test. One morning, going to work on the
State Island ferry, before the Verrazano Narrows bridge, Nyquist had an inspiration
and wrote his work, literally, on the back of an envelope as he rode. Today millions
of control system engineers and students apply the Nyquist stability criterion for the
system design. There is hardly any control system textbook that does not include
the Nyquist stability criterion for the study. This semi-graphical method which is an
alternative to Routh–Hurwitz and root locus methods has the following features.
Advantages of Nyquist Stability Criterion
1. Nyquist stability criterion, like Routh–Hurwitz criterion, gives information about
the absolute stability of the closed loop system. In addition, it gives information
about the relative stability of the system in terms of phase margin and gain
margin. Further, it indicates how to improve the relative stability if required.
Furthermore, if the closed loop system is unstable, it indicates how to stabilize
the system.
2. Since Nyquist stability criterion is developed in the frequency domain, it is easy
to get information about the frequency domain specifications such as resonant
peak Mr , resonant frequency ωr and BW of the closed loop system.
3. Systems with time delay are difficult to handle with accuracy by Routh–Hurwitz
criterion and root locus technique. However, Nyquist criterion can be easily
extended to linear system with pure time delay.
4. The Nyquist plot is based on the open loop frequency response of G H ( jω) and
the location of open loop poles of G H (s). These two things are very easy to
obtain. G H ( jω) is easily plotted using the MATLAB program nowadays.
5. Even though Nyquist stability criterion is developed from Cauchy’s principle of
the argument, which requires the knowledge of the theory of complex variable,
4.4 Nyquist Stability Criterion 545

interpretation of stability condition from the Nyquist plot is very simple and
straightforward.
A Historical Note: Harry Nyquist (1889–1976)

Harry Theodor Nyquist was born on February 7, 1889

in Sweden. His parents had seven children. The family
of Nyquist emigrated to the USA in 1907. He received
the B.S. and M.S. degrees in electrical engineering,
respectively, from North Dakota University and a Ph.D.
degree in Physics in 1917 from Yale University. He
worked at AT&T’s Department of Development and
Research from 1917 to 1934 and he continued his ser-
vice there when it changed its name as Bell Telephone
Laboratories in 1934. He retired in the year 1954 on
superannuation. Nyquist early theoretical work was
on determining the BW requirement for transmitting
information. In 1927, Nyquist formulated sampling theorem according to which the
number of independent pulses that can be sent through a telegraph channel is limited
to twice the bandwidth of the channel.
Nyquist contribution as an engineer at Bell Laboratory was immense. His impor-
tant works include the study of thermal noise, the stability of feedback amplifiers,
telegraphy, facsimile, television and other communication problems. In 1932, he pub-
lished a classical research paper on the stability of feedback amplifiers. The Nyquist
stability criterion can now be found in all textbooks on feedback control theory.
Nyquist was honoured for his “Fundamental Contributions to a Quantitative Under-
standing of Thermal Noise, Data Transmission and Negative Feedback”, “theoretical
analysis and practical inventions in the field of communications systems”, “theories
of telegraph transmission”, “thermal noise in electric conductors”, and many funda-
mental contributions in Engineering. Nyquist lived in Texas state after his retirement
and the great engineer died in Harbinger on April 4, 1976 at the age of 87.

4.4.1 Concepts Used in Nyquist Stability Criterion

Consider the feedback system shown in Fig. 4.14. The closed loop system T.F. is
written as
C(s)
= M(s)
R(s)
G(s)
= (4.21)
[1 + G H (s)]

where 1 + G H (s) = 0 is the characteristic equation. The characteristic equation

1 + G H (s) = 0 in Eq. (4.21) determines the stability of the closed loop system.
546 4 Stability Analysis of Linear Control System

Fig. 4.14 Closed loop

system R(s) C(s)
G(s)

H(s)

Nyquist stability criterion tells whether any roots of the characteristic equation are
in RHP. Before the derivation of Nyquist stability criterion, the following concepts
which are associated with the criterion need to be explained. These concepts include
the following:
1. Relationship between the poles of G H (s) and the poles of 1 + G H (s).
2. The relationship between the poles of the closed loop T.F. M(s) and the zeros
of 1 + G H (s).
3. The concept of mapping points.
4. The concept of mapping contour.

4.4.1.1 Relationship Between the Poles of G H(s) and the Poles of

1 + G H(s)

Consider the following loop T.F.

K (s + z 1 )(s + z 2 ) . . . (s + z m )
G H (s) = (4.22)
(s + p1 )(s + p2 ) . . . (s + pn )

where

(s + z 1 ), (s + z 2 ) . . . (s + z m ) = Open loop zeros

(s + p1 ), (s + p2 ) . . . (s + pn ) = Open loop poles

For simplicity, let us choose H (s) = 1 without loss of generality. The closed loop
T.F. is written as
K (s + z 1 )(s + z 2 ) · · · (s + z m )
(s + p1 )(s + p2 ) · · · (s + pn )
M(s) =  (4.23)
K (s + z 1 )(s + z 2 ) · · · (s + z m )
1+
(s + p1 )(s + p2 ) · · · (s + pn )
K (s + z 1 )(s + z 2 ) · · · (s + z m )
(s + p1 )(s + p2 ) · · · (s + pn )
=  (4.24)
(s + p1 )(s + p2 ) · · · (s + pn ) + K (s + z 1 ) · · · (s + z m )
(s + p1 )(s + p2 ) · · · (s + pn )
4.4 Nyquist Stability Criterion 547

From Eq. (4.24), it is evident that the poles of G H (s) are the same as the poles of
[1 + G H (s)].

Poles of the open loop T.F. = Poles of the characteristic equation (4.25)

4.4.1.2 Relationship Between the Zeros of the Characteristic Equation

and the Closed Loop Poles

Let us rewrite Eq. (4.24) with the zeros of the characteristic equation in factored
form as

(s + p1 )(s + p2 ) · · · (s + pn ) + K (s + z 1 )(s + z 2 ) · · · (s + z m ) = (s + Z 1 )(s + Z 2 ) · · · (s + Z n )

for n > m

K (s + z 1 )(s + z 2 ) · · · (s + z m )
(s + p1 )(s + p2 ) · · · (s + pn )
M(s) =  (4.26)
(s + Z 1 )(s + Z 2 ) · · · (s + Z n )
(s + p1 )(s + p2 ) · · · (s + pn )
K (s + z 1 )(s + z 2 ) · · · (s + z m )
= (4.27)
(s + Z 1 )(s + Z 2 ) · · · (s + Z n )

From Eq. (4.26), (s + Z 1 )(s + Z 2 ) · · · (s + Z n ) are the zeros of the characteristic

equation. From Eq. (4.27), (s + Z 1 )(s + Z 2 ) · · · (s + Z n ) are the poles of the closed
loop T.F.

Poles of the closed loop T.F. = Zeros of the characteristic equation

For the closed loop system to be stable the closed loop poles should all be in the LHP.
The presence of even a single root in RHP will make the system unstable, whereas
the presence of jω axis roots, which are distinct, will make the closed loop system
marginally stable which is BIBO unstable. If the roots in the jω axis are repeated,
the closed loop system is unstable. To tell the result straightaway, the zeros of the
characteristic equation, which are the same as the poles of the closed loop T.F., are
not in the form of factors. The Nyquist stability criterion, aims at, without factorizing
the zeros of the characteristic equations, finding the number of roots of this equation
in RHP, LHP and the jω axis. For the closed loop system to be stable, the roots of

(s + Z 1 )(s + Z 2 ) · · · (s + Z n )

should all be in LHP.

548 4 Stability Analysis of Linear Control System

j Im
s-plane
P 2+j4 Q 14+j32 F(s)-plane

j4 j32

2 14 Re

Fig. 4.15 Mapping a point on the s-plane on to F(s)-plane

4.4.1.3 The Concepts of Mapping Points

Consider the characteristic equation

F(s) = 1 + G H (s) (4.28)

If any complex number in the complex s-plane is chosen and substituted in Eq.
(4.28), another complex number results in the F(s) plane. This is called mapping a
point from s-plane to F(s) plane. s-plane is a complex plane where σ represents the
horizontal axis and jω represents the vertical axis. F(s) plane is represented by Re
as the horizontal axis and Im as the vertical axis.
Consider the following function

F(s) = s 2 + 4s − 10 (4.29)

Let us choose any point s = 2 + j4 in the complex s-plane. Substituting this value
of s in Eq. (4.29), we get

F(s)|s=2+ j4 = (2 + j4)2 + 4(2 + j4) − 10

= −14 + j32

The mapping from s-plane to F(s) plane is shown in Fig. 4.15. The point P in the
s-plane is mapped as point Q in the F(s)-plane.

4.4.1.4 The Concept of Mapping Contour

The concept of mapping a point from s-plane to F(s) plane can be extended to
collection of points. The connection of points in the complex plane is called a contour.
Let us consider the following function.
4.4 Nyquist Stability Criterion 549

j Im

1 j 3 j 1 j
s-plane 1 j
A B A B

C
0 1 F 2 3 1 F 0 C 1 Re

D E E D
1 j 3 j 1 j 1 j

(a) s-plane contour (b) F(s)-plane contour

Fig. 4.16 Mapping of s-plane contour to F(s) plane contour (a zero enclosed inside the contour)

F(s) = (s − 2)

Let us choose a contour in the s-plane as shown in Fig. 4.16a. We choose six points
A, B, C, D, E, F in the s-plane contour. The mapping of these points from s-plane
to F(s) plane is done as follows. The following table is prepared (Table 4.1).
The s-plane contour and F(s) plane contour are shown in Fig. 4.16. For F(s) =
s − 2, a zero is placed inside the contour. The contour ABC D E F encloses the zero.
When this contour is moved in the clockwise direction, the points are mapped as
A , B  , C  , D  , E  , F  in the F(s) plane and forms another contour. This contour
also moves in the same clockwise direction and encircles the origin once. A point
within a contour to the right of the traversal of the contour is considered to be the
point enclosed by the contour. The clockwise traversal of a contour is taken as
positive and the point on the right side of the contour is said to be enclosed.
A point is said to be encircled by a closed path if it is found inside the path (to
the right) during the traversal. The clockwise encirclement is taken as positive, while
encirclement in the counter clockwise is taken as negative.

Table 4.1 Mapping s-plane contour to F(s)-plane contour

Point in the s -plane s F(s) = (s − 2)
A 1+ j A = −1 + j
B 3+ j B = 1 + j
C 3 + j0 C  = 1 + j0
D 3− j D = 1 − j
E 1− j E  = −1 − j
F 1 + j0 F  = −1 + j0
550 4 Stability Analysis of Linear Control System

j Im F(s)-plane

3 j 4 j 1 j 2 j
s-plane
A B A B

C
2 3 F 4 0 1 F C 2 Re
D E E D
3 j 4 j 1 j 2 j

(a) s-plane contour (b) F(s)-plane contour

Fig. 4.17 Mapping of s-plane contour to F(s) plane (a zero not enclosed)

Table 4.2 Mapping s-plane contour to F(s)-plane contour

Point in the s -plane s F(s) = (s − 2)
A 3+ j A = 1 + j
B 4+ j B = 2 + j
C 4 + j0 C  = 2 + j0
D 4− j D = 2 − j
E 3− j E = 1 − j
F 3 + j0 F = 1

Table 4.3 Mapping s-plane contour to F(s)-plane contour

Point in the s -plane s F(s) = 1/(s − 2)
A 1+ j A = −0.5 − 0.5 j
B 3+ j B  = 0.5 − 0.5 j
C 3 + j0 C  = 1 + j0
D 3− j D  = 0.5 + j0.5
E 1− j E  = −0.5 + j0.5
F 1 + j0 F  = −1 + j0

Now let us choose a different contour in the s-plane which does not enclose the
zero at s = 2. This is shown in Fig. 4.17a. The following table is prepared (Table 4.2).
The s-plane contour which does not enclose a zero when mapped into F(s) plane
does not encircle the origin.
Let F(s) = 1/(s − 2). Let the contour ABC D E F enclose the pole at s = 2 in the
s-plane. For the mapping of the points A, B, C, D, E, F from s-plane to F(s)-plane,
the following table is prepared (Table 4.3).
4.4 Nyquist Stability Criterion 551

j Im

1 j 3 j
s-plane 0.5 j0.5 0.5 j0.5
A B
E D
C 1 1
0 1 F 2 3 F 0 C Re
D E A B
0.5 j0.5 0.5 j0.5
1 j 3 j

(a) s-plane contour (b) F(s)-plane contour

Fig. 4.18 Mapping s-plane contour to F(s) plane contour (a pole enclosed inside the contour)

The s-plane contour when mapped in the F(s) plane is as shown in Fig. 4.18.
The F(s) plane contour encircles the origin of F(s) plane in the counter clockwise
direction once. From the above discussions, the following conclusions are derived:
1. An s-plane contour with a zero of F(s) enclosed when mapped from s-plane
to F(s) plane in the clockwise direction will encircle the origin of the s-plane
once in the same clockwise direction. It can be shown that if the s-plane con-
tour encloses two zeros of F(s) when mapped from s-plane to F(s) plane will
encircle the origin of the s-plane twice in the same clockwise direction. For each
additional zero enclosed inside the s-plane contour, there will be one additional
clockwise encirclement of the origin in the same direction.
2. An s-plane contour with no zero or pole enclosed when mapped from s-plane to
F(s) plane will not at all encircle origin of the F(s)-plane.
3. An s-plane contour with a pole of F(s) enclosed when mapped from s-plane to
F(s) plane in the clockwise direction will encircle the origin of the F(s)-plane
once in the anticlockwise direction. For each additional pole enclosed inside the
s-plane contour, there will be one additional encirclement in the F(s)-plane and
the encirclement is in the counter clockwise direction.
4. In general, if there are Z zeros and P poles enclosed by the s-plane contour,
when it is mapped into F(s) plane with the points in the s-plane contour moving
in the clockwise direction, there will be Z − P encirclements around the origin
of F(s) plane in the same clockwise direction provided Z > P. If P > Z , there
will be P − Z encirclements in the counter clockwise direction.
5. Conversely, if there are say 3 encirclements in the F(s)-plane contour, in the
clockwise direction, there are Z zeros and P poles inside the s-plane contour
which satisfies the condition Z − P = 3, where Z > P. For counter clockwise
encirclement, the condition to be satisfied is P − Z = 3, where P > Z .
552 4 Stability Analysis of Linear Control System

(a) s-plane contour (b) F(s)-plane contour

j Im
s-plane s1 F(s)-plane
z1 Ts

z1 TF

p1 p2 F
p1 0 p2 0 Re
z2

z2

Fig. 4.19 Evaluation of net angle of TF

4.4.2 The Principle of the Argument—Cauchy’s Theorem

The Nyquist stability criterion was developed from the theory of conformal mapping
and the principle of the argument which is based on Cauchy’s theorem. Cauchy’s
theorem is based on mapping a function F(s) that has a finite number of poles and
zeros within the contour. Consider F(s) in the following form

K (s + z 1 )(s + z 2 ) · · · (s + z m )
F(s) = (4.30)
(s + p1 )(s + p2 ) · · · (s + pn )

For simplicity, we consider two poles and two zeros in Equation (4.30)

K (s + z 1 )(s + z 2 )
F(s) = (4.31)
(s + p1 )(s + p2 )

The zeros are taken as complex conjugate and poles are real as shown in Fig. 4.19.
Now consider a small contour Ts in the s-plane which has enclosed the zero at
s = z 1 . Take any point s1 on the contour Ts . At point s1 , Eq. (4.31) is written as

F(s1 ) = |F(s1 )|∠θz1 + ∠θz2 − ∠θ p1 − ∠θ p2

By mapping, the net angle in the F(s) plane is

θ F = θz1 + θz2 − θ p1 − θ p2

As s1 traverses along the contour Ts , for a full rotation of 360◦ , the angles of θz2 ,
θ p1 and θ p2 will be zero while θz1 will be 360◦ . Thus, for one zero enclosed by Ts
contour in the s-plane, the contour TF in the F(s) plane will have one encirclement
4.4 Nyquist Stability Criterion 553

around the origin of the F(s)-plane. Suppose there are Z zeros and P poles inside
the s-plane contour Ts . As s is traversed in the clockwise direction, the net angle
of F(s) is 2π(Z − P). If we denote this as N encirclements in the F(s) plane, the
following relation is true.

2π N = 2π(Z − P)

N=Z−P (4.32)

We illustrate below in Table 4.4 the number of poles and zeros inside the s-plane con-
tour and the number of encirclements N made by the F(s) plane contour around the
origin of the F(s) plane using Eq. (4.32). As stated earlier, the clockwise encirclement
is taken as positive, while counter clockwise encirclement is taken as negative.

4.4.3 The Nyquist Stability Criterion

Now consider Eq. (4.26). The closed loop system will be stable if all the roots of the
characteristic equation are in LHP. In Eq. (4.26), Z 1 , Z 2 etc.., should have a negative
real part. Now Eq. (4.32) can be applied to Eq. (4.26) to find the number of zeroes of
the characteristic equation that are in the RHP. For this, the s-plane contour is formed
that will include the entire RHP including the jω axis. However, the contour be
such that it does not pass through poles or zeroes of F(s). Since the poles of open
loop T.F. are the poles of the characteristic equation, the number of poles of F(s) that
are in RHP is known just by the inspection of G H (s). By conformal mapping, the
contour in the s-plane is now mapped to F(s) plane and the number of encirclement
N made around the origin is determined. Then using Eq. (4.32) Z , the number of
zeroes of the characteristic equation that are in RHP is known. The closed loop
system is stable if Z = 0. If Z is not zero, then the closed loop system is unstable.
Relationship Between G H (s) Plane and F(s) Plane
Using Eq. (4.32) the stability of the closed loop system is determined by counting
the number of encirclements N of 1 + G H ( jω) around the origin of the F(s) plane.
As seen from Fig. 4.20, this is equivalent to determine the number of encirclements
of G H ( jω) around (−1, 0) point in the s-plane since plotting of G H ( jω) is much
easier than plotting of 1 + G H ( jω). Hence, the stability condition given by Eq.
(4.32) is modified as given below and is still valid. The following symbols with
suffix ‘r ’ are used in Eq. (4.32) which are defined as follows:
1. Pr = Number of poles of G H (s) in the right half s-plane.
2. Z r = Number of zeros of 1 + G H (s) in the right half s-plane.
3. Nr = Number of clockwise encirclements of the (−1, 0) point when s-plane
Nyquist contour is mapped into G H plane.
Equation (4.32) is rewritten as
554 4 Stability Analysis of Linear Control System

Table 4.4 Mapping of s-plane contour to F(s) plane contour

s-plane contour F(s)-plane contour

j Im

0
Re

Z 1; P 0 N 1
j Im

0
Re
Z 0; P 1 N 1
j Im
0
0
Z 2 0 Re
P 1 N 1
j
Im

0 Re
Z 1
P 2 N 1
j
Im

0
Re
0
Z 2 0
P 2 N 0

j Im

0
Re
Z 1
P 3 N 2
4.4 Nyquist Stability Criterion 555

(a) [1 GH(j )] plot (b) GH(j ) plot

Im Im
F(s)-plane
GH(s) plane

( 1,0)
0 (1,0) Re 0 Re
GH(j )
F(j ) [1 GH(j )] [1 GH(j )]

0 0

Fig. 4.20 [1 + G H ( jω)] in the F(s) plane and G H (s) plane

Z r = Nr + Pr (4.33)

If no pole of the open loop system is in RHP (open loop system is stable), then

Z r = Nr (4.34)

For the closed loop system to be stable Z r = 0. Thus, the problem of determining
Z r is converted to finding Nr . Using Eqs. (4.34) and (4.33), the Nyquist stability
criterion is stated as follows.
Open Loop System is Stable (Pr = 0)
If the open loop system is stable (Pr = 0), then for the closed loop system also to be
stable (Z r = 0), the s-plane Nyquist contour when mapped into G H -plane should
not at all encircle the (−1, 0) point.
Open Loop System is Unstable (Pr = 0)
If the open loop system is unstable, then for the closed loop system to be stable, the
s-plane Nyquist contour when mapped into G H -plane should encircle (−1, 0) point
in the G H plane as many times as there are number of open loop poles in RHP and
the encirclement should be in the counter clockwise direction of the s-plane contour.
How to Count Nr
The following procedure is followed to count Nr :
1. In the Nyquist diagram, put the arrow in the proper direction for the G H ( jω)
plot, mirror image and infinite semi-circles if any.
2. Identify the (−1, 0) point and draw any radial line from this point, directing
preferably in the 3rd quadrant of the G H -plane. This radial line cuts the Nyquist
diagram at different points.
3. For each intersection point, draw a curved arrow which is in the same direction
as in the Nyquist diagram.
556 4 Stability Analysis of Linear Control System

(a) Im (b)
Im
GH-plane GH-plane

( 1,0) 0 1,0 0
0 Re 0 Re

Nr 1
Nr 1

(c) (d)
Im 0 Im
0

GH-plane

( 1,0) ( 1,0) ( 1,0)

Re Re

Nr 2 0 0

Nr 2 Nr 0

Fig. 4.21 Counting of Nr of Nyquist diagram

4. If the arrow is in the clockwise direction, count Nr = 1. If the arrow is in the

anticlockwise direction, count Nr = −1. Sum up all the Nr s which gives the
total Nr for the given Nyquist plot.
Figure 4.21 illustrates the procedure to count Nr for different types of Nyquist plots.

Example 4.20
A certain unity feedback control system has the following open loop T.F. with min-
imum phase. Determine the stability of the closed loop system using the Nyquist
stability criterion.
K
G(s) =
(s + T1 )(s + T2 )

Solution
See Fig. 4.22.
Type 0, Order 2 System
4.4 Nyquist Stability Criterion 557

(a) (b)
j Im
s-plane GH-plane
A Mirror image
RHP
LHP of G(j )
R
0
j /2 ≤ ≤ /2 K
0 Re ( 1,0) T1 T2
0 0 B Re 0 O 0 Re
A BC

Nr 0
C Nyquist contour or
Nyquist path G(j ) plot Nyquist plot or
Pr 0 Nyquist diagram

Fig. 4.22 a Nyquist contour in the s-plane. b Nyquist plot in the G H -plane G(s) =
K /[(s + T1 )(s + T2 )]

1. Since we are interested in whether there is any zero of 1 + G H (s) in the jω axis
and the RHP of the s-plane, a Nyquist contour as shown in Fig. 4.22a is chosen.
This contour covers the entire RHP as R tends to ∞ and varied in the range
π/2 ≤ θ ≤ −π/2. Further, the contour includes the jω axis also to identify
whether there is any zero in the jω axis as well. The s-plane contour is called
Nyquist path.
2. Since the given T.F. is of minimum phase, the poles s1 = −T1 and s2 = −T2
are in LHP. There is no RHP poles of G(s). Hence, Pr = 0. For the closed loop
system to be stable, Z r = 0. Equation (4.34) is applied and Nr should then be
zero.
3. To get Nr , the s-plane Nyquist contour is mapped to G H -plane. This is done
section by section mapping for the given G(s) as explained below.

Section, O A; s = jω; 0 ≤ ω ≤ ∞

K
G( jω) =
( jω + T1 )( jω + T2 )
K ∠− tan−1 Tω1 − tan−1 ω
T2
=
(ω2 + T12 )(ω2 + T22 )

For ω = 0;
K ∠0◦
G( jω) =
T1 T2

For ω = ∞;
558 4 Stability Analysis of Linear Control System

G( jω) = 0∠−180◦

The G( jω) plot starts from +ve react at K /(T1 T2 ) for ω = 0, traverses through
the fourth and third quadrants in the G H -plane and terminates at the origin for
ω = ∞. Section O A in the s-plane is mapped into O  A in the G H -plane as
shown in Fig. 4.22b.
4. Section ABC, s = Re jθ where R → ∞ and π/2 ≤ θ ≤ −π/2.
For the Section ABC
K
G(Re jθ ) = lim
R→∞ (Re jθ + T1 )(Re jθ + T2 )
=0

Section ABC in the s-plane is mapped into the origin of the G H -plane as A B  C  .
5. Section C O, s = − jω; −∞ ≤ ω ≤ −0.
This is the mirror image of G( jω) and is shown as C  O  in dotted line in the
G H -plane.
6. The closed Nyquist contour in the s-plane is mapped into another closed con-
tour in the G H -plane. The G H -plane contour is called Nyquist plot or Nyquist
diagram. The s-plane contour is known as Nyquist path.
7. In Fig. 4.22b, (−1, 0) is identified and from there a radial dotted line is drawn.
This radial line does not intersect the Nyquist plot. Hence, Nr = 0. Using Eqs.
(4.33) or (4.34), we get

Z r = Nr + Pr
= 0+0

Zr = 0

Since Z r = 0, there is no zero of the characteristic equation (closed loop pole)

in RHP and the imaginary axis.

The closed loop system is stable

Note: There is no need for numerical values for K , T1 , and T2 for the
above problem.

Example 4.21

Consider the following loop T.F. with minimum phase of a certain closed loop
system
4.4 Nyquist Stability Criterion 559

K
G H (s) =
s(s + T1 )(s + T2 )

Determine the stability of the closed loop system using Nyquist stability criterion.

(Anna University, 2005)

Solution
K
G H (s) =
s(s + T1 )(s + T2 )

1. Type 1, Order 3, System and Pr = 0

The Nyquist contour of Fig. 4.22a is slightly modified as shown in Fig. 4.23a.
The given G H (s) has a pole at s = 0 and the Nyquist contour of Fig. 4.22a if
not modified, it will not satisfy Cauchy’s theorem. Hence, a small circle around
the origin of s-plane is drawn in the RHP with centre at the origin and radius
r → 0 and −π/2 ≤ φ ≤ π/2. Now, section by section mapping is done from
s-plane to G H -plane.
2. Section AB; s = jω; 0≤ω≤∞

K
G H ( jω) =
jω( jω + T1 )( jω + T2 )
K
|G H ( jω)| =
ω (ω2 + T12 )(ω2 + T22 )
ω ω
∠G H ( jω) = −90◦ − tan−1 − tan−1
T1 T2

For ω = 0
G H ( jω) = ∞∠−90◦

For ω = ∞
G H ( jω) = 0∠−270◦

The polar plot starts from −∞ in the jω axis for ω = 0, traverses through the
third quadrant, cuts the negative real axis for ω = ω pc , and terminates at the
origin of the G H -plane in the second quadrant.
3. The point in the negative real axis at which the G H ( jω) plot cuts is very impor-
tant to assess the stability. This is estimated as described below. On the negative
real axis, the phase angle is 180◦ and the corresponding frequency is called the
phase crossover frequency which is determined as follows:
560 4 Stability Analysis of Linear Control System

ω ω
−90◦ − tan−1 − tan−1 = −180◦
T1 T2
ω
−1 T1
+ Tω2
tan = 90◦
1 − Tω1 T2
2

Taking tan on the both sides, we get

ω ω
+
= tan 90◦ = ∞
T1 T2
ω2
1− T1 T2

√
To satisfy the above condition, ω = T1 T2 which is ω pc . At ω pc , the magnitude
of G H ( jω) is

K
|G H ( jω)|ω=ω pc = √
T1 T2 (T1 T2 + T22 )(T1 T2 + T12 )
K
=
T1 T2 (T1 + T2 )

4. Section BC D; s = Re jθ ; R → ∞; π/2 ≤ θ ≤ −π/2

K
G H ( jω) = lim
R→∞ Re jθ (Re jθ + T1 )(Re jθ T2 )
=0

This is plotted into the origin of the G H -plane.

5. Section D E; s = − jω −∞ ≤ ω ≤ −0
This is the mirror image of section AB. This is drawn in dotted line.
6. Section E F A; s = r e jφ ; r → 0 −π/2 ≤ φ ≤ π/2

K
G H (r e jφ ) =
r e jφ (r e jφ + T1 )(r e jφ + T1 )
= ∞e− jφ

where −π/2 ≤ φ ≤ π/2. This is one semi-circle of ∞ radius in the clockwise

direction. The G H -plane contour becomes automatically closed.
7. The Nyquist diagram shown in Fig. 4.23b is interpreted as follows. Since, the
numerical values for K , T1 and T2 are not given, if

|G H ( jω)|ω pc > 1

or
K
>1
T1 T2 (T1 + T2 )
4.4 Nyquist Stability Criterion 561

(−1, 0) point is enclosed by the Nyquist plot. As seen in Fig. 4.23b,

Nr = +2. Since Pr = 0

Z r = Nr + Pr
=2

There are 2 RHP roots of the characteristic equation in RHP, and the system is
unstable.
8. If
K
<1
T1 T2 (T1 + T2 )

The Nyquist plot does not encircle (−1, 0) point and hence Nr = 0. Since Pr = 0

Z r = Nr + Pr
= 0+0=0

No root of the characteristic equation is in RHP and the imaginary axis and the
system is stable

K
1. > 1; system is unstable
T1 T2 (T1 + T2 )
K
2. < 1; system is stable
T1 T2 (T1 + T2 )
K
3. = 1; system is marginally stable or oscillatory
T1 T2 (T1 + T2 )

One important point which is to be noted in the above 2 examples is that one semi-
circle of infinite radius is drawn in the clockwise direction for Type 1 system so that
the contour in the G H -plane becomes closed.

Example 4.22
Consider the following loop T.F of a certain closed loop system

K
G H (s) =
s 2 (s + T )

where K and T are greater than zero. Determine the stability of the closed loop
system using Nyquist stability criterion.

Solution
1. Type 0, Order 3 system. Pr = 0.
562 4 Stability Analysis of Linear Control System

j Im

Mirror image 0 Semi-circle of

LHP s-plane
radius
B RHP DE
R EFA
/2 ≤ ≤ /2 K
j T1T2(T1 T2)
A
Re ( 1,0)
0
F C Re
T 1T 2 0 B CD
E pc
re j r 0
/2 ≤ ≤ /2 Nr 2 AB

D Nyquist path G(j ) plot

0

Fig. 4.23 a Nyquist path and b Nyquist plot for Example 4.21

K
G H (s) =
s 2 (s + T )

The Nyquist contour of Fig. 4.23a can be used for this also. For T > 0, Pr = 0.
The section by section mapping of s-plane Nyquist contour into G H -plane is
explained below.
2. Section AB; s = jω; 0≤ω≤∞

K
G H (s) =
s 2 (s + T )
K ω
G H ( jω) = √ ∠−180◦ − tan−1
ω2 ω2 + T 2 T
For ω = 0, G H ( jω) = ∞∠−180◦
For ω = ∞, G H ( jω) = 0∠−270◦

The G H ( jω) plot starts from negative real at −∞ in the second quadrant,
traverses through the same quadrant with phase angle of G H ( jω) reduced as ω
increases and ends at the origin of G H -plane with phase angle −270◦ . This is
shown as A B  in the Nyquist plot.
3. Section BC D; s = Re jθ ; R → ∞; π/2 ≤ θ ≤ −π/2

K
G H (Re jθ ) =
Re j2θ (Re jθ + T )
=0

This plots into the origin of G H -plane as B  C  D  .

4. Section D E; s = − jω; −∞ ≤ ω ≤ −0
4.4 Nyquist Stability Criterion 563

Im

GH-plane
Infinite
GH(j ) plot radius circle
(2 semi-circles)

0 A B ( 1,0)
0 Re
B CD
E 0 EFA
Mirror image Pr 0
of GH(j )
D

Nr 2

Fig. 4.24 Nyquist plot for Example 4.22

K
G H (− jω) = √
ω2 − jω + T

This is the mirror image of AB. This is drawn as D  E  .

5. Section E F A; s = r e jφ ; r → 0; −π/2 ≤ φ ≤ π/2

K
G H (r e jφ ) =
r e j2φ (r e j2φ + T)
= ∞e− j2φ

For −π/2 ≤ φ ≤ π/2, section E F A is mapped as a circle (2 semi-circles) of ∞

radius drawn in the clockwise direction.
6. The Nyquist plot gets closed in the G H -plane. Choose point (−1, 0) and draw
a radial line to count Nr . As seen from Fig. 4.24, Nr = 2. Using Eq. (4.33), Z r
is determined.

Z r = Nr + Pr
= 2+0
=2

There are 2R H P roots of characteristic equation.

7.
2 RHP roots
Closed loop system is unstable
for all values of K > 0
564 4 Stability Analysis of Linear Control System

To Summarize
The general procedure to obtain the closed loop system stability using Nyquist cri-
terion was illustrated by the above three examples. We summarize as follows:
1. From the given T.F. G H (s), just by inspection find Pr , the number of open loop
poles in RHP.
2. Draw the G H ( jω) plot for 0 ≤ ω ≤ ∞. (The traversal of the G H ( jω) plot is
decided by phase angle expression ∠G H ( jω). By careful assessment of this,
one can easily sketch the G H ( jω) plot.)
3. Draw the mirror image of G H ( jω) plot by dotted line.
4. For a Type 0 system, the Nyquist plot gets automatically closed. If G H (s) is
a Type 1 system, draw one semi-circle in the clockwise direction. For Type 2
system, draw 2 semi-circles in the clockwise direction. Thus, the number of
semi-circles to be drawn depends on the Type of the system. Now the Nyquist
plot gets closed in the G H -plane.
5. Draw a dotted radial line from (−1, 0) point directing in the 3rd quadrant of the
G H -plane and count Nr .
6. Using Eq. (4.33), determine Z r . For the closed loop system to be stable, Z r = 0.
Otherwise, the system is unstable.
In the example we work out below, straightaway we follow the above six steps.
Section by section mapping of Nyquist contour from s-plane to G H -plane is avoided
as it will be automatically satisfied when one follows the above six steps. However,
in certain special cases, the contour mapping will have to be considered. Further,
it should be noted that the G H ( jω) which gives important information about the
stability of the closed loop system need not be drawn after calculating the magnitude
|G H ( jω)| and the phase ∠G H ( jω) for 0 ≤ ω ≤ ∞. In many practical cases, the
intersection of G H ( jω) on the negative real axis alone will suffice to find the stability.
In the Nyquist method of finding the stability estimation, the only part which requires
some calculation is the plot of G H ( jω). When this is done judiciously by calculating
|G H ( jω)| at ω pc , the stability study becomes simpler and quicker. This is how many
problems in this book are solved.

Example 4.23
Sketch the Nyquist plot for the following open loop T.F. and examine the closed loop
system stability
K
G H (s) =
(s + 1)(s 2 + 2s + 5)

(Anna University, December, 2010)

4.4 Nyquist Stability Criterion 565

Solution
1.
K
G H (s) =
(s + 1)(s 2 + 2s + 5)

Since it is a minimum phase T.F., Pr = 0

K
G H ( jω) =
( jω + 1)( j2ω + (5 − ω2 ))
K
|G H ( jω)| =
(ω2 + 1)[4ω2 + (5 − ω2 )2 ]
2ω
∠G H ( jω) = − tan−1 ω − tan−1
5 − ω2

For ω = 0
K ◦
G H ( jω) = ∠0
5
For ω = ∞

G H ( jω) = 0∠−270◦

Since the given G H (s) is a minimum phase T.F., which is of Type 0 and order
three, it starts from +ve real axis at K /5 when ω = 0, traverses through the
fourth and third quadrants, cuts the negative real axis, passes through the second
quadrant and terminates at the origin when ω = ∞. The G H ( jω) plot for 0 ≤
ω ≤ ∞ is shown in Fig. 4.25.
2. The mirror image of G H ( jω) is drawn in dotted line. The Nyquist plot gets
closed.
3. The phase crossover frequency ω pc and the point of intersection of G H ( jω)
plot with the −ve real axis are determined as follows:

2ω
− tan−1ω − tan−1 = −180◦
5 − ω2
2ω
180 − tan−1 ω = tan−1
5 − ω2

Taking tan on both sides of the above equation, we get

566 4 Stability Analysis of Linear Control System

2ω
tan[180 − tan−1 ω] = tan tan−1
5 − ω2
2ω
−ω =
5 − ω2
√
ω pc = 7
√
Substituting ω = 7 in G H ( jω), we get

K
|G H ( jω)|ω=√7 =
(ω2 + 1)[4ω2 + (5 − ω2 )2 ]
K
= √
8 × (4 + 28)
K
=
16
4. For K /16 > 1, or K > 16, (−1, 0) point is encircled by the Nyquist plot and
Nr = 2. Now

Z r = Nr + Pr
= 2+0=2

The closed loop system has two RHP roots and the system is unstable.
5. For K = 16, the Nyquist plot passes through the (−1, 0) point. The system is
marginally stable or oscillatory.
6. For K < 16, the Nyquist plot does not encircle (−1, 0) point. Nr = 0 and

Z r = Nr + Pr
= 0+0=0

The system is stable.

7.
K < 16, the system is stable.
K = 16, the system is oscillatory.
K > 16, the system is unstable.

Example 4.24
The polar diagram of a conditionally stable system for an open loop gain K = 1 is
shown in Fig. 4.26 in thick lines. The open loop T.F. of the system is known to be
stable. Determine the range of K for which the closed loop system is stable.
4.4 Nyquist Stability Criterion 567

Solution
1. The G( jω) plot drawn in thick line is given. Since the open loop system is stable,
the corresponding T.F. has only LHP poles and hence Pr = 0. For ω = 0, the
G H ( jω) plot starts from −ve jω axis, and therefore, it is a Type 1 system.
2. The mirror image is drawn with dotted line for −∞ ≤ ω ≤ −0. For Type 1
system, one semi-circle of infinite radius is drawn.
3. For K < 1/8, the entire Nyquist plot is to the right of (−1, 0) point and Nr = 0.
Since Pr = 0;

Z r = Nr + Pr = 0 + 0 = 0

The closed loop system is stable.

4. From Fig. 4.26, if (−1, 0) point is within loop 1 and loop 3, Nr = 2 and

Z r = Nr + Pr = 2 + 0 = 2

There will be 2 RHP roots and the closed loop system will be unstable.
5. If (−1, 0) point is within the loop 2, Nr = 0. Since Pr = 0 then Z r = 0 and the
closed loop system will be stable. If K > 0.5 and K < 5, the loop 2 will have
(−1, 0) point inside it.
6. The closed loop system is stable if

1
(1) K < or
8
(2) 0.5 < K < 5

Example 4.25
By Nyquist stability criterion, determine the value of K for the stability of the closed
loop system whose open loop T.F. is given by

K (s + 4)
G H (s) =
(s + 1)(s − 1)

(Anna University, December, 2009)

Solution
1.
K (s + 4)
G H (s) =
(s + 1)(s − 1)

The pole s = 1 is in RHP and hence Pr = 1.

568 4 Stability Analysis of Linear Control System

Im
Mirror image
of GH(j )
GH-plane

K/16
( 1,0) 0 K/5
0 0 Re
pc 7
Pr 0

GH(j ) plot

Nr 2

Fig. 4.25 Nyquist plot for Example 4.23

Im
Semi-circle of raduis
0 GH-plane
Mirror image
Loop 1 Loop 2 Loop 3

0
8 Re

GH(j ) plot
2 0.2

0
Pr 0
Nr 2
Nr 0
Nr 2

Fig. 4.26 Nyquist plot for Example 4.24

2.
K ( jω + 4)
G H ( jω) =
( jω + 1)( jω − 1)
√
K ω2 + 16
|G H ( jω)| =
(ω2 + 1)
ω
∠G H ( jω) = 180◦ + tan−1
4
For ω = 0,
4.4 Nyquist Stability Criterion 569

Fig. 4.27 Nyquist plot for Im

Example 4.25
Mirror image GH-plane
of GH(j )

0 4K ( 1,0)
0 0 Re

Pr 1
GH(j ) plot

Nr 1

G H ( jω) = 4K ∠180◦

For ω = ∞,

G H ( jω) = 0∠270◦

For 0 ≤ ω ≤ ∞, the G( jω) plot starts from −ve real axis at (−4K , 0) for ω = 0,
traverses through the third quadrant and terminates at the origin for ω = ∞.
3. The mirror image is drawn. The Nyquist plot gets closed at ω = −0 and ω = 0.
The complete Nyquist plot is shown in Fig. 4.27. From Fig. 4.27, for K > 0.25,
Nr is counted as −1. Since Pr = 1,

Z r = Nr + Pr = −1 + 1 = 0

The closed loop system is stable.

4. For K = 0.25, the G( jω) passes through (−1, 0) point and the system is oscil-
latory.
5. For K < 0.25, the entire Nyquist path will be to the right of the (−1, 0) point
and hence Nr = 0. In this case

Z r = Nr + Pr = 0 + 1 = 1

There will be 1 RHP root and the closed loop system us unstable.
6.
1. K > 0.25 the system is stable.
2. K = 0.25 the system is oscillatory.
3. K < 0.25 the system is unstable.
570 4 Stability Analysis of Linear Control System

Nyquist plot of G(s) (4s2 s)/(s3 2s 4)

2.5

1.5 GH(j ) plot

1
Mirror image of GH(j )
Imaginary axis

0.5 pc 1.554
0
( 1,0)
0
0
0.5
2.414
1

1.5

2
Nr 2
2.5
3 2.5 2 1.5 1 0.5 0 0.5
Real axis

Fig. 4.28 Nyquist plot for Example 4.26

Example 4.26
A certain feedback control system has the following loop T.F

s(1 + 4s)
G H (s) =
(s 3 + 2s + 4)

Determine the stability of the closed loop system using Nyquist stability criterion.
Give MATLAB program.

Solution
1.
s(1 + 4s)
G H (s) =
(s 3 + 2s + 4)

The system is Type 0 order 3 system

4.4 Nyquist Stability Criterion 571

jω(1 + j4ω)
G H ( jω) =
(4 + jω(2 − ω2 ))
√
ω 1 + 16ω2
|G H ( jω)| =
16 + ω2 (2 − ω2 )2
ω(2 − ω2 )
∠G H ( jω) = 90◦ + tan−1 4ω − tan−1
4

For ω = 0, G H ( jω) = 0∠90◦

For ω = ∞, G H ( jω) = 0∠−90◦

The G H ( jω) starts from the origin of G H -plane in the second quadrant, tra-
verses through the second quadrant, cuts the negative real axis, traverses through
the third quadrant as ω increases and terminates at the origin when ω = ∞.
2. The mirror image of G H ( jω) plot is drawn. Since the problem is drawn using
the MATLAB program, the mirror image is also drawn in thick line.
3. Since the given system is of Type 0, the Nyquist plot gets closed.
4. The point of intersection of G H ( jω) plot with the negative real axis for ω pc is
obtained as explained below

ω(2 − ω2 )
90◦ + tan−1 4ω − tan−1 = 180◦
4
 
−1 ω(2 − ω )
2
−1 ◦
tan 4ω = tan + 90
4

Taking tan on both sides, we get

ω(2 − ω2 )
4ω = − cot tan−1
4

−1
4ω =
tan tan−1 ω(2−ω
2)
4
−4
=
ω(2 − ω2 )
−4ω + 8ω = −4
4 2

4ω4 − 8ω2 − 4 = 0

Substitute ω2 = x
572 4 Stability Analysis of Linear Control System

x 2 − 2x − 1 = 0
x = 2.414
√
ω= x
√
= 2.414
ω pc = 1.554

Substituting this value in the equation for |G H ( jω)|, we get

ω (1 + 16ω2 )
|G H ( jω)| =
16 + ω2 (2 − ω2 )2
√
1.554 1 + 16 × 2.414
|G H ( jω)| =
16 + 2.414(2 − 2.414)2
= 2.414

The Nyquist plot is shown in Fig. 4.28. From Fig. 4.28, Nr is counted as −2.
5. Pr is found from the open loop T.F. The characteristic equation of the open loop
system is s 3 + 2s + 4 = 0. The following Routh’s array is formed.
For ε = 0
3
s 1 2 1 2
+
s 2
ε 4 0 4
s 1 2ε−4
ε
−∞
0
s 4 4
There are two sign changes in the first column. Hence, there are two RHP roots
of the open loop system. Hence, Pr = 2.
6.

Z r = Nr + Pr = −2 + 2 = 0
Zr = 0

The closed loop system is stable.

7. The MATLAB program is shown below.

MATLAB Program

%. . . To find stability using Nyquist plot. . .%

num= [0 4 1 0]
den[1 0 2 4]
4.4 Nyquist Stability Criterion 573

Nyquist plot of G(s) (s 3)(s 10)/(s 4)(s 5)

2
GH(j ) plot GH-plane
1.5
Mirror image
1 of GH(j )

1.5K
0.5
Imaginary axis

pc 4.91
0
( 1, 0)
0
K 0
1.44K
0.5 GH(j )

1.5
Nr 2
Mirror image
2
1.5 1 0.5 0 0.5 1 1.5
Real axis

Fig. 4.29 Nyquist plot for Example 4.27

nyquist(num,den)
axis auto
title (‘Nyquist plot of G(s) = (4s 2 + s)/(s 3 + 2s + 4)’)

Example 4.27
A certain feedback control system has the following loop T.F

K (s + 3)(s + 10)
G H (s) =
(s − 5)(s − 4)

Determine the value of K so that the closed loop system is stable. Use Nyquist
stability criterion (Fig. 4.29).

Solution
1.
K (s + 3)(s + 10)
G H (s) =
(s − 5)(s − 4)
574 4 Stability Analysis of Linear Control System

There are two poles of G H (s) in RHP at s = 5 and s = 4. Hence, Pr = 2.

2.
K ( jω + 3)( jω + 10)
G H ( jω) =
( jω − 5)( jω − 4)

(ω2 + 9)(ω2 + 100)

|G H ( jω)| = K
(ω2 + 25)(ω2 + 16)
ω ω ω ω
∠G H ( jω) = tan−1 + tan−1 + tan−1 + tan−1
3 10 5 4
3. For ω = 0

G H ( jω) = 1.5K ∠0◦

For ω = ∞

G H ( jω) = K ∠0◦

The polar plot starts from +ve real axis at 1.5K for ω = 0. As ω increases,
the G H ( jω) plot traverses through the 1st, 2nd, 3rd and 4th quadrants in the
G H -plane and terminates on the +ve real axis at K for ω = ∞. The point of
intersection of G H ( jω) plot on the negative real axis is calculated as follows.
4. At ω = ω pc

ω ω ω ω
tan−1 + tan−1 + tan−1 + tan−1 = 180◦
3 10 5 4
ω ω ω ω
+ +
tan−1 3 ω102 + tan−1 5 ω42 = 180◦
1 − 30 1 − 20
13ω 9ω
tan−1 = 180◦ − tan−1
30 − ω 2 20 − ω2

Taking tan on both sides

 
13ω −1 9ω
= tan 180 − tan
30 − ω2 20 − ω2
−9ω
=
20 − ω2
13(20 − ω ) = −9(30 − ω2 )
2

22ω2 = 530
4.4 Nyquist Stability Criterion 575

ω2 = 24.1
ω = 4.91
ω pc = 4.91

5.
√
K (24.1 + 9)(24.1 + 100)
|G H ( jω)|ω=ω pc = √
(24.1 + 25)(24.1 + 16)
= 1.44K

6. The mirror image of G H ( jω) is drawn.

7. If 1.44K > 1 or K > 0.694, the (−1, 0) point is within the G H ( jω) loops. A
radial line from (−1, 0) point is drawn and Nr is counted as Nr = −2. Since
Pr = 2

Z r = Nr + Pr
= −2 + 2 = 0

The closed loop system is stable.

8. If K < 0.694, the Nyquist plot does not encircle (−1, 0) point and hence Nr = 0

Z r = Nr + Pr
= 0+2=2

The closed loop system has 2RHP poles and the system is unstable.
9.
K > 0.694 stable.
K = 0.694 marginally stable.
K < 0.694 unstable.

10. The open loop system which is unstable is made stable by feedback.

Example 4.28
A certain control system has the following loop transfer function

K (s + 2)(s + 0.5)
G H (s) =
(1 + 0.2s)(1 − 5s)

Using Nyquist stability criterion, determine the value of K so that the closed loop
system is stable.
576 4 Stability Analysis of Linear Control System

Solution
1.
K (s + 2)(s + 0.5)
G H (s) =
(1 + 0.2s)(1 − 5s)

There is pole at s = 0.2 in RHP. Hence, Pr = 1.

2.
K ( jω + 2)( jω + 0.5)
G H ( jω) =
(1 + 0.2 jω)(1 − jω5)
K (ω2 + 4)(ω2 + 0.25)
|G H ( jω)| =
(1 + 0.04ω2 )(1 + 25ω2 )

ω ω
∠G H ( jω) = tan−1 + tan−1 − tan−1 0.2ω + tan−1 5ω
2 0.5
3. For ω = 0

G H ( jω) = K ∠0◦

For ω = ∞

G H ( jω) = K ∠180◦

For 0 ≤ ω ≤ ∞, the G H ( jω) starts from +ve real at |G H ( jω)| = K , traverses

through the first, second and third quadrants as ω increases and terminates at
−K when ω = ∞ as seen from ∠G H ( jω).
4. The mirror image of G H ( jω) is drawn as shown in Fig. 4.30 for −∞ ≤ ω ≤ −0.
The Nyquist plot gets closed.
5. Since Pr = 1, the Nyquist plot should encircle (−1, 0) point in the counter
clockwise direction once.
6. The phase crossover frequency is calculated as follows. At ω = ω pc

tan−1 0.5ω + tan−1 2ω − tan−1 0.2ω + tan−1 5ω = 180◦

2.5ω 4.8ω
tan−1 + tan−1 = 180◦
(1 − ω2 ) (1 + ω2 )

2.5ω 4.8ω
tan−1 = 180◦ − tan−1
(1 − ω2 ) (1 + ω2 )

Taking tan on both sides, we get

4.4 Nyquist Stability Criterion 577

2.5ω 4.8ω
=−
(1 − ω )
2 (1 + ω2 )

Solving the above equation for ω, we get

ω2 = 3.1739
ω pc = 1.7815

7. At ω = ω pc
√
K (3.1739 + 4)(0.25 + 3.1739)
|G H ( jω)|ω=ω pc = √
(1 + 0.04 × 3.1739)(1 + 25 × 3.1739)
= 0.521K

For loop 1, Nr = 1 and Z r = Nr + Pr = 1 + 1 = 2 unstable.

9. For loop 2 to encircle (−1, 0) point in the anticlockwise direction (Nr = −1),

0.521K > 1
K > 1.92
Z r = Nr + Pr
= −1 + 1 = 0 stable

10.
K > 1.92 system is stable.
K = 1.92 system is oscillatory.
K < 1.92 system is unstable.

Example 4.29
A certain feedback control system has the following loop T.F

K (s + 2)
G H (s) =
(s 3
+ 2s 2 − 5)

Determine the value of K , using Nyquist stability criterion so that the closed loop
system is stable.

Solution
1.
K (s + 2)
G H (s) =
(s 3 + 2s 2 − 5)
578 4 Stability Analysis of Linear Control System

Nyquist diagram
G(s) K(s 2)(s 0.5)/(1 0.2s)(1 5s)
0.8
Loop 1
Loop 2
0.6
pc 1.7815

0.4 GH(j ) plot

0.2
Imaginary axis

K ( 1, 0) ( 1, 0) K 0
0
0 0

0.2

0.4
0.521K Mirror
image
0.6
Nr 1
Nr 1
0.8
1.5 1 0.5 0 0.5 1 0.5
Real axis

Fig. 4.30 Nyquist plot for Example 4.28

Nyquist diagram
G(s) K(s 2)/(s3 2s2 5)
0.2

0.15 Mirror image of

GH(j )
0.1

0.05
Imaginary axis

0 ( 1, 0)
0 0.4K
0

0.05

0.1
GH(j ) plot

0.15 Nr 1

0.2
0.5 0.4 0.3 0.2 0.1 0 0.1
Real axis

Fig. 4.31 Nyquist plot for Example 4.29

4.4 Nyquist Stability Criterion 579

The characteristic equation of the open loop system is

s 3 + 2s 2 − 5 = 0

The following Routh’s array is formed for the above equation.

s3 1 0
s 2
2 −5
s1 2.5
s0 −5
There is one sign change in the first column of Routh’s array. Hence, Pr = 1.
2.
K ( jω + 2)
G H ( jω) =
− jω3 − 2ω2 − 5

K (ω2 + 4)
|G H ( jω)| =
ω6 + (5 + 2ω2 )2
ω ω3
∠G H ( jω) = 180◦ + tan−1 − tan−1
2 (5 + 2ω2 )

3. For ω = 0

G H ( jω) = 0.4K ∠180◦

For ω = ∞

G H ( jω) = 0∠180◦

4. For 0 ≤ ω ≤ ∞, the Nyquist plot starts from negative real axis at −0.4K , tra-
verses through the third quadrant as ω increases and terminates at the origin of
G H -plane as shown in Fig. 4.31.
5. The mirror image of G H ( jω) is drawn for −∞ ≤ ω ≤ −0. The Nyquist plot
gets closed.
6. Since Pr = 1, Nr = −1 for Z r to be zero. The Nyquist plot should encircle
(−1, 0) point once in the anticlockwise direction. For this, 0.4K > 1 or K > 2.5.
7.
K > 2.5 system is stable.
K = 2.5 system is oscillatory.
K < 2.5 system is unstable.
580 4 Stability Analysis of Linear Control System

Nyquist diagram
GH(s) K(s 5)/(1 2s)2
4

3
GH(j ) plot
2

1
Imaginary axis

5K 0 ( 1, 0)
0
0
1

2 Mirror image
of GH(j )

3
Nr 1
4
6 5 4 3 2 1 0 1
Real axis

Fig. 4.32 Nyquist plot for Example 4.30

Example 4.30
Consider the following loop T.F. of a certain feedback control system

K (s − 5)
G H (s) =
(1 + 2s)2

Determine the value of K so that the closed loop system is stable. Use Nyquist
stability criterion.

Solution
1.
K (s − 5)
G H (s) =
(1 + 2s)2

The open loop poles are in LHP and hence Pr = 0

2.
K ( jω − 5)
G H ( jω) =
(1 + 2 jω)2
4.4 Nyquist Stability Criterion 581

K (ω2 + 25)
|G H ( jω)| =
(1 + 4ω2 )
ω
∠G H ( jω) = 180◦ − tan−1 − 2 tan−1 2ω
5
3. For ω = 0

G H ( jω) = 5K ∠180◦

For ω = ∞

G H ( jω) = 0∠−90◦

4. The Nyquist plot starts from negative real at −5K when ω = 0, traverses through
the second and first quadrants as ω increases, cuts the +ve real axis, traverses
through the fourth quadrant and terminates at the origin of the G H -plane as
shown in Fig. 4.32. The point of intersection of G H ( jω) plot with the +ve real
axis is not exceed.
5. For −∞ ≤ ω ≤ −0, the mirror image of G H ( jω) plot is drawn. The Nyquist
plot gets closed.
6. Since Pr = 0, for the closed loop system to be stable, Nr = 0. As seen from
Fig. 4.32, the Nyquist plot should not encircle (−1, 0) point. This is possible if
5K < 1 or K < 0.2.
7.
K < 0.2 system is stable.
K = 0.2 system is oscillatory.
K > 0.2 system is unstable.

Example 4.31
A certain closed loop control system has the following open loop T.F

K
G H (s) =
(s + 5)3

Determine the value of K so that the closed loop system is stable. Use Nyquist
stability criterion.
582 4 Stability Analysis of Linear Control System

Nyquist diagram
10 3 GH(s) K/(s 5)3
6

4 Mirror image
of GH(j )

2
Imaginary axis

0.001K 0
0
0
8K 10 3
2 8.66
pc

4 GH(j ) plot

6
4 2 0 2 4 6 8 10
Real axis 103

Fig. 4.33 Nyquist plot for Example 4.31

Solution
1.
K
G H (s) =
(s + 5)3

All the three open loop poles are in LHP and Pr = 0.

2.
K
G H ( jω) =
( jω + 5)3
K
|G H ( jω)| = 2
(ω + 25)3/2
ω
∠G H ( jω) = −3 tan−1
5
3. For ω = 0

G H ( jω) = K 8 × 10−3 ∠0◦

For ω = ∞
4.4 Nyquist Stability Criterion 583

G H ( jω) = 0∠−270◦

4. For 0 ≤ ω ≤ ∞, the Nyquist plot starts from +ve real at K 8 × 10−3 , traverses
through the fourth and third quadrants of the G H -plane, cuts the negative real
axis when ω = ω pc and terminates at the origin in the second quadrant as shown
in Fig. 4.33.
5. For −∞ ≤ ω ≤ −0, the mirror image of G H ( jω) is drawn and the Nyquist plot
gets closed.
6. The magnitude |G H ( jω)|ω=ω pc is very important which decides the stability of
the system and is calculated as follows:
ω
−3 tan−1 = −180◦
5
ω pc = 8.66
K
G H ( jω)|ω=ω pc = 2
(ω + 25)3/2
= 0.001K

7. For the system to be stable, the Nyquist plot should not encircle (−1, 0) point.
In otherwords, 0.001K < 1 or K < 1000.
8.
K < 1000 system is stable.
K = 1000 system is oscillatory.
K > 1000 system is unstable.

Example 4.32
A certain feedback system has the following loop T.F

(1 + K s)
G H (s) =
s(s − 4)

Investigate the stability of the closed loop system using Nyquist stability criterion.
Determine K , where K > 0.

Solution
1.
(1 + K s)
G H (s) =
s(s − 4)

There is one pole of G H (s) at s = 4 in RHP. Pr = 1.

584 4 Stability Analysis of Linear Control System

Im

0 GH-plane

GH(j ) plot

Mirror image
of GH(j )
K
4
( 1,0)
Re
2
pc K

Nr 1

Fig. 4.34 Nyquist plot for Example 4.32

2.
(1 + j K ω)
G H ( jω) =
jω( jω − 4)
√
1 + K 2 ω2
|G H ( jω)| = √
ω ω2 + 16
ω
∠G H ( jω) = 90◦ + tan−1 K ω + tan−1
4
3. For ω = 0

G H ( jω) = ∞∠90◦

For ω = ∞

G H ( jω) = 0∠270◦

The G H ( jω) plot starts from +ve imaginary at ∞ when ω = 0, traverses through
the second quadrant, cuts the negative real axis when ω = ω pc , passes through
4.4 Nyquist Stability Criterion 585

the third quadrant and terminates at the origin of G H -plane when ω = ∞ as

shown in Fig. 4.34.
4. For −∞ ≤ ω ≤ −0, the mirror image of G H ( jω) is drawn.
5. Since the Type of the system is 1, a semi-circle of ∞ radius in the clockwise
direction is drawn. The Nyquist plot gets closed in the G H -plane.
6. The magnitude of G H ( jω pc ) is obtained as follows:

ω
90 + tan−1 K ω + tan−1 = 180◦
4
ω
tan−1 K ω + tan−1 = 90◦
4
ω
Kω +
tan−1 4
= 90◦
1− ω
K 2
4

Taking tan on both sides, we get

ω
Kω + 4
= tan 90◦ = ∞
1− ω
K 2
4

This is possible if

K 2
1− ω =0
4
4
ω2 =
K
2
ω pc = √
K

7.

1 + K 2 K4
|G H ( jω pc )| =
√2+ 16
K
4
K
√
K (1 + 4K )
= √
2 4 + 16K
K
=
4
For the system to be stable, Nr should be −1. For this

K
>1
4
586 4 Stability Analysis of Linear Control System

8.
K > 4 system is stable.
K = 4 system is oscillatory.
K < 4 system is unstable.

Example 4.33
A certain feedback control system has the following loop T.F.

(1 + K s)2
G H (s) =
s3
Find the value of K so that the closed loop system is stable. Use Nyquist stability
criterion.

Solution
1.

(1 + K s)2
G H (s) =
s3

No pole of G H (s) is in RHP. Hence, Pr = 0.

2.

(1 + j K ω)2
G H ( jω) =
( jω)3
(1 + K 2 ω2 )
|G H ( jω)| =
ω3
∠G H ( jω) = −270◦ + 2 tan−1 K ω

3. For ω = 0

G H ( jω) = ∞∠−270◦

For ω = ∞

G H ( jω) = 0∠−90◦

4. The G H ( jω) plot starts from + j∞ when ω = 0, traverses through the second
quadrant, cuts the negative real axis passes through the third quadrant as ω
increases and terminates at the origin of G H -plane when ω = ∞.
5. For −∞ ≤ ω ≤ −0, the mirror image of G H ( jω) is drawn as shown in Fig. 4.35.
6. Since the Type of the open loop system is 3, three semi-circles of infinite radius
as shown in Fig. 4.35 is drawn in dotted lines in the clockwise direction.
4.4 Nyquist Stability Criterion 587

7. For ω = ω pc , the G H ( jω) plot cuts the negative real axis. The point of inter-
section is calculated as follows. At ω = ω pc ,

∠G H ( jω pc ) = 180◦
−270◦ + 2 tan−1 K ω = 180◦
2 tan−1 K ω = 90◦
Kω = 1
1
ω pc =
K
8.

|G H ( jω pc )| = (1 + 1)K 3 = 2K 3

9. When (−1, 0) point is within loop 2, Nr = 2 and since Pr = 0, Z r = 2. The

system will be unstable. If (−1, 0) point is within loop 1, Nr = 0 and Z r = 0
and the system will be stable. So (−1, 0) point should be to the right of −2K 3 .
For this

2 K3 > 1
K > 0.7937

10.
K > 0.7937 system is stable.
K = 0.7937 system is oscillatory.
K < 0.7937 system is unstable.

Example 4.34
A certain control system has the following open loop T.F.

9
G H (s) =
(s 2 + 16)

Determine the stability of the closed loop system using Nyquist stability criterion.
If the system is oscillatory, find the frequency of oscillation.
588 4 Stability Analysis of Linear Control System

Im

GH(j )
2K3 plot
Mirror
image
( 1,0) ( 1,0)
Re
pc
1
K
loop 1
loop 2

0
Nr 2

Nr 0

Fig. 4.35 Nyquist plot for Example 4.33

Solution
1.
9
G H (s) =
s2 + 16

G H (s) poles are at s1 = + j4 and s2 = − j4. No pole is in RHP. Hence, Pr = 0.

As per Cauchy’s theorem, the s-plane contour should not pass through any pole.
Hence, these poles are by-passed by a small circle as shown in Fig. 4.36a. The
section by section mapping of s-plane contour is discussed below.
2. Section O A, s = jω; 0≤ω≤4

9
G H ( jω) =
−ω2 + 16
9
For ω = 0, G H ( jω) = ∠0◦
16
For ω = 4, G H ( jω) = ∞∠0◦

This is shown as O  A in G H -plane.

3. Section AB, s = j4 + r e jφ ; r → 0; −π/2 ≤ φ ≤ π/2
4.4 Nyquist Stability Criterion 589

9
G H (s) =
s2 + 16
9
=
(s + j4)(s − j4)
9
G H ( j4 + r e jφ ) =
( j4 + r e + j4)( j4 + r e jφ − j4)
jφ

9
=
( j8 + r e jφ )r e jφ

As r → 0

G H ( j4 + r e jφ ) = ∞e− jφ − π/2 ≤ φ ≤ π/2

This represents a semi-circle of infinite radius in the clockwise direction. This

is shown as A B  in the G H -plane.
4. Section BC; s = jω; 4≤ω≤∞

−9
G H ( jω) =
− 16)(ω2
For ω = 4, G H ( jω) = −∞
For ω = ∞, G H ( jω) = 0∠180◦

This is plotted as B  C  on the negative real axis in the G H -plane.

5. Section C D E; s = Re jθ ; R → ∞; π/2 ≤ θ ≤ −π/2

9
G H (s) =
+ 16s2
9
G H (Re ) =
jθ
(Re + 16)
jθ

= 0 for R = ∞

This is plotted into the origin of G H -plane.

6. Section E F; s = − jω; −∞ ≤ ω ≤ −4
This is the mirror image of BC and is shown as E  F  is Fig. 4.36b.
7. Section F G; s = − j4 + r e jφ ; r → 0; −π/2 ≤ φ ≤ π/2
This is the mirror image of AB and is shown as F  G  .
8. Section G O; s = jω; −4 ≤ ω ≤ −0
This is the mirror image of O A. This is shown as G  O  . The Nyquist plot is
closed. From Fig. 4.36b, it is observed that the Nyquist plot passes through
(−1, 0). Hence, the closed loop system is oscillatory.
In Fig. 4.36a, for clarify, the Nyquist plot in the G H -plane on the real axis is
slightly shifted above and below the real axis to differentiate the overlapping
part.
590 4 Stability Analysis of Linear Control System

(a) j (b) Im
C s-plane GH-plane
re j
B
j4
CDE
A j F > 4
Re ≤ 4 0G
0 D B 0 O A
re j >4 ≤4
G ( 1,0)
9
j4 16
F
r 0
E R

Fig. 4.36 a Nyquist contour in the s-plane and b Nyquist plot in G H -plane

The system is oscillatory.

9. In Fig. 4.36b, the section B  C  passes, through (−1, 0) point where the frequency
range is 4 ≤ ω ≤ ∞ and s = jω

9
G H ( jω) =
−ω2 + 16
= −1
ω − 16 = 9
2

ω=5

The system oscillates with a frequency of

ω = 5 rad/sec.

Step-by-Step Procedure to Draw Nyquist Plot for Poles of G H ( jω) on the jω

Axis
1. Draw G H ( jω) plot for 0 ≤ ω ≤ ω1 where ω1 is the location of the pole on the
jω axis.
2. Draw the infinite semi-circle in the clockwise direction.
3. Draw the G H ( jω) plot for the frequency range ω1 ≤ ω ≤ ∞.
4. Draw the mirror image of step 3.
5. Draw the semi-circle of infinite radius in the clockwise direction.
6. Draw the mirror image of step 1.
7. The Nyquist plot gets closed if the Type of the system is zero. If the Type of the
system is one, draw a semi-circle of infinite radius in the clockwise direction,
and for Type two system, two semi-circles and so on.
4.4 Nyquist Stability Criterion 591

8. From G H (s), find the number of open loop poles in RHP. Pr is known.
9. Draw a radial line from (−1, 0) point and count Nr .
10. Using Z r = Nr + Pr , find Z r . If Z r = 0, the closed loop system is stable. Oth-
erwise, it is unstable.

Example 4.35
A certain feedback control system has the following loop T.F

(s + 9)
G H (s) =
(s 2 + 9)(s + 2)

Using Nyquist stability criterion, determine the stability of the closed loop system.

Solution
1.
(s + 9)
G H (s) =
(s 2
+ 9)(s + 2)
(s 2 + 9) = (s + j3)(s − j3)

Two roots at jω axis are at s = j3 and s = − j3. No pole of G H (s) is in RHP

and hence Pr = 0. For s-plane Nyquist contour, refer to Fig. 4.36a.
2. For 0 ≤ ω < 3

( jω + 9)
G H ( jω) =
(9 − ω2 )( jω + 2)
(ω2 + 81)
|G H ( jω)| = √
(9 − ω2 ) ω2 + 4
ω ω
∠G H ( jω) = tan−1 − tan−1
9 2
∠G H ( jω)|ω=3 = 6◦ − 26◦ = −20◦
For ω = 0, G H ( jω) = 0.5∠0◦
For ω = ∞, G H ( jω) = ∞∠−20◦

The plot O  A is drawn.

3. For the point j3, a semi-circle of infinite radius is drawn as A B  .
4. For 3 ≤ ω ≤ ∞
592 4 Stability Analysis of Linear Control System
√
ω2 + 81
|G H ( jω)| = √
(ω2 − 9) ω2 + 4)
ω ω
∠G H ( jω) = 180◦ + tan−1 − tan−1
9 2
For ω ≥ 3, |G H ( jω)| = ∞
∠G H ( jω) = 160◦
For ω = ∞, |G H ( jω)| = 0
∠G H ( jω) = 180◦

The plot B  C  is drawn.

5. For −∞ < ω ≤ −3, the mirror image of step 4 is drawn as E  F  .
6. For the point − j3 the mirror image of step 3 is drawn. This is F  G  .
7. For −0 ≤ ω ≤ −∞, the mirror image of step 2 is drawn. This is G  O  . The
complete Nyquist plot is shown in Fig. 4.37.
8. From Fig. 4.37, Nr = 2. Since Pr = 0, Z r = Nr + Pr = 2 + 0 = 2. There are
three RHP roots.

The system is unstable.

Example 4.36
A certain feedback control system has the following loop T.F.

(s + 1)
G H (s) =
(s 2 + 9)(s + 2)

By using Nyquist stability criterion, test whether the closed loop system is stable.

Solution
1.
(s + 1)
G H (s) =
(s 2 + 9)(s + 2)

There are two poles on the jω axis at s = j3 and s = − j3. There are no poles
in RHP. Hence, Pr = 0. For s-plane Nyquist contour, refer to Fig. 4.36a.
4.4 Nyquist Stability Criterion 593

Re
3

≤3
Mirror image

A
G

raduis semi-circle
semi-circle

GH(j )
raduis

0
0

Asymptote
20º

0.5
O

20º
CDE
Im

0
Asymptote

Nr 2
GH(j )

( 1,0)

Asymptote
3
≤
F
3B

Fig. 4.37 Nyquist plot for Example 4.35

2.
( jω + 1)
G H ( jω) =
(9 − ω2 )( jω + 2)
√
ω2 + 1
|G H ( jω)| = √
(9 − ω2 ) ω2 − 4
ω
∠G H ( jω) = tan−1 ω − tan−1
2
594 4 Stability Analysis of Linear Control System

Re
Asymptote

18.4

3
≤3

G'
Mirror image
GH(j ) plot

of GH(j )
O' 1/18
0

0
Im

C'D'E'

Semi circle of
Semi circle of

radius
Mirror image
radius

of GH(j )

GH(j ) plot
( 1,0)

B'
F

3
3

Fig. 4.38 Nyquist plot for Example 4.36

3. For the frequency range 0 ≤ ω ≤ 3

1
For ω = 0, G H ( jω) = ∠0◦
18
For ω = 3, G H ( jω) = ∞∠45◦ − 26.6◦
= ∞∠18.4◦

The G H ( jω) plot is drawn as O  A as shown in Fig. 4.38.

4.4 Nyquist Stability Criterion 595

4. For the frequency ω = + j3, a semi-circle with infinite radius is drawn in the
clockwise direction as A B  .
5. For the frequency range 3 ≤ ω ≤ ∞
√
ω2 + 1
|G H ( jω)| = √
(ω2 − 9) ω2 + 4
ω
G H ( jω) = 180◦ + tan−1 ω − tan−1
2
For ω = 3, |G H ( jω)| = ∞∠198.4◦
For ω = ∞, G H ( jω) = 0∠180◦

The G H ( jω) plot is drawn as B  C  as shown in Fig. 4.38.

6. The ∞ semi-circle BC D in s-plane is mapped into the origin as B  C  D  in the
G H -plane.
7. For the frequency range −∞ ≤ ω ≤ −3, the mirror image of step 5 is drawn
and shown as D  E  in the Fig. 4.38.
8. For the frequency ω = −3, a semi-circle of ∞ radius is drawn in the clockwise
direction and is shown as E  F  . This is also the mirror image of step 4.
9. For the frequency range −3 ≤ ω ≤ −0, the mirror image of step 3 is drawn. The
Nyquist plot in the G H -plane is closed and complete.
10. From (−1, 0) point, a radial line is drawn and Nr is counted. From Fig. 4.38, the
Nyquist plot does not encircle (−1, 0) point and hence Nr = 0.
11.

Z r = Nr + Pr
= 0+0=0

There is no closed loop root in RHP and the system is stable.

Closed loop system is stable.

Example 4.37
Consider the following open loop T.F. of a certain feedback control system

K s(s + 2)
G H (s) =
(s 3 + 64)

Determine the value of K so that the closed loop system is stable. Use Nyquist
stability criterion.
596 4 Stability Analysis of Linear Control System

Solution
1.
K s(s + 2)
G H (s) =
(s 3 + 64)

The characteristic equation of the open loop system is

s 3 + 64 = 0
s 3 = −64
s = 4∠(2n + 1)π/3

where n = 0, 1, 2,

s1 = 4∠60◦ = 4 cos 60◦ + j4 sin 60◦

= 2 + j3.464
s2 = 4∠180◦ = −4
s3 = 4∠300◦ = 4 cos 300◦ + j4 sin 300◦
= 2 − j3.464

The poles s1 and s3 are in RHP. Hence, Pr = 2.

2.
K s(s + 2)
G H (s) =
s 3 + 64
K jω( jω + 2)
G H ( jω) =
− jω3 + 64
K ω (ω2 + 4)
|G H ( jω)| =
ω6 + (64)2
ω ω 3
∠G H ( jω) = 90◦ + tan−1 + tan−1
2 4
3. Frequency range 0 ≤ ω ≤ ∞
From phase angle expression, it is evident that it varies from 90◦ to 270◦ . There-
fore, the G H ( jω) plot lies in the second and 3rd quadrant in the G H -plane. For
ω=0

G H ( jω) = 0∠90◦

For ω = ∞

G H ( jω) = 0∠270◦
4.4 Nyquist Stability Criterion 597

The G H ( jω) plot starts from the origin for ω = 0, traverses through the sec-
ond quadrant, cuts the negative real axis, passes through the third quadrant and
terminates at the origin when ω = ∞.
4. The phase crossover frequency ω pc and the point of intersection of the G H ( jω)
plot on the −ve real axis are calculated as follows. For ω = ω pc

ω ω 3
90◦ + tan−1 + tan−1 = 180◦
2 4
ω 3 ω
tan−1 = 90◦ − tan−1
4 2
Taking tan on both sides, we get

ω 3 ω
= cot tan−1
4 2
1 2
= ω =
tan tan−1 2
ω
ω4 = 128
ω = 3.36
ω pc = 3.36

At ω pc the magnitude is

K 3.36 (3.36)2 + 4
|G H ( jω pc )| = = 0.1765K
(3.36)6 + (64)2

5. The mirror image of G H ( jω) is drawn for the frequency range −∞ ≤ ω ≤ −0.
This is shown in dotted line in Fig. 4.39.
6. From G H (s), Pr is obtained as 2. Hence, the Nyquist plot should have Nr = −2.
From Fig. 4.39, this is possible if

−0.1765K > −1
K > 5.66

7.
K > 5.66 the system is stable.
K = 5.66 the system is oscillatory.
K < 5.66 the system is unstable.
598 4 Stability Analysis of Linear Control System

GH(j ) plot Im

GH-plane

0.1765 K

( 1, 0) C'D'E' 0;
0 Re
pc 3.36

Mirror image
of GH(j )
Nr 2

Fig. 4.39 Nyquist plot for Example 4.37

Example 4.38
The loop T.F. of a certain control system is given by

K (s + 5)(s + 2)
G(s) =
(s 3 − 20)

Using Nyquist stability criterion, determine the value of K so that the closed loop
system is stable.

Solution
1.
K (s + 5)(s + 2)
G H (s) =
(s 3 − 20)

The characteristic equation of the open loop system is

s 3 − 20 = 0
√
s = 20 ∠2nπ/3
3

where n = 0, 1, 2
4.4 Nyquist Stability Criterion 599

Mirror image Im
of GH(j )

GH-plane

0.5 K
0 ( 1, 0)
0 Re

GH(j ) plot

Nr 1

Fig. 4.40 Nyquist plot for Example 4.38

s1 = 2.7144∠0◦
s2 = 2.7144∠120◦
s3 = 2.7144∠240◦

One pole s1 = 2.7144∠0◦ is in RHP. Hence, Pr = 1.

2.
K ( jω + 5)( jω + 2)
G H ( jω) =
(− jω3 − 20)
K (ω2 + 25)(ω2 + 4)
|G H ( jω)| = √
ω6 + 400
ω ω ω3
∠G H ( jω) = 180◦ + tan−1 + tan−1 − tan−1
5 2 20
3. For ω = 0

G H ( jω) = −0.5K

For ω = ∞

G H ( jω) = 0∠270◦
600 4 Stability Analysis of Linear Control System

For small values of ω, tan−1 ω5 + tan−1 ω2 will contribute more phase than
tan−1 ω20 and the phase angle is in the 3rd quadrant of G H -plane. For high
3

value of ω, each of them will contribute 90◦ which results in total phase 270◦ .
4. The G H ( jω) plot starts from −ve real at 0.5K for ω = 0, traverses through the
3rd quadrant as ω increases and terminates at the origin for ω = ∞ as shown in
Fig. 4.40 in thickline.
5. The mirror image of G H ( jω) for −∞ ≤ ω ≤ −0 is drawn in dotted line. The
Nyquist plot is closed.
6. Since Pr = 1, Nr should be −1 for Z r to be zero. This is possible if −0.5K is
to the left of (−1, 0) point. For the system to be stable 0.5K > 1 or K > 2.
7.
K > 2, the system is stable.
K = 2, the system is oscillatory.
K < 2, the system is unstable.

Application of Nyquist stability criterion for system with improper and

strictly proper T.F.

Example 4.39
The loop T.F. of a certain feedback control system is

e−0.1 s
G H (s) =
s(s + 1)(s + 2)

Determine whether the closed loop system is stable. Use Nyquist stability criterion.
Is the system stable without time delay?

Solution
1.

5e−0.1 s
G H (s) =
s(s + 1)(s + 2)

The system open loop pole in RHP is zero. Thus, Pr = 0.

2.

5e−0.1 jω
G H ( jω) =
jω( jω + 1)( jω + 2)
5
|G H ( jω)| =
ω (ω2 + 1)(ω2 + 4)

ω
∠G H ( jω) = −5.73ω − 90◦ − tan−1 ω − tan−1
2
4.4 Nyquist Stability Criterion 601

Im
(a)
0
Mirror image
of GH(j )

1.06
( 1,0)

Re

pc 1.25

GH(j ) plot

Nr 2
0 Semi circle of
radius

Im
(b) 0
Mirror image
of GH(j ) Semi circle of
radius

0.8333

Re
pc 2
( 1,0)
Pr 0
Nr 0
GH(j ) plot Zr 0
System is stable
0

Fig. 4.41 a Nyquist plot of system with transportation lag (Example 4.39). b Nyquist plot for
G H (s) = s(s+1)(s+2)
5
602 4 Stability Analysis of Linear Control System

3. The transportation lag does not make any change in the magnitude of the system
without delay. However, at each frequency, it adds a phase of −5.73ω. Hence,
the G H ( jω) plot of the original system is only phase shifted.
4. Frequency range 0 ≤ ω ≤ ∞. For ω = 0

G H ( jω) = ∞∠−90◦

For ω = ∞

G H ( jω) = 0∠0◦

The G H ( jω) plot starts from −ve jω axis for ω = 0, traverses through the 3rd
quadrant, cuts the −ve real axis for ω = ω pc , traverses through the 2nd and first
quadrants as ω increases and terminates at the origin of the G H -plane when
ω = ∞. This is shown in thick line in Fig. 4.41a.
5. Frequency range −∞ ≤ ω ≤ −0
For the above frequency range, the Nyquist plot is the mirror image of G H ( jω)
of step 4.
6. Since the type of the system is one, a semi-circle of infinite radius in the clockwise
direction is drawn and the Nyquist plot in the G H -plane gets closed.
7. The phase crossover frequency ω pc and the point of intersection of G H ( jω) plot
with the −ve real axis are important to calculate the stability of the closed loop
system. They are calculated as follows. For ω = ω pc , the phase angle is 180◦ .
Thus
ω
−5.73ω − tan−1 ω − tan−1 − 90◦ = 180◦
2
ω
tan−1 ω + tan−1 + 5.73ω = 90◦
2
By trial and error, ω is calculated to satisfy the above equation. ω is obtained as
1.25 rad/sec
ω pc=1.25

At ω pc , the magnitude of G H ( jω) is

5
|G H ( jω pc )| =
1.25 [(1.25)2 + 1][(1.25)2 + 4]
= 1.06

8. From Fig. 4.41, Nr is calculated as 2

Z r = Nr + Pr
= 2+0
4.4 Nyquist Stability Criterion 603

There are 2 RHP roots of the characteristic equation of the closed loop system.

The system is unstable.

System Without Time Delay

The T.F. of the system without time delay is

5
G H (s) =
s(s + 1)(s + 2)
5
G H ( jω) =
jω( jω + 1)( jω + 2)
5
|G H ( jω)| =
ω (ω + 1)(ω2 + 4)
2

∠G H ( jω) = −90◦ − tan−1 ω − tan−1 0.5ω

At phase crossover frequency, ∠G H ( jω pc ) = 180◦

−90 − tan−1 ω − tan−1 0.5ω = 180◦

tan−1 ω + tan−1 0.5ω = 90◦
1.5ω
tan−1 = 90◦
1 − 0.5ω2

Taking tan on both sides, we get

1.5ω
= tan 90◦ = ∞
1 − 0.5ω2
1 − 0.5ω2 = 0
√
ω = ω pc = 2

Substituting this value in |G H ( jω)|, we get

5
|G H ( jω pc ) = √ = 0.833
2×3×6

The Nyquist plot is shown in Fig. 4.41b.

From Fig. 4.41b, it is observed that the G H ( jω) plot does not encircle (−1, 0)
point and hence Nr = 0. Since Pr = 0,

Z r = Nr + Pr
= 0+0=0
The system without delay is stable.
604 4 Stability Analysis of Linear Control System

Im

Mirror image
of GH(j )

K
2520 ( 1,0)
0
0 K Re
pc 7.416
1000

GH(j ) plot

Nr 2

Fig. 4.42 Nyquist plot for Example 4.40

Example 4.40
A certain feedback control system has the following loop T.F

K
G H (s) =
(s 2 + 3s + 10)(s 2 + 3s + 100)

Determine the value of K so that the closed loop system is stable. Use Nyquist
stability criterion.

Solution
1.
K
G H (s) =
(s 2 + 3s + 10)(s 2 + 3s + 100)

Since the roots of quadratic equation with +ve coefficients have negative real
part, all the four poles of G H (s) are in LHP. Hence, Pr = 0.
2.
K
G H ( jω) =
(−ω2 + j3ω + 10)(−ω2 + j3ω + 100)
K
|G H ( jω)| =
[(10 − ω2 )2 + 9ω2 ][(100 − ω2 )2 + 9ω2 ]
3ω 3ω
∠G H ( jω) = − tan−1 − tan−1
(10 − ω2 ) (100 − ω2 )
4.4 Nyquist Stability Criterion 605

3. Frequency range 0 ≤ ω ≤ ∞.
Since G H (s) is of Type 0 order 4 system, the G H ( jω) plot starts from +ve real
axis at K /1000 when ω = 0, traverses through the 4th, 3rd, 2nd and 1st quadrants
of G H -plane as ω increases, and terminates at the origin when ω = ∞. This is
shown in Fig. 4.42.
4. Frequency range −∞ ≤ ω ≤ −0.
The mirror image of G H ( jω) is drawn in dotted lines. This completes the
Nyquist plot.
5. The phase crossover frequency ω pc and the point of intersection of G H ( jω) on
the negative real axis are determined as follows. At the phase crossover frequency
ω pc

∠G H ( jω pc ) = 180◦
3ω 3ω
− tan−1 − tan−1 = 180◦
(10 − ω2 ) (100 − ω2 )
3ω 3ω
tan−1 = 180 − tan−1
(10 − ω2 ) (100 − ω2 )

Taking tan on both sides, we get

3ω −3ω
=
(10 − ω2 ) (100 − ω2 )
100 − ω2 = ω2 − 10
ω2 = 55
ω = ω pc
= 7.416
K
|G H ( jω pc )| =
(452 + 9 × 55)(452 + 9 × 55)
K
=
2520
6. For the system to be stable, Nr should be zero since Pr = 0. For this, the following
condition is to be satisfied.
K
<1
2520
7.
K < 2520, the system is stable.
K = 2520, the system is oscillatory.
K > 2520, the system is unstable.
606 4 Stability Analysis of Linear Control System

(a) (b)
Im Im
Mirror
Mirror
image
image
K K
8 ( 1,0) 0 K 4 ( 1,0) 0 K
0 Re 0 Re
pc
6

0.5
0.86

GH(j )
plot
pc

GH(j )
plot
Nr 2 Nr 2

Fig. 4.43 Nyquist plots for Example 4.41. a G H (s) = K

(1+2s)3
; and b G H (s) = K
(1+2s)4

Example 4.41
A certain feedback control system has the following loop T.F. Determine the value of
K in each case for the closed loop system to be stable. Use Nyquist stability criterion

K
I. G H (s) =
(1 + 2s)3
K
II. G H (s) =
(1 + 2s)4

Solution
I. 1.
K
G H (s) =
(1 + 2s)3

All the 3 poles of G H (s) are at s = − 21 in LHP. No RHP pole of G H (s).

Hence, Pr = 0.
2.
K
G H ( jω) =
(1 + j2ω)3
K
|G H ( jω)| =
(1 + 4ω2 )3/2
∠G H ( jω) = −3 tan−1 2ω
4.4 Nyquist Stability Criterion 607

3. For the frequency range 0 ≤ ω ≤ ∞. For ω = 0

G H ( jω) = K ∠0◦

For ω = ∞

G H ( jω) = 0∠−270◦

The given G H (s) is a Type 0 and order 3 system with minimum phase. The
G H ( jω) plot starts from +ve real at K for ω = 0, traverses through the 4th
and 3rd quadrants, cuts the −ve real axis when ω = ω pc , passes through the
2nd quadrant and terminates at the origin of G H -plane when ω = ∞. This
is shown in thick line in Fig. 4.43a.
4. For the frequency range −∞ ≤ ω ≤ −0, the Nyquist plot is the mirror image
of G H ( jω) of step 3 and is shown in dotted line in Fig. 4.43a. The Nyquist
plot gets closed now.
5. The phase crossover frequency ω pc and the magnitude of G H ( jω) at this
frequency is important to calculate the stability of the closed loop system.
This is calculated as explained below.
6. For ω = ω pc

∠G H ( jω pc ) = −180◦
−3 tan−1 2ω = −180◦
tan−1 2ω = 60◦

Taking tan on both sides, we get

2ω = tan 60◦ = 1.732

ω = ω pc = 0.866

Substituting this value, the magnitude of G H ( jω) is calculated

K K
|G H ( jω pc )| = =
[1 + 4(.866)2 ]3/2 8

7. If K /8 is to the left of (−1, 0) point, Nr = 2. Since Pr = 0,

Z r = Nr + Pr
= 2+0=2

The system will be unstable. For the system to be stable, K < 8.

608 4 Stability Analysis of Linear Control System

8.
K < 8, the system is stable.
K = 8, the system is oscillatory.
K > 8, the system is unstable.

II. 1.
K
G H (s) =
(1 + 2s)4

All the 4 poles of G H (s) are at s = −1/2 in LHP. No poles of G H (s) is in

RHP. Hence, Pr = 0.
2.
K
G H ( jω) =
(1 + jω2 )4
K
|G H ( jω)| =
(1 + 4ω2 )2
∠G H ( jω) = −4 tan−1 2ω

3. For the frequency range 0 ≤ ω ≤ ∞. For ω = 0

G H ( jω) = K ∠0◦

For ω = ∞

G H ( jω) = 0∠360◦

The given G H (s) is a Type 0 order 4 system T.F. Hence, the G H ( jω)
plot starts from +ve real at K for ω = 0, traverses through the 4th and 3rd
quadrants, cuts the negative real axis for ω = ω pc , traverses through the 2nd
and 1st quadrants and terminates at the origin of G H -plane when ω = ∞.
This is shown in thick line in Fig. 4.43b.
4. For the frequency range −∞ ≤ ω ≤ −0
This is the mirror image of G H ( jω) of step 3. This is drawn is dotted line
in Fig. 4.43b. The Nyquist plot now gets closed. The calculation of ω pc and
the corresponding magnitude of G H ( jω pc ) is important for the stability
analysis. They are calculated as follows.
5. At ω = ω pc
4.4 Nyquist Stability Criterion 609

∠G H ( jω pc ) = −180◦
−4 tan−1 2ω = −180◦
tan−1 2ω = 45◦
ω = ω pc = 0.5

The magnitude at this frequency is calculated as

K
|G H ( jω pc )| =
(1 + 4 × 0.25)2
K
=
4
−K
6. If 4
is to the left of (−1, 0) point, Nr = 2 as seen in Fig. 4.43b and

Z r = Nr + Pr
= 2+0=2

The closed loop system will be unstable. Hence, K < 4 for the system to be
stable.
7.
K < 4, the system is stable.
K = 4, the system is oscillatory.
K > 4, the system is unstable.

1. Stability concept for linear time invariant continuous time system is explained.
Zero input response, zero state response, zero input stability, asymptotic (inter-
nal) stability, marginal (neutral) stability, bounded input bounded output stability
and relative stability are defined.
2. BIBO stability via impulse response and the characteristic roots location in the
s-plane are explained.
3. If the linear is asymptotically stable, it is also BIBO stable. But all BIBO stable
systems need not be asymptotically stable.
4. For BIBO and asymptotic stability, no root of the characteristic equation should
be in RHP.
5. Presence of distinct characteristic roots in the jω axis implies that the system
is marginally stable, but BIBO and asymptotically unstable. If the roots are
repeated in the jω axis, the system is unstable.
610 4 Stability Analysis of Linear Control System

6. Stability analysis is carried out in the time as well as frequency domains. In the
time domain, Routh–Hurwitz and root locus methods are extensively used, while
Nyquist stability criterion is the most popular method in the frequency domain.
7. Using Routh’s stability criterion, absolute and relative stability can be obtained.
Further, the method can be used to find the gain that can be increased before the
system becomes unstable. Furthermore, in few cases, the characteristic polyno-
mial can be factorized and the root locations in the s-plane can be determined.
8. In the frequency response method, Nyquist stability criterion is applied. The
method applies Cauchy’s mapping theorem and the principle of the arguments
to establish the stability criterion.
9. Nyquist stability analysis has many advantages over Routh–Hurwitz method.
By using Routh–Hurwitz method, we find the answer as yes or no regarding
the stability of the system. If the system is unstable, it does not suggest how
to improve the system stability. However, in the Nyquist stability method, in
addition to the information we get from the Routh–Hurwitz method, it suggests
methods to improve the system stability if it is unstable.
10. Nyquist stability method can be applied to determine frequency response spec-
ifications of closed loop system in addition to stability of the system.
11. Nyquist stability criterion can be applied to system whose open loop T.F. is of
minimum phase, non-minimum phase, improper and strictly proper type. It can
also be applied to system with transportation lag.

Short Answer Type Questions

1. What do you understand by stability of a linear time invariant system?

What is instability?
When a linear time invariant system is subjected to disturbances the total response
of the system is the sum of the forced and zero input (natural) responses. A linear
time invariant system is said to be stable if the zero input response approaches
zero as time approaches infinity. The system is said to be unstable if the zero
input response grows without bound as time approaches infinity.
2. Does the stability of a linear system depend upon the input?
The stability of a linear system is independent of the input. It depends on only
the locations of the characteristic roots of the system.
3. What happens when the system becomes unstable?
If the system is unstable, the output grows without bound and causes severe
damages not only to the system itself, but also to the environment and human life.
4.4 Nyquist Stability Criterion 611

4. What are the different methods available for the stability study of linear
system?
Time domain and frequency domain approaches are available for the stability
study of linear system. Routh–Hurwitz method, root locus technique and Lia-
punov methods are the widely used time domain methods while the method
proposed by Nyquist is extensively used in the frequency domain.
5. Define asymptotic stability as applied to the linear system?
For a linear system, zero state is the equilibrium state. When the system is given
a small disturbance, the zero initial state is changed to non-zero initial state. If
the system left to itself returns back to zero state as time tends to infinity, the
system is said to be asymptotically stable.
6. What are the different names given to asymptotic stability?
The different names given to asymptotic stability are zero input stability, internal
stability and simply stability. It is also called as stability in the sense of Liapunov.
7. Define marginal stability of a system.
A linear system when disturbed, gets its equilibrium state shifted. Some char-
acteristic modes of the system neither decay to zero nor grow indefinitely, but
remains constant. Such systems are said to be marginally stable with neutral
equilibrium. Marginal stability is also called as neutral stability.
8. What is bounded input and bounded output (BIBO) stability?
A linear time invariant system is said to be BIBO (externally) stable if every
bounded input yields a bounded output.
9. What is the condition that guarantees BIBO stability?
For a linear system to be BIBO stable, that all the transfer function poles should
be in LHP.
10. How impulse response of a system is correlated with BIBO stability?
A linear time invariant system is BIBO stable if the area under the impulse
response curve is finite.
11. If the system is marginally stable, will it be BIBO stable?
A marginally stable system is not BIBO stable since the area under the impulse
response curve of a marginally stable system is not finite.
12. What is the common requirement for the linear system to be asymptotically,
marginally and BIBO stable?
For the linear system, if the T.F. has all the poles in LHP, the system is said to
be asymptotically, marginally and BIBO stable.
13. What is the relationship between asymptotic stability and BIBO stability?
If the linear system is asymptotically stable, then it is also BIBO stable. However,
if the linear system is BIBO stable, it does not guarantee asymptotic stability.
14. Under what condition, BIBO stability and asymptotic stability are equiva-
lent?
In the overall T.F., if there is no pole–zero cancellation in RHP, then BIBO sta-
bility is equivalent to asymptotic stability. The above condition implies that the
given system should be controllable and observable.
612 4 Stability Analysis of Linear Control System

15. The T.F. of a linear system has repeated roots in the jω axis. Is the system
marginally stable or unstable?
If the linear system has repeated roots in the jω axis, the output grows indefinitely
with respect to time. Hence, the system is unstable.
16. The T.F. of a linear system has a pole at the origin of s-plane. How the sta-
bility is termed?
If the pole at the origin is intentionally introduced, the system is said to be stable.
Otherwise, it is termed as marginally stable.
17. What is relative stability of a linear system?
If the linear system is stable, it is said to be absolutely stable. How far it is closer
to the stability limit is called relative stability.
18. What are the measures of the relative stability?
In the frequency domain, the relative stability is measured in terms of phase mar-
gin and gain margin. Higher the phase margin and gain margin, higher the relative
stability. In the time domain, it is measured by the root location in LHP. If the near-
est root in LHP is placed far away from the jω axis, the relative stability is high.
19. What is a conditionally stable system?
A linear system may be stable for a range of values of a system parameter.
For other values, the same system will be unstable. Such systems are called
conditionally stable systems.
20. Mention any three applications of Routh’s–Hurwitz stability criterion?
Routh’s–Hurwitz criterion can be applied to determine the absolute stability,
relative stability and the system gain that can be increased before the system
becomes unstable. It can be used to find out the number of roots of the character-
istic polynomial in RHP, LHP and the jω axis. In few cases, the exact locations
of the characteristic roots in s-plane can be determined.
21. Mention any two disadvantages of Routh–Hurwitz method.
Routh–Hurwitz criterion is applicable only to characteristic polynomial which
are real. If the coefficients are complex, the method is not applicable. The jω axis
is considered as the stability boundary for Routh–Hurwitz criterion. For other
boundaries, it is not applicable. When repeated roots are present in the jω axis, it
cannot be found using Routh–Hurwitz criterion and the erroneous conclusion that
the system is marginally stable is derived, whereas the system is actually unstable.
22. What are the necessary conditions to apply Routh–Hurwitz stability condi-
tion?
To apply Routh–Hurwitz stability criterion, it is necessary that all the coefficients
of the characteristic polynomial are positive real and no coefficient should be
zero.
4.4 Nyquist Stability Criterion 613

23. State Routh–Hurwitz criterion.

The Routh–Hurwitz criterion states that the number of roots of the characteristic
equation with +ve real parts is equal to the number of changes in sign of first
column of the Routh array.
24. For simplifying the numerical calculation, in the Routh array what method
is followed?
For simplifying the numerical calculations, the entire row may be multiplied or
divided by a positive number by which the stability condition is unaffected.
25. What are the difficulties encountered when Routh’s array is prepared?
When Routh’s array is prepared, the coefficients of complete array becomes
zero and the Routh’s array is broken. Routh’s array is also broken when the first
column alone becomes zero.
26. When the complete row becomes zero, how the Routh’s array is completed?
When the complete row has zero coefficients in all the columns of that row, the
equation of the previous row is considered and is taken as the auxiliary equation.
The auxiliary equation is differentiated with respect to s, and the coefficients of
this equations are taken for the row which has zeros in all the columns.
27. What are the properties of the auxiliary equation of even polynomial?
The even polynomial of auxiliary equation has the following properties:
1. The roots of the even polynomial are symmetric about the origin and real.
2. They could be on the jω axis as simple roots or repeated roots.
3. The roots are quadrantal (complex roots symmetrically placed with respect to
the jω axis).

28. When the complete row becomes zero in the Routh’s array, does it imply
the presence of roots in the jω axis?
When the coefficients of a complete row becomes zero, the even polynomial
may have any one property mentioned in Question 27. If the even polynomial
possesses property one or property three, only symmetrically placed real and
complex roots exist both in RHP as well as in LHP. In these cases, there won’t
be any root in the jω axis.
29. If the characteristic polynomial has jω axis roots, when the Routh’s array
is formed will it have complete row with zero coefficients?
If the characteristic polynomial has jω axis roots, Routh’s array will have a
complete row with zero coefficients.
30. How stability is interpreted when complete row becomes zero?
The Routh’s array is completed as explained in Question 26. The array is divided
in to two parts: one below the auxiliary equation and the other above the auxil-
iary equation. For the rows of even (auxiliary) polynomial, the number of sign
changes is determined which gives RHP roots. In this polynomial, equal number
of LHP roots exist. The remaining roots are in the jω axis. For the other poly-
nomial, which is above the even polynomial, the number of sign changes are
determined and the RHP roots found. The remaining roots of this polynomial
are in LHP. There won’t be any root in the jω axis for the other polynomial.
614 4 Stability Analysis of Linear Control System

Thus, for the whole characteristic polynomial, LHP, RHP and jω axis roots are
determined. For the system to be stable, there should not any RHP root.
31. For a system with complete row becoming zero, there is no sign change in
the first column of Routh array. Does it mean the system is stable?
No. When the rows of even polynomial are interpreted, it is to be ensured that
there is no root being repeated more than once in jω axis. If there are repeated
roots in the jω axis, the system is unstable.
32. How to identify the presence of repeated roots in the jω axis?
The auxiliary equation is a factor of the characteristic polynomial. By long
division of the characteristic polynomial with the auxiliary polynomial, the exact
root locations can be found out not only in the jω axis, but also in RHP and LHP.
From this, we can interpret whether there exits repeated roots in the jω axis.
33. If the first column of Routh array becomes zero, what are the methods
available to complete the Routh’s array?
When the first column becomes zero, the Routh’s array is broken. The Routh’s
array is completed using any one of the following methods:

1. By forming the reciprocal of the given polynomial.

2. A small value ε (epsilon) is assigned to replace the zero.
3. The given polynomial is multiplied by a factor (s + a) where a > 0 and the
Routh’s array is formed for the new polynomial.

34. If the Routh’s array does not have the coefficients of complete row becoming
zero, how the number of roots in RHP, LHP and jω axis are determined?
For the given characteristic polynomial the Routh’s array is formed. The number
of sign changes in the first column of Routh’s array gives the number of roots in
RHP. The characteristic polynomial F(−s) is formed from F(s) and the Routh’s
array is formed. The number of sign changes in the first column, now gives the
number of roots of F(s) in LHP. By subtracting LHP and RHP root number
from the order of the polynomial of F(s), the number of roots in the jω axis is
determined.
35. How to find relative stability from Routh’s array?
Suppose the required relative stability unit is −σ . In the characteristic polynomial
F(s), s is replaced as (s − σ ) and the new characteristic polynomial is formed
and the Routh’s array is prepared. If there is no sign change in the first column
of Routh’s array, the system possesses the given relative stability limit.
36. Using Routh–Hurwitz criterion, the critical gain of the system K is found
before the system becomes unstable. If K is found to be complex, how the
stability is decided?
If the gain K of the system is found to be complex using Routh’s stability
criterion, the system is unstable.
37. What is the relationship between the poles of open loop T.F. G H (s) and the
poles of the characteristic equation 1 + G H (s)?
The poles of open loop T.F. and the poles of the characteristic equation are one
and the same.
4.4 Nyquist Stability Criterion 615

38. How the zeros of the characteristic equation 1 + G H (s) are related to the
poles of closed loop T.F.?
The zeros of the characteristic equation 1 + G H (s) are the same as the poles of
closed loop T.F.
39. What is the necessary and sufficient condition for the closed loop system to
be stable?
The necessary and sufficient condition for the closed loop system to be stable,
the poles of the closed loop system (zeros of the characteristic equation) should
be placed in LHP of the s-plane.
40. What do you understand by s-plane contour?
The connection of points in the complex s-plane is called s-plane contour.
41. What do you understand by a point being enclosed and being encircled?
A point in the complex plane is considered to be the point enclosed by the contour
if it lies to the right of the traversal. A point is said to be encircled by a closed
path if it is found inside the path (to the right) during the traversal.
42. There are P poles and Z zeros inside a contour in the s-plane. For clockwise
traversal of the s-plane contour, how many encirclements around the origin
are made when mapped into F(s) plane in the following cases?

1. P = 5; Z = 3
2. P = 3; Z = 5
3. P = 3; Z = 3

1. Two encirclements in the counter clockwise direction.

2. Two encirclements in the clockwise direction.
3. No encirclement.

43. On what principle, the Nyquist stability criterion is based upon?

Nyquist stability criterion is based upon the conformal mapping and the principle
of the argument which are derived by the application of Cauchy’s theorem.
44. What is Nyquist stability criterion?
Nyquist stability criterion is stated in two forms:
i Open loop system is stable: If the open loop system is stable, then for the
closed loop system also to be stable, the s-plane contour when mapped into
G H -plane should not at all encircle (−1, 0) point.
ii The open loop system is unstable: If the open loop system is unstable then
for the closed loop system to be stable, the s-plane Nyquist contour when
mapped into G H -plane should encircle (−1, 0) point as many times as there
are number of open loop poles in RHP and the encirclement should be in
the counter clockwise direction of the s-plane contour.

45. How mathematically Nyquist stability condition is expressed?

Nyquist stability condition is mathematically expressed as follows:

Z r = Nr + Pr
616 4 Stability Analysis of Linear Control System

where

Z r = Number of zeros of the characteristic equation located in RHP.

Pr is the number of poles of the open loop T.F. located in RHP.
Nr is the number of counter clockwise encirclements of the (−1, 0) point when
the s-plane clockwise Nyquist contour is mapped into G H -plane.

For the closed loop system to be stable, Z r = 0.

46. What is Nyquist contour and what is Nyquist path?
The s-plane closed contour which includes the +ve and −ve jω axes and the
entire right half s-plane is called Nyquist contour. Such a contour should avoid
any pole or zero in the jω axis by-passing it with a small circle in RHP to satisfy
Cauchy’s theorem. This Nyquist contour is called Nyquist path.
47. Why Nyquist contour encloses the entire RHP and not the LHP?
The Nyquist contour encloses the entire RHP in search of whether there is any
zero of the characteristic equation 1 + G H (s) = 0 is present in RHP. As far as
stability is concerned, it is immaterial whether there is any zero of 1 + G H (s) =
0 is in LHP or not. That is why Nyquist contour does not cover LHP.
48. A certain system has the following poles of G H (s) s = ± j4 and s = 0. How
Nyquist contour bypasses these poles?

1. For s = j4, the contour is s = j4 + r e jθ where r → 0 and −π/2 ≤ θ ≤ π/2.

2. For s = − j4, the contour is s = − j4 + r e jθ where r → 0 and −π/2 ≤ θ ≤
π/2.
3. For s = 0, the contour is s = r e jθ where r → 0 and −π/2 ≤ θ ≤ π/2.

49. What is Nyquist plot or Nyquist diagram? Why is it called a frequency

response method?
The s-plane contour when mapped into G H -plane, the contour so obtained is
called Nyquist plot or Nyquist diagram. Since the part of the Nyquist diagram
is nothing but the polar plot of G H ( jω), it is a frequency response method.
50. For a stable open loop system, how Nyquist criterion is simplified?
If the open loop system is stable, then for the closed loop system also to be stable,
the Nyquist plot should not encircle (−1, 0) point.
51. For an unstable open loop system, how Nyquist criterion is simplified?
If the open loop system is unstable, then for a closed loop system to be stable,
the Nyquist plot should encircle (−1, 0) as many times as these are number of
open loop poles in RHP and the number of encirclement should be in the counter
clockwise direction.
52. How an oscillatory system is identified when Nyquist stability criterion is
applied?
If the Nyquist plot passes through (−1.0) point, the closed loop system gives
sustained oscillations. Such systems as per Nyquist criterion, are oscillatory
systems.
4.4 Nyquist Stability Criterion 617

53. What does Nyquist criterion tell us?

By applying Nyquist criterion, we get the information about the number of zeros
of the characteristic equation that are in RHP. For the closed loop system to be
stable, there should not be any zero of the characteristic equation in RHP.
54. What is the advantage of having the Nyquist plot in the G H -plane rather
than (1 + G H )-plane?
A part of Nyquist plot requires polar plot of G H ( jω) or 1 + G H ( jω). Plotting
of G H ( jω) for the Nyquist path is much easier than plotting of 1 + G H ( jω).
Sometimes one can experimentally get G H ( jω). That is why plotting of
G H ( jω) and counting the number of encirclements around (−1, 0) in the G H -
plane is preferred rather than plotting 1 + G H ( jω) and counting the encir-
clements around the origin.

Long Answer Type Questions

1. The characteristic equation of a feedback control system is

s 4 + 20s 3 + 15s 2 + 2s + K = 0

Determine the range of values of K for the system to be stable. Can the system
be marginally stable? if so, find the required value of K and the frequency of
sustained oscillations. (Anna University, May, 2011)
Ans:
1. For 0 < K < 1.49 the system is stable.
2. For K = 1.49, the system is marginally stable. The frequency of sustained
oscillation is ω = 0.316 rad/sec.
2. A certain system has the following characteristic equation

F(s) = s 6 + 3s 5 + 4s 4 + 6s 3 + 5s 2 + 3s + 2

Examine the stability of the system using Routh–Hurwitz criterion.

(Anna University, December, 2010).

Ans: There are 2 LHP roots and 4 roots in jω axis. They are repeated twice at
s = + j and s = − j. The system is unstable even though there is no sign change
in the first column of Routh’s array.
3. Find the range of values of K so that the system with the characteristic equation

s(s 2 + s + 1)(s + 4) + K = 0

will be stable using Routh’s criterion. (Anna University, May, 2009).

Ans: 0 < K < 3.36

618 4 Stability Analysis of Linear Control System

4. For a system with

F(s) = s 4 + 22s 3 + 10s 2 + s + K

obtain the marginal value of K and the frequency of oscillations at that value of
K. (Anna University, May, 2007).

Ans: K = 0.452; ω = 0.2132 rad/sec.

5. Test the stability using Routh’s criterion for the polynomial

s 4 + 2s 3 + 3s 2 + 4s + 5 = 0

Determine RHP, LHP and jω axis roots.

Ans: RHP roots =2; LHP roots=2; jω axis roots=nil

6. Determine the range of K for the system to be stable if the characteristic equation
is

s 4 + 3s 3 + 3s 2 + 2s + K = 0

Ans: K < 1.555

7. For the characteristic equation

s 5 + 2s 4 + 24s 3 + 48s 2 − 25 − 50 = 0

Perform complete Routh stability analysis. (Anna University, May, 2007)

Ans: There are 1 RHP root, 2 LHP roots and two jω axis roots. The roots in the
jω axis are distinct and the system is marginally stable.
8. Using Routh’s criterion, discuss the stability of the system whose characteristic
equation is

s 3 + 10s 2 + 50s + 500 = 0

Ans: 1 LHP root; 2 jω axis roots. The system is marginally stable. No sign
change in the first column of Routh’s array.
9. Test their stability of the system using Routh–Hurwitz criterion. Also, comment
on the location of roots for the systems whose characteristic equations are

(a) s 6 + 3s 5 + 2s 4 + 9s 3 + 5s 2 + 12s + 20 = 0
(b) s 5 + s 4 + 2s 3 + 3s + 5 = 0

(Anna University, December, 2009).

Ans: (a) Unstable. RHP roots=2; LHP=4 and (b) Unstable. RHP roots=2; LHP
roots =3.
4.4 Nyquist Stability Criterion 619

Im
GH-plane
Mirror image

0 ( 1, 0)
0 Re

GH(j ) plot Pr 1

Nr 1

Fig. 4.44 Nyquist diagram for Question 11

10. Consider the following characteristic equation

s 4 + 6s 2 − 16 = 0

Using Routh–Hurwitz criterion, determine the stability of the system and also
the locations of RHP, LHP and jω axis roots.
Ans: system√ is unstable.
√ RHP root=1;
√ LHP root= 1; √ jω axis roots =2. The roots
are at s = 2, s = − 2, s = + j 8 and s = − j 8.
11. The Nyquist plot of G H (s) is shown in Fig. 4.44. G H (s) has one pole in RHP.
Complete the Nyquist diagram and determine the stability of the closed loop
system.

Ans: Pr = 1; Nr = −1; Z r = 1 − 1 = 0. The closed loop system is stable.

12. The loop T.F. of a certain feedback control system is given by
s
G H (s) =
(s + 200)3

Sketch the Nyquist plot and determine the stability of the closed loop system.

Ans: Pr = 0; Nr = 0; Z r = 0 + 0 = 0. The closed loop system is stable.

13. The loop T.F. of a certain feedback control system is given as

K s2
G H (s) =
(s 3 + 5s + 25)
620 4 Stability Analysis of Linear Control System

Using Nyquist criterion, determine the value of K so that the closed loop system
is stable.

Ans: For the given system Pr = 2; Nr should be −2 for Z r = 0. For this, K > 5.
For K > 5, the closed loop system is stable.
14. The forward path T.F. of a certain unity feedback system is given as

K (2s + 1)(s + 1)
G(s) =
(5s + 1)(s − 1)

Using Nyquist stability criterion, determine the value of K so that the closed
loop system is stable.

Ans: Pr = 1; For K > 1.33, Nr = −1; Z r = 0. The closed loop system is stable
for K > 43 .
15. A certain feedback control system has the following loop T.F

K (s + 3)(s + 2)
G H (s) =
(1 + 5s)(s − 2)

By applying the Nyquist stability criterion, determine the value of K so that the
closed loop system is stable.

Ans: Pr = 1; Nr should be −1 so that Z r = 0. For this, K > 1.8.

16. A certain control system has the following forward path T.F

K (s + 3)(s + 5)
G H (s) =
(s − 2)(s − 4)

By applying Nyquist stability criterion, determine the value of K so that the

closed loop system is stable.

Ans: For the given system, Pr = 2; For Z r to be zero Nr should be −2. This is
possible if 43 K > 1 or K > 0.75.
17. Consider the following loop T.F. of a certain closed loop system

K
G H (s) =
(1 + s)(1 + 2s)(1 + 3s)

Draw the Nyquist plot and determine the value of K so that the closed loop
system is stable.

Ans: K < 10 for the closed loop system to be stable. For the given G H (s), Pr =
0. For the closed loop system to be stable Nr = 0. For this, at ω pc , −0.1K should
be to the right of (−1, 0) point. Hence, K < 10.
4.4 Nyquist Stability Criterion 621

18. Consider the following loop T.F. of a certain closed loop system

K (6s − 1)
G H (s) =
(s 2 + 2s + 2)

Draw the Nyquist plot and determine the value of K so that the closed loop
system is stable.

Ans: Pr = 0; Nr should be zero for the closed loop system to be stable. K < 2
for the system to be stable.
19. A certain feedback control system has the following open loop T.F.

K (s − 2)2
G H (s) =
(s + 2)2 (s + 4)2

Draw the Nyquist plot and determine the value of K so the closed loop system
is stable.

Ans: K < 18 for the closed loop system to be stable. Since Pr = 0, Nr should
be zero. This implies K < 18.
20. A certain closed loop system has the following open loop T.F.

K (s + 2)
G H (s) =
(s 3 + 8s 2 − 10)

Sketch the Nyquist diagram and determine the value of K so that the closed loop
system is stable.

Ans: Pr = 1; Nr should be −1 so that Z r = 0. For this, 0.2K > 1 or K > 5 for

the closed loop system to be stable.
21. A certain feedback control system has the following loop T.F

K
G H (s) =
(s 2 − 16)(s + 2)

Investigate the stability of the closed loop system, using Nyquist stability crite-
rion.
1. Pr = 1; If K > 32, Nr = 1 and Z r = 2. The system is unstable.
2. Pr = 1; If K < 32, Nr = 0 and Z r = 1. The system is unstable.
622 4 Stability Analysis of Linear Control System

Im

Mirror image
of GH(j )

( 1,0) 0 0
Re

GH(j ) plot

Fig. 4.45 Nyquist diagram for Question 12

Nyquist Diagram
GH(s) Ks2/(s3 5s 25)
0.4

GH(j ) plot
0.3

0.2

( 1,0)
0.1
Imaginary axis

pc 2.24
( 1,0) 0
0
0
0.2K
0.1 Mirror image of
GH(j )
0.2

0.3 Nr 0
Nr 2
0.4
0.4 35 0.3 0.25 0.2 0.15 0.1 0.05 0 0.05 0.01
Real axis

Fig. 4.46 Nyquist diagram for Question 13

4.4 Nyquist Stability Criterion 623

Nyquist plot of
G(s) K(2s 1)(s 1)/(5s 1)(s 1)
0.5

0.4

0.3 Mirror image

of GH(j )
0.2
1
pc 4.78
Imaginary axis

0.1
0 ( 1, 0)
0 0
0.1 3K 0.4K
K
4
0.2
GH(j ) plot
0.3

0.4 Nr 1
0.5
1 0.8 0.6 0.4 0.2 0 0.2 0.4
Real axis

Fig. 4.47 Nyquist diagram for Question 14

624 4 Stability Analysis of Linear Control System

Nyquist diagram
G(s) K(s 3)(s 2)/(1 5s)(s 2)
1.5
GH(j ) plot

0.5 pc 1.138
5K
Imaginary axis

9
0 ( 1, 0) 0.2K
0
0
3K

0.5

1
Nr 1
Mirror image
of GH(j )
1.5
3.5 3 2.5 2 1.5 1 0.5 0 0.5
Real axis

Fig. 4.48 Nyquist diagram for Question 15

Im

GH(j ) plot

15
K
4K 8
3 ( 1,0)
0
K 0
Re
pc 3.3166

Mirror
image

Nr 2

Fig. 4.49 Nyquist diagram for Question 16

4.4 Nyquist Stability Criterion 625

Im

0.1 K
( 1,0)
0
K 0 Re
pc 1

Nr 2

Fig. 4.50 Nyquist diagram for Question 17

Im
GH(j ) plot

0 K/2 ( 1,0) 7/3

0 3 Re

Mirror image of
GH(j ) plot
Nr 1

Fig. 4.51 Nyquist diagram for Question 18

626 4 Stability Analysis of Linear Control System

Im

Mirror image
of GH(j )

K
18 ( 1,0) 0
0
1.4 Re
pc
K
16

GH(j ) plot
Nr 2

Fig. 4.52 Nyquist diagram for Question 19

Im
Mirror image
of GH(j )

0.2 K
0 ( 1,0)
0 Re

GH(j ) plot
Nr 1

Fig. 4.53 Nyquist diagram for Question 20

4.4 Nyquist Stability Criterion 627

Im

GH(j ) plot
K 32
( 1, 0)
0 ( 1, 0)
0 Re

Nr 0 Mirror image
of GH(j ) plot
Nr 1

Fig. 4.54 Nyquist diagram for Question 21

Im
Mirror image
of GH(j )

10 K ( 1, 0) 0
0 Re

K8
GH(j ) plot

Nr 1

Fig. 4.55 Nyquist plot for Question 22

628 4 Stability Analysis of Linear Control System

Mirror image of GH(j )

Im

GH-plane
GH(j )

K/50
( 1,0) 0
0 Re
pc 1

GH(j ) plot

Nr 2

Fig. 4.56 Nyquist plot for Question 23

22. A certain closed loop T.F. has the following T.F. which is strictly proper. Using
Nyquist stability criterion determine the value of K so that the closed loop system
is stable

K (1 − 10s 2 )
G H (s) =
(s + 2)(s + 4)

K < 0.1, the system is stable.

K = 0.1, the system is oscillatory.
K > 0.1, the system is unstable.
4.4 Nyquist Stability Criterion 629

23. A certain feedback control system has the following open loop T.F

K s3
G H (s) =
(s + 2)2 (s + 3)2

Determine the value of K so that the closed loop system is stable (Figs. 4.45,
4.46, 4.47, 4.48, 4.49, 4.50, 4.51, 4.52, 4.53, 4.54, 4.55, 4.56).

K < 50, the system is stable.

K = 50, the system is oscillatory.
K > 50, the system is unstable.
Chapter 5
Root Locus Method for Analysis

Learning Objectives

After completing this chapter, you should be able to

 Understand the root locus concept.

 Understand the magnitude and angle criteria of root locus.
 Understand the step-by-step procedure and the rules to draw the root locus.
 Draw the root locus for variation of system parameters.
 Determine the closed loop roots for the given gain K .
 Determine transient response for the desired damping.
 Determine the phase margin and gain margin from root locus.
 Draw the root locus for the negative value of gain.
 Draw the root locus for the positive feedback system.
 Study the effect of the addition of poles and zeros.

5.1 Introduction

In the previous chapters, stability and relative stability of linear systems were dis-
cussed in the time and frequency domains. In the time domain approach, it was
seen that the system stability and relative stability can be improved by adjusting the
gain of the system. It was further established that a linear time invariant system is
said to be absolutely stable if all the roots of the characteristic equation are in LHP.
Furthermore that if these poles in the LHP are placed farther from the imaginary

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 631
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_5
632 5 Root Locus Method for Analysis

axis of the complex s-plane, the relative stability of the system increases. In the fre-
quency domain, the absolute stability of the system was discussed using the Nyquist
stability criterion. Using Bode plots and Nichols chart, the relative stability of the
system was discussed. However, these methods do not give information about the
exact root locations of the closed loop system which give useful information about
time domain specifications. The trajectories of the roots of the characteristic equation
when investigated carefully as the gain of the system is varied give all information
about absolute and relative stabilities like the other methods discussed in previous
chapters. In addition, the method gives the exact root locations of the closed loop
system. The method is called the root locus method, and the trajectories of the closed
loop poles in the s-plane are called root loci. The root locus is the path of the
characteristic equation traced out in the s-plane as the system parameter is
varied. The method developed by W. R. Evans in 1948 is a very powerful method
and extensively used in control engineering analysis and design. The root locus is a
way of representing graphical information about system behaviour. It has the virtue
of being a good design tool for continuous time systems when we work in the s-plane
and discrete time system when we work in the z-plane. The concept of root locus is
very simple enough, but the problem is that the closed loop poles can exhibit many
strange and wonderful movements in the s-plane when the system is even a little bit
complex.
A root locus plot is simply a plot of the s zeros and the s poles on a graph with
real and imaginary coordinates. It represents a curve of the location of the poles
of a transfer function as some parameter is varied. The locus of the roots of the
characteristic equation of the closed loop system as the gain varies from −∞ to ∞
gives the name of the method. Such a plot shows clearly the contribution of each open
loop pole or zero to the locations of closed loop poles. General rules to construct the
root locus exist and if the designer follows them, sketching of the root loci becomes
a simple matter. However, a simple MATLAB program is also available to draw
the root locus very accurately, and all necessary information can be easily obtained
from the graph so drawn. The information include absolute stability, relative stability,
locations of closed loop poles and time and some frequency domain specifications.
Thus, root locus when used has the problem of plenty but without plenty of
problems.

5.1.1 Advantages of Root Locus Method

1. Absolute and relative stabilities can be determined using root locus.

2. For the given system, the gain of the system can be determined for the system to be
marginally stable. The corresponding frequency of oscillation can be determined.
3. For a given value of the system gain K , the closed loop pole locations can be
determined. Similarly, for the required damping ratio of the dominant closed
loop system, the closed loop pole locations can be determined. From this, the
5.1 Introduction 633

time response of the closed loop system for the given input can be obtained.
Further, it is possible to determine the closed loop frequency response also.
4. For the given gain of the system, time domain specifications can be determined.
5. From the root locus, the frequency domain specifications such as phase margin
and gain margin can be obtained, which are the measures of relative stability.
6. The root locus shows clearly the contribution of each open loop pole or zero to
the locations of the closed loop poles which ultimately decide the absolute and
relative stabilities and also give information about the time and frequency domain
specifications.
7. Like other methods, a very simple MATLAB program is available to draw root
locus which makes the drawing of root loci simpler. That is why root locus is
extensively used in the design of compensators in the time domain.

5.2 The Concept of Root Locus

Consider the closed loop system shown in Fig. 5.1. The closed loop T.F. is obtained
as
C(s) K G(s)
= (5.1)
R(s) 1 + K G(s)

where K is a real constant and G(s) is a function of s with numerator and denominator
polynomials of any order. The characteristic equation of Eq. (5.1) is

F(s) = 1 + K G(s) = 0 (5.2)

K G(s) = −1
|K G(s)| = 1 (5.3)
∠K G(s) = 180◦ + n × 360◦ (5.4)

where n = 0, ±1, ±2, ± . . ..

The root locus is the path of the roots of the characteristic equation that moves
in the complex s-plane as the system parameter varies in the range −∞ ≤ K ≤
∞.

Fig. 5.1 Closed loop system R(s) C(s)

with variable parameter K K G(s)
634 5 Root Locus Method for Analysis

Table 5.1 Variation of roots with respect to gain K

K 0 2 4 6 8 ∞
s1 0 −0.586 −2 −2 + j2 −2 + j4 −2 + j∞
s2 −4 −3.414 −2 −2 − j2 −2 − j4 −2 − j∞

At present, we restrict our discussion by allowing K to vary in the range 0 ≤

K ≤ ∞. The root locus should always satisfy Eqs. (5.3) and (5.4), where Eq. (5.3)
is called magnitude criterion and Eq. (5.4) is called angle criterion. Let us consider
the following simple example to trace the root locus as K is varied from 0 to ∞. It
is evident from Eqs. (5.4) and (5.3) that angle criterion is used to trace the root locus
while magnitude criterion is used to calibrate the root locus in terms of K .
The angle criterion states that for any point s1 in the s-plane to be a point on the
root locus, the sum of the angles of the vectors drawn from all the zeros of G H (s)
minus the sum of the angles of the vectors drawn from all the poles of G H (s) to the
point s1 should be an odd multiple of 180◦ .
According to the magnitude condition, at any point s1 in the s-plane where root
locus exists, the value of K is calculated as

Product of the lengths of the vectors drawn from the poles of G H (s) to s1
K =
Product of the lengths of the vectors drawn from the zeros of G H (s) to s1

Now consider the following open loop T.F

K
K G(s) =
s(s + 4)
F(s) = s 2 + 4s + K = 0 (5.5)

Equation (5.5) is solved for the two roots as

√
s1 = −2 + 4 − K
√
s2 = −2 − 4 − K

For a few values of K , the roots s1 and s2 as given by the above equation are calculated
and are shown in Table 5.1.
The poles of G H (s) are at s = 0 and s = −4. From Table 5.1, for K = 0, s1 = 0
and s2 = −4. Thus, the root locus starts from the poles of G(s) and the value of K at
these points is zero. As K increases, s1 moves towards the left and s2 moves towards
the right on the negative real axis. For K = 4, s1 = −2 and s2 = −2. The root loci
originating from the respective poles meet each other at s = −2 and break away.
For further increases of K , the roots are complex conjugate with real part remaining
constant at −2 and the imaginary part increasing. As K tends to ∞, the two root loci
also tend to ∞. The root locus plot of Table 5.1 is shown in Fig. 5.2.
5.2 The Concept of Root Locus 635

K 2 j
j
K 8 2 j4

s1 locus
s-plane
K 6 2 j2
Break
K 0 away
point K 0
K 4 2 0
4
K 6 2 j2

s2 locus
K 8 2 j4
K 2 j

K
Fig. 5.2 Root locus for G(s) =
s(s + 4)

From Fig. 5.2, we observe that

1. The root loci start from poles of G(s) where K = 0.
2. The root loci terminate at ∞ and −∞ as K → ∞ in the absence of zeros of
G(s).
3. There are as many root loci as the order of the system.
4. At any point on the root locus, Eqs. (5.3) and (5.4) are satisfied.
The above procedure is presented to illustrate the concept of root locus. However,
for complex systems, where G(s) has many poles and zeros, it is more or less impos-
sible to draw the root locus. Further, using Eqs. (5.3) and (5.4) alone, one cannot draw
the root locus. The path of the root is decided by the angle criterion of Eq. (5.4). It is
extremely difficult to search for a point in the vast area complex s-plane which will
satisfy the angle criterion of root locus. To make the drawing of root locus simpler,
a set procedure is available and they are to be followed. The procedure includes the
properties of root locus or the rules of root locus. They are discussed below with
illustrations.

A Historical Note: Walter R. Evans (1920–1999)

Walter Richard Evans a noted American control
theorist was born on 15 January 1920, in St.
Louis, Missouri. Young Evan wanted to become
an engineer like his father. He received his B.E.
Degree in Electrical Engineering from Washington
University in St. Louis, Missouri, in 1941 and
his M.S. in Electrical Engineering from the Uni-
versity of California, Los Angeles, in 1951. He
started his career as a trainee in General Elec-
636 5 Root Locus Method for Analysis

tric Company in 1944 and worked as an instructor at Washington Univer-

sity from 1946 to 1948. Evans’ principal contribution to the automatic control
theory was his invention of the Evans Root Locus Method in 1948. Evans’ root locus
method has widespread application in the design of control systems. Now, root locus
is included as a standard chapter in feedback control systems textbooks. Evans had
worked as an engineer in several companies including General Electric Rockwell
International and Ford Automatic Company. He published a book titled “Control
System Dynamics” with McGraw Hill in 1954. He was awarded the Prestigious
Rufus Oldenburger Medal in 1967 and Richard E. Bellman Control Heritage Award
in 1988 in recognition of his outstanding contributions in the fields of automatic
control theory. Although a genius, he never looked down on others. His ambition in
life was to teach, instruct, be a model and serve. Evans was married to Arline and
the couple had four children. The world lost a great technological genius on 10 July
1999.

5.3 Properties of the Root Loci (Rules of the Root Loci)

The construction of root locus manually is made simpler using the properties of
the root loci. These properties are also called rules of root loci. Using these rules,
certain key points in the s-plane are indentified. These points lie on the root loci. The
connection of these points gives approximate or asymptotic root loci. However, by
smoothening the asymptotic plots, root loci which are closer to the actual root loci
are drawn. These rules are discussed below.
Rule 1: The Starting Points of the Root Loci
Consider the characteristic equation given in Eq. (5.2)

1 + K G H (s) = 0
1
G H (s) = − (5.6)
K
As the magnitude K approaches zero, |G H (s)| = ∞. This implies that s must
approach the poles of G H (s). Since

(s + z 1 )(s + z 2 ) · · · (s + z m )
G H (s) = (5.7)
(s + p1 )(s + p2 ) · · · (s + pn )

the root loci start from the poles of G H (s) where the value of K = 0.
Rule 2: The Ending Points of Root Loci
Starting from the poles of G H (s), as the value of K increases, the root loci move
in the s-plane. For K = ∞, Eq. (5.6) becomes zero. This is possible if s approaches
the zeros of G H (s) or ∞ of the s-plane.
5.3 Properties of the Root Loci (Rules of the Root Loci) 637

j s-plane
K K 0 K K 0 K K 0
0
4 3 2 1 1

Locus 3 Locus 2 Locus 1

Fig. 5.3 Root locus showing the number of root loci, starting and ending parts and the value of K

The root loci end at the zeros of G H (s) or at ∞ where the value of K = ∞.
From Fig. 5.1, it can be seen that the two root loci start at s = 0 and s = −4 where
the value of K = 0. Since there are no zeros of G H (s), they all end at ∞ where
K = ∞.
Rule 3: Number of Root Loci of 1+GH(s)
The number of root loci of the characteristic equation 1 + G H (s) = 0 is equal to
the order of the polynomial of Eq. (5.7). Consider the following loop T.F.

K (s + 1)(s + 3)
G H (s) =
(s − 1)(s + 2)(s + 4)

The numerator polynomial is of order two while the denominator polynomial is of

order three. Hence, there should be three root loci as shown in Fig. 5.3.
Rule 4: Symmetry of the Root Loci
The root loci are symmetrical with respect to the real axis of the s-plane. The rule is
derived based on the property that for real coefficients of characteristic polynomial,
the roots must be real or in complex conjugate form. Consider the following loop
T.F
K
G H (s) =
s(s + 4)(s + 2)

The root locus is shown in Fig. 5.4. Loci 1 and 2 are symmetrical with complex
conjugate root loci with respect to the real axis.
Rule 5: Angle of the Asymptote of the Root Loci
If there are P number of poles and Z number of zeros of G H (s), and if P = Z , all
the P root loci originating from their respective poles will end at the zeros where
the value of K = ∞. However, if P > Z , then (P − Z ) loci will end at ∞ where
K = ∞. The root loci near ∞ in the s-plane are described by the asymptotes of
the root loci. They are giving guidance to the actual root locus. The angles of the
asymptotes are given by

(2n + 1) × 180◦
θ = (5.8)
(P − Z )
638 5 Root Locus Method for Analysis

Asymptote j
Symmetry
K

s-plane

1
K K 0 K 0 60º K 0
3 2 0
4 2 Break 0.667
60º away
point

Symmetry
Asymptote K

Fig. 5.4 Root locus which illustrates complex symmetry

where
P = Finite number of poles of G H (s).
Z = Finite number of zeros of G H (s).
n = 0, 1, 2, . . . , (P − Z − 1).
For the example illustrated in Fig. 5.3, the angles of the asymptotes are θ = 60◦ , 180◦
and −60◦ .
Rule 6: Intersect of the Asymptote (Centroid)
The intersect of the asymptotes of the root loci lies on the real axis of the s-plane at
 
real part of poles of G H (s) − real part of zeros of G H (s)
σ = (5.9)
(P − Z )

The intersect of the asymptotes σ is the centre of gravity of the root loci, and hence
it is also called centroid. The value of σ is always a real number. For the example
illustrated in Fig. 5.3, the centroid is

0−2−4−0
σ =
3
= −2

Rule 7: Root Loci on the Real Axis

At any point s1 on the real axis, the angles of the vectors drawn from the complex
poles and zeros of G H (s) add up to zero. Only the poles and zeros of G H (s) on the
real axis contribute angle to any point s1 in the real axis. The angle contributions by
the poles and zeros of G H (s) to the left of the point s1 is zero. However, each pole
to the right of s1 will contribute −180◦ , while each zero of G H (s) to the right of
5.3 Properties of the Root Loci (Rules of the Root Loci) 639

the point s1 will contribute +180◦ (+180◦ and −180◦ are same). Thus, to satisfy the
angle criterion of root locus, given by Eq. (5.4), any point s1 on the real axis to be a
point on the root locus, the total number of poles and zeros of G H (s) to the right of
that point should be odd. Three poles and one zero will make a total number 4 which
is even. Two poles and one zero will make the total number 3 which is odd.
Consider the root locus drawn in Fig. 5.3. Consider any point s1 between s = −1
and +1. To the right of this point, one pole at s = 1 exists which is an odd number.
Therefore, root locus on the real axis exists between s = 1 and s = −1. Now consider
any point s1 between s = −1 and s = −2 on the real axis. To the right of s1 even
number of zeros and poles exists between s = −1 and s = −2 on the real axis, and
there won’t be any root locus.
Rule 8: Angle of Departure of the Root Locus
The angle of departure of root locus at a pole of G H (s) is nothing but the angle of
the tangent to the locus near the pole. The angle of departure is given by

θd = (2n + 1) × 180◦ − (θ P − θ Z ) (5.10)

where θ P and θ Z are the net angle contribution to the concerned pole by all other
poles and zeros of G H (s), respectively. This is illustrated in Fig. 5.4 for the root
locus drawn for the loop T.F.

K (s + 6)
G H (s) =
(s + 3)(s 2 + 2s + 2)

The root locus plot is shown in Fig. 5.5. The root loci from s = −1 + j and
s = −1 − j depart with a curved line. The angle of departure at s = −1 + j is
calculated by drawing vectors from poles and zeros of G H (s) as shown in Fig. 5.5.
Using Eq. (5.10), we get

j K

d 74.7° K 0
1 j
C K 34
d
6.48
K 11.3° 26.6° 90°
6 3K 0 90° 0 1
d
d 74.7°
K 0 1 j

K (s + 6)
Fig. 5.5 Illustration of angle of departure for G H (s) =
(s + 3)(s 2 + 2s + 2)
640 5 Root Locus Method for Analysis

1 j j
K
a 90º

135º
K 0 45º
2 1 K 0

90º
a 90º
K
1 j

K (s 2 + 2s + 2)
Fig. 5.6 Illustration of angle of arrival for G H (s) =
s(s + 2)

θd = 180◦ + (90◦ + 26.6◦ − 11.3◦ )

= 285.3◦
= −74.7◦

Similarly, for the conjugate pole at s = −1 − j, we get

θd = 74.7◦

Rule 9: Angle of Arrival of the Root Locus

The angle of arrival of the root locus at the zero of the root locus is represented by
θa , and it is the angle of the tangent to the locus near the zero. The angle of arrival
is given by

θa = (2n + 1) × 180◦ − (θ Z − θ P ) (5.11)

where the symbols θ Z and θ P have the same meaning but now drawn to the point of
the concerned zero. The angle of arrival is illustrated in Fig. 5.6.
Figure 5.6 shows the root locus for the loop T.F.

K (s 2 + 2s + 2)
G H (s) =
s(s + 2)

The two root loci starting from the poles s = 0 and s = −2 breakaway at s = −1.0
and arrive at the zeros at s1 = −1 + j1 and s2 = −1 − j1. The angle of arrival at s1
is determined using Eq. (5.11)
5.3 Properties of the Root Loci (Rules of the Root Loci) 641

θa = 180◦ + 45◦ + 135◦ − 90◦

= −90◦

The angle of arrival at s2 is θa = 90◦ .

Rule 10: Breakaway and Break-in Points or Saddle Points on the Root Locus
The point at which more than one root loci meet and then depart is called the break-
away point. At the breakaway point, double roots are present. A root locus may
have more than on breakaway point. Breakaway points may be real and complex
conjugate. The breakaway point is obtained by satisfying the condition

dK
=0 (5.12)
ds

in the characteristic equation 1 + G H (s) and solving for s. Further, the point s so
obtained must also be a point on the root loci for some real value of K . Consider the
loop T.F.

K
G H (s) =
s(s + 4)(s + 2)
1 + G H (s) = s 3 + 6s 2 + 8s + K
=0

Differentiating the above equation with respect to s and equating d K /ds = 0, we

get

dK
3s 2 + 12s + 8 + =0
ds
3s 2 + 12s + 8 = 0

s1 = −3.33 and s2 = −0.667. The root locus is shown in Fig. 5.4. The breakaway
point s1 = −3.33 is not a point on the root locus, and hence it is not a breakaway
point. However, s2 is a point on the root locus. Hence, the breakaway point is at
s2 = −0.667.
When more than one root locus approach a real axis and depart, such a point
is called the break-in point. The conditions applied to breakaway points apply to
break-in points also.
Rule 11: Angle of Arrival and Departure of Root Loci at the Breakaway Points
The angles at which the root loci arrive or depart from a breakaway or break-in point
(Fig. 5.7b) depend on the number of locus breakaway or break-in. In Fig. 5.7a, b,
two loci meet and breakaway and break-in, respectively. In these cases, the angle of
arrival and departure is 90◦ . However, in Fig. 5.7c there are four root loci that meet
at the breakaway point and depart with an angle of 45◦ . Thus, in general if there are
n root loci that meet at the breakaway point, they arrive or depart at 180◦ /n.
642 5 Root Locus Method for Analysis

(a) (b)
K

K 0 K 0 K K
Break-in
point
Breakaway
K
point

(c)
K 0 j

K K

K 0 45 45

Breakaway 45 K 0
point
45

K K

K 0

Fig. 5.7 Representation of angle of breakaway points

Rule 12: Intersection of Root Loci with the Imaginary Axis

For some systems, as the parameter K is increased, the root loci cross the imaginary
axis. The value of K at this point c is calculated using the Routh–Hurwitz criterion.
The system is now marginally stable and has sustained oscillations, and the oscillation
frequency can also be determined by solving the auxiliary equation in Routh’s array.
Consider the following loop T.F.:

K (s + 6)
G H (s) =
(s + 3)(s 2 + 2s + 2)

The root locus for the above T.F. is shown in Fig. 5.5. The critical value or marginal
value of K when the root loci cross from LHP to RHP is calculated by forming
Routh’s array for the following characteristic equation:
5.3 Properties of the Root Loci (Rules of the Root Loci) 643

K (s + 6)
1+ =0
(s + 3)(s 2 + 2s + 2)
(s + 3)(s 2 + 2s + 2) + K (s + 6) = 0
s 3 + 5s 2 + (K + 8)s + 6(K + 1) = 0

The following Routh’s array is formed.

s3 1 (K + 8)
s2 5 6(K + 1)
s1 −K + 34
s0 (K + 1)

For the system to be marginally stable, −K + 34 = 0 or K = 34. The auxiliary

equation from Routh’s array is written as

5s 2 + 6(34 + 1) = 0

In the jω axis, s = jω and substituting this above equation we get

5ω2 = 42
ω = 6.48

This is represented in Fig. 5.5. The properties or the rules of root locus are summarized
in Table 5.2.

5.4 Procedure to Construct Root Locus

The following procedure is followed to construct root locus:

1. Express the given open loop T.F. in pole–zero form as

K (s + z 1 )(s + z 2 ) · · · (s + z m )
G H (s) =
(s + p1 )(s + p2 ) · · · (s + pn )

Locate the poles and zeros of G H (s) on the s-plane.

2. The root loci start from the poles of G H (s) where the value of K = 0 and end
at the zeros of G H (s) or at ∞ if the order of the numerator polynomial m is
less than or equal to the order of the denominator polynomial n. If m > n, then
(m − n) loci will start from ∞ and end at the zeros of G H (s). The value of K
at the zeros is zero.
3. Identify the root loci on the real axis. For any point s1 on the root locus to be on
the real axis, to the right of that point the sum total of poles and zeros should be
odd.
644 5 Root Locus Method for Analysis

Table 5.2 Summary of properties (rules) of root locus

1. Magnitude criterion K = (Product of directed distances drawn
from
 the poles of G H (s) to the point s1 )
all
Product of directed distances drawn from
all
the zeros of GH(s) to the point s1
2. Angle criterion (Sum of all the angles of vectors drawn
from all the zeros of G H (s) to any point
s1 -sum of all the angles of vectors drawn
from all the poles of G H (s) to any point
s1 ) = odd multiple of 180◦
3. The starting points of the root loci The root loci start from the poles of G H (s)
where the value of K = 0
4. The ending points of the root loci The root loci end at the zeros of G H (s) or
at ∞ where the value of K = ∞
5. Number of root loci of 1 + G H (s) The number of root loci of the
characteristic equation 1 + G H (s) = 0 is
equal to the order of the numerator
polynomial or the denominator polynomial
of G H (s) whichever is higher
6. Symmetry of the root loci The root loci is symmetrical with respect to
the real axis of the s-plane
7. Angle of the asymptote of the root loci The angles of the asymptotes of the root
◦
loci are given by θ = (2n+1)180
(P−Z ) , where
P = finite number of poles of G H (s); Z =
finite number of zeros of G H (s);
n = 0, 1, 2, . . . , (P − Z − 1)

8. Intersect of the asymptote (Centroid) σ =  part of poles of G H (s)
Real
−
 Real part of zeros of
G H (s) (P − Z )

9. Root loci on the real axis For any point s1 on the real axis to be a
point on the root locus, the total number of
poles and zeros of G H (s) to the right of
that point should be odd
10 The angle of departure of root locus θd = (2n + 1) × 180◦ − (θ P − θ Z ) where
θ P and θ Z are the net angle contribution to
the concerned pole by all other poles and
zeros of G H (s), respectively
11. The angle of arrival of root locus θa = (2n + 1) × 180◦ − (θ Z − θ P )
12. Breakaway and break-in points on the root The breakaway and break-in points on the
locus root locus are obtained by satisfying the
condition d K /ds = 0 in the characteristic
equation and solving for s
(continued)
5.4 Procedure to Construct Root Locus 645

Table 5.2 (continued)

13. Angle of arrival and departure of root loci If there are n root loci that meet at
at the breakaway or break-in points breakaway or break-in points, they arrive or
depart at 180◦ /n
14. Intersection of the root loci with the The root loci move from LHP to RHP as
imaginary axis the value of K increases and when it
intersects the imaginary axis, the system
becomes marginally stable. The critical
value of gain is determined by forming
Routh’s array. From the auxiliary equation
the frequency of oscillation is determined

4. Determine the angles of the asymptotes using

(2n + 1) × 180◦
θ=
(P − Z )

5. Determine the centroid which is given by

 
real part of poles of G H (s) − real part of zeros of G H (s)
σ =
(P − Z )

Using steps 4 and 5, draw the asymptotes.

6. Find the breakaway and break-in points of the root loci. For this, express the
characteristic equation as

P(s) + K Q(s) = 0
−Q(s)
K =
P(s)

Differentiate K with respect to s and put d K /ds = 0. Solve for s which gives
the breakaway or break-in points. However, only those points which lie only on
the root loci alone are to be taken.
7. Determine the angle of arrival and angle of departure of the root loci using
Eqs. (5.10) and (5.11), respectively.
8. Find the intersection of the root loci with the imaginary axis where the system
is marginally stable using the Routh–Hurwitz criterion.
9. By joining the points obtained from steps 6 to 8, the root locus is sketched. The
root loci coincide with the asymptote as K tends to ∞. However, in the neigh-
bourhood of the origin of the s-plane, root loci are to be checked whether they
satisfy the angle criterion. By following the procedure cited above reasonability
accurate root loci are drawn. However, if very accurate root loci are needed,
MATLAB may be used rather than hand calculation.
646 5 Root Locus Method for Analysis

Root Locus
GH(s)=K/s(s+1)(s+2)
2
j K
Kc 6
1.5
Asymptote 1

s1
1

Locu
Angle of asymptotes 2
0.5
Imaginary axis

K K 0 60°
K 0 K 0
1
Locus 3 2 1 60° 0.42260
K 0.385
Centroid Break away
0.5
point

Lo
1 Asymptote 2

cu
s2
1.5
K
2
3 2.5 2 1.5 1 0.5 0 0.5 1
Real axis

Fig. 5.8 Root locus for Example 5.1

10. After the root loci are drawn, the value of K at some intervals in the root loci
are obtained using magnitude condition given by Eq. (5.3).

Example 5.1

A certain negative feedback control system has the following loop T.F.

K
G H (s) =
s(s + 1)(s + 2)

Draw the root locus for 0 ≤ K ≤ ∞.

(Anna University, November, 2009)

Solution

1.
K
G H (s) =
s(s + 1)(s + 2)
5.4 Procedure to Construct Root Locus 647

G H (s) is in pole–zero form. There are 3 poles at s = 0, s = −1 and s = −2.

There are no zeros of G H (s). The poles are located as shown in Fig. 5.8.
2. The order of the numerator polynomial of G H (s) is zero and that of the denom-
inator polynomial is 3. Hence, there should be three separate root loci.
3. The root loci on the real axis are identified between s = 0 and s = −1 and
between s = −2 and s = −∞.
4. The root loci start from the poles of G H (s) where K = 0 and end at ∞ since
there are no zeros of G H (s). At s = ∞, the value of K = ∞. (There are ∞
number of poles and zeros at s = ∞).
5. The angle of the asymptote is given by

(2n + 1) × 180◦
θ=
(P − Z )

where P = 3 and Z = 0; n = 0, 1, 2

θ = 60◦ , 180◦ and 300◦

6. The centroid (intersection of the asymptote with the real axis) is calculated from
 
real part of poles of G H (s) − real part of zeros of G H (s)
σ =
(P − Z )

0−1−2−0
σ =
3
= −1

The two asymptotes with θ = 60◦ and θ = 300◦ or −60◦ are drawn in dashed
lines. The third asymptote with θ = 180◦ coincides with the negative real axis.
7. The root loci starting from s = 0 and s = −1 move on the negative real axis
between s = 0 and s = −1, meet each other and breakaway. The breakaway
point is calculated as follows. The characteristic equation of the system is

1 + G H (s) = 0
K
1+ =0
s(s + 1)(s + 2)
s 3 + 3s 2 + 2s + K = 0

Differentiating the above equation with respect to s, we get

dK
3s 2 + 6s + 2 + =0
ds
648 5 Root Locus Method for Analysis

Putting d K /ds = 0, we get

3s 2 + 6s + 2 = 0

Solving for s, we get the breakaway points at

s = −51.5774 and s = −0.4226

Since at s = −1.5774, there is no root locus, the point s = −0.4226 alone is

considered. At the breakaway point, the gain is calculated as

K = 0.4226 × 1.5774 × 0.5774

= 0.385

8. The critical value of K which is denoted as K c when the root loci cross from
LHP to RHP is determined along with the frequency ω at that point as follows.
The characteristic equation of the system is

s 3 + 3s 2 + 2s + K = 0

The following Routh’s array is formed.

s3 1 2
s2 3 K
(6−K )
s1 3
s0 K

The system becomes oscillatory when K = 6. Hence, the critical value of K is

K c = 6. The auxiliary equation is

3s 2 + K c = 0

Substituting s = jω and K c = 6 in the above equation, we get

−3ω2 + 6 = 0
√
ω= 2

9. The third root loci which starts from s = −2 moves towards the left on the −ve
real axis and goes to −∞, where K = ∞.
10. The asymptotes, the breakaway point and the intersection of the root loci with the
imaginary axes give guidance for the movements of the root loci. They are joined
by a smooth curve and the complete root locus diagram is shown in Fig. 5.8.
5.4 Procedure to Construct Root Locus 649

Example 5.2

Consider the following loop T.F. of a certain closed loop control system.

K
G H (s) =
s(s + 1)(s + 2)(s + 3)

Draw the root locus for 0 ≤ K ≤ ∞.

(Anna University, April, 2010)

Solution

1.
K
G H (s) =
s(s + 1)(s + 2)(s + 3)

There are four poles and no zeros of G H (s). Hence, there should be four root
loci.
2. The poles are at s = 0; s = −1; s = −2; and s = −3. The root loci start from
these four poles where K = 0, and they all end at ∞ in the s-plane.
3. On the real axis, root loci exist between s = 0 and s = −1 and between s = −2
and s = −3. In these regions, to the right of any point odd number of poles exist.
4. The angles of asymptotes are calculated as

(2n + 1) × 180◦
θ =
(P − Z )
(2n + 1) × 180◦
=
4
where n = 0, 1, 2, 3

θ = 45◦ , 135◦ , 225◦ and 315◦

5. The centroid is calculated as

 
real part of poles G H (s) − real parts of zeros of G H (s)
σ =
(P − Z )
0−1−2−3
=
4
= −1.5
650 5 Root Locus Method for Analysis

6. The breakaway points are calculated using the characteristic equation which is
given by

s(s + 1)(s + 2)(s + 3) + K = 0

s 4 + 6s 3 + 11s 2 + 6s + K = 0

Differentiating the above equation with respect to s and putting d K /ds = 0, we

get
4s 3 + 18s 2 + 22s + 6 = 0

Solving the above equation for s, we get

s1 = −2.5
s2 = −0.3
s3 = −1.7

At s = −1.7, there is no root locus and it is not a breakaway point. Hence, s1

and s2 are taken as breakaway points.
7. The intersections of the root loci with the imaginary axis is calculated from
Routh’s array formed by the characteristic equation

s4 1 11 K
s3 6 6
s2 10 K
60−6K
s1 10
s0 K

From Routh’s array the critical value of gain K c = 10. The auxiliary equation is

10s 2 + K c = 0

Substituting s = jω and K c = 10, we get ω = 1. The value of K c obtained can

be verified using magnitude condition. According to the magnitude condition,
at any point on the root locus, the value of K is

Product of directed distances drawn from all the poles to that point
K =
Product of directed distances drawn from all the zeros to that point.

Since the root loci intersect the imaginary axis at ω = 1,

√ √ √
Kc = 1 × 2 × 5 10 = 10

The complete root locus is shown in Fig. 5.9.

5.4 Procedure to Construct Root Locus 651

Root Locus
GH(s)=K/s(s+1)(s+2)(s+3)
2
K j K
Asymptote 2
1.5
Kc 10
Asymptote 1

s3
1
Locus 1

Locu
0.5 1
Imaginary axis

45°
K 0 K 0 K 0
1
3 2.5 2 1 0.3 0
K 0
Break away
0.5 1.5 Break
point Centroid away
point
1 Locus 4
Locus 2
Asymptote 4
1.5 Asymptote 3
K K
2
4 3.5 3 2.5 2 1.5 1 0.5 0 0.5 1
Real axis

Fig. 5.9 Root locus for Example 5.2

To Determine Closed Loop System Poles for the given Gain K

Example 5.3

A certain closed loop system has the following loop T.F.:

K
G H (s) =
s(s + 4)(s 2 + 4s + 20)

Draw the root locus for 0 ≤ K ≤ ∞. Identify the following:

1. Breakaway points and the value of K at these points.
2. Critical value of K and the oscillation frequency.
3. The angle of departure and the angle of arrival if any.
4. Find the closed loop system poles for K = 3000 and also the closed loop T.F.
5. Is the system stable for this value of K ?

(Anna University, December, 2010)

652 5 Root Locus Method for Analysis

Solution

1.
K
G H (s) =
s(s + + 4s + 20)
4)(s 2
(s 2 + 4s + 20) = (s + 2 + j4)(s + 2 − j4)
K
G H (s) =
s(s + 2)(s + 2 + j4)(s + 2 − j4)

There are four poles and no zero for the T.F. There should be four root loci. The
poles of G H (s) are located in the s-plane as shown in Fig. 5.10.
2. The root loci start from the poles s = 0, s = −2, s = −2 + j4 and s = −2 − j4
where the value of K = 0 and end at ∞ where K = ∞.
3. The root loci, on the real axis, exist in between s = 0 and s = −4 where odd
number of poles exists to the right of any point in the negative real axis.
4. The angles of the asymptotes are given by

(2n + 1) × 180◦
θ =
(P − Z )
(2n + 1) × 180◦
=
4
where n = 0, 1, 2, 3

θ = 45◦ , 135◦ , 225◦ and 315◦

5. The centroid is given by

 
real part of poles of G H (s) − real part of zeros of G H (s)
σ =
(P − Z )
0−2−2−4
=
4
= −2

6. The breakaway points of the root loci are obtained from the characteristic equa-
tion 1 + G H (s) = 0.

s 4 + 8s 3 + 36s 2 + 80s + K = 0

Differentiating the above equation with respect to s, we get

dK
4s 3 + 24s 2 + 72s + 80 + =0
ds
5.4 Procedure to Construct Root Locus 653

(a)
Root Locus
GH(s)=K/s(s+4)(s2+4s+20)
15
K j K

10
Kc 260
s1 3 j5.4
7 j5.4 s
3 2 j4 K 3000
5
K 3000
Imaginary axis

K 0 10
K 0
0
4 0
Kc 260
K 0
5 K 3000
2 j4
7 j5.4
s2 3 j5.4
K 3000 s4 2 j2.45
10
K
K
15
15 10 5 0 5 10 15
Real axis
(b) (c)
j 2 j4
j
2 j4 1.55 d
90°
d K 100 2 j2.45
10
6.45
10

64° 116° 116°

64°
4 0 4 2 K 64 0

K 100 2 j2.45
90°

2 j4 d 90°
2 j4

Fig. 5.10 Root locus for Example 5.3

654 5 Root Locus Method for Analysis

Putting d K /ds = 0 and solving for s, we get

s1 = −2
s2 = −2 + j2.45
s3 = −2 − j2.45

7. The angle of departure of the root locus from the pole at s = −2 + j4 is obtained
by drawing vectors from s = 0, s = −4 and s = −2 − j4. Thus, as shown in
Fig. 5.10c:

θd = 180◦ − (90◦ + 116◦ + 64◦ )

= −90◦

For the conjugate pole at s = −2 − j4, we will get the conjugate of the other
pole as θd = +90◦ .
8. The intersection of the root loci with the imaginary axis is obtained as follows.
The following Routh’s array is formed for the characteristic equation.

s4 1 36 K
s3 1 10
s2 26 K
s1 260 − K
s0 K

From Routh’s array, the critical value of gain is obtained as K c = 260. At this
value of K , the root locus crosses from LHP to RHP. The point at which it is
critical is determined by taking the auxiliary equation, which is obtained from
Routh’s array as
26s 2 + K c = 0

Substituting s = jω and K c = 260, we get

√
ω= 10

9. The root loci are drawn as follows. The root loci starting from s = 0 and s =
−4 meet each other at s = −2 and break away and move in the complex s-
plane, one vertically upward and another vertically downward. These loci and
the loci starting from the poles s = −2 ± j4 meet each other and break away.
The breakaway points as determined in step 6 are at s = −2 ± j2.45. These
breaking loci move towards the right and towards the left taking guidance from
the asymptotes as shown in Fig. 5.10a.
10. To find the roots for K = 3000, choose any point s1 on any one root locus. Draw
vectors from all the poles of G H (s) to the point s1 . The value so obtained gives
the value of K . This is obtained by trial and error only. For one such point
s1 = 3 + j5.4, the value of K is calculated as
5.4 Procedure to Construct Root Locus 655

K = 6.1 × 8.8 × 10.6 × 5.3

= 3000

For this value of K , for the conjugate root locus, the root is at

s2 = −3 − j5.4

By searching in the third and fourth root loci, by trial and error for K = 3000,
we get

s3 = −7 + j5.4
s4 = −7 + j5.4

11. At the breakaway points, the values of K are calculated by referring to

Fig. 5.10c. At the breakaway point s = −2

K = 2×2×4×4
= 64

At the breakaway point s = −2 + j2.45,

√ √
K = 10 × 10 × 6.45 × 1.55
= 100

12. The closed loop transfer function is obtained as

C(s) G(s)
=
R(s) 1 + G(s)
K
=
s4 + 8s 3 + 36s 2 + 8s + K

For K = 3000,

C(s) 3000
=
R(s) [(s + 7 + j5.4)(s + 7 − j5.4)(s − 3 + j5.4)(s − 3 − j5.4)]
3000
= 2
(s + 14s + 78.2)(s 2 − 6s + 38.2)

13. Since there are 2RHP roots for K = 3000, the closed loop system is unstable.
656 5 Root Locus Method for Analysis

Answers:

1. Breakaway points − 2 withK = 64 and − 2 ± j2.48 with K = 100

√
2. K c = 260 and ω = 10
3. θd = −90◦ and + 90◦ . No angle of arrival
4. Closed loop poles for K = 3000
s1 = 3 + j5.4
s2 = 3 − j5.4
s3 = −7 + j5.4
s4 = −7 − j5.4
5. The closed loop T.F. for K = 3000
C(s) 3000
= 2
R(s) (s + 14s + 78.2)(s 2 − 6s + 38.2)
6. The closed loop system is unstable

To Get Root Locus by MATLAB Program

Example 5.4

Consider the following loop T.F. of a certain closed loop system

K (s 2 + 2s + 10)
G H (s) =
s(s + 2)(s + 4)

Draw the root locus for 0 ≤ K ≤ ∞ using the MATLAB program. From the root
locus so obtained, verify the angle of arrival and the breakaway point. At the break-
away point, what is the value of K ? For this value of K , determine the closed loop
T.F.

Solution

K (s 2 + 2s + 10)
G H (s) =
s(s + 2)(s + 4)

The following MATLAB program is used to plot the root locus.

MATLAB Program
To draw Root Locus
5.4 Procedure to Construct Root Locus 657

Root locus
GH(s)=K(s2+2s+10)/s(s+2)(s+4)
4
j
1 j3
3 K
a 34°

1
Imaginary axis

K 0 K 0 72°
K B 56° A 108°
0
4 2 0.86 K 0 0
4.6
1 K 0.342 K 0.342

K a
34°
3
1 j3

4
6 5 4 3 2 1 0 1
Real axis

Fig. 5.11 Root locus for Example 5.4

clc;
% Let K = 1
num=[0 1 2 10];
den=conv([1 0], conv([1 2], [1 4]);
rlocus(num,den);
title(‘Root Locus’);
xlabel(‘Real axis’);
ylabel(‘Imaginary axis’);

1.

K (s 2 + 2s + 10)
G H (s) =
s(s + 2)(s + 4)
(s + 2s + 10) = (s + 1 + j3)(s + 1 − j3)
2

K (s + 1 + j3)(s + 1 − j3)
G H (s) =
s(s + 2)(s + 4)

The poles and zeros are marked in the s-plane as shown in Fig. 5.11.
2. The root loci on the real axis exist between s = −4 and s = −∞ and between
s = 0 and s = −2 as shown in Fig. 5.11. The root locus starting from s = −4
658 5 Root Locus Method for Analysis

moves towards the left and ends at −∞. The root loci starting from s = 0 and
s = −2 meet each other and break away and end at the zeros at s = −1 + j3
and s = −1 − j3, where the value of K = ∞.
3. The breakaway point is obtained from the characteristic equation which is given
by

(s 3 + 6s 2 + 8s) + K (s 2 + 2s + 10) = 0
−(s 3 + 6s 2 + 8s)
K =
(s 2 + 2s + 10)

Differentiating K with respect to s, we get

dK
=0
ds
= s 3 + 8.5s 2 + 30s + 20

Solving the above equation, we get

s1 = −0.86
s2 = −3.83 + j3
s3 = −3.83 − j3

4. The angle of arrival of the root locus at the zero at s = −1 + j3 is obtained as

θa = 180◦ − (90◦ − 108◦ − 72◦ − 56◦ )

= −34◦

and at s = −1 − j3 is obtained as

θa = +34◦

Since there are no root loci at s2 and s3 , as seen in Fig. 5.11, these points are
not the breakaway points. Hence, there is only one breakaway point which is at
s = −0.86. The gain K at point A marked in Fig. 5.9 is calculated as follows.
 Product of directed distances drawn from poles of G H (s) to point A

K =
A Product of directed distances drawn from zeros of G H (s) to point A
0.86 × 1.14 × 3.14
=
3.01 × 3.01
= 0.342

5. There are two closed loop roots at point A where the value of the gain K = 0.342.
In the third root locus which lies between s = −4 and s = −∞, there will be
5.4 Procedure to Construct Root Locus 659

another root. This is searched by trial and error. This is obtained at point B in
Fig. 5.11. The point B is obtained from
 0.6 × 4.6 × 2.6

K =
B 4.6 × 4.6
= 0.342

The three closed loop poles are at s = −0.86, s = −0.86 and s = −4.6. The
closed loop T.F. is obtained from

C(s) G(s)
=
R(s) 1 + G H (s)
K (s + 1 + j3)(s + 1 − j3)
= 3
(s + 6s 2 + 8s) + K (s 2 + 2s + 10)

For K = 0.342,

C(s) 0.342(s + 1 + j3)(s + 1 − j3)

=
R(s) (s + 0.86)2 (s + 4.6)

C(s) 0.342(s 2 + 2s + 10)

=
R(s) (s + 0.86)2 (s + 4.6)

To Determine the Closed Loop Poles for the given Damping Ratio of Dominant
Poles

Example 5.5

A certain closed loop system has the following open loop T.F.

K (s + 2)
G H (s) =
s(s + 1)(s + 4)

Draw the root locus for 0 ≤ K ≤ ∞. For the damping ratio of ζ = 0.707, determine
the value of K and find closed loop poles and the T.F.

Solution

1.
K (s + 2)
G H (s) =
s(s + 1)(s + 4)
660 5 Root Locus Method for Analysis

There are three poles at s = 0, s = −1 and s = −4 and one zero at s = −2.

There will be three root loci. They start from the poles of G H (s). One locus
ends at the zero and the remaining two end at ∞.
2. The centroid of the root locus is
0−1−4+2
σ = = −1.5
2
The angles of the asymptotes are

(2n + 1) × 180◦
θ = = 90◦ , 270◦
2
3. The characteristic equation is

s(s + 1)(s + 4) + K (s + 2) = 0
−s(s + 1)(s + 4)
K =
(s + 2)
−(s 3 + 5s 2 + 4s)
=
(s + 2)

Differentiating K with respect to s, we get

dK
(s + 2)(3s 2 + 10s + 4) − (s 3 + 5s 2 + 4s) =
ds
where d K /ds = 0.

s 3 + 5.5s 2 + 10s + 4 = 0
(s + 0.55)(s + 2.45 + j1.313)(s + 2.45 − j1.313) = 0

Since root loci do not break away in the complex s-plane in this particular case,
the breakaway point at s = −0.55 is taken. At this point, the value of the gain
is
0.55 × 0.45 × 3.45
K =
1.45
= 0.59

4. Smooth curves from the breakaway point are drawn in the complex s-plane,
taking guidance from the asymptote. It is checked at a few points that these two
loci satisfy the angle criterion of root locus. That is at any point on the root locus,
the total sum of the angles contributed by all the zeros minus the total sum of
all the angles contributed by the poles should be an odd multiple of 180◦ . The
5.4 Procedure to Construct Root Locus 661

K j

0.707
line A
cos 1

K 1.5
0.62 j0.62 45°

K 0 K 1.5 K K 0 K 0
4 3.75 2 1.5 1 0
K 0.59

0.62 j0.62 K 1.5

Asymptote

K
K (s+2)
Fig. 5.12 Root locus for G H (s) = s(s+1)(s+4)

third root locus which starts from s = −4 moves towards the right and ends at
the zero at s = −2.
5. Damping factor ζ = 0.707 corresponds to φ = cos−1 0.707 = 45◦ . The straight
line O A with φ = 45◦ is drawn as shown in Fig. 5.12. It cuts the root locus at
s = −0.62 + j0.62. The value of K at this point is calculated using magnitude
criterion as
0.87 × 0.73 × 3.45
K =
1.49
= 1.5
662 5 Root Locus Method for Analysis

6. In the third root loci, by trial and error it is calculated that at s = −3.75, the
value of K = 1.5.
0.25 × 2.75 × 3.75
K =
1.75
= 1.5

Thus, corresponding to K = 1.5, the three closed loop roots are

s1 = −0.62 + j0.62
s2 = −0.62 − j0.62
s3 = −3.75

7. The closed loop T.F. is

C(s) G(s)
=
R(s) 1 + G(s)
K (s + 2)
=
s(s + 1)(s + 4) + K (s + 2)

For K = 1.5
C(s) 1.5(s + 2)
=
R(s) (s + 3.75)(s + 0.62 + j0.62)(s + 0.62 − j0.62)

C(s) 1.5(s + 2)
=
R(s) (s + 3.75)(s 2 + 1.24s + 0.7688)

To Find the Transient Response of the System

Example 5.6

Consider the following loop T.F. of a certain unity feedback system

K
G H (s) =
(s + 4)(s 2 + 2s + 2)

Draw the root locus for 0 ≤ K ≤ ∞. For K = 20, find the closed loop system poles
and hence determine the transient response for a unit step input.
5.4 Procedure to Construct Root Locus 663

K
Kc 1.5

0.44 j2.3 K 20
A

d 71.6° 10
K 0
1 j

K 20 K 0 60°
18.4°
K 5.1 4 2 0
C 60°

K 0
1 j

d 71.6°

0.44 j2.3 K 20
B

Fig. 5.13 Root locus for G H (s) = K

(s+4)(s 2 +2s+2)

Solution

1.
K
G H (s) =
(s + + 2s + 2)
4)(s 2
K
=
(s + 4)(s + 1 + j)(s + 1 − j)

Since the denominator polynomial order is three, there will be three root loci
and they start from the poles of G H (s) where the value of K = 0 and end at
s = ∞ where K = ∞. The open loop poles are located as shown in Fig. 5.13.
2. The root locus on the real axis exists between s = −4 and s = −∞. The other
two root loci are complex conjugate.
3. The centroid of the root loci is at
664 5 Root Locus Method for Analysis

−4 − 1 − 1
σ =
3
= −2

The angles of the asymptotes are

(2n + 1) × 180◦
θ =
3
= 60◦ , 180◦ and 300◦

4. The characteristic equation is

1 + G H (s) = 0
s + 6s + 10s + (K + 8) = 0
3 2

The following Routh’s array is formed.

s3 1 10
s2 6 8+K
s1 52 − K
s0 8+K

For the system to be critically stable, K c = 52. The auxiliary equation is

6s 2 + (8 + K ) = 0

At s = jω, and K = K c = 52, the above equation is written as

6ω2 = 60
√
ω = 10

5. The breakaway point is obtained from the characteristic equation

K = −(s 3 + 6s 2 + 10s + 8)
dK
=0
ds
= s 2 + 4s + 3.3

Solving the above equation for s, we get

s1 = −2.837
s2 = −1.16

In these points, there is no root locus. Hence, there is no breakaway point.

5.4 Procedure to Construct Root Locus 665

6. The angles of departure from the complex poles are obtained as follows. For the
pole at s = −1 + j,

θd = 180◦ − (θ P − θ Z )
= 180◦ − (90◦ + 18.4◦ )
= 71.6◦

For the pole at s = −1 − j, θd = −71.6◦ .

7. From the information obtained above and taking guidance from the asymptotes,
the three root loci are drawn. The closed loop roots for the given K are determined
from each locus by trial and error using the magnitude condition

K = Product of directed vector distances drawn from the poles of G H (s)

Any point A is chosen in one root locus and vectors are drawn from all the poles
of G H (s), and K is found

K = 4.3 × 1.4 × 3.4

= 20

Point A corresponds to s = −0.44 + j2.3. By conjugate property at point B,

s = −0.44 − j2.3. These are the two closed loop poles. In this locus at point C,
by trial and error the value of K is calculated at s = −5.1.

K = 4.3 × 4.3 × 1.1

= 20

The third root is at s = −5.1.

Corresponding to K = 20, the three closed loop roots are obtained as (s +
5.1)(s + 0.44 + j2.3)(s + 0.44 − j2.3).
8. The closed loop T.F. is

C(s) G(s)
=
R(s) 1 + G(s)
K
=
(s + 4)(s 2 + 2s + 2) + K

For K = 20,

C(s) 20
=
R(s) (s + 5.1)(s + 0.44 + j2.3)(s + 0.44 − j2.3)

For unit step input, R(s) = 1/s

666 5 Root Locus Method for Analysis

20
C(s) =
s(s + 5.1)(s + 0.44 + j2.3)(s + 0.44 − j2.3)

This is put into partial fraction as

A1 A2 A3 A4
C(s) = + + +
s (s + 5.1) (s + 0.44 + j2.3) (s + 0.44 − j2.3)
20 = A1 (s + 5.1)(s 2 + 0.88s + 5.5) + A2 s(s 2 + 0.88s + 5.5)
+A3 s(s + 5.1)(s + 0.44 − j2.3) + A4 s(s + 5.1)(s + 0.44 + j2.3)

For s = 0,
A1 = 0.714

For s = −5.1,

20
A2 =
(−5.1)((−5.1)2 − 0.88 × 5.1 + 5.5)
= −0.145

For s = −0.44 − j2.3,

20
A3 =
(−.44 − j2.3)(−.44 − j2.3 + 5.1)(− j4.6)
20
=
2.38∠79 5.2∠−26◦ × 4.6∠90◦
◦

= 0.35∠−143◦
A4 = 0.35∠143◦
◦ ◦
0.714 0.145 e− j143 0.35 e+ j143 0.35
C(s) = − + +
s (s + 5.1) s + 0.44 + j2.3 s + 0.44 − j2.33

Taking inverse Laplace transform, we get

◦ ◦
c(t) = 0.714 − 0.145e−5.1t + 0.35e−0.44t [e− j143 e− j2.3t + e j143 e j2.3t ]
= 0.714 − 0.145e−5.1t + 0.7e−0.44t cos(2.3t + 143◦ )

c(t) = 0.714 − 0.145e−5.1t − 0.7e−0.44t sin(2.3t + 0.93rad)

To Find Phase Margin and Gain Margin from Root Locus

To Find Gain Margin G m
The root locus for a certain loop T.F. is shown in Fig. 5.14. Suppose K d is the desired
gain margin as shown in Fig. 5.14. The gain margin is the amount by which K d is
increased so that the system is not unstable. This is nothing but the ratio of K c /K d .
5.4 Procedure to Construct Root Locus 667

K
Kc
Kd 1
c
b a
90º
K K 0 K 0
2 1 0
2 1 K 0

Fig. 5.14 To find phase margin and gain margin from root locus

Kc
Gm =
Kd

To Find Phase Margin

For the given K d , find ω1 in the jω axis which satisfies the magnitude condition
of root locus. From Fig. 5.14, K d = abc at ω1 . Draw vectors from poles and zeros
of G H ( jω1 ) and find sum total of angles drawn from zeros of G H (s)− Sum total
of angles drawn from poles of G H (s) to the point ω1 . Let it be φ. Then the phase
margin is given by

φm = 180◦ + G H ( jω1 )
= 180◦ + φ

In Fig. 5.14, φ is given by

φ = −(90◦ + θ1 + θ2 )

Example 5.7

The characteristic equation of a certain unity feedback control system is given by

F(s) = s(s + 4)(s 2 + 6s + 10) + K

668 5 Root Locus Method for Analysis

K Kc 120
cos 1 0.5

K 0 3 j Kd 24.5 2
E d 117° 0.42 j0.73
A B 1 0.6
Kd 24.5
K 0 45° K 11.7 60° K 0
D 4 2.5 0.43 0

d 117°
C
K 0 3 j

Fig. 5.15 Root locus to determine phase margin and gain margin

Draw the root locus for 0 ≤ K ≤ ∞. For the damping ratio ζ = 0.5 for the dominant
roots find K . For this value of K , find the phase margin and gain margin.

Solution
1.

F(s) = s(s + 4)(s 2 + 6s + 10) + K

K
G(s) =
s(s + 4)(s 2 + 6s + 10)
K
=
s(s + 4)(s + 3 + j)(s + 3 − j)

The poles are located as shown in Fig. 5.15. There are four poles and there should
be four root loci.
5.4 Procedure to Construct Root Locus 669

2. The centroid of the root loci is

0−4−3−3
σ =
4
= −2.5

The angles of the asymptotes are

(2n + 1) × 180◦
θ =
4
where n = 0, 1, 2, 3

θ = 45◦ , 135◦ , 225◦ , 315◦

3. From the characteristic equation, we get

K = −(s 4 + 10s 3 + 34s 2 + 40s)

dK
= 0 = 4s 3 + 30s 2 + 68s + 40
ds
Solving the above equation, we get

s1 = −0.43
s2 = −3.5 + j1.22
s3 = −3.5 − j1.22

For the breakaway point s1 = −0.43, root loci exist. By intuition, the other two
breakaway points do not have a root locus.
4. For the characteristic equation, the following Routh’s array is prepared.

s4 1 34 K
s3 1 4
s2 30 K
s1 120 − K
s0 K

The critical value of K = 120. The auxiliary equation is

30s 2 + K c = 0

Substituting s = jω and K c = 120, we get

30ω2 = 120
ω=2
670 5 Root Locus Method for Analysis

5. For the pole at s = −3 + j, the angle of departure is obtained as

θd = 180◦ − (162◦ + 90◦ + 45◦ )

= −117◦

For the pole at s = −3 − j, the angle of departure is

θd = 117◦

6. Starting from the poles of G(s) where the value of K = 0, the root loci from the
complex poles move in the direction of the asymptotes with an angle of departure
−117◦ and +117◦ . The root loci, starting from s = −4 and s = 0, move on the
−ve real axis, break at s = −0.43 and move in the complex s-plane. They cross
the jω axis with K c = 120 and ω = 2 rad/sec. A smooth graph is drawn.
7. For a damping ratio ζ = 0.5, θ = cos−1 0.5 = 60◦ line as shown in Fig. 5.15
is drawn. It cuts the root locus at point A which is at s = −0.42 + j0.73. The
value of K at point A is the desired value which is denoted by K d . At this point
using magnitude condition, K d is calculated as
−→ −→ −→ −→
K d = | O A|| D A|| E A||C A|
= 0.836 × 3.67 × 2.58 × 3.09
= 24.5

8. The gain margin is given by

Kc
Gm =
Kd
120
= = 4.9
24.5
9. To determine the phase margin, on the jω axis, point B is chosen where the gain
is K d = 24.5. At ω1 ,
−→ −→ −→ −→
K d = | O B||C B|| D B|| E B|
= 0.6 × 3.3 × 4.1 × 3
= 24.5

This is done by trial and error. ω1 is calculated as 0.6 rad/sec. From G(s)

24.5
G( jω) =
( jω)( jω + 4)(−ω2 + j6ω + 10)
ω 6ω
∠G( jω) = −90◦ − tan−1 − tan−1
4 10 − ω2
5.4 Procedure to Construct Root Locus 671

Substituting ω = 0.6,

0.6 6 × 0.6
φ = ∠G( jω)ω=0.6 = −90◦ − tan−1 − tan−1
4 10 − 0.36
= −90◦ − 8.5◦ − 20.5◦
= −119◦

Hence, phase margin

φm = 180◦ − 119◦
= 61◦

The same result can also be obtained graphically by drawing vectors from all
the poles of G(s) to point B, and total sum is subtracted from 180◦ to get the
phase margin
−→ −→ −→ −→
φ = ∠ O B + ∠C B + ∠ D B + ∠ E B
= 90◦ + 29◦ + 9◦ − 9◦
= 119◦
φm = 180◦ − 119◦
= 61◦

10.
Gain margin, G m = 24.5
Phase margin, φm = 61◦

Example 5.8

Consider the following loop T.F. of a certain unity feedback control system

K (s + 3)
G H (s) =
s(s + 2)(s 2 + 4s + 13)

(a) Draw the root locus for 0 ≤ K ≤ ∞.

(b) Identify whether there are any repeated roots.
(c) For each of the cases in (b), find the closed loop T.F.

(Anna University, May, 2010)

672 5 Root Locus Method for Analysis

j K
G

3
5

j4.
10
K

0.8
K 0 Kc 36.8
2 j3
D 79°
d

H 1.8 j2.8
d 79° K 5.13
3.3

Breakaway point
E 1.2
3.8
K B A 60°
F
K 105 3 2 60° 0
K K 0 K
Break-in point 5.13

1.8 j2.8
K 5.13
K 0 d 79°
C
2 j3

0.8 j4.3
K 105 K

K (s+3)
Fig. 5.16 Root locus for G H (s) = s(s+2)(s 2 +4s+13)

Solution

1.
K (s + 3)
G H (s) =
s(s + 2)(s 2 + 4s + 13)
K (s + 3)
=
s(s + 2)(s + 2 + j3)(s + 2 − j3)

The poles and zeros are located as shown in Fig. 5.16.

2. There are 4 poles and 1 zero. Hence, there should be four root loci which originate
from the four poles where K = 0. One of them ends at the zero s = −3 and
remaining end at s = ∞.
3. The centroid of the asymptote is given by
5.4 Procedure to Construct Root Locus 673

−0 − 2 − 2 − 2 + 3
σ =
3
= −1

The angles of the asymptotes are

(2n + 1) × 180◦
θ =
3
where n = 0, 1, 2

θ = 60◦ , 180◦ , 300◦

4. The characteristic equation is

1 + G H (s) = 0
K (s + 3)
1+ =0
s(s + 2)(s 2 + 4s + 13)
s(s + 2)(s 2 + 4s + 13) + K (s + 3) = 0
−(s 2 + 2s)(s 2 + 4s + 13)
K =0=
(s + 3)

dK
=0
ds
= (s + 3)(4s 3 + 18s 2 + 42s + 26)
−(s 4 + 6s 3 + 21s 2 + 26s) − s 4 + 8s 3 + 25s 2 + 41s + 26 = 0

Solving by trial and error, we get

s1 = −3.8
s2 = −1.2
s3 = −1.5 + j1.77
s4 = −1.5 − j1.77

5. Root loci on the negative real axis exist between s = 0 and s = −2 and s = −3
and s = −∞. The breakaway points s1 and s2 are in the root loci on the −ve
real axis. It can be shown that for s3 and s4 , angle criterion of root loci will not
be satisfied
K (s + 3)
G H (s) =
s(s + 2)(s + 2 + j3)(s + 2 − j3)
674 5 Root Locus Method for Analysis

At s = −1.5 + j1.77, the phase angle of G H ( jω) is


G H ( jω) = K (−1.5 + j1.77 + 3) (−1.5 + j1.77)(−1.5 + j1.77 + 2)

×(−1.5+ j1.77+2+ j3)(−1.5+ j1.77+2− j3)

K (1.5 + j1.77)
=
(−1.5 + j1.77)(0.5 + j1.77)(0.5 + j4.77)(−0.5 − j1.23)
1.77 1.77 1.77
∠G H ( jω) = tan−1 + 180◦ + tan−1 − tan−1
1.5 1.5 0.5
4.77 1.23
− tan−1 − tan−1 + 180◦
0.5 0.5
= 50◦ + 50◦ − 74◦ − 84◦ − 68◦
= −126◦

This is not an odd multiple of 180◦ . So, this point is not a point on the root locus.
By conjugate property, s = −1.5 − j1.77 is also not a point on the root locus.
At point D

θd = 180◦ − (90◦ + 124◦ + 90◦ − 45◦ )

= −79◦

6. There are two closed loop roots at the breakaway point s = −1.2 and the value
of K at this point is
−→ −→ −→ −→
| O E||C E|| AE|| D E|
K = −→
| B E|
1.2 × 3.1 × 0.8 × 3.1
=
1.8
= 5.13

7. There are two closed loop roots at the breakaway point at s = −3.8. The value
of K at this point is
−→ −→ −→ −→
| O F|| AF||C F|| D F|
K = −→
| B F|
3.8 × 1.8 × 3.5 × 3.5
=
0.8
= 105

8. The characteristic equation is

s 4 + 6s 3 + 21s 2 + (26 + K )s + 3K = 0
5.4 Procedure to Construct Root Locus 675

The following Routh’s array is formed.

s4 1 21 3K
s3 6 26 + K
s2 100 − K 18K
s1 −K 2 − 34K + 2600
s0 K

From Routh’s array, the following conditions are derived for stability:

1. K < 100
2. −K 2 − 34K + 2600 > 0
(−K + 36.8)(K + 71.0) > 0
K > −71 and K < +36.8
3. K >0

The condition K < 36.8 satisfies the stability condition K c = 36.8. The oscil-
lating frequency is obtained from the auxiliary equation which is given by

(100 − 36.8)s 2 + 18 × 36.8 = 0

Putting s = jω, we get

63.2ω2 = 18 × 36.8
ω = 3.3

The root locus is drawn using the data obtained above.

9. There are two closed loop roots at the breakaway point at s = −3.8. The value
of K at this point is
−→ −→ −→ −→
| O F|| AF||C F|| D F|
K = −→
| B F|
3.8 × 1.8 × 3.5 × 3.5
=
0.8
= 105

10. Corresponding to K = 105, a root is searched at point G by trial and error

−−→ −→ −→ −→
| OG|| AG||C G|| DG|
K = −→
| BG|
4.4 × 5.13 × 7.83 × 3.09
=
5.2
= 105
676 5 Root Locus Method for Analysis

The point is at s = 0.8 + j4.3. In the fourth root locus, at the conjugate point
for K = 105, the root is located at s = 0.8 − j4.3. Hence, corresponding to
K = 105, the four closed loop roots are at

s1 = −3.8
s2 = −3.8
s3 = 0.8 + j4.3
s4 = 0.8 − j4.3

Since two roots s3 and s4 are in RHP, the closed loop system is unstable. The
closed loop T.F. is

C(s) 105(s + 3)
=
R(s) (s + 3.8)2 (s 2 − 1.6s + 19.1)

11. Corresponding to K = 5.13, a root is searched at point H by trial and error

−−→ −→ −→ −−→
| O H || AH ||C H || D H |
K = −→
|B H |
3.33 × 2.8 × 5.9 × 0.2828
=
3.05
= 5.13

The point H is at s = −1.8 + j2.8. The fourth root is at s = −1.8 − j2.8 in the
fourth root locus. The four closed loop poles are at

s1 = −1.2
s2 = −1.2
s3 = −1.8 + j2.8
s4 = −1.8 − j2.8

All the four roots are in LHP, and the closed loop system is stable. The closed
loop T.F. for K = 5.13

C(s) 5.13(s + 3)
=
R(s) (s + 1.2)2 (s 2 + 3.6s + 11)
5.4 Procedure to Construct Root Locus 677

Circular Root Locus

Example 5.9

A certain feedback control system has the following loop T.F.

K (s + 4)(s + 5)
G H (s) =
(s + 1)(s + 3)

Show that part of root locus is a circle for 0 ≤ K ≤ ∞. Draw the actual root locus.
Find the repeated roots of the closed loop system and the corresponding value of K
and the closed loop T.F. in each case.

(Anna University, December, 2007)

Solution

1.
K (s + 4)(s + 5)
G H (s) =
(s + 1)(s + 3)

There are two poles at s = −1 and s = −3 and two zeros at s = −4 and s = −5.
The two root loci originating from the poles where the value of K = 0 end at the
zeros where the value of K = ∞. In such problems, we will not get guidance
from the asymptotes. However, theoretically it can be established that part of the
root loci is a circle and from that exact root locus can be obtained as explained
below.
2. Substituting s = σ + jω in G H (s), we get

K (σ + jω + 4)(σ + jω + 5)
G H (σ + jω) =
(σ + jω + 1)(σ + jω + 3)
ω ω ω ω
∠G H ((σ + jω)) = tan−1 + tan−1 − tan−1 − tan−1
σ +4 σ +5 σ +1 σ +3

3. Applying angle criterion to the above equation, we get

ω ω ω ω
tan−1 + tan−1 = 180◦ + tan−1 + tan−1
σ +4 σ +5 σ +1 σ +3
ω ω ω ω+1
+ +
tan−1 σ +4 ωσ2+5 = 180◦ + tan−1 σ σ +3
ω2
[1 − (σ +4)(σ +5) ] [1 − (σ +1)(σ +3)
]
ω(2σ + 9) ω(2σ + 4)
tan−1 = 180◦ − tan−1 2
(σ 2 + 9σ + 20 − ω )
2 (σ + 4σ + 3 − ω2 )

Taking tan on both sides, we get

678 5 Root Locus Method for Analysis

Root locus
GH(s) K(s 4)(s 5)/(s 3)(s 1)
1.5
j

Break away
Imaginary axis

0.5 point
K K K 0 K 0
0 3.4
5 4 3 2.4 1
0.5 K 0.2
4.4
Break in r 1
point Circle
K 19.83
1

1.5
6 5 4 3 2 1 0 1
Real axis

Fig. 5.17 Root locus for Example 5.9

(2σ + 9) (2σ + 4)
= 2
σ 2 + 9σ + 20 − ω2 (σ + 4σ + 3 − ω2 )
(2σ + 9)(σ 2 + 4σ + 3 − ω2 ) = (2σ + 4)(σ 2 + 9σ + 20 − ω2 )
σ 2 + 6.8σ + ω2 + 10.6 = 0
(σ + 3.4)2 + ω2 = (1)2

This is the equation of a circle. The centre of the circle is at (−3.4, 0) and radius
r = 1.
4. On the real axis, root loci exist between s = −1 and s = −3 and between s = −4
and s = −5. The root loci starting from poles s = −1 and s = −3 break at
s = −2.4, traverse through the complex s-plane and break—in at s = −4.4 and
end at s = −4 and s = −5. The root locus with centre at s = −3.4 and radius
r = 1 is drawn as shown in Fig. 5.17.
5. At the breakaway and break-in points, there are repeated roots. At s = −2.4, the
gain K is

0.6 × 1.4
K =
1.6 × 2.6
= 0.2
5.4 Procedure to Construct Root Locus 679

The closed loop T.F. now is

C(s) G(s)
=
R(s) 1 + G(s)
K (s+4)(s+5)
(s+1)(s+3)
= (s+4)(s+5)
1 + K(s+1)(s+3)

For K = 0.2,

C(s) 0.2(s + 4)(s + 5)

=
R(s) (s + 1)(s + 3) + 0.2(s + 4)(s + 5)
0.2(s + 4)(s + 5)
=
(1.2s 2 + 5.8s + 7)
0.2(s + 4)(s + 5)
=
1.2(s 2 + 4.8s + 5.8)

C(s) 0.167(s + 4)(s + 5)

=
R(s) (s + 2.4)2

6. At the break-in point, the value of K is

3.4 × 1.4
K =
0.4 × 0.6
= 19.83

The closed loop T.F. at this value of K is

C(s) 0.952(s + 4)(s + 5)

=
R(s) (s + 4.4)2

Example 5.10

A certain closed loop system has the following open loop T.F.

K (s + 5)
G H (s) =
s(s + 3)

Draw the root locus for 0 ≤ K ≤ ∞. Determine the following:

680 5 Root Locus Method for Analysis

(a) The breakaway and break-in points.

(b) The minimum value of damping that the system can provide and the correspond-
ing value of K and the closed loop roots.
(c) The repeated roots and the corresponding closed loop T.F.
(d) The range of values of K for which the system is under-damped.
(e) The value of K for critical damping ratio.

Solution

1.
K (s + 5)
G H (s) =
s(s + 3)

Substitute s = σ + jω.

K (σ + jω + 5)
G H (σ + jω) =
(σ + jω)(σ + jω + 3)
ω ω ω
∠G(σ + jω) = tan−1 − tan−1 − tan−1
σ +5 σ σ +3

2. Root locus satisfies angle criterion of odd multiple of 180◦

ω ω ω
tan−1 − tan−1 − tan−1 = 180◦
σ +5 σ σ +3
ω ω ω
tan−1 − tan−1 = 180◦ + tan−1
σ +5 σ +3 σ
−1 ω(−2) −1 ω
tan = 180 + tan
(σ + 5)(σ + 3) + ω2 σ

Taking tan on both sides, we get

−σ 2 = (σ + 5)(σ + 3) + ω2
−2σ = σ 2 + 8σ + 15 + ω2
σ 2 + 10σ + 15 + ω2 = 0
(σ + 5)2 + ω2 − 10 = 0
√
(σ + 5)2 + ω2 = ( 10)2
√
This is the equation of a circle with centre at (−5, 0) and radius r = 10 = 3.16.
3. On the negative real axis, the root loci lie in between s = 0 and s = −3 and
s = −5 and s = −∞. The complete root locus is shown in Fig. 5.18.
4. The breakaway point is obtained in between s = 0 and s = −3 as −1.84 and
the corresponding value of K at this point is
5.4 Procedure to Construct Root Locus 681

−→ −→
| O E|| B E|
K = −→
|C E|

1.84 × 1.16
K =
3.16
= 0.675

5. The break-in point is obtained in between s = −5 and s = −∞ as −8.16. The

corresponding value of K at this point is
−−→ −→
| O D|| B D|
K = −→
|C D|
8.16 × 5.16
=
3.16
= 13.32

6. A radial line from the origin is drawn which is tangential to the circle which
is part of the root locus. It is tangential at point A in the circle. The angle θ is
measured. This point gives the minimum value of damping ratio that the system
can provide. The damping ratio is given by

ζ = cos θ
= cos 38◦
= 0.788

At this point, the value of K is calculated as

−→ −→
| O A|| B A|
K = −→
CA
3.8 × 2.3
=
3.1
= 2.82

7. The repeated roots occur at the breakaway and break-in points. From Fig. 5.18,
at the breakaway point the repeated roots are at s = −1.84. The repeated poles
are (s + 1.84)2 . The closed loop T.F. corresponding to this point where the value
of K = 0.675 is
682 5 Root Locus Method for Analysis

3 j2.3
r 10 A K 2.82
cos 38°
0.788

K D K 13.32 C 3B E 38° K 0
8.16 5 K 0 1.84 K 0 0
Break-in K K 0.675
point Breakaway
point

Circular root
locus

K (s+5)
Fig. 5.18 Root Locus for G H (s) = s(s+3)

C(s) G(s)
=
R(s) 1 + G(s)
0.675(s+5)
s(s+3)
=
1 + 0.675(s+5)
s(s+3)

C(s) 0.675(s + 5)
=
R(s) (s + 1.84)2

The break-in point is at s = −8.16. The value of K at this point is K = 13.32.

Corresponding to this point, the repeated roots are (s + 8.16)2 . The closed loop
T.F. is given by
13.32(s+5)
C(s) s(s+3)
=
R(s) 1 + 13.32(s+5)
s(s+3)

C(s) 13.32(s + 5)
=
R(s) (s + 8.16)2
5.4 Procedure to Construct Root Locus 683

8. If the root loci are complex, the damping ratio ζ < 1. At the breakaway and
break-in points, ζ = 1 and on the real axis ζ > 1. The system is critically damped
(ζ = 1) for the value of K = 0.675 and K = 13.32.
9. (a) Breakaway point s = −1.84
Break-in point s = −8.16
(b) Minimum value of ζ = 0.788
K = 2.82
(c) At the breakaway point at (s + 1.84)2 , poles are repeated

C(s) 0.675(s + 5)
=
R(s) (s + 1.84)2

At the break-in point, the roots are repeated as (s + 8.16)2 . The closed loop
T.F. is
C(s) 13.32(s + 5)
=
R(s) (s + 8.16)2

(d) The range of values of K for which the system is under-damped is

0.675 ≤ K ≤ 13.32

(e) The values of K for which ζ = 1 is

K = 0.675
K = 13.32

Example 5.11

A certain closed loop control system has the following open loop T.F.:

K (s + 4)(s + 3)
G H (s) =
(s + 2)(s − 2)

Draw the root locus for 0 ≤ K ≤ ∞. Show that part of the root locus is a circle.
Determine the value of K for which the system is stable. At this value of K , find the
closed loop system roots and hence the closed loop T.F. What are the repeated roots
and the corresponding values of K ? Also, show that s = −2.29 + j1.11 is on the
root locus.

(Anna University, December, 2003)

684 5 Root Locus Method for Analysis

Solution

1.
K (s + 4)(s + 3)
G H (s) =
(s + 2)(s − 2)

Substitute s = σ + jω

K (σ + jω + 4)(σ + jω + 3)
G H (σ + jω) =
(σ + jω + 2)(σ + jω − 2)
ω ω
∠G H (σ + jω) = tan−1 + tan−1
σ +4 σ +3
ω ω
− tan−1 − tan−1
σ +2 σ −2

To satisfy the angle criterion of root locus,

ω ω ω ω
tan−1 + tan−1 − tan−1 − tan−1 = 180◦
σ +4 σ +3 σ +2 σ −2
ω ω ω ω
tan−1 + tan−1 = 180◦ + tan−1 + tan−1
σ +4 σ +3 σ +2 σ −2
ω ω ω ω
+ +
tan−1 σ +4 ωσ2+3 = 180◦ + tan−1 σ +2 ωσ2−2
1 − (σ +4)(σ +3) 1 − (σ +2)(σ −2)
ω(2σ + 7) ω2σ
tan−1 = 180◦ + tan−1 2
σ 2 + 7σ + 12 − ω2 σ − 4 − ω2

Taking tan on both sides, we get

(2σ + 7)(σ 2 − 4 − ω2 ) = 2σ (σ 2 + 7σ + 12 − ω2 )
σ 2 + 4.58σ + ω2 + 4 = 0
(σ + 2.29)2 + ω2 = (1.11)2

This is the equation of a circle with centre (−2.29, 0) and radius r = 1.11.
2. On the real axis, root loci exist between s = 2 and s = −2 and between s = −3
and s = −4 where odd number of poles and zeros exist to the right of any point
on the root loci.
3. The root loci originate from s = 2 and s = −2 where the value of K = 0, move
on the real axis and breakaway. These loci traverse through the complex s-plane
in a circular path, break in on the negative real axis and end at the zeros at s = −3
and s = −4 where the value of K = ∞.
4. The circular path of the root loci is drawn with centre at −2.29 and radius
r = 1.11. The complete root locus is shown in Fig. 5.19.
5.4 Procedure to Construct Root Locus 685

2.29 j1.1
I K 0.333
H 1.75

1
1.1
A F

r
E K 31.5 D B K 0.333
G C
4 3 2.29 2 1.18 0 2
3.4 K K 0 K 0.51
K K 0

Circular root
loci

K (s + 3)(s + 4)
Fig. 5.19 Root locus for G H (s) =
(s + 2)(s − 2)

5. The root locus crosses the imaginary axis at point s = 0. At this point, the value
of K is calculated as
−−→−−→
|F O||B O|
K = −−−→−−→
|D O||E O|
2.0 × 2.0
K = = 0.333
3.0 × 4.0

For K > 0.333, the system is stable. At this point, the value of s = 0 and for
K = 0.333, there is another root. This root is determined by trial and error.
Choose any point H = −1.75. At this point, the value of K is calculated as
−−−→−−−→
|F H ||B H |
K = −−−→−−−→
|D H ||E H |
3.75 × 0.25
= = 0.333
1.25 × 2.25

Hence, corresponding to K = 0.333, the other root is at s = −1.75. The closed

loop T.F. for this value of K is given by
(s+3)(s+4)
C(s) G(s) 0.333 (s+2)(s−2)
= =
R(s) 1 + G(s) 1 + 0.333(s+3)(s+4)
(s+2)(s−2)
686 5 Root Locus Method for Analysis

C(s) 0.25(s + 3)(s + 4)

=
R(s) s(s + 1.75)

6. The breakaway point as seen in Fig. 5.19 is at s = −1.18. The repeated roots are
(s + 1.18)2 . The value of K at this point is
−−→−−→
|F A||B A|
K = −−→−−→
|D A||E A|
3.18 × 0.82
=
1.82 × 2.82

K = 0.51

Repeated roots occur at break-in point also. This occurs at point G where
s = −3.4. The value of K is calculated as
−−→−−→
|F G||BG| 5.4 × 1.4
K = −−−→−−→ =
|DG||E G| 0.4 × 0.6

K = 31.5

7. Let the point I be denoted as −2.29 + j1.11. Using angle criterion, at point I
−→ − → − → − →
∠G H (−2.29 + j1.11) = D I + E I − F I − B I
= 58◦ + 34◦ − 168◦ − 104◦ = −180◦

This is an odd multiple of 180◦ . Hence, the point s = −2.29 + j1.11 is on

the root locus. The above result can be obtained analytically by substituting
s = −2.29 + j1.11 in G H (s) as explained below
(−2.29 + j1.11 + 3)(−2.29 + j1.11 + 4)
G H (s)|s=−2.29+ j1.11 =
(−2.29 + j1.11 + 2)(−2.29 + j1.11 − 2)
(0.71 + j1.11)(1.71 + j1.11)
=
(−0.29 + j1.11)(−4.29 + j1.11)
1.11 1.11 1.11
φ = tan−1 + tan−1 + 180◦ + tan−1 + 180◦
0.71 1.71 0.29
1.11
+ tan−1
4.29
= 58◦ + 33◦ + 76◦ + 13◦ = 180◦

The phase angle is an odd multiple of 180◦ . Hence, the point −2.29 + j1.11 is
on the root locus.
5.4 Procedure to Construct Root Locus 687

Root locus
G(s) K(s 2)2/s3
4
j

Kc 1
2 C

1 2
Imaginary axis

B K 13.5 A
0 xxx
6 0 K 0
K 2
1 K

2 Break-in point

4
8 7 6 5 4 3 2 1 0 1 2
Real axis

Fig. 5.20 Root locus for Example 5.12

Example 5.12

Consider the following loop T.F.

K (s + 2)2
G H (s) =
s3
Draw the root locus for 0 ≤ K ≤ ∞. Determine the breakaway and break-in points
if any and find the corresponding value of K . Determine the range of K for which
the system is stable.

Solution

1.

K (s + 2)2
G H (s) =
s3

There are three poles and two zeros of G H (s). The poles and zeros are located
as shown in Fig. 5.20. There should be three root loci.
688 5 Root Locus Method for Analysis

2. The root loci exist in the entire negative real axis since to any point in the negative
real axis, an odd number of poles and zeros exist. There won’t be any root loci
on the positive real axis.
3. There are three poles at the origin. One such locus lies between s = 0 and
s = −2. The other two loci traverse in the complex s-plane symmetrically with
respect to the real axis and again break in on the negative real axis. One of them
ends at s = −2 where K = ∞ and the other ends at s = −∞ where K = ∞ as
shown in Fig. 5.20.
4. The break-in point is obtained as follows. The characteristic equation is

1 + G H (s) = 0
K (s + 2)2
1+ =0
s3
s3
K =−
(s + 2)2
dK
= 0 = (s + 2)2 (3s) − s 3 (2)(s + 2)
ds
3(s + 2) = 2s
s = −6

The break-in point is at

s = −6

At the break-in point, the value of K is calculated as

−−−→
|O B|3 63
K = −−−→ = 2 = 13.5
|AB|2 4

K = 13.5

5. At point C, the root locus passes from RHP to LHP. This point can be determined
using Routh’s criterion. The characteristic equation is

1 + G H (s) = 0
K (s + 2)2
1+ =0
s3
s 3 + K s 2 + 4K s + 4K = 0
5.4 Procedure to Construct Root Locus 689

The following Routh’s array is formed

s3 1 4K
s2 K 4K
s1 4K (K − 1)
s0 K

From Routh’s array, it is evident that K > 1 for the system to be stable. Hence,
K c = 1. The auxiliary equation is

K s 2 + 4K = 0

s2 + 4 = 0
ω=2

6.
The break-in point is at s = −6
The value of K at this point is K = 13.5
The system is stable for K > 1

Example 5.13

The forward path T.F. of a certain feedback control system is given by

K (s + 2)(s + 6)
G H (s) =
s 2 (s + 4)

Draw the root locus for 0 ≤ K ≤ ∞. Choose any point P on the root locus which is
complex and graphically show that it satisfies the angle criterion. Calculate the value
of K at this point using magnitude condition. Also calculate the value of K at the
break-in point (Fig. 5.21).

Solution
1.
K (s + 2)(s + 6)
G H (s) =
s 2 (s + 4)

There are three poles and two zeros. Hence, there will be three root loci. They
start from the poles at s = 0 (repeated) and s = −4 where the value of K = 0.
690 5 Root Locus Method for Analysis

Root locus
GH(s) K(s 2)(s 6)/s2(s 4)
5
j
4

3 P 9.1 j3
K 16.5
2
Imaginary axis

1
K K 0 K K 0
0
K 10.3 D 6 4 2 0
1 C B A
K 18.7
2

5
15 10 5 0 5
Real axis

Fig. 5.21 Root locus for Example 5.13

They end at the zero at s = −2 and s = −6 where K = ∞. The third root locus
ends at −∞ where the value of K = ∞.
2. On the real axis, root loci exist between s = −4 and s = −2 and between s = −6
and s = −∞. The root loci originating from the origin move in the complex s-
plane symmetrically with respect to the real axis. They break in on the real axis.
One locus ends at s = −6 and the other at s = −∞.
3. The break-in point is calculated as follows. The characteristic equation is given
by

K (s 2 + 8s + 12)
1+ =0
(s 3 + 4s 2 )
−(s 3 + 4s 2 )
K =
(s 2 + 8s + 12)
dK
= 0 = (s 2 + 8s + 12)(3s 2 + 8s) − (s 3 + 4s 2 )(2s + 8)
ds
s 3 + 16s 2 + 68s + 96 = 0

Solving the above equation, we get

s1 = −10.3 and s2,3 = −2.85 ± j1.08

5.4 Procedure to Construct Root Locus 691

The break-in point occurs at s = −10.3. The value of K at this point is obtained
from
−−−→−−−→−−→
|O D||O D||B D| 10.3 × 10.3 × 6.3
K = −−→−−→ = = 18.7
|AD||C D| 8.3 × 4.3

K = 18.7

4. The point P is chosen at s = −9.1 + j3

−→ −→ −→ −→
∠G H (−9.1 + j3) = ∠ A P + ∠C P − 2∠ O P − ∠ B P
= 158◦ + 136◦ − 2 × 162◦ − 150◦

φ = −180◦

This is an odd multiple of 180◦ . Hence, point P is in the root locus. The value
of K at point P is calculated as
−−→2 −−→
|O P| |B P| (9.6)2 × 5.9
|G H (−9.1 + j3)| = −−→−−→ =
|A P||C P| 7.7 × 4.3

K = 16.5

Example 5.14

Consider the following loop T.F. of a certain feedback control system:

K (s 2 + 6s + 25)
G H (s) =
s(s + 3)(s 2 + 8s + 25)

Draw the root locus for 0 ≤ K ≤ ∞. Find the breakaway and break-in points and
the corresponding values of K (Fig. 5.22).
692 5 Root Locus Method for Analysis

Root locus
GH(s) K(s2 6s 25)/s(s 3)(s2 8s 25)
K
10
2.8 j5.4
K 57.3

5
K 0 3 j4
Imaginary axis

4 j3 K
K 2.9
0 2.5 1.3
3 K 0
K 0
K 0 3 j4
4 j3 K
5

2.8 j5.4
Asymptote
K 57.3
10
K
5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0
Real axis

Fig. 5.22 Root locus for Example 5.14

Solution

1.

(s 2 + 6s + 25)
G H (s) =
s(s + 3)(s 2 + 8s + 25)
s + 6s + 25 = (s + 3 + j4)(s + 3 − j4)
2

s 2 + 8s + 25 = (s + 4 + j3)(s + 4 − j3)
K (s + 3 + j4)(s + 3 − j4)
G H (s) =
s(s + 3)(s + 4 + j3)(s + 4 − j3)

The poles and zeros are located as shown in Fig. 5.22.

2. There are four poles and two zeros. Hence, there should be four separate root
loci. Four root loci start from s = 0, s = −3 and s = −4 + j3 and s = −4 − j3.
Two of them end at the zeros at s = −3 + j4 and s = −3 − j4. The remaining
two root loci end at ∞.
3. The root loci on the real axis exist between s = 0 and s = −3 only. The root
loci starting from s = 0 and s = −3 move on the negative real axis and break
away in the complex plane symmetrically with respect to the real axis. These
root loci and the root loci originating from the other poles s = −4 + j3 and
5.4 Procedure to Construct Root Locus 693

s = −4 − j3 meet and break away. Two of them end at the zeros at s = −3 + j4

and s = −3 − j4. The remaining two root loci go to ∞.
4. The centroid of the root loci
0−3−4−4+3+3
σ =
2
= −2.5

The angle of the asymptote

(2n + 1) × 180◦
θ =
2
where n = 0, 1

θ = 90◦ , 270◦

The asymptotes are drawn as shown in Fig. 5.22.

5. We can expect three breakaway points, one of them on the negative real axis
and the remaining two in the complex s-plane. They are calculated as explained
below. The characteristic equation of the system is

K (s 2 + 6s + 25)
1+ =0
s(s + 3)(s 2 + 8s + 25)
(s 2 + 3s)(s 2 + 8s + 25)
K =−
s 2 + 6s + 25
−(s + 11s 3 + 49s 2 + 75s)
4
=
(s 2 + 6s + 25)
dK
=0
ds
= (s 2 + 6s + 25)(4s 3 + 33s 2 + 98s + 75)
−(s 4 + 11s 3 + 49s 2 + 75s)(8s + 6)
0 = s 5 + 14.5s 4 + 116s 3 + 522s 2 + 1225s + 938

Solving the above equation is by trial and error, we get

s1 = −1.3
s2 = −2.8 ± j5.4
s3 = −3.8 ± j4.5

From the root locus on the real axis, it is seen that s = −1.3 is a point on the root
locus. Hence, s1 is a breakaway point. Now we see whether s = −2.8 ± j4.5
are in the root locus which satisfy the angle criterion
694 5 Root Locus Method for Analysis

K (s + 3 + j4)(s + 3 − j4)
G H (s) =
s(s + 3)(s + 4 + j3)(s + 4 − j3)

Substituting s = −2.8 + j5.4, we get

G H (−2.8 + j5.4) = K (−2.8 + j5.4 + 3 + j4)(−2.8 + j5.4 + 3 − j4)


(−2.8 + j5.4)(−2.8 + j5.4 + 3)(−2.8 + j5.4 + 4 + j3)

(−2.8 + j5.4 + 4 − j3)
K (0.2 + j9.4)(0.2 + j1.4)
=
[(−2.8 + j5.4)(0.2 + j5.4)(1.2 + j8.4)(1.2 + j2.4)]

9.4 1.4 5.4 5.4
φ = tan−1 + tan−1 + 180◦ + tan−1 − tan−1
0.2 0.2 2.8 0.2
8.4 2.4
− tan−1 − tan−1
1.2 1.2
= 180◦ + 89◦ + 82◦ + 63◦ − 88◦ − 82◦ − 64◦
= 180◦

This is an odd multiple of 180◦ . Hence, s = −2.8 ± j5.4 are the breakaway
points and they lie in the root locus. Now, we see whether s = −3.8 ± j4.5 are
in root loci and satisfy angle criterion

G H (−3.8 + j4.5) = K (−3.8 + j4.5 + 3 + j4)(−3.8 + j4.5 + 3 − j4)


(−3.8 + j4.5)(−3.8 + j4.5 + 3) + (−3.8 + j4.5 + j3)

(−3.8 + j4.5 + 4 − j3)
K (−0.8 + j8.5)(−0.8 + j0.5)
=
[(−3.8 + j4.5)(−0.8 + j4.5)(0.2 + j7.5)(0.2 + j1.5)]

8.5 0.5 4.5
φ = 180◦ − tan−1 + 180◦ − tan−1 180◦ + tan−1
0.8 0.8 3.8
4.5 7.5 1.5
+180◦ + tan−1 − tan−1 − tan−1
0.8 0.2 0.2
= −85◦ − 32◦ + 50◦ + 80◦ − 89◦ − 82◦
= −158◦

This is not an odd multiple of 180◦ . Hence, s = −3.8 + j4.5 is not a breakaway
point since it is not in the root locus.
6. The values of K at the breakaway points are calculated as given below. At
s = −1.3

1.3 × 2.7 × 4 × 4
K =
4.4 × 4.4
5.4 Procedure to Construct Root Locus 695

K = 2.9

At s = −2.8 ± j5.4,

6.08 × 5.4 × 2.7 × 8.5

K =
1.4 × 9.4

K = 57.3

Example 5.15

A certain feedback control system has the following forward path T.F.:

K (1 + 2s)
G H (s) =
s 2 (s + 4.5)

Draw the root locus for 0 ≤ K ≤ ∞ and find the repeated roots if any. What is the
value of K at this point? For this value of K , determine the closed loop T.F (Fig. 5.23).

Solution

1.
K (1 + 2s)
G H (s) =
s 2 (s + 4.5)

The given T.F. is rewritten in the following pole–zero form:

2K (s + 0.5)
G H (s) =
s 2 (s + 4.5)
K  (s + 0.5)
= 2
s (s + 4.5)

where K  = 2K . The poles and zeros are as shown in Fig. 5.23. The variable
parameter is K  .
2. There are three poles and one zero. Hence, there should be three root loci.
3. On the real axis, the root loci exist between s = −0.5 and s = −4.5. The two
root loci originating from the origin move through the complex s-plane, and
break in on the negative real axis. One locus ends at the zero at s = −0.5 and
other two go to ∞.
4. The centroid of the root locus is
696 5 Root Locus Method for Analysis

K j

Asymptote

Triple roots

K 0 K K 0
2 0.5
4.5

1.5
K 3.375
Asymptote

Fig. 5.23 Root locus for Example 5.15

−0 − 0 − 4.5 + 0.5
σ =
2
= −2

The angle of the asymptote is

(2n + 1) × 180◦
θ =
2
where n = 0, 1

θ = 90◦ , 270◦

5. The characteristic equation is given by

1 + G H (s) = 0
K  (s + 0.5)
1+ 2 =0
s (s + 4.5)
−(s 3 + 4.5s 2 )
K =
(s + 0.5)
5.4 Procedure to Construct Root Locus 697

dK
= 0 = (s + 0.5)(3s 2 + 9s) − (s 3 + 4.5s 2 )
ds
s 2 + 3s + 2.25 = 0
(s + 1.5)2 = 0
s = −1.5

The breakaway point is at s = −1.5. At the breakaway point, the value of K  is

obtained as
1.5 × 1.5 × 3
K = = 6.75
1
6.75
K = = 3.375
2

K = 3.375

6. The closed loop roots for K = 3.375 are (s + 1.5)3 . The closed loop T.F. is

C(s) 6.75(s + 0.5)

=
R(s) (s + 1.5)3

Example 5.16

Consider the following loop T.F. of a certain closed loop system:

K (s + 2)
G H (s) =
(s 2 + 2s + 2)

Show that part of the root locus is a circle. Determine the break-in point and the value
of K at this point. For this value of K , determine the closed loop T.F (Fig. 5.24).

Solution

1.
K (s + 2) K (s + 2)
G H (s) = =
(s 2 + 2s + 2) (s + 1 + j)(s + 1 − j)

There are two poles and one zero. Hence, there should be two root loci. The part
of the root locus is a circle which can be proved as follows.
2. Substitute s = σ + jω in G H (s)
698 5 Root Locus Method for Analysis

Root locus
GH(s) K(s 2)/(s2 2s 2)
1.5
j

K 0
1 1 j
Imaginary axis

0.5
K K
0
3.414 2 0
0.5 K 4.82

1 j
1 K 0

1.5
6 4 3 2 1 0 1
Real axis

Fig. 5.24 Root locus for Example 5.16

K (σ + jω + 2)
G H (σ + jω) =
(σ + jω)2 + 2(σ + jω) + 2
K (σ + jω + 2)
= 2
(σ + 2σ + 2 − ω2 ) + j2ω(σ + 1)
ω 2ω(σ + 1)
∠G H (σ + jω) = tan−1 − tan−1 2
σ +2 (σ + 2σ + 2 − ω2 )

To satisfy the angle criterion of root locus, we get

ω 2ω(σ + 1)
tan−1 − tan−1 2 = 180◦
σ +2 (σ + 2σ + 2 − ω2 )
ω 2ω(σ +1)
tan−1 = 180◦ + tan−1 2
σ +2 (σ +2σ +2−ω2 )

Taking tan on both sides and cross multiplying, we get

σ 2 + 2σ + 2 − ω2 = 2σ 2 + 6σ + 4
σ 2 + 4σ + ω2 + 2 = 0
(σ + 2)2 + ω2 = 2
5.4 Procedure to Construct Root Locus 699
√
This is the equation of a circle with centre at (−2, 0) and radius 2.
3. The root loci on the negative real axis exists between s = −2 and s = −∞. The
root loci starting from the complex conjugate poles break in on the −ve real
axis. One locus goes to the zero at s = −2 and the other goes to −∞.
4. The break-in point is calculated as −3.414. The value of K at this point is

2.61 × 2.61
K = = 4.82
1.414

K = 4.82

5. At the break-in point, the roots are repeated as (s + 3.414)2 . The closed loop
T.F. is obtained as
C(s) G(s)
=
R(s) 1 + G(s)
4.82(s + 2)
= 2
(s + 2s + 2) + 4.82(s + 2)

C(s) 4.82(s + 2)
=
R(s) (s + 3.414)2

1. Root loci are the trajectories of the characteristic equation when a system param-
eter is varied. It is an extensively used technique for the design of a compensator
for continuous and discrete time systems. However, in this chapter, it is con-
structed only for linear time invariant continuous time systems.
2. Magnitude criterion and angle criterion are used to construct root locus. The
angle criterion is used to locate the points in the root locus. Using the magnitude
criterion, the values of the variable parameter in the root loci are found.
3. Searching of points in the vast s-plane which lie in the root loci trajectories is a
very tedious process. However, by using certain properties of the root locus, the
job is very much simplified. These properties are also called rules to construct
root loci.
4. The properties of the root loci include the following:
(a) Starting and termination of root loci.
700 5 Root Locus Method for Analysis

(b) Number of root loci.

(c) Symmetry of the root loci.
(d) The centroid and angles of the asymptotes.
(e) Root loci on the real axis.
(f) Angle of arrival and angle of the departure of root loci.
(g) Intersection of the root loci with the imaginary axis and determination of
critical value of the gain.
(h) Breakaway and break-in points.
5. After drawing the root loci, using the properties, exact root loci are obtained by
choosing points closer to these loci and verifying whether they satisfy the angle
criterion.
6. From root loci, we can obtain the closed loop system poles and hence the closed
loop T.F. for any given value of K .
7. When time domain specifications are given, one can identify the point in the
dominant root loci. The value of K at this point is determined. For this value of
K , in the other root loci, closed loop roots are determined using the magnitude
criterion by trial and error.
8. It is possible to find the frequency domain specifications such as phase margin
and gain margin.
9. Using root locus, not only can we find the absolute stability of the system but
also the relative stability in time and frequency domains.
10. To draw the root locus, for positive feedback, the angle criterion is different. Some
of the properties are also different. By following the modified angle criterion and
the properties, root loci for a positive feedback system can be drawn.
11. The root locus drawn for −∞ ≤ K ≤ 0 is called complementary root locus.
Here, the angle criterion is the same as for the positive feedback system. Most
of the properties also resemble those for the positive feedback system.
12. It is not true that one can always use the properties of the root loci and construct
them. For certain transfer functions, the properties of root loci will not be of any
help to us.
13. For a few T.F.s, a part of root loci will be a circle. For such T.F.s, analytically
it can be derived that the root loci have the equation of a circle with centre and
radius. Knowing these two, one can quickly construct the exact root locus.
14. For a linear system with transportation lag, root locus can be drawn by expressing
the time lag in terms of a single pole and a single zero which is approximate
only.
15. A simple MATLAB program is available to draw exact root loci for any complex
T.F. We can obtain all required information from the plots so drawn.
5.4 Procedure to Construct Root Locus 701

Short Answer Type Questions

1. What is root locus?
Root locus is the trajectory of the closed loop poles as the gain of the system is
varied in the range 0 ≤ K ≤ ∞.
2. What is complementary root locus?
The root locus drawn for the variation of the system parameter K in the range
−∞ ≤ K ≤ 0 is called complementary root locus.
3. What criteria are used to construct root locus?
Magnitude and angle criteria are used to construct root locus.
4. What criterion decides the shape of the root locus trajectories?
Angle criterion decides the shape of the root locus trajectories.
5. How magnitude criterion is used in the root locus plot?
The root loci are drawn for the variation of the gain parameter K . The values of
K at different points of the root loci are determined using magnitude criterion.
At any point s1 in the s-plane, the K value is calculated from

Product of directed distances drawn from all the poles to the point s1
K =
Product of directed distances drawn from all the zeros to the point s1

6. How is angle criterion used to construct root loci?

The angle criterion is used to construct root locus in the cases described as
follows::
(a) Negative feedback; 0 ≤ K ≤ 0: For any point s1 in the complex s-plane
to be a point on the root locus, the difference of angles drawn from all the
zeros and the poles of G H (s) to the point s1 should be an odd multiple of
180◦ .
(b) Negative feedback; −∞ ≤ K ≤ 0: As above in (a) but the angle differ-
ence should be even multiple of 180◦ which includes 0◦ also.
(c) Positive feedback; 0 ≤ K ≤ ∞: As in the case of (b).
7. Are the root loci continuous curves or is any of them broken with disconti-
nuity?
Root loci are continuous curves. They are not broken in the middle.
8. What is the nature of symmetry of root locus in the complex s-plane?
Root loci are symmetrical with respect to the real axis of the complex s-plane.
They have the complex conjugate property.
9. What are the important properties of root loci?
Some of the important properties of root loci include the starting of the root
loci, the ending of the root loci, the number of root loci, the centroid, the angle
of intersection of the asymptote, breakaway and break-in points, the points at
702 5 Root Locus Method for Analysis

which root loci cross the jω axis, root loci on the real axis, angle of departure
and angle of arrival.
10. What information do we get from the properties of root loci?
The properties of root loci give guidance for the approximate root loci and not
the exact root loci.
11. Given the loop T.F. G H (s), how do you determine the total number of root
loci?
For the given G H (s), the number of root loci is determined from the number of
poles or zeros of G H (s) whichever is greater.
12. What are the starting and ending points of the root loci?
The root loci start from poles of G H (s) where the value of K = 0 and end at
zeros or ∞ where the value of K = ∞ if P, the number of poles of G H (s) is
greater than or equal to Z the number of zeros of G H (s).
13. There are three poles and five zeros. From where do the root loci start and
where do they end?
There will be five root loci. Three of them will start from the three poles and end
at three zeros. The remaining two root loci start from ∞ and end at the remaining
two zeros.
14. Do the zeros of the system change their positions as a result of variation of
the gain K ?
If the system gain K is varied, the positions of the zeros of the system in the
s-plane are unaffected.
15. How do you find the existence of root locus on the real axis for −∞ ≤ K ≤
∞?
For 0 ≤ K ≤ ∞, on the real axis root loci exist in the region to the right of which
an odd number of poles and zeros are present. For −∞ ≤ K ≤ 0, on the real
axis root loci exist in the region to the right of which an even number of poles
and zeros are present.
16. What are asymptotes of root loci?
The asymptotes of the root loci give guidance to the actual root loci at s → ∞.
17. How do you get the asymptotes for the given G H (s)?
The asymptotes are identified by their intersection and the angles with the real
axis. The intersect is given by
 
Real part of pole of G H (s) − Real part of zeros of G H (s)
σ =
(P − Z )

The angles of the asymptotes are given by

(2n + 1) × 180◦
θ=
(P − Z )

where n = 0, 1, 2, . . . , (P − Z − 1).
5.4 Procedure to Construct Root Locus 703

18. What is centroid of root locus?

The intersect σ of the asymptote is the centre of gravity of the root loci. The
intersect of the asymptote on the real axis is called the centroid.
19. How many asymptotes will be there for the given G H (s)? Will they intersect
at any other point in the s-plane other than the real axis?
There are as many asymptotes as there are the number of root loci. Since σ the
intersect of the asymptotes is a real number, asymptotes intersect the real axis
only.
20. What are breakaway and break-in points of root loci?
When two or more than two branches of the root loci meet at one point, they
depart from that point and reach their destiny. These points are called breakaway
or break-in points.
21. What are the properties of the breakaway and break-in points?
(a) For the given G H (s), there may be more than one breakaway or break-in
point and they may simultaneously present.
(b) These points may be complex conjugate in the s-plane.
(c) At these points, multiple order roots of the characteristic equation are
present.
22. What are the necessary conditions that the breakaway points must satisfy?
Will it be sufficient also?
The necessary condition that the breakaway point must satisfy is that

dK d
=0= [G H (s)]
ds ds
The above condition is not sufficient because the point obtained may not be a
point in the root locus for some real value of K . Hence, it is further necessary
and sufficient that the value of s so obtained is a point on the root locus.
23. What are the methods available to determine the point of intersection of the
root loci with the jω -axis?
(a) By using the Routh–Hurwitz stability criterion, the critical value K c can be
determined. Using this value in the auxiliary equation, the point of intersec-
tions of the root loci with the jω axis can be determined.
(b) By trial and error, any point ω1 in the jω-axis is chosen that satisfies the
angle criterion. By using the magnitude criterion, the critical value K c is
obtained.
24. How is the absolute stability of the closed loop system determined using root
locus plot?
For the given value of K , the roots of the characteristic equation are found in
their respective root locus. If all the roots so found are in LHP, the closed loop
system is absolutely stable.
704 5 Root Locus Method for Analysis

25. How to determine the relative stability from the root locus plot?
There are two methods to determine the relative stability from the root locus
plot.
(a) For the given value of K , determine the roots of the characteristic equations.
Of these roots, the roots which are closer to the jω-axis give the relative
stability. If they are away from the jω-axis and in LHP, the relative stability
is high.
(b) By finding the phase margin and gain margin, the relative stability can be
determined. The higher the phase margin and gain margin are and positive,
the higher will be the relative stability.
26. How do you determine phase margin and gain margin from root locus plot?
(a) Let K = K d be the desired value. Then the gain margin is given by

Kc
Gm =
Kd

where K c is the critical value of K when the root loci cross the jω-axis.
(b) Determine any point ω1 on the jω-axis that will satisfy the magnitude crite-
rion for the value of K = K d . At this point ω1 , find the phase angle difference
from zeros and poles. Let it be φ. Then phase margin is given by

φm = 180◦ − φ

27. When is the centroid not used and why?

When the difference between the number of poles and the number of zeros
(P − Z ) is either zero or one, the centroid is not used. For these values, the
closed loop poles do not move off the real axis on the way towards infinity.
28. What is the effect of addition of pole to G H (s)?
By addition of a pole to G H (s), the order of the system is increased by one.
Now the root loci bend towards the right and in general the system becomes less
stable.
29. What is the effect of addition of zero to G H (s)?
By addition of zero to G H (s), the root loci bend towards left. In General, the
system becomes more stable.
30. Write down the expression for the angle of departure and angle of arrival.
The angle of departure is given by

θd = 180◦ − (θ P − θ Z )
θa = 180◦ − (θ Z − θ P )

where θ P is the sum of angles of vectors drawn from other poles to the pole
concerned and θ Z is the sum of angle of vectors drawn from other zeros to the
pole concerned.
5.4 Procedure to Construct Root Locus 705

31. Can root locus be drawn for a system with delay? If not, how is it accounted
for to draw the root locus?
Root locus is drawn for rational polynomials of the T.F. When the time delay
e−T s is considered, the polynomial becomes irrational and root locus cannot be
applied. However, using Pade approximation e−T s is approximated as

(s − T2 )
e−T s = −
(s + T2 )

32. How do you find the angle of arrival and angle of departure of root loci at
the breakaway and break-in points?
If there are n root loci that meet at breakaway or break-in points, they arrive or
◦
depart at 180
n
.
33. For the givenK , how do you find the closed loop poles and the closed loop
T.F.?
If the value of K is given, in each root locus, using magnitude criterion, deter-
mine the roots by trial and error. These roots are nothing but the roots of the
characteristic equation or the poles of the closed loop system. The closed loop
T.F. is determined using the equation

C(s) G(s)
=
R(s) 1 + G(s)H (s)

The numerator of [1 + G H (s)] is represented by the poles of the characteristic

equation.
34. For a given damping ratio ζ , how do you find the closed loop poles from
root locus?
For the dominant complex conjugate poles, the damping factor is defined as
φ = cos−1 ζ . Draw a radial line from the origin of the s-plane with an angle
(180◦ − φ) which cuts the dominant root loci. Using magnitude criterion, find
the value of K at this point. This point and the conjugate of this give the closed
loop roots. For the remaining root loci, the roots are found, by trial and error for
the value of K fixed on the dominant root loci. All these poles represent closed
loop poles.
35. For the following Loop T.F., sketch the root locus for 0 ≤ K ≤ ∞.

K (s − 2)
(a) G H (s) =
(s + 2)(s + 3)
K (s + 4)(s + 2)
(b) G H (s) =
(s 2 + 4)
K (s + 2)(s + 4)
(c) G H (s) =
(s − 2)(s − 4)
K
(d) G H (s) = 2
(s + 4)
706 5 Root Locus Method for Analysis

K
(e) G H (s) =
+ 4)
s(s 2
K
(f) G H (s) = 2
(s − 4)

36. Consider the following loop T.F. of a certain feedback control system.

K (s + a)
G H (s) =
s(s + b)

Under what condition is part of the root loci a circle? If this condition is not
satisfied, what will be the nature of damping available in the system?
For part of the root loci to be circle a > b. If a < b, the root loci exist only on the
negative real axis, and the system is over-damped for all values of K (Figs. 5.25,
5.26, 5.27, 5.28, 5.29, 5.30, 5.31, 5.32, 5.33, 5.34, 5.35, 5.36, 5.37, 5.38, 5.39
and 5.40).
Long Answer Type Questions
1. A certain closed loop unity feedback control system has the following open loop
T.F.:
K
G(s) =
s(s + 5)(s + 10)

Draw the root locus for 0 ≤ K ≤ ∞. For K = 70, find the closed loop roots.
For this value of K , determine the closed loop T.F.
Ans: Closed loop poles are at s1 = −11.05; s2,3 = −2 ± j1.6

C(s) 70
=
R(s) (s + 11.05)(s 2 + 4s + 6.6)

2. Consider the following open loop T.F. of certain closed loop system.

1
G(s) =
s(1 + 0.5s)(1 + 0.2s)

Determine the damping ratio and natural frequency of oscillations of the domi-
nant roots. Find the closed loop T.F. using the root locus method.
Ans: ζ = cos 60◦ =0.5; ωn = 1.5 rad/s. For K = 10, s1 = −5.5; s2,3 = −0.75 ±
j1.3

C(s) 10
=
R(s) (s + 5.5)(s 2 + 1.5s + 2.3)

3. Consider the following open loop T.F. of a certain closed loop system:
5.4 Procedure to Construct Root Locus 707

(a) j (b) j
K 0
j2

K K 0 K 0 K K K
3 2 2 4 2 0

2.73 j2
K 0

(c) (d) K j
j
K 0 j2
K 0

K K K 0 K 0
4 2 0 2 4 0

K 0 j2
K
(e) j K (f) K

K 0
j2
K 60° K 0 K 4 K 0
0 K 0 2 0 2
K 0
j2

K
K

Fig. 5.25 Root locus for Question 35

K (s + 9)
G H (s) =
s(s 2 + 4s + 11)

Draw the root locus for 0 ≤ K ≤ ∞. Determine the value of K to have a damping
ratio of ζ = 0.3 for the dominant roots. For this value of K , determine gain
margin. What is the closed loop T.F.?
Ans: K = 2.26; Gain margin, G m = 2.2

C(s) 2.26(s + 9)
=
R(s) (s + 2)(s 2 + 2s + 10)
708 5 Root Locus Method for Analysis

7.07
Kc 750

K 70 2 j1.6

K K 70 K 0 K 0 60° K 0
11.05 10 5 K 48 2.1 0

K 70 2 j1.6

K
Fig. 5.26 Root locus for G H (s) =
s(s + 5)(s + 10)

4. Consider the following loop T.F. of a certain closed loop system:

K (s + 1)
G H (s) =
s(s + 4)(s 2 + 6s + 10)

Draw the root locus for 0 ≤ K ≤ ∞. For K = 18, determine the closed loop
system roots. For this value of K , determine the phase margin and gain margin.
Ans: K = 18 and the closed loop system roots are s1 = −5.7; s2 = −0.39; s3 =
−1.4 + j2.4; and s4 = −1.4 − j2.4, φm = 93◦ positive and G m = 12 positive.
5. Consider the following open loop system of a certain unity negative feedback
control system:

K (s + 2)
G H (s) =
(s + 1)2
5.4 Procedure to Construct Root Locus 709

10
Kc 70

0.75 j1.3
K 10
n
1.5

K K 10 K 0 K 0 60° K 0
5.5 5 2 K 4 0.88 0

K 10 0.75 j1.3

K
Fig. 5.27 Root locus for G H (s) =
s(s + 2)(s + 5)

Draw the root locus for 0 ≤ K ≤ ∞. Show that part of the root locus is a circle.
Find the centre and radius of the circle.
Ans: Centre is at c = (−2, 0) and radius r = 1. Break-in point is at s = −3,
where K = 4.
6. Consider the following loop T.F. of a certain closed loop system. Sketch the root
locus. Identify the value of K for which the system is stable.

K
G H (s) =
s 2 (s 2 + 4s + 10)

Ans: For all values of K , where 0 ≤ K ≤ ∞, there are two roots in the RHP.
The system is always unstable.
7. A certain closed loop control system has the following open loop T.F.:

K (s + 2)(s 2 + 4)
G H (s) =
(s 2 − 16)
710 5 Root Locus Method for Analysis

j K

1 j3
K 2.26

Kc 4.9
3.32
K 0
2 j2.65

cos 10.3 72.5

K K 0
9 K 2.26 0 2.5
2

K 2.26
Gm 2.2 K 0
2 j2.65

K (s + 9)
Fig. 5.28 Root locus for G(s) =
s(s 2 + 4s + 11)

Draw the root for 0 ≤ K ≤ ∞. For what value of K is the system stable? For
what value of K does the system have repeated poles and what are these pole
locations? Is the system stable for this value of K ?
Ans:
1. The system is stable for K > 2.
2. The breakaway point is at s = −6.67.
5.4 Procedure to Construct Root Locus 711

j K
5.05
Kc 215.6

K 18
1.4 j2.4

K 18 j
3 j
K 0
1 0.5
K 18 K 0 60º K 0
K 5.7 4 3 1 0.39 0
K 18
3 j
K 0

1.4 j2.4

Gm 12 K 18

1 0.5 rad/sec
m 95

K (s + 1)
Fig. 5.29 Root locus for G(s) =
s(s + 4)(s 2 + 6s + 10)

3. The repeated poles are (s + 6.67)2 and the corresponding value of

K = 0.126.
4. The system is unstable for this value of K since K < 2.
8. A certain control system has the following open loop T.F.:

K (s − 2)(s − 3)
G H (s) =
s(s + 2)

Draw the actual root locus for 0 ≤ K ≤ ∞. Show that part of the root locus is
a circle. Find the centre and radius of the circle. Find the value of K for which
712 5 Root Locus Method for Analysis

Root locus
GH(s) K(s 2)/(s 1)2
1
j
0.8

0.6

0.4
Imaginary axis

0.2

K K 4 K K 0
0
2 1 0

0.2 3
0.4

0.6

0.8

1
6 4 3 2 1 0 1
Real axis

Fig. 5.30 Root locus for Question 5

the damping is zero. For what value of K is the system critically damped? For
this value of K , determine the closed loop poles and hence closed loop T.F.
Ans:
1. Centre of the circle, c = 0.86.
2. Radius of the circle, r = 1.55.
3. The damping is zero for K = 0.4.
4. For K = 0.091, the system is critically damped.
5. For K = 0.091, the closed loop poles are (s + 0.69)2 .
6. The closed loop T.F. for the above value of K is

C(s) 0.0834(s − 2)(s − 3)

= (5.13)
R(s) (s + 0.69)2

9. A certain feedback control system has the following loop T.F.:

K (s 2 − 4s + 5)
G H (s) =
(s + 1)(s + 2)

Draw the root locus for 0 ≤ K ≤ ∞. Find the value of K for which the system
is critically stable. Determine the value of K for which the closed loop system
5.4 Procedure to Construct Root Locus 713

Root locus
GH(s) K/s2(s2 4s 10)
10
K j K
8

4
K 0
Imaginary axis

2 2 j2.5

0
K 0
2 K 0
2 j2.5
4

8
K K
10
10 8 6 4 2 0 2 4 6 8
Real axis

K
Fig. 5.31 Root locus for G H (s) =
s 2 (s 2 + 4s + 10)

poles are repeated. What are these repeated roots? Is any part of the root locus
circular. If so find the centre and radius.
Ans:
1. The critical value of K is K c = 0.75.
2. At K = 0.0192, the root loci break away, and there exist double roots at
s = −1.44.
3. The root locus is a circle with centre at s = 0.38 and radius r = 1.82.
10. A certain closed loop system has the following open loop T.F.:

K (s 2 − 4s + 5)
G H (s) =
s(s + 1)(s + 2)

Draw root locus for 0 ≤ K ≤ ∞. Determine the breakaway point and the value
of K at that point. What is the critical value of K for the system to be stable?
How does the addition of a pole at s = 0 affect the stability of the system.
Ans:
1. The breakaway point is at s = −0.34 and the corresponding value of K =
0.058.
714 5 Root Locus Method for Analysis

Root locus
15
j

K
10 j2

5
Imaginary axis

K K 0.126 K 0 K K 0
0 K 2
6.67 2 0 4
4

10
j2
K

15
15 10 5 0 5
Real axis

K (s + 2)(s 2 + 4)
Fig. 5.32 Root locus for G H (s) =
(s 2 − 16)

2. The critical value of K is K c = 0.4 which is 0.75 when a pole at s = 0 is

not added. Thus, the addition of a pole reduces the stability margin.
11. Consider the following loop T.F. of a certain closed loop system:

K (s − 2)(s − 3)
G H (s) =
(s + 2)(s + 3)

Draw the root locus for 0 ≤ K ≤ ∞. In this case, the pole at the origin of
the system described in Question 8 is shifted to s = −3. Compare the critical
stability limit and the transient response of these two systems.
Ans:
1. The critical value of stability is increased from K c = 0.4 to K c = 1.
2. For 0.01 < K < 1, the system is under-damped. However, the system
described in Question 8 is under-damped for 0.091 < K < 0.4. For any
given value of gain K , the system described in Question 11 is better damped
than the one given in Question 8.
12. Consider the system described in Question 11. The locations of the poles and
zeros are interchanged and the loop T.F. is
5.4 Procedure to Construct Root Locus 715

Root locus
1
j
0.8 j1.3
Kc 0.4
0.6
r 1.55
0.4
Imaginary axis

0.2
K 0 K 0 K K 0
0
2 0 0.86 2 3

0.2 0.69 2.41

K 0.091 K 44
0.4

0.6

0.8

1
6 2 1 0 1 2 3 4
Real axis

K (s − 2)(s − 3)
Fig. 5.33 Root locus for G H (s) =
s(s + 2)

Root locus
2
j
Kc 0.75
1.5
j1.81 K 0
K
r 1.82
1
2 j

0.5
Imaginary axis

K 0 K 0
0
2 1 0 0.38

0.5
1.44
K 0.0192
K
1
2 j
1.5

2
2.5 4 3 2 1 0 1 2 3 4 0
Real axis

K (s 2 − 4s + 5)
Fig. 5.34 Root locus for G H (s) =
(s + 1)(s + 2)
716 5 Root Locus Method for Analysis

Root locus
1.5
j K
2 j
Kc 0.4 0.73
1
Imaginary axis

0.5
K K 0 K 0
0
2 1 0
0.5
0.34
K 0.058

1
K
2 j
1.5
6 5 4 3 2 1 0 1 2 1
Real axis

K (s 2 − 4s + 5)
Fig. 5.35 Root locus for G H (s) =
s(s + 1)(s + 2)

Root locus
3
K 0 j
j2.45
2 Kc 1

1 r 2.5
Imaginary axis

K 0 K 0 K 0 K
0
3 2 0 2 3
2.5 2.5
1 K 0.01 K 100

3
4 3 2 1 0 1 2 3 0
Real axis

K (s − 2)(s − 3)
Fig. 5.36 Root locus for G H (s) =
(s + 2)(s + 3)
5.4 Procedure to Construct Root Locus 717

Root locus
3
j
j2.45
2 Kc 1

1
Imaginary axis

K K K 0 K 0
0
3 2 0 2 3

2.5 2.5
1 K 0.01
K 100

3
4 3 2 1 0 1 2 3 0
Real axis

K (s + 2)(s + 3)
Fig. 5.37 Root locus for G H (s) =
(s − 2)(s − 3)

Root locus
5 j
4 j4
4 j3.2 K
K 0.375
3

2
Imaginary axis

1
K 0 K 0 39º
0
6 5 0
1
5.5
2 K 5.4 10 3

3 K 0.375 4 j3.2

4 j4
K
5
5 6 5 4 3 2
Real axis

K (s 2 + 16)
Fig. 5.38 Root locus for G H (s) =
(s + 5)(s + 6)
718 5 Root Locus Method for Analysis

Root locus
4

K 1.25
3 j3
1.5 j3 K

1
Imaginary axis

K 0 K 1.25
0
3 0.85 K 0

1.5 j3 K
3 K 1.25 j3

4
4 3 2.5 2 1.5 1 0.5 0 0.5
Real axis

K s(s 2 + 9)
Fig. 5.39 Root locus for G H (s) =
(s + 3)3

K (s + 2)(s + 3)
G H (s) =
(s − 2)(s − 3)

Draw the root locus for 0 ≤ K ≤ ∞. Comment on the stability limit and transient
response of these two systems.
Ans:
1. The critical value of K for stability in both the cases is K c = 1. However, for
K > 1, the system becomes more stable and for K < 1, it becomes unstable
which is the reverse process of the system in Question 11.
2. As the gain is increased in the range 1 ≤ K ≤ 100, the system is under-
damped and hence one can choose any value of K in this range which is
wider so that the desired damping ratio is achieved.
13. Consider the following loop T.F. of a closed loop system:

K (s 2 + 16)
G H (s) =
(s + 5)(s + 6)

Draw the root locus for 0 ≤ K ≤ ∞. Find the roots of the characteristic equation
for a damping factor of ζ = 0.78.
5.4 Procedure to Construct Root Locus 719

Root locus
2.5
K
2
j2
1
1.5

1 1 j
K 0
0.5
Imaginary axis

K K 0 K 0 K
0
3 2 0
0.5
K 0
1 1 j

1.5

2 j2
K
2.5
4 3.5 3 2.5 2 1.5 1 0.5 0 0.5 1
Real axis

K s(s 2 + 4)
Fig. 5.40 Root locus for G H (s) =
(s + 2)(s + 3)(s 2 + 2s + 2)

Ans: For ζ = 0.78, K = 0.375. The roots of the characteristic equations are
s1 = −4 + j3.2 and s = −4 − j3.2.
14. Consider the following open loop T.F.:

K s(s 2 + 9)
G H (s) =
(s + 3)3

Draw the root locus for 0 ≤ K ≤ ∞. For K = 1.25, determine the closed loop
roots and hence the closed loop T.F.
Ans:
1. For K = 1.25, the closed loop roots are s1 = −0.85, s2 = −1.5 + j3 and
s3 = −1.5 − j3.
2. The closed loop T.F. is

C(s) 0.555 s(s 2 + 9)

=
R(s) (s + 0.85)(s 2 + 3s + 11.25)

15. Consider the following open loop T.F. of a certain closed loop system:
720 5 Root Locus Method for Analysis

Root locus
15
K j

10

5 K 0
Imaginary axis

1 j
K
0
2 0
1 j
K 0
5 Double zeros
Asymptote

10

15 K
4 2 1.5 1 0.5 0 0.5
Real axis

Fig. 5.41 Root locus for G H (s) = K s 2 /(s 2 + 2s + 2)2

Root locus
2
j

1.5 1 24º
d

1
1 j
K 0
0.5
Imaginary axis

K K 0 Double zeros
0 3 0
4
K
0.5
1 j
K 0
1

1.5
d 1 24º
2
8 7 6 5 4 3 2 1 0 1
Real axis

Fig. 5.42 Root locus for G H (s) = K (s + 3)2 /(s + 4)(s 2 + 2s + 2)

5.4 Procedure to Construct Root Locus 721

K s(s 2 + 4)
G H (s) =
(s + 2)(s + 3)(s 2 + 2s + 2)

Draw the root locus for 0 ≤ K ≤ ∞.

16. A certain closed loop system has the following open loop T.F.:

K s2
G H (s) =
(s 2 + 2s + 2)2

Draw the root locus for 0 ≤ K ≤ ∞ (see Fig. 5.41).

17. A certain closed loop control system has the following open loop T.F.:

K (s + 3)2
G H (s) =
(s + 4)(s 2 + 2s + 2)

Draw the root locus for 0 ≤ K ≤ ∞ (see Fig. 5.42).

Chapter 6
Design of Compensators

Learning Objectives

After completing this chapter, you should be able to:

 Understand the performance criteria of compensators.
 Design compensator in the frequency and time domains.
 Design lead compensator in the frequency and time domains.
 Design lag compensator in the frequency and time domains.
 Design lag–lead compensator in the frequency and time domains.

6.1 Introduction

It is the intention of the control engineer to design a system which gives optimum
performance. However, very often it happens that the specifications offered by dif-
ferent users vary, and therefore, it becomes necessary to do some adjustments either
in the design of the system or insert some external devices to obtain the required
performance. The control system design engineer has used his ingenuity to design
the system or process to give optimum performance in general. Therefore, the given
system has been altered as much as possible to meet the performance specifications
of many users. Further alterations of the plant parameters may not give optimum
performance in general. Now the option is to insert some devices according to the
requirement of the individuals so that the already designed system together with
the additional devices inserted gives optimum performance. These additional com-
ponents or devices which are inserted into the system equalize or compensate the
performance deficiency. These compensating devices may include electric mechani-
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 723
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_6
724 6 Design of Compensators

cal, hydraulic or pneumatic components and they are called compensators. Thus, the
process of adjustment or alteration of a control system to get the desired performance
specification is called compensation. The additional components or circuits that are
inserted into the system in order to compensate for a performance deficiency are
called compensators. In this chapter, we discuss the design procedure for electric
circuit compensators.

6.2 Performance Criteria for Compensators

A control engineer uses his ingenuity to have a well designed control system that must
have less transient and steady state error, good accuracy, good speed of response,
good relative stability and good damping so that undue overshoot, etc. can be avoided.
The first attempt to achieve the above requirements is by the adjustment of the gain.
However, by adjustment of gain alone it is very difficult to achieve satisfactory
result. A higher value of the gain improves the steady state behaviour but at the
cost of transient response which becomes poor. Thus, the task becomes challenging
which requires adding external devices called compensators to fulfil the specification
requirements of the individual.
Control system design problem involves two basic steps. The first step involves
the choice of the components that should be inserted into the system and the second
step is about the location of these devices. Generally, compensators are inserted in the
forward path, feedback path or the input path of the feedback loop or the combination
of the above. Having chosen the location of the compensator in the control loop,
the problem is to design the components that act together to compensate for the
performance deficiency of the original system. Care should also be taken to use
minimum components since the use of more comments becomes more expensive.
The compensators are located in a feedback system as:
1. Series compensator
2. Parallel compensator
3. Series-parallel compensator
A series or cascade compensator with the T.F. G c (s) connected in series with the
plant whose T.F. is G(s) is shown in Fig. 6.1a. In this scheme, the signal flows from
a lower energy level to a higher energy level. Therefore, additional amplifiers are
required to increase the gain as well as to provide accessory isolation. It has more
number of components and hence it is expensive.
The location of the parallel compensator is shown in Fig. 6.1b. In this scheme, the
flow of energy is from higher energy level to lower energy level. Therefore, additional
amplifiers are not required in this scheme. However, series compensation is simpler
than parallel (feedback) compensation.
The location of the series-parallel compensator is shown in Fig. 6.1c. In this
scheme, the advantage of series and parallel schemes are achieved.
In control system design, the transient response specifications are more important
and ultimate measures of performance indices. These specifications include: peak
6.2 Performance Criteria for Compensators 725

(a)

R(s) C(s)
Gc(s) G(s)

H(s)

(b)

R(s) C(s)
G1(s) G2(s)

Gc(s)

H(s)

(c)

R(s) C(s)
Gc(s) G(s)

Gc(s)

H(s)

Fig. 6.1 a Location of a series compensator; b Parallel compensator and c Series-parallel compen-
sator

overshoot, rise time, settling time and the steady state error. The root locus method
is used to design compensators which satisfy time domain specifications. By this
method, the shape of the root locus is altered by the addition of poles and zeros of
the compensator so that the required performance specifications are met.
Although a variety of frequency response methods are available for the system
analysis, the design of electric network compensator is carried out using either Bode
diagram or Nichols chart. The one big advantage of using Bode diagram in preference
to Nichols chart is that, while designing a compensator in the frequency domain, if the
given specifications are not met, then it is necessary to reshape only that particular part
of the Bode diagram which is contributed by the pole or the zero of the compensator.
726 6 Design of Compensators

It is easy to identify this in the Bode diagram, whereas in the Nichols chart, the entire
frequency response plot has to be redrawn. Thus, we choose, Bode diagram to design
the compensator in the frequency domain.

6.3 Time and Frequency Domain Approaches

We choose the root locus method to design a compensator to meet the time domain
specifications. The poles and zeros of the compensator reshape the root locus of the
system so that the closed loop poles are placed in locations in the s-plane that give
desired transient response. In the frequency domain approach, the Bode diagram is
used for the design of the compensator. When the compensator is added to the Bode

low Middle frequency High frequency

region (transient response)
+dB frequency region (relative stability)
−20dB/dec.
region
(steady state)
ωgc
0
log ω
−dB Gm

−40dB/dec.

−60dB/dec.

φ
ωpc
0
log ω

−90º
−φ φm

−180º

−270º

Fig. 6.2 Frequency response characteristic illustrating, low, middle and high frequency regions
6.3 Time and Frequency Domain Approaches 727

plot of the original system, only the portion that corresponds to the poles and zeros
of the compensator need to be reshaped in the original Bode diagram. Further, if
the open loop gain is varied, the magnitude curve of the Bode plot is shifted up or
down without changing the slope of any part of Bode diagram while retaining the
original phase curve. In the Bode diagram, the frequency range which is far below
the gain crossover frequency is termed as low frequency region. This region indicates
the steady state behaviour of the closed loop system. The region nearer to the gain
crossover frequency is called medium frequency region which indicates the relative
stability of the closed loop system. The region far above the gain crossover frequency
is called high frequency region and indicates the complexity of the system. In this
region, it is difficult to control the transient response of the system. Therefore, to
achieve steady state accuracy, the gain in the low frequency region should be as
high as possible. To get satisfactory relative stability, the slope of the log-magnitude
curve in the Bode diagram near the gain crossover frequency should be −20 dB/dec
and to achieve the desired phase margin, this slope should be extended over a wide
frequency range. To improve the transient response and to suppress the effects of
noise, the gain should be attenuated in the high frequency region of the Bode diagram.
This is illustrated in Fig. 6.2.

6.4 Compensator Design in the Frequency Domain

We describe below the design methodology for series (cascade) compensators in the
frequency domain. The most commonly used series compensators are:
1. Lead compensator
2. Lag compensator
3. Lag–lead compensator.
We discuss below, the design of the above compensators in the frequency domain
using electric components.

6.4.1 Design of Lead Compensator

A typical lead compensator electric network is shown in Fig. 6.3. The transfer func-
tion of the lead compensator is determined as given below. Let Z 1 be the equivalent
impedance of the parallel combination of R1 and C

R1
Z1 =
(1 + R1 Cs)

The T.F. of the compensator is

728 6 Design of Compensators

Fig. 6.3 Lead compensator

electric network
C

R1
ei(t) R2 eo(t)

Eo R2
(s) =
Ei R2 + (1+RR11 Cs)
R2 (1 + R1 Cs)
=
[(R1 + R2 ) + R1 R2 Cs]

Eo R2 (1 + R1 Cs)
(s) =  
Ei (R1 + R2 ) 1 + R2 R1 Cs
(R1 +R2 )

Eo (1 + T s)
(s) = α (6.1)
Ei (1 + αT s)

where
R2
α= <1
R1 + R2
T = R1 C

The sinusoidal T.F. of Equation (6.1) is written as

Eo α(1 + jωT )
( jω) =
Ei (1 + jαωT )

Eo α (1 + ω2 T 2 )
( jω) =  ∠tan−1 ωT − tan−1 αωT (6.2)
Ei (1 + α 2 ω2 T 2 )

From Eq. (6.2), it is evident that the phase contribution of the lead network over the
frequency range 0 ≤ ω ≤ ∞ is from 0◦ to 90◦ . In other words, the output phase leads
the input and that is why it is called a lead network.
The lead network is inserted into the system so that the system performance
deficiency is compensated. For this, appropriate values for α and T are to be chosen.
6.4 Compensator Design in the Frequency Domain 729

α and T depend on network parameters R1 , R2 and C. Some useful information is

obtained from the frequency response characteristics of the lead network which are
given below.
The pole–zero diagram is shown in Fig. 6.4a. For the small value of α, the pole
is located far to the left in LHP. The smaller the value of α, the higher is the angle
contribution from the lead compensator. However, the lower limit of α is normally
restricted to α = 0.07 from the view point of physical realization. Further, if 1/α
becomes high, noise is amplified. For α = 0.07, the maximum phase lead contribu-
tion is restricted to θm = 60◦ . If the phase lead requirement exceeds 60◦ , then it is
advisable to connect two-phase lead compensators in a cascade.
The polar plot of lead compensator is drawn using Eq. (6.2) and is shown in
Fig. 6.4b. From Fig. 6.4b, the following relationships are obtained.
Let ωm be the frequency at which the phase angle contribution from the lead
network is maximum. Let this be denoted by θm . From Eq. (6.2), for ω = 0,
|G( jω)| = α∠0◦ . For ω = ∞; G( jω) = 1. Thus, the polar plot is a semi-circle with
centre (1 + α)/2 and radius r = (1 − α)/2. √ From the right angle triangle O AB, the
value of G( jω) at point A is obtained as α. Further

(1 − α)/2 (1 − α)
sin θm = =
(1 + α)/2 (1 + α)

(1 − sin θm )
α= (6.3)
(1 + sin θm )

The frequency ωm at which the maximum phase angle is contributed is obtained as

follows:

θm = tan−1 ωm T − tan−1 αωm T

ωm (1 − α)T
= tan−1
(1 + ωm2 T 2 α)
ωm (1 − α)T
tan θm =
(1 + ωm2 T 2 α)

From Fig. 6.4b and the triangle O AB, we get

(1 − α)
tan θm = √
2 α
(1 − α) ωm (1 − α)T
√ =
2 α (1 + ωm2 T 2 α)
730 6 Design of Compensators

(a) j s-plane

0
1/ T 1/T

(b) Im

m
A
Gc(j ) plot
(1 )/2

m 0
0 B 1 Re

(1 )/2

(c)
+dB
1/T 1/ T
log ω

20dB/dec.
20 log
−dB

−90º

φ θm

0
ωm log ω

Fig. 6.4 a Pole–zero diagram of lead compensator. b Polar plot of lead compensator. c Bode
diagram of lead compensator
6.4 Compensator Design in the Frequency Domain 731

Solving the above equation, we get

1
ωm = √ (6.4)
αT

The Bode diagram for the lead network is drawn using Eq. (6.1). The corner frequen-
cies are 1/T and 1/αT . The log-magnitude versus ω and phase angle φ versus ω are
shown in Fig. 6.4c. From the magnitude curve, it is evident that the low frequencies
are attenuated and the high frequencies are passed and acts as a high-pass filter. From
the phase angle characteristics, it is observed that the maximum phase θm contributed
by the lead network occurs at ω = ωm which is the geometric mean of the two corner
frequencies and is given by Eq. (6.4). The geometric mean is given by
 
1 1 1
log ωm = log + log
2 T αT
1
ωm = √
αT

The same result is derived from the polar plot shown in Fig. 6.4b.

6.4.2 Design Procedure for Lead Compensator in the

Frequency Domain

The phase lead compensator is inserted into the system to compensate for the defi-
ciencies in the frequency response specifications such as phase margin, gain margin,
resonant peak and steady state error. The primary objective of inserting the phase lead
network is to reshape the frequency response of the system by providing phase lead
angle which offsets the excessive phase lag that is inherently present in the original
system. By doing so, the deficiencies in the frequency response specifications are
compensated. The step-by-step procedure is given below:
1. Let G(s) be the T.F. of the uncompensated system and G c (s) be the T.F. of the
lead compensator as given in Eq. (6.1). Let K be the gain of the uncompensated
system. Determine the gain K which satisfies the steady state error requirement
which is normally given in terms of static error constants.
2. For this value of K obtained in step 1, draw the frequency response plot for G( jω)
in the Bode diagram and determine the phase margin.
3. If the phase margin determined in step 2, it is not sufficient enough to meet the
given phase margin in the specification, determine the amount of deficiency which
is to be contributed by the lead compensator. Let it be φ.
4. When the lead compensator is inserted into the system, the gain crossover fre-
quency of the composite system (original system + compensator) is shifted to the
732 6 Design of Compensators

right and the phase margin is now decreased. To account for this, add an additional
phase angle of roughly about 7◦ to 12◦ in φ obtained in step 3. This gives θm .
5. √ determine the attenuation factor α.
Using Eq. (6.3),
6. At ωm = 1/ αT , the maximum phase shift θm occurs. At this frequency, the
magnitude of G( jω) is −20 log( √1α ). Knowing α, determine ωm . This frequency
is the gain crossover frequency ωgc of the compensated system G c ( jω)G( jω).
Make ωm = ωgc .
7. Knowing α from step 5 and ωgc = ωm in step 6, determine T using Eq. (6.4).
Thus, α and T , the lead compensator parameters are determined. The zero of the
compensator is at s = −1/T and the pole of the compensator is at s = −1/αT .
8. For the compensated system, determine whether the given specifications are sat-
isfied. If not, change the locations of the poles and zeros of the compensator and
repeat the design steps from 3 to 7.

Example 6.1

The open loop T.F. of an uncompensated system is

5
G(s) =
s(s + 2)

Design a lead compensator for the system so that the static velocity error constant
K v ≥ 20 sec−1 , the phase margin is at least 55◦ and the gain margin is at least 12 dB.

(Anna University, December, 2005)

Solution
5
G(s) =
s(s + 2)
s5
K v = lim sG(s) = lim = 2.5
s→0 s→0 s(s + 2)

But the required K v = 20. Hence, an amplifier with a gain of 8 is to be connected in

cascade with the compensator. Now the T.F. becomes

20
G(s) =
s(1 + 2s )

(To draw the Bode plot, the T.F. is to be in time constant form). The Bode plot of the
uncompensated system is shown in Fig. 6.5a.
The Bode diagram for the uncompensated system is sketched as shown in Fig. 6.5a.
From magnitude plot, the gain crossover frequency ωgc is obtained as 6.32. Corre-
sponding to this the phase angle is
6.4 Compensator Design in the Frequency Domain 733

(a)

−20dB/dec.

20 −40dB/dec.
+dB
ωgc 6.32

0
2 ωm 9.8 log ω

−7.65
−dB

0
log ω
−φ

−90º
−162.4º Phase plot
φm 17.6º
−180º

(b)
R(s) + 5.82(s + 4.06) 5 C(s)
8 Gc(s) =
− (s + 23.63) s(s + 2)
G (s )

(c)
−20dB/dec.

20
−40dB/dec.
+dB −20dB/dec.
ωgc ωm 9.85
7.7
0
2 4.06 23.63 log ω
−7.6

−dB −40dB/dec.

0
log ω

−φ
−90º
−123.5º
φm 56.5º
−180º

Fig. 6.5 a Bode diagram of uncompensated system. b Block diagram of the compensated system.
c Block diagram of the compensated system
734 6 Design of Compensators

6.32
φ = −90 − tan−1 = −162.4◦
2
Hence, the phase margin of the uncompensated system is 180◦ − 162.4◦ = 17.6◦ .
However, the required phase margin is 55◦ . Hence, the additional phase lead to be
provided by the lead compensator is 55◦ − 17.6◦ = 37.4◦ . Adding another 7.6◦ for
the effect of ωgc shifted to the right, θm = 37.4◦ + 7.6◦ = 45◦ . Using Eq. (6.3), we
get

(1 − sin θm ) (1 − sin 45◦ )

α= = = 0.172
(1 + sin θm ) (1 + sin 45◦ )

As stated in step 6
√ √
20 log α = 20 log 0.172 = −7.65 dB

Corresponding to −7.65 dB, the new gain crossover frequency which is nothing but
ωm is obtained from the G( jω) magnitude plot as
ωm
40 log = 7.65
6.32
ωm = 9.8

Using Eq. (6.4), we get

1 1
T = √ = √ = 0.246
ωm α 9.8 0.172
1 1
= = 23.63
αT 0.172 × 0.246
αT = 0.0423

The compensator T.F. is

α(1 + T s) 0.172(1 + 0.246s)

G c (s) = =
(1 + αT s) (1 + 0.0423s)

Taking 0.172 into the over all gain of the system which will ultimately satisfy the
given K v , the T.F. of the compensated system is obtained as

20(1 + 0.246s)
G c (s)G(s) =
(1 + 0.0423s)s(1 + 2s )

The block diagram of the compensated system is shown in Fig. 6.5b

6.4 Compensator Design in the Frequency Domain 735

T = 0.246
α = 0.172

Now we draw the Bode diagram for the compensated system

20(1 + 0.246s)
G c (s)G(s) =
(1 + 0.0423s)s(1 + 2s )
ω
φ = tan−1 0.246ω − 90 − tan−1 − tan−1 0.0423ω
2
The following table is prepared for the phase angle:

ω 0 1 2 3 5 10 ∞
φ −90◦ −105◦ −114◦ −117◦ −119◦ −124◦ −180◦

The Bode diagram for the compensated system is shown in Fig. 6.5c.
The phase margin is checked as follows.
At ω = 9.85
9.85 9.85 9.85
φ = −90◦ − tan−1 − tan−1 + tan−1
2 23.63 4.06
= −90◦ − 78.5◦ − 22.6◦ + 67.5◦
= −123.5◦

Phase margin, φm = 180◦ − 123.5◦

φm = 56.5◦

The required phase margin is 55◦ . Hence, the required phase margin is met. Since the
phase angle of the compensated system is always less than 180◦ , the gain margin is
∞. The required gain margin is 12 dB only. Thus, the lead compensator is designed.

Example 6.2
Design cascade phase lead compensator for a system whose open loop T.F. is

10
G H (s) =
s(1 + s)

to have a phase margin of 45◦ using Bode plot.

(Anna University, December, 2003)

736 6 Design of Compensators

Solution
1. The Bode plot of the uncompensated system is shown in Fig. 6.6a. The gain
crossover frequency is obtained as

40 log ωgc = 20
ωgc = 3.16

Corresponding to ωgc = 3.16, the phase angle of the uncompensated system is

(a)

−20dB/dec.

20 −40dB/dec.
+dB
ωgc 3.16

0
1 ωm 4.38 log ω

−5.67
−dB

0
log ω
−φ

−90º

−162.5º
φm 17.5º
−180º

(b)

R(s) + (s + 2.28) 10 C(s)

3.69 Gc(s) = G(s) =
− (s + 8.41) s(s +1)

Fig. 6.6 a Bode plot of uncompensated system. b Block diagram of compensated system. c Bode
diagram for the compensated system
6.4 Compensator Design in the Frequency Domain 737

(c)

−20dB/dec.

20
−40dB/dec.
+dB −20dB/dec.
ωm 4.38
5.7
0
1 2.28 8.41 log ω
−5.6

−dB −40dB/dec.

0
log ω

−φ
−90º
−132º
φm 48º Phase plot

−180º

Fig. 6.6 (continued)

∠G( jωgc ) = −90◦ − tan−1 ωgc

= −90◦ − tan−1 3.16 = −162.5◦

The phase margin of the uncompensated system, therefore, is

φm = 180◦ − 162.5◦ = 17.5◦

2. The required phase margin is 45◦ . The phase angle deficiency is 45◦ − 17.5◦ =
27.5◦ . An allowance of say 7.5◦ is given for the shifting of the new gain crossover
frequency when the lead network is inserted. Thus, the maximum phase angle con-
tribution of the lead compensator becomes θm = 27.5 + 7.5◦ = 35◦ . The attenu-
ation factor α is determined using Eq. (6.3)

(1 − sin θm )
α=
(1 + sin θm )
(1 − sin 35◦ )
=
(1 + sin 35◦ )
738 6 Design of Compensators

α = 0.271

3. The gain crossover

√ frequency ωgc is now shifted to the right by adding a magnitude
of 20 log α. The new gain crossover frequency now becomes ωm .
√ √
20 log α = 20 log 0.271
= −5.67

From Fig. 6.6a, the following equation is written for the magnitude plot
ωm
40 log = 5.67
3.16
ωm = 4.38

4. Using Eq. (6.4), the time constant of the zero of the lead compensator is obtained
as
1 √
= ωm α
T √
= 4.38 0.271
= 2.28

T = 0.4386

5. The T.F. of the lead compensator becomes

(s + T1 )
G c (s) =
(s + αT
1
)
(s + 2.28)
=
(s + 8.41)

6. The forward path T.F. of the compensated system now becomes

(s + 2.28) 10
G c (s)G(s) =
s(s + 8.41) (s + 1)

The overall system gain is reduced by a factor α. Hence, an amplifier with a gain
of
1 1
= = 3.69
α 0.271
is to be introduced in the forward path. The block diagram of the compensated
system is shown in Fig. 6.6b.
6.4 Compensator Design in the Frequency Domain 739

7. The Bode diagram of the compensated system is shown in Fig. 6.6c. At the new
gain crossover frequency ωm = 4.38, the phase angle of the compensated system
is obtained as
ωm ωm
φ = −90 − tan−1 ωm − tan−1 + tan−1
8.41 2.28
◦ −1 −1 4.38 −1 4.38
= −90 − tan 4.38 − tan + tan
8.41 2.28
= −90◦ − 77.1◦ − 27.5◦ + 62.5◦
= −132◦

The phase margin of the compensated system is

φm = 180◦ + φ
= 180◦ − 132◦

φm = 48◦

The required phase margin is only 45◦ . Thus, the given specifications are met and
the compensator designed.

Example 6.3
The open loop T.F. of a unity feedback control system is

K
G(s) =
s(s + 2)

Design a suitable lead compensator to meet the following specifications:

1. Velocity error constant K v ≥ 12 s−1
2. Phase margin φm ≥ 45◦

Solution

K v = lim sG(s)
s→0
sK
K v = lim
s→0 s(s + 2)
= 12

K = 24
740 6 Design of Compensators

12
G(s) =
s(1 + 2s )

The Bode plot of the uncompensated system is shown in Fig. 6.7a. The sinusoidal
T.F. is
12 ω
G( jω) =  ∠−90◦ − tan−1
ω 1+ ω2 2
4

(a)

−20dB/dec.

15.6
+dB

ωgc ωm
0
2 4.9 6.65 12 log ω

−5.3
−dB
−40dB/dec.

0
log ω
−φ

−90º
Phase plot
−162.5º
φm 22º
−180º

(b)
R(s) + (s + 3.61) 12 C(s)
3.39 Gc(s) = G(s) =
− (s + 12.28) s(1+s/2)
Amplifier

12
Fig. 6.7 a Bode plot for G(s) = . b Bode diagram for the compensated system. c Bode
s(1 + 2s )
diagram for the compensated system
6.4 Compensator Design in the Frequency Domain 741

(c)

−20dB/dec.

15.6
−40dB/dec.
+dB −20dB/dec.
5.34 ωm 6.68
0 12.28
2 3.61 log ω
−5.3
−dB −40dB/dec.

0
log ω

−φ
−90º
−130º Phase plot
φm 50º
−180º

Fig. 6.7 (continued)

The gain crossover frequency is obtained as

ωgc
40 log = 15.6
2
ωgc = 4.9

The angle contribution when ω = ωgc is

4.9
φ = −90◦ − tan−1
2
= −158◦

Hence, the phase margin is obtained as

φm = 180◦ − 158◦
= 22◦
742 6 Design of Compensators

(Since the phase plot varies from −90◦ to −180◦ for 0 ≤ ω ≤ ∞, the general shape
of phase plot is known, and hence the phase angle at ω = ωgc alone need to be
determined. For other values of ω, the phase angle φ need not be known).
The required phase margin is 45◦ . Therefore, the angle contribution from the lead
network is (45◦ − 22◦ ) = 23◦ and an allowance of 10◦ is added to compensate for
the decrease in phase margin when the lead network is inserted and ωgc is shifted
to the right. The angle deficiency now becomes 23◦ + 10◦ = 33◦ . This is taken as
θm = 33◦ . Using Eq. (6.3), we get the attenuation factor as

(1 − sin θm )
α=
(1 + sin θm )
(1 − sin 33◦ )
=
(1 + sin 33◦ )
α = 0.295

When the lead network is inserted, the gain crossover frequency is shifted to the right
and becomes ωgc = ωm . The amount of attenuation required is
√ √
20 log α = 20 log 0.295
= −5.3 dB

Corresponding to −5.3 dB, attenuation ωm is calculated from Fig. 6.7a as

ωm
40 log = 5.3
4.9
ωm = 6.65

Using Eq. (6.4), we get

1 √
= ωm α
T √
= 6.65 0.295
= 3.61

T = 0.276
1 1
=
αT 0.295 × 0.276
= 12.28
1
αT =
12.28
= 0.0814

The T.F. of the compensator is

6.4 Compensator Design in the Frequency Domain 743

(s + T1 )
G c (s) =
(s + αT
1
)
(s + 3.61)
=
(s + 12.28)

To nullify the effect of attenuation factor α = 0.295, an amplifier with a gain of

1 1
= = 3.39
α 0.295
is connected in cascade with the compensator. The complete block in Fig. 6.7b.
The T.F. of the compensated system is obtained as

3.39(s + 3.61) 12
G c (s)G(s) =
s(s + 12.285) (1 + 2s )
12(1 + 3.61
s
) 1
=
s(1 + 12.28 ) (1 + 2s )
s

The velocity error constant of the compensated system is easily verified as 12 from
the above equation. The corner frequencies of the above T.F. are 2, 3.61 and 12.28.
The Bode magnitude and phase plots are shown in Fig. 6.7c.
From Fig. 6.7c, the phase margin of the given system is checked as follows. From
the magnitude plot, the gain crossover frequency ωm = ωgc is obtained as 6.68.
At this frequency, the phase angle of the compensated system is (this alone need
to be calculated)
ωm ωm ωm
φ = −90◦ − tan−1 − tan−1 + tan−1
2 12.28 3.61
6.68 6.68 6.68
= −90◦ − tan−1 − tan−1 + tan−1
2 12.28 3.61
= −90◦ − 73.3◦ − 28.3◦ + 61.6◦
= −130◦

The phase margin is

φm = 180◦ − 130◦

φm = 50◦

All the given specifications are met with the compensator designed, thus making the
completion of the design.
744 6 Design of Compensators

Two Stage Compensator Design

Example 6.4
A certain unity feedback control system has an open loop T.F

K
G(s) =
s 2 (1 + 0.2s)

Design a lead compensator to meet the following specifications:

(a) Acceleration error constant K a = 10.
(b) Phase margin φm = 35◦ .

(Anna University, December, 2006)

Solution
K
G(s) =
s 2 (1
+ 0.2s)
s 2 10
K a = lim s 2 G(s) = lim 2 = 10
s→0 s→0 s (1 + 0.2s)
10
G(s) = 2
s (1 + 0.2s)

The Bode plot of the uncompensated system is shown in Fig. 6.8a. The gain crossover
frequency is
√
ωgc = 10 = 3.16

At this frequency the phase angle of G( jω) is

∠G( jωgc ) = −180◦ − tan−1 3.16 × 0.2 = −212.3◦

The phase margin is

φm = 180◦ − 212.3◦ = −32.3◦

which is negative and the uncompensated system is unstable. However, the required
phase margin is 35◦ . To achieve this, the total phase lead to be contributed by the
lead network is 35◦ + 32.3◦ = 67.3◦ . When the lead network is inserted, the gain
crossover frequency 3.16 is shifted to the right and further phase lead to be contributed
as is evident from Fig. 6.8a. An allowance of 22.7◦ is added and the required phase
lead contribution now becomes 90◦ . If a single phase lead network is used, the
attenuation factor α becomes
6.4 Compensator Design in the Frequency Domain 745

(a)
+dB
ωgc 6.6
−40dB/dec.

0 10
3.16 5 log ω
−dB
−8

−15.3

−26 −60dB/dec.

0
log ω
−φ

−180º
φm 32.3º
−212.3º

−270º

(b)
R(s) + (s + 2.82) (s + 2.82)
− 5.81 IA 5.81
(s + 16.4) (s + 16.4)
Gc1(s) Isolation Gc2(s)
amplifier

C(s) 10
s2(1+0.2s)

10
Fig. 6.8 a Bode diagram for an uncompensated system G(s) = . b Two stage com-
s 2 (1 + 0.2s)
pensator system. c Bode diagram for the compensated system

(1 − sin 90◦ )
α= =0
(1 + sin 90◦ )

To meet the steady state error requirement, an amplifier with a gain α1 = ∞ is needed
which is impractical. As stated earlier, the minimum value of α is restricted to 0.07
for a lead network, so that the noise level is within the permissible value. Therefore,
a two stage lead network with an isolating amplifier is used to achieve compensation.
746 6 Design of Compensators

(c)

−40dB/dec.
+dB
1.99
ωgc ωm 6.3
16.4
0
2.82 3.16 5 log ω
−20dB/dec.
−dB
−8.32

−60dB/dec.
φ

0
log ω
−90º
−142º Phase plot
−φ φm 38º
−180º

−270º

Fig. 6.8 (continued)

The maximum phase θm of each phase is chosen as 45◦ . Thus, α of a single stage
network is calculated as
(1 − sin 45◦ )
α=
(1 + sin 45◦ )
α = 0.172

To make ωm as the new gain crossover frequency, when the two stage compensator
is introduced the attenuation provided by both the stages are to be considered. Thus,
the attenuation in dB is calculated as
√ √
2 × 20 log α = 2 × 20 log 0.172 = −15.3 dB

Corresponding to −15.3 dB, the gain crossover frequency ωgc = ωm is calculated

from Fig. 6.8a as
6.4 Compensator Design in the Frequency Domain 747

ωgc
60 log = 7.3
5
ωgc = ωm = 6.6
1
ωm = √
αT
1 √ √
= αωm = 0.172 × 6.6 = 2.82
T
T = 0.354

αT = 0.172 × 0.354 = 0.061

1
= 16.4
αT
The T.F. of a single stage lead compensator is

(s + T1 ) (s + 2.82)
G c (s) = =
(s + 1
αT
) (s + 16.4)

To maintain the steady state error, an amplifier with a gain of

1 1
= = 5.81
α 0.172
is to be connected in cascade with each stage. The block diagram of the compensated
system is shown in Fig. 6.8b.

(1 + s 2
)
G c (s) = G c1 (s)G c2 (s) = 2.82
(1 + s 2
16.4
)
(1 + s 2
) 10
G c (s)G(s) = 2.82
(1 + s 2
16.4
) s 2 (1 + 0.2s)

The Bode diagram of the compensated system is shown in Fig. 6.8c. From Fig. 6.8c,
the gain crossover frequency ωgc = ωm is obtained as Fig. 6.3. At this frequency, the
phase angle is

6.3 6.3
φ = −180◦ − tan−1 0.2 × 6.3 + 2 tan−1 − 2 tan−1
2.82 16.4
= −180◦ − 51.6◦ + 131.5◦ − 42◦ = −142◦

The phase margin is

φm = 180◦ − 142◦ = 38◦

748 6 Design of Compensators

The required phase margin is 35◦ only. The given specifications are met by the two
stage compensator.

Example 6.5
The T.F. of a phase lead compensator is given by

(1 + 3T s)
G c (s) =
(1 + T s)

where T > 0. Determine the maximum phase shift provided by the compensator. At
what frequency it occurs?

Solution
(1 + 3T s)
G c (s) =
(1 + T s)
(1 + j3T ω)
G c ( jω) =
(1 + j T ω)
3T ωT ω
∠G c ( jω) = θ = (tan−1 3T ω − tan−1 T ω) = tan−1
1 + 3T 2 ω2
2ωT
tan θ =
(1 + 3T 2 ω2 )

Let tan θ = N
2T ω
N =
(1 + 3T 2 ω2 )

Differentiating N with respect to ω, and equating to zero to get maxima we get

dN 2T ω(T 2 ω)6 − (1 + 3T 2 ω2 )T 2
= =0
dω (1 + 3T 2 ω2 )2
12T 2 ω2 − 2 − 6T 2 ω2 = 0
6T 2 ω2 = 2
1
ω = ωm = √
3T
1
ωm = √
3T

Substituting this value of ω in N , the maximum phase shift is obtained

6.4 Compensator Design in the Frequency Domain 749

√2T
3T 1
N = =√
(1 + 1) 3
1
tan θm = √
3
θm = 30◦

6.4.3 Lag Compensator

The electric network lag compensator is shown in Fig. 6.9a. From Fig. 6.9

Z 1 (s) = R1
1 (1 + R2 Cs)
Z 2 (s) = R2 + =
Cs Cs

(a)
R1
R2
ei(t) eo(t)

(b)
+dB
1/T 1/αT
0
log ω
−dB

20 log α
Μagnitude plot
φ

0
log ω
−φ Phase plot

−90º

Fig. 6.9 a A lag compensator electric network. b Bode diagram for lag compensator. c Bode
diagram with lag compensator
750 6 Design of Compensators

(c)
−20dB/dec.
+dB
ωgc2 ωgc1
0
log ω

−dB
−40dB/dec.
ωm

G1(jω)
φ G2(jω)

−90º log ω
Phase plot
φm2
φm1
−180º

+dB
0
log ω
−dB

20 log α

Fig. 6.9 (continued)

The T.F. of the network is

Eo
= G c (s)
Ei
Eo Z 2 (s)
=
Ei Z 1 (s) + Z 2 (s)
(1+R2 Cs)
= Cs
R1 + (1+R 2 Cs)
Cs
(1 + R2 Cs)
=
[1 + (R1 + R2 )Cs]
6.4 Compensator Design in the Frequency Domain 751

(1 + αT s)
G c (s) = (6.5)
(1 + T s)

where
R2
α= <1
(R1 + R2 )

and

T = (R1 + R2 )C

Equation (6.5) which is in time constant form can be expressed in pole–zero form as

α(s + z)
G c (s) = (6.6)
(s + p)

where
1 1
z= and p =
R2 C (R1 + R2 )C

and

p = αz (6.7)

The Bode diagram for Eq. (6.5) is shown in Fig. 6.9b. The values of 1/T and 1/αT
are chosen small so that they are closer to the origin. This provides attenuation for a
wide range of frequencies.
The Bode diagram of a certain second order system is shown in Fig. 6.9c. G 1 ( jω)
plot shows the magnitude plot when the compensator is not inserted. The gain
crossover frequency is ωgc1 and the phase margin is φm1 which may not be suffi-
cient enough to meet the specification requirement. When the lag compensator is
introduced, the phase shift is not very much affected. However, the attenuation effect
of the lag compensator shifts the G 1 ( jω) plot downwards and is shown as G 2 ( jω)
in Fig. 6.9c. The gain crossover frequency is now shifted towards left and denoted as
ωgc2 . At this frequency, the phase angle contribution of G c ( jω) is negligibly small
and the phase margin is estimated from G 2 ( jω) phase plot and is measured as φm2 .
Therefore, the design problem is approached in the reverse way. For the given phase
margin, the phase contribution of the lag compensator, roughly 10◦ phase angle is
added and the gain crossover frequency with this phase margin is determined and
the design is completed as described below. The important things to be noted are:
752 6 Design of Compensators

1. In the design of the lead compensator, phase lead is contributed by the lead
network at the gain crossover frequency.
2. In the design of the lag compensator, attenuation is provided at the gain crossover
frequency.
The following step-by-step procedure is followed to design the lag compensator:
1. Determine the value of K of the uncompensated system which meets the steady
state error requirement.
2. Determine the phase margin of the uncompensated system. If the phase margin
requirement is not met, go to step 3.
3. To the given phase margin, add (say) 10◦ . For this phase margin, determine the
frequency in the Bode diagram. This frequency becomes the gain crossover fre-
quency.
4. The zero of the lag compensator is arbitrarily fixed at one tenth of the gain
crossover frequency determined in step 3. By doing so, the effect of attenuation
is shifted towards the left where the phase margin is high.
5. At the gain crossover frequency, the attenuation provided by the lag compensator
is 20 log α in dB. The magnitude of the compensator is α. This makes

α|G( jωgc )| = 1 (6.8)

Using Eq. (6.8), α is calculated.

6. To place the zero of the compensator at one tenth of the gain crossover frequency,
we get

1 ωgc
= (6.9)
αT 10
Knowing α and ωgc , T is calculated using Eq. (6.9).
7. With the values of α and T determined in steps 5 and 6, the lag compensator is
designed.
8. The Bode diagram for the compensated system is drawn and it is verified that the
compensated system satisfies the required specifications. If not, the procedure is
repeated from step three by adding more phase angle which is supposed to have
been contributed by the lag network at the gain crossover frequency. With the
new gain crossover frequency, the procedure is repeated.

Example 6.6
A unity feedback control system has an open loop T.F

4
G(s) =
s(1 + 2s)
6.4 Compensator Design in the Frequency Domain 753

It is desired to obtain a phase margin of 38◦ without sacrificing K v . Design a suitable

lag network.

(Anna University, April, 2005)

Solution The Bode plot of uncompensated system is shown in Fig. 6.10a. The gain
crossover of the uncompensated system is obtained from
ωgc
40 log = 18
0.5
√
ωgc = 2
√
At ωgc = 2, the phase angle of the uncompensated system is

φ = −90◦ − tan−1 2ω
√
= −90◦ − tan−1 2 2
= −161◦

The phase margin is (180◦ − 161◦ ) = 19◦ . But the required phase margin is 38◦ . To
account for the phase contribution of the lag network when inserted, 12◦ is added to
the required phase margin. The total phase angle at the new gain crossover frequency
is φ = 180◦ − (38◦ + 12◦ ) = −130◦ . This point is identified at A in phase plot. The
new gain crossover frequency is calculated from

−90◦ − tan−1 2ω = −130◦

tan−1 2ω = 40◦
ω = 0.42

ωgc = 0.42

α|G( jωgc )| = 1
4
α  =1
ωgc 1 + 4ωgc2

0.42 
α= 1 + 4(0.42)2
4
α = 0.137

Using Eq. (6.9), we get

754 6 Design of Compensators

(a) −20dB/dec.
18

+dB

ωgc ωgc = 2
0
0.42 0.5 4 log ω
−dB
−40dB/dec.

0
log ω

−90º
A
−130º Phase plot
φm 50º
φm 19º
−180º

(b)
+dB
−20dB/dec.
56.8
−40dB/dec.
−20dB/dec.
22.3

ωgc 0.523
0.8 log ω
4
0 0.00575 0.042 0.5
−40dB/dec.
−35.3
−dB −60dB/dec.

0
log ω

−φ
−90º
−140º Phase plot
φm 40º
−180º

Fig. 6.10 a Bode plot for G(s) = 4

s(1+2s) b Bode diagram for compensated system
6.4 Compensator Design in the Frequency Domain 755

1 ωgc
= = 0.042
αT 10
αT = 23.8
23.8
T =
0.137
T = 173.8
(1 + αT s)
G c (s) =
(1 + T s)
(1 + 23.8s)
G c (s) =
(1 + 173.8s)

The T.F. of the compensated system is given by

(1 + 23.8s)4
G c (s)G(s) =
(1 + 173.8s)s(1 + 2s)

The Bode diagram drawn for the above T.F. is shown in Fig. 6.10b. From Fig. 6.10b,
the gain crossover frequency is obtained from
ωgc
40 log = 0.8
0.5
ωgc = 0.523

The phase angle corresponding to this frequency is

φ = −90◦ + tan−1 23.8ωgc − tan−1 173.8ωgc − tan−1 2ωgc

= −90◦ + tan−1 23.8 × 0.523 − tan−1 173.8 × 0.523 − tan−1 2 × 0.523

φ = −90◦ + 85.4◦ − 89.4◦ − 46.0◦

= −140◦

The phase margin is

φm = 180◦ − 140◦

φm = 40◦

The given specification is satisfied with the designed controller.

756 6 Design of Compensators

Example 6.7
Consider the unity feedback system whose open loop T.F. is

K
G(s) =
s(s + 1)(1 + 0.5s)

Design a lag compensator for the system so that the static velocity error constant
K v ≥ 5 s−1 , the phase margin is at least 38◦ and the gain margin is at least 10 dB.

(Anna University, December, 2004)

Solution Let
K
G(s) =
s(1 + s)(1 + 0.5s)
K v = lim sG(s)
s→0
sK
= lim
s→0 s(1 + s)(1 + 0.5s)
=K

K =5

Hence, an amplifier with a gain of K = 5 is chosen for the open loop system to
satisfy the given velocity error constant. The T.F. of the uncompensated system is

5
G(s) =
s(1 + s)(1 + 0.5s)

The Bode plot of the uncompensated system is shown in Fig. 6.11a. The gain
crossover frequency of the uncompensated system is calculated as
ωgc
60 log =2
2
ωgc = 2.2

At this frequency, the phase angle of the uncompensated system is

φ = −90◦ − tan−1 ωgc − tan−1 0.5ωgc

= −90◦ − tan−1 2.2 − tan−1 0.5 × 2.2
= −90◦ − 66.6◦ − 47.7◦
= −204.3◦
φm = 180◦ − 204.3◦
6.4 Compensator Design in the Frequency Domain 757

(a)
−20dB/dec.
14
+dB −40dB/dec.
−60dB/dec.
2 ωgc 2.2
log ω
0.52
0 1 2
−35.3
−dB

φ
ωgc
0
log ω

−φ
−90º
−132º A
φm 48º
−180º φm 24.3º
Phase plot

−270º

Fig. 6.11 a Bode diagram for G(s) = s(s+1)(1+0.5s)

5
. b Bode diagram for the compensated system.
c Block diagram representation of compensated system

= −24.3◦

If the phase margin is negative, the uncompensated system is unstable. The required
phase margin is 38◦ . Adding an allowance of say 10◦ , the phase angle of the system
is −180◦ + (38◦ + 10◦ ) = −132◦ . Corresponding to this phase, point A in the phase
plot is identified and the new gain crossover frequency is calculated. Analytically
this is calculated as follows:

−90◦ − tan−1 ω − tan−1 0.5ω = −132◦

tan−1 ω + tan−1 0.5ω = 42◦
1.5ω
tan−1 = 42◦
1 − 0.5ω2

Taking tan on both sides, we get

758 6 Design of Compensators

(b)

−20dB/dec.

58
−40dB/dec.
+dB −20dB/dec.
21
ωgc 0.59
log ω
0
0.0063 0.0526 1 2
−4.6 −40dB/dec.
−10.6
−16.6 −60dB/dec.
−dB

φ
ωpc 2
0
−φ log ω

−90º
−140º
φm 40º
−180º

−270º

(c)

R(s) + (1 + 19s) 1 C(s)

5 Gc(s) =
− (1 + 159s) s(1+s)(1+0.5s)
Amplifier Compensator System

Fig. 6.11 (continued)

1.5ω
= 0.9
(1 − 0.5ω2 )
ω2 + 3.33ω − 2 = 0

Solving for ω, we get

ω = 0.52
6.4 Compensator Design in the Frequency Domain 759

This frequency is taken as the new gain crossover frequency. Thus

ωgc = 0.52

From Eq. (6.8), we get

α|G( jωgc )| = 1
5
α  =1
ωgc (1 + ωgc
2 )(1 + 0.25ωgc
2 )

0.52 √
α= 1.27 × 1.0676
5

α = 0.121

The zero of the compensator is placed at one tenth of this gain crossover frequency

1
= 0.052
αT
1
= 0.052 × 0.121
T

T = 159.0

1
αT =
0.052
= 19.0

The T.F. of the lag compensator is

(1 + 19s)
G c (s) =
(1 + 159s)

The T.F. of the compensated system is

5(1 + 19s)
G c (s)G(s) =
s(1 + 159s)(1 + s)(1 + 0.5s)

The Bode diagram for the compensated system is shown in Fig. 6.11b. From
Fig. 6.11b, the actual gain crossover frequency is obtained from
760 6 Design of Compensators

ωgc
20 log = 21
0.0526

ωgc = 0.59

Corresponding to this frequency, the phase angle of the compensated system is

obtained as

φ = −90◦ − tan−1 ωgc − tan−1 0.5ωgc − tan−1 159ωgc + tan−1 19ωgc

= −90◦ − tan−1 0.59 − tan−1 0.5 × 0.59 − tan−1 159 × 0.59 + tan−1 19 × 0.59
= −90◦ − 30◦ − 16◦ − 89◦ + 85◦
= −140◦

The phase margin of the compensated system is

φm = 180◦ − 140◦

φm = 40◦

The required phase margin is 38◦ . The actual phase margin is obtained as 40◦ . The
gain margin is calculated as follows. The phase crossover frequency is calculated
from

−90◦ − tan−1 ω pc − tan−1 0.5ω pc − tan−1 159ω pc + tan−1 19ω pc = −180◦

ω pc is greater than ωgc . Hence, at this frequency, tan−1 159ω p is roughly equals
tan−1 19ω pc . In other words, the phase angle contribution of the lag network and
phase crossover frequency is negligibly small. Hence, we write

−90◦ − tan−1 ω pc − tan−1 0.5ω pc = −180◦

tan−1 ω pc + tan−1 0.5ω pc = 90◦
1.5ω pc
= tan 90◦
1 − 0.5ω2pc
(1 − 0.5ω2pc ) = 0

√
ω pc = 2

At this frequency, the dB is calculated as

6.4 Compensator Design in the Frequency Domain 761
√
dB|ω pc = −4.6 − 40 log 2
= −10.6

G m = 10.6 dB

The gain margin obtained with the compensator is more than the required gain margin
of 10 dB. Hence, all the prescribed specifications are not with the compensator
designed. The block diagram of the compensated system is shown in Fig. 6.11c.

Example 6.8
A unity feedback system has an open loop T.F

K
G(s) =
s(1 + 2s)

Design a suitable lag compensator so that the phase margin is 40◦ and the steady
state error for ramp input is less than or equal to 0.2

(Anna University, December, 2009)

Solution
K
G(s) =
s(1 + 2s)

The velocity error coefficient K v is

K v = lim sG(s)
s→0
sK
= lim
s→0 s(1 + 2s)
=K

The steady state error is

R
ess =
Kv
1
=
K
= 0.2

K =5
762 6 Design of Compensators

5
G(s) =
s(1 + 2s)

The Bode diagram is drawn as shown in Fig. 6.12a. The gain crossover frequency of
the uncompensated system is given by
ωgc
40 log = 20
0.5
ωgc = 1.6

Corresponding to this frequency, the phase angle is

φ = −90◦ − tan−1 2 × 1.6

= −162◦

The phase margin of the uncompensated system is

φm = 180◦ − 162◦
= 18◦

The above phase margin does not meet the given specification. The required phase
margin is 40◦ . An additional phase angle of 10◦ is added to this to compensate
the phase contribution of the lag compensator. Thus, to have a phase margin of
(40◦ + 10◦ ) = 50◦ , the new gain crossover frequency is obtained as follows:

−90◦ − tan−1 2ω = −130◦

tan−1 2ω = 40◦
ω = 0.42
ωgc = 0.42

At this gain crossover frequency, the magnitude of the compensated system is

α|G( jωgc )| = 1
5
α  =1
ωgc (1 + 4ωgc
2 )

0.42 
α= 1 + 4 × (0.42)2
5

α = 0.11
6.4 Compensator Design in the Frequency Domain 763

(a)

+dB −20dB/dec.
20 ωgc 1.6

log ω
0 5
0.5
−35.3
−dB −40dB/dec.

φ
ωgc 0.42
0
log ω

−φ
−90º
−130º Phase plot
φm 50º φm 18º
−180º

(b)

+dB −20dB/dec.

61
−40dB/dec.
−20dB/dec.
22.7
ωgc 0.53
1.1 log ω
0.0046 0.0417 0.5
−40dB/dec.

0
−φ log ω

−90º
−140º
φm 40º
−180º

Fig. 6.12 a Bode diagram for G(s) = 5

s(1+2s) . b Bode diagram for the compensated system
764 6 Design of Compensators

The zero of the compensator is placed at one tenth of the new gain crossover fre-
quency

1 1
= 0.42
αT 10
αT = 24
24
T =
0.11

T = 216

The T.F. of the compensator is

(1 + αT s)
G c (s) =
(1 + T s)

(1 + 24s)
G c (s) =
(1 + 216s)

The T.F. of the compensated system is

(1 + 24s) 5
G c (s)G(s) =
(1 + 216s)s (1 + 2s)

The Bode diagram for the above T.F. is shown in Fig. 6.12b. From Fig. 6.12b, the
gain crossover frequency ωgc is determined as follows:

ωgc
40 log = 1.1
0.5
ωgc = 0.53

At ω = 0.53, the phase angle is calculated as follows:

φ = −90◦ − tan−1 2ωgc − tan−1 216ωgc + tan−1 24ωgc

= −90◦ − tan−1 2 × 0.53 − tan−1 216 × 0.53 + tan−1 24 × 0.53
= −90◦ − 46.2 − 89.3◦ + 85.5◦ = −140◦

The phase margin of the compensated system is

φm = 180◦ − 140◦
6.4 Compensator Design in the Frequency Domain 765

φm = 40◦

The designed lag compensator satisfies the given specifications.

6.4.4 Lag–Lead Compensator

As discussed in Sects. 6.4.1 and 6.4.2, the lead and lag compensators have their own
advantages and disadvantages. A lag–lead compensator is designed which provides
the advantages of these two compensators simultaneously. The lag–lead compensat-
ing network is shown in Fig. 6.13a. The T.F. of the compensator is determined as
described below. Let
R1
Z 1 = Parallel combination of R1 and C1 =
1 + R1 C 1 s

Z 2 = Series combination of R2 and C2

1
= R2 +
C2 s
(1 + R2 C2 s)
=
C2 s
Eo Z2
(s) =
Ei Z1 + Z2
(1 + R2 C2 s)/C2 s
= R1
1+R1 C1 s
+ (1+R 2 C 2 s)
C2 s
(1 + R1 C1 s)(1 + R2 C2 s)
=
(1 + R1 C1 s + R2 C2 s + R1 C2 s + R1 C1 R2 C2 s 2 )

Eo (1 + α1 T1 s)(1 + α2 T2 s)
(s) = G c (s) = (6.10)
Ei (1 + T1 s)(1 + T2 s)

where

α1 T1 = R2 C2
α2 T2 = R1 C1
T1 + T2 = R1 C1 + R2 C2 + R1 C2
α1 α2 = 1
α1 < 1
766 6 Design of Compensators

(a)
C1

R1 R2
ei eo

C2

(b) j

0
1/T2 1/ 2T2 1/ 1T1 1/T

(c) Im
Lead portion

1
1 0
0 Re

Lag portion

Fig. 6.13 a Lag–lead compensating network. b Pole–zero locations of G c (s). c Frequency response
characteristic of lag–lead compensator in polar plot. d Bode diagram of lag–lead compensator

α2 > 1 (6.11)
1 1 1 1
< < < (6.12)
T1 α1 T1 α2 T2 T2

The T.F
(1 + α1 T1 s)
G c1 (s) = (6.13)
(1 + T1 s)

represents the lag portion of the compensator while

6.4 Compensator Design in the Frequency Domain 767

(d)
+dB

1/T1 1/ 1T1 1/ 1T2 ωm 1/T2

0
log ω
−dB
−20dB/dec.
+20dB/dec.
−10 log 2
20 log 1
= −20 log 2

90º
θm
+φ
45º
0
−φ log ω

−45º

Fig. 6.13 (continued)

(1 + α2 T2 s)
G c2 (s) = (6.14)
(1 + T2 s)

represents the lead portion. The sinusoidal T.F. of the lag–lead compensator is given
by

(1 + jωα1 T1 )(1 + jωα2 T2 )

G c ( jω) =
(1 + jωT1 )(1 + jωT2 )


(1 + ω2 α12 T12 )(1 + ω2 α22 T22 )
|G c ( jω)| =  (6.15)
(1 + ω2 T12 )(1 + ω2 T22 )
∠G c ( jω) = tan−1 α1 T1 ω + tan−1 α2 T2 ω − tan−1 T1 ω − tan−1 T2 ω (6.16)

The poles and zeros of G c (s) are shown in Fig. 6.13b. The polar plot of the lag–lead
compensator is shown in Fig. 6.13c.
The Bode diagram of Eq. (6.10) is shown in Fig. 6.13d. From Fig. 6.13d, it is
evident that the phase angle contribution at high frequency is due to the lead network
768 6 Design of Compensators

and the lag network contributes negligibly small phase. The maximum phase angle
contribution φm occurs at ωm which is in between

1 1
ω= and ω =
α2 T2 T2

Now consider the lead portion of the compensator which is given in Eq. (6.14). The
sinusoidal T.F. of G c2 is written as

(1 + jωα2 T2 )
G c2 ( jω) =
(1 + jωT2 )


1 + ω2 α22 T 2
|G c2 ( jω)| = √
1 + ω2 T 2
θ = ∠G c2 ( jω) = tan ωα2 T2 − tan−1 ωT2
−1
(6.17)
ωT2 (α2 − 1)
= tan−1 (6.18)
(1 + ω2 α2 T22 )

Taking tan on both sides we get

ωT2 (α2 − 1)
tan θ =
(1 + ω2 α2 T22 )

Let us define tan θ = N

ωT2 (α2 − 1)
N =
(1 + ω2 α2 T22 )
dN
= 0 = ωT2 (α2 − 1)(2ωα2 T22 ) − (1 + ω2 α2 T22 )T2 (α2 − 1)
dω
2ω2 α2 T22 = 1 + ω2 α2 T22

1
ωm = √ (6.19)
T2 α2

The maximum phase angle at ω = 1

√
T2 α2
is obtained from Eq. (6.18) as

√ 1
θm = tan−1 α2 − tan−1 √
α2
√
( α2 − √α2 )
1
= tan−1
(1 + 1)
6.4 Compensator Design in the Frequency Domain 769

Taking tan on both sides, we get

(α2 − 1)
tan θm = √
2 α2

The above equation is represented below.

From the triangle, we can write

(α2 − 1)
sin θm =
(α2 + 1)

(1 + sin θm )
α2 = (6.20)
(1 − sin θm )

From the magnitude plot of Fig. 6.13d, at ωm , the magnitude in dB is obtained as

1
dB|ω=ωm = 20 log
T2 ωm
= −10 log α2

or
1
|G c2 ( jω)| = √
α2

If ωm is taken as the new gain crossover frequency, then the following relationship
holds good

|G( jωgc )|
√ =1
α2

√
|G( jωgc )| = α2 (6.21)

Using Eqs. (6.19), (6.20) and (6.21), the lag–lead compensator is designed in the
frequency domain following the step-by-step procedure given below.
Step-by-Step Procedure to Design Lag–Lead Compensator

1. To satisfy the steady state error requirement, determine the value of open loop
gain K . For this value of K , draw the Bode diagram.
2. Determine the phase margin of the uncompensated system and find the angle
deficiency that must be contributed by the compensator. Using Eq. (6.20), find α2 .
770 6 Design of Compensators

3. Using Eq. (6.21), determine the gain crossover frequency ωgc . Make ωgc = ωm .
Using Eq. (6.19), find T2 . Now the lag portion of the compensator is to be designed.
4. Using the relationship α1 α2 = 1 as derived in Eq. (6.11), determine α1 .
5. Arbitrarily place the zero of the lag portion at one tenth of the zero of the lead
portion. Thus, we get

1 1
= 0.1 (6.22)
α1 T1 α2 T2

From this, find T1 . Now the lag portion of the compensator is designed.
6. Draw the Bode plot for the compensated system and verify whether all the given
specifications are satisfied. If not satisfied, go to step 2 and change θm and follow
the procedure. If necessary, modify Eq. (6.22) by appropriate factor to get the
desired result.

Example 6.9
The open loop T.F. of a unity feedback control system is

K
G(s) =
s(s + 2)(s + 20)

Design a lag–lead compensator to obtain the following specifications:

(a) Static velocity error = 0.1
(b) Phase margin φm ≥ 50◦
(c) Gain margin G m ≥ 10 dB

(Anna University, December, 2011)

Solution

K v = lim sG(s)
s→0
sK
= lim
s→0 s(s + 2)(s + 20)

K
Kv =
40
1
ess = = 0.1 or K v = 10
Kv

K = 400
6.4 Compensator Design in the Frequency Domain 771

10
G(s) =
s(1 + s
2
)(1 + s
20
)

The corner frequencies are 2 and 20. The Bode magnitude and phase plots are shown
in Fig. 6.14a. From Fig. 6.14a, the gain crossover frequency of uncompensated system
is obtained as
ωgc
40 log = 14
2
ωgc = 4.47

The phase angle at this frequency is

4.47 4.47
φ = −90◦ − tan−1 − tan−1
2 20
= −90◦ − 66◦ − 13◦

(a) −20dB/dec.
14
−40dB/dec.
+dB
ωgc = 4.47
0
2 20 log ω
−dB

−26

φ −60dB/dec.

0
log ω
−φ

−90º

−169º
φm 11º
−180º Phase plot

−270º

Fig. 6.14 a Bode diagram for uncompensated system with G(s) = 10

s .
s(1+ 2s )(1+ 20 )
b Bode diagram
for the compensated system
772 6 Design of Compensators

(b)

−20dB/dec.
50.5
+dB −40dB/dec.
24.8 0dB/dec.
−20dB/dec. −20dB/dec.
ωgc ωm 3.5
4.8 log ω
5.8 20
0 0.03 0.132 1.32 2
−4.5
−40dB/dec.
−dB
−16.5

−26
−60dB/dec.
φ
ωpc 11.6
0
log ω

−φ
−90º
−123º φm 57º
−180º
Phase plot
−270º

Fig. 6.14 (continued)

= −169◦

The phase margin

φm = 180 − 169◦ = 11◦

But the required phase margin is 50◦ . The balance phase that should be contributed
by the lag–lead compensator is

θm = 50◦ − 11◦ = 39◦

α2 is calculated using Eq. (6.20)

(1 + sin θm )
α2 =
(1 − sin θm )
(1 + sin 39◦ )
=
(1 − sin 39◦ )
6.4 Compensator Design in the Frequency Domain 773

α2 = 4.4

The gain crossover frequency ωgc is calculated using Eq. (6.21)

√
|G( jωgc )| = α2
400 √
 = 4.4
ωgc (ωgc
2 + 4)(ωgc
2 + 400)

ωgc
2
(ωgc
2
+ 4)(ωgc
2
+ 400) = 36363.6

Substituting ωgc
2
= x, we get

x(x + 4)(x + 400) = 36363.6

x 3 + 404x + 1600 = 36363.6

Solving for x, we get x = 7.65

√
ωgc = 7.65

ωgc = 2.77

ωm = ωgc = 2.77

Using Eq. (6.19), we get

1
T2 = √
ωm α2
1
= √
2.77 4.4

T2 = 0.172

1 1
=
α2 T2 4.4 × 0.172
= 1.32
774 6 Design of Compensators

Using Eq. (6.22), we get

1 0.1
=
α1 T1 α2 T2
= 0.132
α1 T1 = 7.575

Using Eq. (6.11), we get

α1 α2 = 1
1
α1 =
α2
1
=
4.4

α1 = 0.227

7.575
T1 =
α1
7.575
=
0.227

T1 = 33.33

Thus, the lag–lead compensator design is completed as

(1 + 7.575s)(1 + 0.757s)
G c (s) =
(1 + 33.33s)(1 + 0.172s)

The T.F. of the compensated system is given by

(1 + 7.575s)(1 + 0.757s)10
G c (s)G(s) =
(1 + 33.33s)(1 + 0.172s)s(1 + 2s )(1 + s
20
)

The corner frequencies are 0.03, 0.132, 1.32, 2, 5.8 and 20. The Bode diagram for
the compensated system is shown in Fig. 6.14b. From the magnitude plot, the gain
crossover frequency of the compensated system is obtained as
ωgc
20 log = 4.8
2
6.4 Compensator Design in the Frequency Domain 775

ωgc = 3.5

The actual gain crossover frequency is slightly shifted to the right when we
take into account the lag portion of the compensator. Without lag portion, the gain
crossover frequency was obtained as 2.77. Now, with ωgc = 3.5, we calculate below
the phase angle of the compensated system

φ = tan−1 7.575ωgc + tan−1 0.757ωgc − 90◦ − tan−1 33.33ωgc

− tan−1 0.172ωgc − tan−1 0.5ωgc − tan−1 0.05ωgc .

Substituting ωgc = 3.5, we get

φ = 87.8◦ + 69.7◦ − 90◦ − 89.5◦ − 31◦ − 60◦ − 10◦

= −123◦

The phase margin of the compensated system is

φm = 180◦ + φ
= 180◦ − 123◦

φm = 57◦

The required phase margin is at least 50◦ . However, we get phase margin as 57◦
which satisfies the phase margin requirement. Now we calculate the gain margin. As
seen from Fig. 6.14b, the phase crossover frequency ω pc > ωgc . As seen from the
calculation of phase angle of the system with the compensator, at higher frequency
the phase angle contribution is approximately written as

φ = tan−1 0.757ω − 90◦ − tan−1 0.172ω − tan−1 0.5ω − tan−1 0.05ω

At ω = ω pc

φ = 180◦
90◦ = tan−1 0.757ω − tan−1 0.172ω
−(tan−1 0.5ω + tan−1 0.05ω)
tan−1 0.757ω − tan−1 0.172ω = 90◦ + (tan−1 0.5ω + tan−1 0.05ω)
0.585ω 0.55ω
tan−1 = 90◦ + tan−1
(1 + 0.13ω )
2 (1 − 0.025ω2 )

Taking tan on both sides, we get

776 6 Design of Compensators

(0.585ω) (1 − 0.025ω2 )
=
(1 + 0.13ω2 ) 0.55ω

The above equation is simplifier as

ω4 − 133ω2 − 312.5 = 0
ω2 = 135.7
√
ω pc = 135.7

ω pc = 11.6

At the phase crossover frequency ω pc = 11.6, in Fig. 6.14b, the dB is calculated as

11.6
dB|ω pc = −4.5 − 40 log
5.8
= −16.5

The gain margin is, therefore

G m = 16.5

All the given specifications are satisfied and thus the design is validated.
Summary of Performance Characteristic of Lead, Lag and Lag–Lead Compen-
sators
I. Lead compensator
1. The phase lead compensator adds phase lead at the gain crossover frequency which
improves the phase margin of the system. This increases the overall damping of
the closed loop system. Further, the rise time and setting time are reduced and
thus the transient response is faster.
2. As a result, increased phase margin and gain margin are achieved when a lead
compensator is inserted, thereby increasing the relative stability.
3. The bandwidth of the closed loop system is increased.
4. Since the phase lead compensator provides lead phase angle at higher frequency,
the steady state error is not corrected.
5. One of the limitations of the lead compensator is that high frequency noise enters
the system.
6. To meet the steady state requirement, an amplifier with a gain of 1/α is to be
inserted which is costly.
7. In practical applications, the maximum phase lead provided by the compensator
is between 50◦ and 55◦ . If the phase lead requirement is more than this, a single
stage lead compensator is not enough. In such cases, a two stage lead compensator
with an isolation amplifier is designed, which increases the cost.
6.4 Compensator Design in the Frequency Domain 777

II. Lag compensator

1. Lag compensator provides attenuation at the new gain crossover frequency which
is shifted towards the low frequency region of the Bode diagram of the uncom-
pensated system. This provides the required phase margin.
2. The phase angle curve of the uncompensated system is not appreciably affected
when the lag network is introduced. The steady state error is appreciably reduced.
3. The system gain at high frequencies is reduced which reduces the bandwidth of
the system. As a result, the transient response is very slow with increased rise
time and settling time.
III. Lag–Lead compensator

1. Basically, the lead compensator is used to improve the transient response which
has no effect on steady state error. Similarly, the lag compensator is used to
improve the steady state error. However, with the lag compensator, the transient
response becomes slow. By using the lag–lead compensator, the good effects of
lag and lead compensators are achieved. It provides a good transient response
and improves the steady state error. However, the additional components used
increase the cost of the compensator.

6.5 Compensator Design in the Time Domain

In the control system analysis, transient response specifications are more informative
than the frequency response specifications. The time domain specifications include
peak overshoot, rise time, settling time and the locations of the dominant roots which
are expressed in terms of damping ratio and undamped natural frequency of oscil-
lation. Root locus technique gives a viable solution to the design problem, in that
it reshapes the root loci of the uncompensated system so that the desired transient
response specifications are met in addition to satisfying the steady state error require-
ment. We discuss below the design methodologies for lead, lag and lag–lead com-
pensators using root locus method.

6.5.1 Design of Lead Compensator Using Root Locus

From Eq. (6.1), the T.F. of the lead compensator is given by

α(1 + T s)
G c (s) =
(1 + αT s)

The above equation is written in pole–zero form as

778 6 Design of Compensators

Fig. 6.15 Pole and zero of j

lead compensator

s-plane

0
1/ T 1/T

s+ 1
G c (s) = T
(6.23)
(s + 1
αT
)

The pole and zero of Equation (6.23) are represented in Fig. 6.15.
Step-by-Step Procedure to Design Lead Compensator

1. For the given uncompensated system determine the value of K which satisfies the
steady state error requirement. Locate the poles and zeros of the uncompensated
system in the s-plane as shown in Fig. 6.16.
2. Identify the point P in the s-plane that satisfies the given time domain specifica-
tion. Draw the line O P. Draw the line P Q which is horizontal to the negative
real axis.
3. Draw the line PC which is the bisector of O P Q.
4. Draw the vectors from all the poles and zeros of the uncompensated system to
the point P. If P is to be a point on the root locus, then the total angle of all the
vectors drawn from the zeros minus the total angle of all the vectors drawn from
the poles of the uncompensated system should be an odd multiple of 180◦ . If this
phase angle condition is not satisfied, then determine the phase angle deficiency
φ which is to be contributed by the lead network.

j
Q P

φ/2 φ/2 s-plane

θ1 θ2 θ
A C B 0

Fig. 6.16 Lead compensator design using root locus

6.5 Compensator Design in the Time Domain 779

5. Draw on either side of the bisector line PC, straight lines P A and P B which
make φ/2 with the bisector line PC as reference. These lines cut the negative
real axis at point A and at point B, respectively.
6. It can be shown that by locating the pole of the compensator at point A and the
zero at point B, the compensator provides the phase angle deficiency of φ when
inserted in cascade with the uncompensated system.
7. θ2 is the phase angle contribution of the zero of the compensator and θ1 is the
phase angle contribution of the pole. It is to be proved that (θ2 − θ1 ) = φ. From
triangle AB P, in Fig. 6.16

θ1 + (180◦ − θ2 ) + φ = 180◦
(θ2 − θ1 ) = φ

8. It is to be noted that the bisector need not be always drawn. One may fix point B
where the zero of the lead compensator is placed and making angle ∠B P A = φ,
the point A is fixed where the pole of the lead compensator is placed. This may
be necessary to satisfy the magnitude criterion of root locus at point P and also
static error constants simultaneously.

Example 6.10
The forward path T.F. of a unity feedback control system is

K
G(s) =
s(s + 1)

Design a lead compensator to meet the following specifications:

1. Damping ratio ζ = 0.7
2. Settling time ts = 1.4 s
3. Velocity error constant K v ≥ 2 s−1

(Anna University, December, 2006)

Solution
1.
K
G(s) =
s(s + 1)
K v = lim sG(s)
s→0
sK
= lim
s→0 s(s + 1)
2=K
780 6 Design of Compensators

K =2

2. For 2% error tolerance in the steady state

ts = 3.91T

where
1
T =
ζ ωn
1.4 = 3.91T
1.4
T =
3.91
3.91
ζ ωn =
1.4
3.91
ωn =
1.4 × 0.7

ωn = 4

3.

ωd = ωn 1 − ζ 2
√
= 4 1 − 0.49
= 2.8
ζ ωn = 0.7 × 4 = 2.8

The point P is located at s = −2.8 + j2.8.

4. From point P, the horizontal straight line P Q is drawn. The open loop poles are
placed at s = 0 and s = −1. From Fig. 6.17a, the phase angle contribution from
these open loop poles is (134◦ + 120◦ ) = 254◦ . For the point P to be a point on
the root locus, the lead compensator should contribute +74◦ (−180◦ + 254◦ ).φ =
74◦ .
5. The bisector of O P Q is drawn as PC. The lines P A and P B are drawn with angle
37◦ from PC as shown in Fig. 6.17a. At point A, the pole of the compensator is
placed. It is at s = −7.5. At point B, the zero of the compensator is placed. It is
at s = −2.
Thus
1
=2
T
6.5 Compensator Design in the Time Domain 781

(a) j
K 25
Q P 2.8 j2.8

37º 37º s-plane

d
=
5
5.

c=

4
b = 2.9
=
a

3.3
120º 134º

−7.5 −2 B −1 0

A
C

(b)
R(s) + (s + 2) 2 C(s)
12.5 Gc(s) = G(s) =
− (s + 7.5) s(s+1)
Amplifier

Fig. 6.17 Pole–zero locations for Example 6.10. b Block diagram of compensated system

T = 0.5

1
= 7.5
αT
1
α=
7.5 × 0.5

α = 0.267

6. The T.F. of the lead compensator is

(s + 2)
G c (s) =
(s + 7.5)

7. From Fig. 6.17a, the gain at point P is calculated as

782 6 Design of Compensators

acd
K =
b
5.5 × 3.3 × 4
=
2.9
= 25

Already the gain available is 2. Hence, an amplifier with a gain of 12.5 is to be

provided. Under this condition, the velocity error constant K v is

s(s + 2) 2 × 12.5
K v = lim
s→0 (s + 7.5) s(s + 1)
50
=
7.5
= 6.67 s−1

This is more than the given specification of K v ≥ 2 sec−1 . The block diagram of
the compensated system is shown in Fig. 6.17b.

Example 6.11
Design a suitable lead compensator for a system with unity feedback and having the
open loop T.F.
K
G(s) =
s(s + 1)(s + 4)

to meet the following specifications:

1. Damping ratio ζ = 0.5.
2. Undamped natural frequency of oscillation ωn = 2.
3. The velocity error constant K v ≥ 1.5.

Solution
K
G(s) =
s(s + 1)(s + 4)
K v = 1.5
K v = lim sG(s)
s→0
sK
= lim
s→0 s(s + 1)(s + 4)
K
1.5 =
4
6.5 Compensator Design in the Time Domain 783

(a) j
K 24
Q P 1 j 3

30º s-plane
12 30º
c
a

3
2

b
A 30º 90º 60º 120º
−4 B −1 0

(b)

R(s) + (s + 1) 6 C (s )
4 Gc(s) =
− (s + 4) s(s+1)(s+4)
Amplifier G(s)

Fig. 6.18 a Pole–zero diagram for Example 6.11. b Block diagram of the compensated system for
Example 6.11

K =6

Hence
6
G(s) =
s(s + 1)(s + 4)

The poles of G(s) are placed as shown in Fig. 6.18a. Since ωn = 2 and ζ = 0.5

ωd = ωn 1 − ζ 2
√
= 2 0.75
√
= 3
ζ ωn = 0.5 × 2
=1
√
The point P is fixed at −1 + j 3. The horizontal line P Q is drawn. The bisector
PC of O P Q is drawn. The phase lead to be contributed by the compensator is φ =
−180◦ + 120◦ + 90◦ + 30◦ = 60◦ . The lines P A and P B are drawn with ∠A PC =
∠B PC = 30◦ . The point A is located at −4 where the pole of the compensator is
placed. The point B is located at s = −1 where the zero of the compensator is placed.
The value of K at point P is calculated as
784 6 Design of Compensators

a×a×b×c
K =
b
= 12 × 2

K = 24

(s + 1)
G c (s) =
(s + 4)
1
The zero of the compensator is T
. Hence

1
=1
T

T =1

The pole of the compensator is 1/αT . Hence

1
=4
αT

α = 0.25

The T.F. of the compensated system is

K a (s + 1)6
G c (s)G(s) =
(s + 4)s(s + 1)(s + 4)
6K a
=
s(s + 4)2

Here, pole–zero cancellation takes place. In addition to the system gain of K = 6, an

amplifier with a gain of K a = 4 is to be connected so that at point P, the magnitude
criterion of gain = 24 is satisfied. For the compensated system, the velocity error
constant is
s24
K v = lim
s→0 s(s + 4)2
K v = 1.5
6.5 Compensator Design in the Time Domain 785

The specification is satisfied. The block diagram of the compensated system is shown
in Fig. 6.18b.

Example 6.12
Design a lead compensator for a control system having a T.F. as

1
G(s) =
s(s + 1)(s + 5)

to meet the following specifications. (a) Over-shoot for step input is 25%. (b) Settling
time of ts = 5 s

(Anna University, December, 2003)

Solution
1. For 25% overshoot due to dominant roots, we have

−ζ π
%M p = e  × 100
1 − ζ2
−ζ π
e × 100 = 25
1 − ζ2

Solving for ζ , we get

ζ = 0.4

The settling time ts is given (for 2% error tolerance) as

ts = 3.91T
5 = 3.91T
T = 1.28
1
T =
ζ ωn
1
ωn =
Tζ
1
=
1.28 × 0.4

ωn = 2 rad/s
786 6 Design of Compensators

ωd = ωn 1 − ζ 2

= 2 1 − (0.4)2 = 1.833

2. The point P is located at s = −0.8 + j1.833. The horizontal line P Q is drawn.

Vectors from the poles of G(s) from s = 0, s = −1 and s = −5 are drawn to
point P. The angles contributed by these poles are measured. The sum of these
angle is φ = 110◦ + 80◦ + 20◦ = 210◦ . The phase lead to be contributed by the
compensator is 210◦ − 180◦ = 30◦ .
3. The bisector of OPQ is drawn as PC. Lines P A and P B are drawn with ∠A PC =
∠B PC = 15◦ . The pole of the compensator is located at point A which is at
s = −3 and the zero of the compensator is placed at point B which is at s = −1.5.
The T.F. of the compensator is

(s + 1.5)
G c (s) =
(s + 3)

4. The gain of the compensated system at point P is calculated as

4.8 × 2.9 × 1.9 × 2

K =
2

K = 26.5

The block diagram of the compensated system with an amplifier gain as 26.5
(Since the gain of the original system is 1) is shown in Fig. 6.19b.
5. The velocity error constant of the uncompensated system is

K v = lim sG(s)
s→0
1
= lim s
s→0 s(s + 1)(s + 5)
= 0.2

The velocity error constant of the compensated system is

K v = lim sG c (s)G(s)
s→0
(s + 1.5)26.5
= lim s
s→0 (s + 3)s(s + 1)(s + 5)
K v = 2.65

The increased velocity error constant reduces the steady state error.
6.5 Compensator Design in the Time Domain 787

(a) K 26.5 j
Q 0.8 j1.833
P

15º 15º s-plane

.8

d
4

=
a=

c=

2
2.9

b=2
a=

1.9
80º
20º 70º
−5 −3 −1.5 −1 0
A

C B

(b)
R(s) + (s + 1.5) 1 C (s )
26.5 Gc(s) =
− (s + 3) s(s+1)(s+5)
Amplifier G(s)

Fig. 6.19 a Pole–zero diagram for Example 6.12. b Block diagram of the compensated system for
Example 6.12

6.5.2 Design of Lag Compensator Using Root Locus

To improve the steady state error, the lag network is inserted in cascade with the
given system. By this method, for the given time domain specification, the gain of
the system is increased as much as possible without sacrificing the transient response
characteristics. For this, the pole and zero of the lag network are placed quite nearer
to the origin of the s-plane which influence the low frequency response and hence
the steady state characteristic of the system. From Eqs. (6.6) and (6.7), we write

α(s + z)
G c (s) =
(s + p)

where

p = αz

and

α<1

The pole–zero diagram of the lag compensator is shown in Fig. 6.20.

788 6 Design of Compensators

Fig. 6.20 Pole–zero j

diagram of a lag s-plane
compensator
0
z (1/ T) p 1/T

Step-by-Step Procedure to Design Lag Compensator

1. Locate the poles and zeros of the uncompensated system in the s-plane. Draw the
root locus for the uncompensated system.
2. Identify the point P in the root locus which is chosen to satisfy the given time
domain specifications for the dominant roots.
3. Determine the gain of the uncompensated system at the point P using magnitude
criterion. Let this gain be K u .
4. Determine the value of K of the open loop T.F. that satisfies the steady state error
requirement.
5. When the compensator is inserted in the forward path with the T.F

α(s + z)
G c (s) =
(s + p)

the forward path gain together with the gain of the uncompensated system gain
becomes K α. This should satisfy the magnitude criterion of root locus at point P
chosen in step 2. However, the introduction of the compensator slightly modifies
this gain roughly by a factor of 1.1. Thus, the gain at point P is 1.1K α. This
should be equal to K u determined in step 3. Thus, we write

1.1K α = K u

Ku
α= (6.24)
1.1K

Knowing K u and K , calculate α.

6. Place the zero of the compensator on the negative real axis which is roughly about
one tenth of the distance of the second pole of the uncompensated system pole
from the origin.
7. The pole of the compensator is determined using p = αz. Thus, the lag compen-
sator design is completed.
8. Draw the root locus for the compensated system and verify whether the given
specifications are satisfied by the compensated system. If not, go to step 5 and
repeat the procedure till the required specifications are obtained.
6.5 Compensator Design in the Time Domain 789

Example 6.13
The forward path T.F. of a certain unity negative feedback control system is given
by

K
G(s) =
s(s + 2)(s + 4)

The system is to have a damping factor ζ = 0.6 for the dominant poles. The steady
state error for unit ramp input should be limited to 0.2 rad. Design a suitable lag
network.

Solution
1.
K
G(s) =
s(s + 2)(s + 4)
K v = lim sG(s)
s→0
sK K
= lim =
s→0 s(s + 2)(s + 4) 8
1
ess =
Kv
8
0.2 =
K

K = 40

2. The root locus plot for G(s) is shown in Fig. 6.21. For the damping ratio, ζ = 0.6

cos θ = ζ
θ = cos−1 ζ
= cos−1 0.6
= 53◦

The damping factor line for ζ = 0.6 is drawn. This line makes 53◦ from −ve real
axis and cuts the root locus at point P. The point P is identified as −0.75 + j1.1
in the s-plane. The gain at the point P is calculated using magnitude criterion of
root locus. Thus, we obtain

K u = 1.3 × 1.75 × 3.5

790 6 Design of Compensators

K 48

0.6

Ku 8 8
0.75 j1.1 P
5

1.3
3.5 1.7
K K 0 53° K 0
K 0 60°
4 2 60° 0.76 0

Fig. 6.21 Root locus plot for uncompensated system

Ku = 8

3. Using Eq. (6.24), the value of α is calculated as

Ku
α=
1.1K
8
=
1.1 × 40

α = 0.182

4. The second pole of G(s) is at s = −2. The zero of the compensator is placed at
one tenth of this value. Thus
6.5 Compensator Design in the Time Domain 791

z = 0.2

5. Using the relation p = αz, we get

p = 0.18 × 0.2

p = 0.036

6. Thus, the compensator T.F. is obtained as

α(s + z)
G c (s) =
(s + p)
0.182(s + 0.2)
G c (s) =
(s + 0.036)

7. The T.F. of the compensated system is written as

0.182(s + 0.2)40
G c (s)G(s) =
(s + 0.036)s(s + 2)(s + 4)
K v = 5.05
1
ess =
5.05
= 0.198

which is less than the specified value.

Example 6.14
Design a lag compensator for a system with open loop T.F

K
G(s) =
s(s + 1)(s + 4)

The system is to be compensated to meet the following specifications:

(a) Damping ratio ζ = 0.4.
(b) Settling time for 5% tolerance ts = 10 s.
(c) Velocity error constant K v ≥ 5 s−1 .
(d) The specifications in the frequency domain are: Phase margin φm = 43◦ .

(Anna University, May, 2009)

792 6 Design of Compensators

Solution The design is first carried out in the time domain using root locus and for
the compensated system, the frequency domain specifications are verified.
1.
K
G(s) =
s(s + 1)(s + 4)
K v = lim sG(s)
s→0
sK
= lim
s→0 s(s + 1)(s + 4)
K
5=
4

K = 20

The settling time for 5% error tolerance is

ts = 3T

where
1
T =
ζ ωn
3
ωn =
0.4 × 10

ωn = 0.75


ωd = ωn 1 − ζ 2

= 0.75 1 − (0.4)2
= 0.7

2. The root locus for the given open loop T.F. is drawn as shown in Fig. 6.22a. The
point P corresponds to −ζ ωn + jωd = −0.3 + j0.7 or ζ = 0.4 line.
3. The point P should satisfy magnitude criterion of root locus. The gain of the
uncompensated system at point P is

Directed distances from all poles to P

Ku =
Directed distances from all zeros to P
6.5 Compensator Design in the Time Domain 793

(a) j

K 20

0.4

Ku 2 2
0.3 j0.7 P

0.7
3.76 1

5
K K 0 60° 66.4° K 0
4 1.67 60° K
1
0 0

0.465

(b)

−20dB/dec.

51.8
−40dB/dec.
+dB −20dB/dec.
16.1
ωgc 0.637
log ω
0
0.0128 0.1 1 4
−3.92 −40dB/dec.

−28 −60dB/dec.
−dB

K
Fig. 6.22 a Root locus for G(s) = . b Bode magnitude plot of the compensated
s(s + 1)(s + 4)
system
794 6 Design of Compensators

K u = 0.75 × 1.0 × 3.76

K u = 2.82

4. Using Eq. (6.23), the value of α is calculated as

Ku
α=
1.1 × K
2.82
=
1.1 × 20

α = 0.128

5. Put the zero of the lag compensator at one tenth value of the second pole nearer
to the origin
z = 0.1

6. The pole of the compensator is given by

P = αz
= 0.128 × 0.1

P = 0.0128

7. The T.F. of the lag compensator is

α(s + z)
G c (s) =
(s + p)
0.128(s + 0.1)
G c (s) =
(s + 0.0128)

8. The T.F. of the compensated system is

0.128(s + 0.1)20
G c (s)G(s) =
s(s + 0.0128)(s + 1)(s + 4)

In time constant form, the above equation is written as

5(1 + 0.1
s
)
G c (s)G(s) =
s(1 + 0.0128 )(1 + s)(1 + 4s )
s
6.5 Compensator Design in the Time Domain 795

The corner frequencies are 0.0128, 0.1, 1 and 4. The Bode magnitude plot is
shown in Fig. 6.22b.
9. From Fig. 6.22b, the gain crossover frequency is obtained as ωgc = 0.637. At this
frequency, the phase angle is

0.637 0.637 0.637

φ = −90◦ − tan−1 − tan−1 0.637 − tan−1 + tan−1
0.0128 4 0.1
= −90◦ − 50◦ − 32.5◦ − 9◦ + 81.5◦
= −100◦

Phase margin, φm = 180 + φ = 180◦ − 100◦

φm = 80◦

6.5.3 Design of Lag–Lead Compensator Using Root Locus

The lag compensator with its pole and zero located closer to the origin in the s-plane
improves the steady state response of the system and has little effect on the transient
response. On the other hand, the lead network, which contributes phase lead at the
gain crossover frequency, improves the transient response and has little effect in
improving the steady state error. To accomplish the advantages of both lag and lead
compensators, the lag–lead network is designed. The T.F. of the lag compensator as
given in Eq. (6.10) is

(1 + α1 T1 s)(1 + α2 T2 s)
G c (s) =
(1 + T1 s)(1 + T2 s)

The above equation is written in pole–zero form as

(s + 1
α1 T1
)(s + 1
α2 T2
)
G c (s) = (6.25)
(s + T1 )(s
1
+ 1
T2
)

The poles and zeros of Equation (6.25) are as shown in Fig. 6.23. As already
stated, the pole and zero of the lag portion of the compensator is located closer to the
origin. Further, for any point chosen on the root locus, the magnitude contribution
of the lag network is almost one and the phase contribution is legibly small. If these
conditions are satisfied, by arbitrarily placing the pole and zero of the lag portion,
then the lead portion of the compensator is designed exactly by the method described
in Sect. 6.5.1. The step-by-step procedure to design the lag–lead compensator is given
below.
796 6 Design of Compensators

Fig. 6.23 Pole–zero j

diagram of lag–lead
compensator s-plane

1/ 2T2 1/ 2T1 1/T1 0

1/T2

Step-by-Step Procedure to Design Lag–Lead Compensator

1. Determine the value of K which satisfies the steady state error requirement.
2. Locate the dominant pole location in the s-plane that satisfies the time domain
specification. Let this point be P. Join O P. Draw horizontal line P Q.
(s+ α 1T )
3. The lead portion (s+ T1 )
2 2
contributes maximum phase angle to the point P and
2
(s+ α 1T )
the lag portion 1 1
(s+ T1 )
contributes negligibly small phase. Further, it is assumed
1
that
 
 (s + 1 ) 
 α1 T1 
  =1
 (s + T1 ) 
1 s= p


(s + )
α1 T1 
1
∠ = 0◦
(s + T11 ) s= p

4. Calculate magnitude and phase angle contributions of G(s), to the point P. The
angle deficiency should be contributed by the lead network. If φ is the angle
deficiency draw lines P A and P B such that ∠A P B = φ as shown in Fig. 6.24.
5. Locate the pole of the lead portion at A and zero at point B. The magnitude ratio
|P A|
of |P B|
should be such that the gain of the system with the lead portion is the same
as that calculated in step 1. Otherwise, the locations of A and B are adjusted such
that the gain requirement is satisfied noting that the angle requirement ∠A P O is
always φ which satisfies the phase angle requirement of root locus.

Example 6.15
The open loop T.F. of a certain unity feedback control system is given by

K
G(s) =
s(s + 5)(s + 10)

Design a lag–lead compensator to meet the following specifications:

6.5 Compensator Design in the Time Domain 797

j
Q P

s-plane

A
1/T2 1 / 2T 2 B 0

Fig. 6.24 Pole–zero location of the lead portion of the lag–lead compensator

1. The velocity error constant K v ≥ 4.8 sec−1 .

2. The damping ratio for the dominant poles should be 0.6.
3. The undamped natural frequency of oscillation is 5 rad/s.

Solution
1.

K v = lim sG(s)
s→0
sK
= lim
s→0 s(s + 5)(s + 10)
K
4.8 =
50

K = 240

2.

ζ ωn = 0.6 × 5 = 3

ωd = ωn 1 − ζ 2
√
= 5 1 − 0.36
= 4 rad/s.

The point P is at −3 + j4 and the gain at that point should be 240 to satisfy K v
requirement.
3. The lines O P and P Q are drawn. The value of K at point P is 240. The poles
of G(s) are located as shown in Fig. 6.25. The angle contribution from the poles
798 6 Design of Compensators

j
K 240
Q P 3 j4

s-plane
8 50º c
5.8

b 4
4.1 5
90º 127º
30º 40º 73º

−10 −7.46 −5 −3 0

A
B

Fig. 6.25 Pole–zero locations of Lag–lead compensator for Example 6.15

of G(s) to point P is

θ = 127◦ + 73◦ + 30◦ = 230◦

4. For the point P to be the point on the root locus, it should satisfy angle criterion
of an odd multiple of 180◦ . Hence, the phase lead compensator should provide
a phase angle lead of 230◦ − 180◦ = 50◦ .
5. The gain calculated at the point P with the uncompensated system is obtained
by the product of directed distances from the poles of G(s) to the point P. As
seen from Fig. 6.25 it is 8 × 4.1 × 5 = 164. However, the required gain is 240.
6. Draw any line P A where it cuts the negative real axis at A. Draw the line P B
such that ∠A P B = 50◦ . Now the phase angle criterion of root locus at point P
is satisfied. But the required magnitude at P is 240, while the actual magnitude
is 164. By introducing the lead network, the magnitude should be increased by
|P A|
a factor of 240
164
= 1.46. That is |P B|
= 1.46.
|P A|
7. By trial and error, the points A and B are shifted such that |P B|
= 1.46 and
◦
keeping ∠P AB = 50 . When these two conditions are satisfied, the pole of the
lead portion is fixed at point A and the zero at point B. The point A is at s = −7.6
(s+4)
and the point B is at s = −3. Thus, the lead portion is designed with (s+7.6) . The
gain at point P is obtained as

8 × 5.8 × 4.1 × 5
K = = 240
4
8.
1
= 7.6
T2
6.5 Compensator Design in the Time Domain 799

T2 = 0.1315

1
=4
α2 T2
1 7.6
α2 = =
3T2 3

α2 = 2.53

9.

α1 α2 = 1
1
α1 =
α2
1
=
2.53

α1 = 0.394

The pole of the lag portion is chosen arbitrarily closer to the origin. Thus, we
choose
1
= 0.01
T1

T1 = 100

The zero of the lag portion is then obtained as

1 0.01
=
α1 T1 0.394
= 0.0253

10. The T.F. of the lag–lead compensator is

(s + 0.0253)(s + 3)
G c (s) =
(s + 0.01)(s + 7.6)

The T.F. of the compensated system is

800 6 Design of Compensators

(s + 0.0253)(s + 3)240
G c (s)G(s) =
(s + 0.01)(s + 7.6)(s)(s + 5)(s + 10)

SUMMARY

1. Compensators and controllers are intentionally inserted into the system for
improving the performance deficiency.
2. Series compensators, parallel compensators and series-parallel compensators are
normally used. However, the series (cascade) compensators are extensively used.
The series compensator design is discussed in this chapter.
3. In control system performance specifications, transient response specifications
are more important and they are the ultimate measure of performance indices.
Since the root locus method gives direct information about the transient response
of the closed loop system, it is preferred for the design of compensators in the
time domain.
4. The frequency response approach is a very powerful tool to design compensators
even though we get the information about the transient response performance in
an indirect way. Bode plot is the choice for the design of the compensators in the
frequency domain.
5. The compensators that are normally used are lag, lead and lag–lead compensators.
The design methodologies of these compensators in the frequency domain and
time domain are explained in detail with illustrative examples.

EXERCISE

Short Answer Type Questions

1. What do you understand by compensation?
The process of adjustment or alteration of a control system to get desired per-
formance specifications is called compensation.
2. What is a compensator?
The additional compensates or circuits that are inserted into the system in order
to compensate for a performance deficiency are called compensators.
3. What are the performance criteria specified for compensators?
Transient and steady state errors, good accuracy good speed of response, high
relative stability and good damping are the performance criteria that are com-
monly specified.
4. What are the performance specifications given in the frequency domain?
Phase margin, gain margin, resonant peak, phase crossover frequency, resonant
frequency, gain crossover frequency, bandwidth and static error constants are
some of the specifications given in the frequency domain.
6.5 Compensator Design in the Time Domain 801

5. How frequency domain specifications are correlated to time domain speci-

fications?
Phase margin, gain margin and resonant peak in the frequency domain give the
rough estimate of the system damping while gain crossover frequency reso-
nant frequency and bandwidth are closely associated with the speed of transient
response.
6. What are the different methods used to connect the compensator in the
system?
The compensators are connected in
1. Series or cascade
2. Parallel
3. Series-parallel
7. List out the advantages and disadvantages of the different methods of con-
necting the compensators?
In the series-connected scheme, the signal flows from lower energy level to higher
energy level. Therefore, additional amplifiers are required to increase the gain as
well as to provide necessary isolation. Thus, it has more number of components
and hence it is expensive. In the parallel scheme, the flow of energy is from higher
level to lower level, and hence additional components are not required. However,
series compensation is simpler than parallel compensation. The advantages of
series and parallel scheme are incorporated in the series-parallel scheme.
8. Why Root locus method is used to design compensators to meet time domain
specifications?
To meet the time domain specifications root locus method is used in which the
poles and zeros of the compensator can reshape the root locus of the system such
that the closed loop poles are placed in locations in the s-plane that give desired
transient response.
9. Does the transient response and the stability of a system depend only on
pole location?
The stability of the system solely depends upon the pole locations of the system.
The transient response also, to a great extent, depends upon pole locations.
However, the zeros also have their impact on the transient response.
10. Why Bode plot is preferred to design compensators in the frequency
domain?
When the compensator is added to the Bode plot of the original system, only
the portions that correspond to the poles and zeros of the compensator need
to be reshaped. Further, if the open loop gain is varied, the magnitude curve
of the Bode plot is shifted up or down without changing the slope of any part
of the Bode diagram while retaining the original phase curve. However, when
bandwidth requirement is specified, it is preferable to use the Nichols plot.
11. In the Bode diagram, how the frequency range is classified?
In the Bode plot, the frequency region is classified as low frequency region,
medium frequency region and high frequency region.
802 6 Design of Compensators

12. How low frequency region is identified in the Bode plot and what is its
implication?
In the Bode plot, the frequency range which is far below the gain crossover
frequency is termed as low frequency region and it indicates the steady state
behaviour of the system.
13. How medium frequency region is identified in the Bode plot and what is its
implication?
The region nearer to the gain crossover frequency is called the medium frequency
region which indicates the relative stability of the system.
14. How high frequency region is identified in the Bode plot and what is its
implication?
The region far above the gain crossover frequency is called the high frequency
region and it determines the transient response of the system.
15. How steady state accuracy is achieved in the low frequency region?
By increasing the gain at low frequency, the steady state accuracy is achieved.
16. How relative stability is achieved in the medium frequency region?
The slope of the lag-magnitude curve in the Bode diagram near the gain crossover
frequency should be −20 dB/s and to achieve the desired phase margin and high
relative stability, this slope is extended over a wide frequency range.
17. How improved transient response is achieved in the high frequency region?
To improve the transient response and to suppress the effects of noise, the gain
at high frequency region should be attenuated.
18. What is the advantage of the frequency response method of compensator
design over that of time domain approach using root locus?
In the frequency response method, it is easy to implement the design. In the Bode
plots, the required gain is easily found from the asymptotic plots, whereas in the
root locus, it requires several calculations.
19. How does a phase lead network perform?
The phase lead network provides a phase lead angle which offsets the excessive
phase lag which is inherently present in the original system. It is so designed
that the maximum phase contribution occurs near the gain crossover frequency,
thereby increasing the phase margin.
20. What is the maximum phase lead that can be practically obtained from a
lead compensator?
The maximum phase lead that can be obtained from a lead compensator in
practice is about 50◦ to 60◦ .
21. If more than 60◦ phase lead is required, how a compensator is designed?
If more than 60◦ phase lead is required, a two stage lead network with an isolation
amplifier is designed.
22. What is the effect of phase lead compensator on the steady state error?
The lead compensator will not have much effect on steady state error since it
provides attenuation at low frequency region.
23. What is the effect of lead compensator on the system bandwidth, rise time
and settling time?
By introducing a lead compensator, the phase margin is increased. This increases
6.5 Compensator Design in the Time Domain 803

the damping and the bandwidth. The time response is faster and therefore the
rise time and settling time are reduced.
24. What is the limiting value of the attenuation factor α in the lead compen-
sator?
The limiting value of the attenuation factor α is normally restricted to 0.07 from
the view point of physical realization. Further, if 1/α becomes very high, noise
is amplified.
25. How does a lag compensator perform?
In the phase lag compensator, attenuation is provided at the gain crossover fre-
quency which shifts the Bode magnitude plot downwards without affecting the
phase plot. The new gain crossover frequency is shifted to the left where the
enlarged phase margin is obtained which is required in the specifications.
26. What is the effect of lag compensator on steady state error?
Lag compensator, being a low pass filter, provides high gain at low frequencies.
This reduces the steady state error.
27. What is the effect of the lag compensator on system bandwidth, rise time
and settling time?
The introduction of the lag compensator shifts the gain crossover frequency to
the low frequency region of the Bode plot which diminishes the bandwidth of
the system. Thus, the rise time and settling time with phase lag compensator are
longer.
28. How does a lag–lead network perform?
The lag portion of the compensator provides attenuation near and above the
gain crossover frequency which increases the overall gain of the system at low
frequency region. This gives improved steady state performance of the system.
The phase lead portion provides phase lead angle at the gain crossover frequency
which increases the phase margin.
29. What is the effect of lag–lead network on bandwidth, rise time and settling
time?
Because of the increased phase margin, the bandwidth is increased. The rise time
and settling time are reduced which increases the fastness of response.
Long Answer Type Questions

1. The forward path T.F. of a certain unity feedback control system is given by

K
G(s) =
s(s + 10)

It is required to have a velocity error constant K v ≥ 20, phase margin is at least

45◦ and the gain margin is more than 20 dB. Design a suitable lead compensator.
Answer:
804 6 Design of Compensators

(s + 11.48)
G c (s) =
(s + 17.1)
φm = 47◦
Gm = ∞

Additional amplifier gain A = 1.5.

2. The forward path T.F of a unity feedback control system is given by

K
G(s) =
s(s + 2)(s + 10)

The steady state error for a unit step input is 0.5. The required phase margin is at
least 48◦ and the gain margin is at least 5 dB. Design a phase lead compensator.
Answer:
(s + 1.463)
G c (s) =
(s + 2.215)
φm = 50◦
G m = 5.86 dB
ωgc = 5.3
K = 60.5

3. The forward path T.F. of a certain unity feedback system is

K (s + 0.2)
G(s) =
s(s + 1)(s + 2)

The velocity error constant is K v ≥ 6 s−1 . The required phase margin is more
than 40◦ and gain margin more than 8 dB. Design a suitable lag compensator.
Answer:
(1 + 3s)
G c (s) =
(1 + 13.8s)
K = 60
α = 0.22
T = 13.8
φm = 41◦
Gm = ∞
ωgc = 3.3

4. The forward path T.F. of a certain unity feedback system is

6.5 Compensator Design in the Time Domain 805

K
G(s) =
s(s + 2)(s + 3)

It is required that the steady state error is less than 0.2 for unit step input. The
phase margin should be more than 40◦ and gain margin is at least 5 dB. Design a
suitable lag compensator.
Answer:
(1 + 11.32s)
G c (s) =
(1 + 66.2s)
K = 30
φm = 49◦
G m = 5.5 dB
α = 0.171
T = 66.2
ωgc = 0.776

5. The forward path T.F. of a certain unity feedback system is

K
G(s) =
s(s + 2)(s + 5)

It is required that the steady state error for unit step input should not exceed
0.2. The phase margin and gain margin should be greater than 45◦ and 7 dB
representatively. Design a suitable lag–lead network.
Answer:
(s + 0.3467)(s + 1.2)
G c (s) =
(s + 0.1)(s + 6.85)
K = 83
φm = 53.3◦
G m = 7.5 dB
ωgc = 5.62

6. The forward path T.F. of a certain unity feedback control system is

K
G(s) =
s(s + 2)(s + 8)

The velocity error constant K v ≥ 25. The required phase margin and gain margin
should be greater than 30◦ and 10 dB, respectively. Design a lag–lead compen-
sator.
Answer:
806 6 Design of Compensators

(s + 1.216)(s + 3)
G c (s) =
(s + 0.1)(s + 36)
K = 400
φm = 33.4◦
G m = 21.5 dB

7. Consider the following open loop T.F. of a certain unity feedback closed loop
system
K
G(s) =
s(s + 2)(s + 5)

It is required that the velocity error constant K v ≥ 4 sec−1 . The phase margin
should be greater than 40◦ and the gain margin is at least 5 dB. The gain crossover
frequency ωgc > 4 rad/s Design a lag–lead compensator.
Answer:
(s + 0.2317)(s + 1.944)
G c (s) =
(s + 0.1)(s + 4.5)
K = 40
φm = 49.1◦
G m = 5.2 dB
ωgc = 4.66 rad/s.
Chapter 7
State Space Modelling and Analysis

Learning Objectives

After completing this chapter, you should be able to

 Represent linear continuous time system by state equation.
 Model mechanical, electromechanical and electrical systems by state equa-
tion.
 Convert differential equation to state equation.
 Obtain state equation from transfer function and vice versa.
 Find the solution of state equation by the Laplace transform method and by
the general form.
 Determine the state transition matrix and the solution of state equation for
typical inputs.
 Define and determine state controllability, output controllability and observ-
ability of linear time invariant continuous time systems.
 Represent the discrete time system by state equation.
 Model the discrete time system by state equation.
 Find the solution of state equation using the z-transform method.
 Define state controllability, output controllability and observability of dis-
crete time systems.
 Develop a MATLAB program to solve state equations for continuous and
discrete time systems.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 807
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3_7
808 7 State Space Modelling and Analysis

7.1 Introduction

The transfer function (T.F.) model, for quite a long time, was used for the analysis
and design of linear time invariant continuous time systems. However, this model has
many limitations in that it is expressed as a ratio of output to input variables, and thus
the internal behaviour of the system is hidden. Further, the T.F. method is valid only
for linear systems with initial conditions being zero. It is powerless for non-linear,
time varying and multi-input and multi-output (MIMO) systems. It is also difficult
to handle large-scale complex systems with a transfer function model. Furthermore,
system design modelling by T.F. is based on the trial and error procedure which in
general will not lead to optimal control systems. All these limitations are overcome by
representing the system in the state space model. This is a differential (or difference)
equation model which is expressed in n first order equations which are written in a
specific format. The model is valid for linear, non-linear and time varying systems and
also when initial conditions are not zero. Unlike the T.F. model, it gives a complete
description of the internal behaviour of all physical variables in the system. In this
chapter, we first develop a state space model of mechanical systems and electrical
networks followed by the conversion of the T.F. model to the state space model. The
solution of the state equation is derived and illustrated by simple examples. Finally,
the discrete time system represented by the difference equation is converted into
state equations. Comparison of classical and modern control approaches are given
in Table 7.1.

7.2 The State of a System and State Equation

of Continuous Time System

For a linear continuous time system, the state of a system is defined as the minimum
number of initial conditions that must be specified at any initial time t0 so that
the complete dynamic behaviour of the system at any time t > t0 is determined
when the input u(t) is known.
When the input u(t) is applied, the future states of the system, for t > t0 , also
change and we can uniquely determine these states. Since the states of the system
vary with respect to time, we call these variables state variables.
The number of state variables depend on the dynamic model selected to describe
the physical systems. For a system described by the nth order differential equation,
there will be n state variables. If these n state variables form the coordinates of an n
dimensional vector space, it is known as state space. For a continuous time system
described by nth order differential equation if the variables which represent the states
are chosen less than n, then the system is not fully represented and information about
the missing states will not be known. Similarly, if the states are chosen more than n,
then some of the states chosen are redundant and they can always be expressed in
terms of other known states. Hence, for an nth order model of a system it is necessary
7.2 The State of a System and State Equation of Continuous Time System 809

Table 7.1 Comparison of classical and modern control theories

S. No. Classical control theory (Root locus Modern control theory
and frequency response methods)
1. Generally applicable to linear time Modern control theory can be
invariant system having single-input successfully applied in all these cases
and single-output It is powerless for
time varying, non-linear and
multi-input and multi-output systems
2. Not applicable to optimal and Can be successfully applied in these
adaptive control systems cases
3. It is difficult to handle large-scale It is not difficult to handle large-scale
complex systems complex systems with the help of
modern computers
4. This is a complex frequency domain This is essentially a time domain
approach approach
5. System design is based on trial and Optimal control system can be
error procedure which in general will designed with ease for the given
not lead to optimal control systems performance index
6. Design is carried out for a specific Design can be carried out for a class
input function such as impulse step of inputs
and ramp functions
7. Initial condition is to be zero Initial condition need not be zero
8. Information about one output at a Complete information about all the
time is known variables is known at a time

strictly to choose only n appropriate states and they are represented by n first order
equations together with the input.

7.3 Vector Matrix Differential Equation of Continuous

Time System

State variable equations, whether linear or non-linear, are expressed in the time
domain by using compact vector matrix notations. These equations are called vec-
tor matrix differential equations. The standard form of representing these state
equations for a continuous time system is

ẋ(t) = Ax(t) + Bu(t) (7.1)

y(t) = C x(t) + Du(t) (7.2)

Equation (7.1) is called the state equation and Eq. (7.2) is called output equation.
Equations (7.1) and (7.2) together describe the system dynamics and they are called
vector matrix differential equations.
810 7 State Space Modelling and Analysis

u(t) ˙
x(t) x(t) y(t)
B ∫ dt C

Fig. 7.1 Block diagram representation of state Eqs. (7.1) and (7.2)

In Eqs. (7.1) and (7.2),

x(t) = State vector (n × 1 dimension);
y(t) = Output vector ( p × 1 dimension);
u(t) = Input vector (r × 1 dimension);
A = State matrix (n × n dimension);
B = Input matrix (n × r dimension);
C = Output matrix ( p × n dimension);
D = Direct transmission matrix ( p × r dimension).
A block diagram representation of Eqs. (7.1) and (7.2) is shown in Fig. 7.1.
Depending upon the dimensions of the vectors x, u and y, the appropriate dimen-
sions of the matrices, A, B and C are chosen. In most practical applications, the
direct transmission of input u(t) to the output y(t) is not done and hence D = 0. In
forming the state Eq. (7.1), it is to be observed that only the first derivative of x(t)
appears on the left side of the equation and no derivative of x(t) appears on the right
side. The right side of the equation contains only the states and input. The following
examples illustrate the method of forming state equations.

7.3.1 State Equations for Mechanical Systems

The dynamic equations of mechanical systems are written from the free body dia-
gram. The physical variables such as displacement and velocity are chosen as the
states and for each state variable, the equation for its first derivative is obtained and
converted in the format of Eq. (7.1).

Example 7.1
For the mechanical system shown in Fig. 7.2a, form the state equation.
7.3 Vector Matrix Differential Equation of Continuous Time System 811

(a) (b)

p
p
K d2p
M
dt2
M f(t) dp
B M f(t)
dt

B Kp

Mechanical System Free Body Diagram

Fig. 7.2 Mechanical system and its free body diagram

Solution
1. The mechanical system is shown in Fig. 7.2a. Its free body diagram is shown in
Fig. 7.2b.
2. p is the displacement of mass M and f (t) is the force applied. In the free body
diagram, the opposing forces act in the direction opposite to the direction of
motion.
3. From Fig. 7.2b, the following dynamic equations for the given mechanical system
are written:
d2 p dp
M 2 +B + K p = f (t) (7.3)
dt dt
4. Let us choose the displacement p, which is the physical variable in the mechan-
ical system as one state variable. Thus,

p = x1 (t) (7.4)

5.
dp
= ẋ1 (t)
dt
dp
Let us choose the velocity dt
as the second state variable. Thus,

dp
= ẋ1 (t) = x2 (t) (7.5)
dt
6. Equation (7.3) gives complete description of the given mechanical system and
it is a second order system. Therefore, there should be two states and two state
equations. Equation (7.5) represents one state equation. Similar to that, we should
obtain an equation for ẋ2 (t).
7. Now consider Eq. (7.3). By substituting p = x1 (t), dp
dt
= x2 (t) and f (t) = u(t),
we get
812 7 State Space Modelling and Analysis

M ẋ2 (t) + Bx2 (t) + K x1 (t) = u(t)

Solving for ẋ2 (t), we get

K B 1
ẋ2 (t) = − x1 (t) − x2 (t) + u(t) (7.6)
M M M
8. From Eqs. (7.5) and (7.6), the following vector matrix differential equation is
formed:
     0
ẋ1 (t) 0 1 x1 (t)
= [ẋ(t)] = K B + 1 u(t) (7.7)
ẋ2 (t) − − x2 (t)
 M  M M
 
A B

In Eq. (7.7),
   
0 1 0
A= K B and B = 1
− −
M M M
Note that the A matrix is a square matrix. In the examples to follow, for conve-
nience, the state variables are denoted as x instead of x(t).

Example 7.2
Consider the mechanical system shown in Fig. 7.3a. Form the state equation. The
displacement plus velocity of mass M are taken as the output.

Solution
1. The mechanical system is represented in Fig. 7.3a and its free body diagram in
Fig. 7.3b. The mass M is given a displacement of p2 . The point A moves with a
displacement p1 .
2. From the free body diagram, the following equations are written:

M p̈2 + B( ṗ2 − ṗ1 ) + K 2 p2 = f (t) (7.8)

B( ṗ1 − ṗ2 ) + K 1 p1 = 0 (7.9)

The following state variables are chosen:

p2 = x 1
ṗ2 = x2
p1 = x 3

The equations for the first derivatives of x1 , x2 and x3 are to be obtained.

7.3 Vector Matrix Differential Equation of Continuous Time System 813

(a) (b)

f(t) ˙˙ 2
Mp B(p˙ 2 p˙ 1)
K1
f(t) A p1
B
M
p2
M

K2 p2

K2 K 1 p1 B(p˙ 1 p˙ 2)

A
Mechanical System Free Body Diagram

Fig. 7.3 Mechanical system and its free body diagram

ẋ1 = ṗ2 = x2
M ẋ2 + K 1 x3 + K 2 x1 = u(t)

K2 K1 1
ẋ2 = − x1 − x3 + u(t)
M M M
B ẋ3 − Bx2 + K 1 x3 = 0
K1
ẋ3 = x2 − x3
B
Thus
⎡ ⎤
0 1 0 ⎡ ⎤ ⎡ 0
⎤
⎢ K2 K1 ⎥ x1
⎢− 0− ⎥⎣ ⎦ ⎢ 1 ⎥
ẋ(t) = ⎢ M M ⎥ x2 + ⎣ ⎦ u(t)
⎣ ⎦ x M
K1 3 0
0 1−  
  B
A B

The output y is given by

⎡
⎤
x1
y = [1 1 0] ⎣ x2 ⎦
 
x3
C
814 7 State Space Modelling and Analysis

(a)
p2 p1

B2

K1
M2 M1 f(t)

K2

Mechanical System
(b)

K2 p2 M1˙˙p1
K1(p2 p1)
M2 f(t) M1
B2 p˙ 2

M2 p˙˙2 K1(p1 p2)

Free Body Diagrams

Fig. 7.4 Mechanical system and its free body diagrams

Example 7.3
Consider the mechanical system shown in Fig. 7.4a. Form the state equation.

Solution
1. The mechanical system is shown in Fig. 7.4a and its free body diagram in
Fig. 7.4b.
2. From Fig. 7.4b, the following dynamic equations are written:

M2 p̈2 + B2 ṗ2 + K 2 p2 + K 1 ( p2 − p1 ) = f (t) (7.10)

M1 p̈1 + K 1 ( p1 − p2 ) = 0 (7.11)

3. The following state variables are chosen:

7.3 Vector Matrix Differential Equation of Continuous Time System 815

x 1 = p1
x2 = ẋ1 = ṗ1
x 3 = p2
x4 = ẋ3 = ṗ2

4. The first derivatives ẋ1 and ẋ2 are known. The first derivatives ẋ3 and ẋ4 are
obtained from Eqs. (7.11) and (7.10), respectively, and they are given as follows:

K1 K1
ẋ2 = − x1 + x3 (7.12)
M1 M1
K1 (K 1 + K 2 ) B2 1
ẋ4 = x1 − x3 − x4 + f (t) (7.13)
M2 M2 M2 M2

5. The vector matrix differential equation is thus obtained as

⎡ ⎤ ⎡ ⎤
0 1 0 0 0
⎢ K1 K1 ⎥ ⎢ 0 ⎥
⎢− 0 0 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
ẋ(t) = ⎢ M1 M1 ⎥x(t) + ⎢ 0 ⎥ u(t)
⎢ 0 0 0 1 ⎥ ⎢ ⎥
⎣ K1 (K 1 + K 2 ) B2 ⎦ ⎣ 1 ⎦
0− −
M2 M2 M2 M2

Example 7.4
For the mechanical system shown in Fig. 7.5, obtain the state space model.

K1 r
T(t) 1 J

B2
2
B1

B3 K3

Fig. 7.5 A hybrid mechanical system

816 7 State Space Modelling and Analysis

Solution
1. A hybrid mechanical system which is a combination of translational and rota-
tional systems is shown in Fig. 7.5. Just by inspection, the following dynamic
equations are written.
2.

T (t) = B1 (θ̇1 − θ̇2 ) + K 1 (θ1 − θ2 ) (7.14)

Also

T (t) = J θ̈2 + B2 θ̇2 + r (M p̈ + B3 ṗ + K 2 p) (7.15)

Substituting p = r θ2 in Eq. (7.15), we get

T (t) = (J + r 2 M)θ̈2 + (B2 + r 2 B3 )θ̇2 + r 2 K 3 θ2 (7.16)

3. The following state variables are chosen:

x1 = θ1
x2 = θ2
x3 = θ̇2 = ẋ2

4. The derivatives of x1 and x2 are obtained from Eqs. (7.14) and (7.16) and are
given as

K1 K1 1
ẋ1 = − x1 + x2 + x3 + T (t)
B1 B1 B1
ẋ2 = x3
K eq Beq 1
ẋ3 = − x2 − x3 + T (t)
Jeq Jeq Jeq

where Jeq = (J + r 2 M), Beq = (B2 + r 2 B3 ) and K eq = r 2 K 3 .

⎡ K K1 ⎤ ⎡ 1 ⎤
1
− 1
⎢ B1 B1 ⎥ ⎢ B1 ⎥
⎢ 1 ⎥ ⎢ ⎥
ẋ(t) = ⎢ 0 0 ⎥x(t) + ⎢ 0 ⎥ u(t)
⎣ K eq Beq ⎦ ⎣ 1 ⎦
0 − −
Jeq Jeq Jeq
7.3 Vector Matrix Differential Equation of Continuous Time System 817

7.3.2 State Equations for Electrical Circuits

The number of independent energy-storing elements in the electrical circuit deter-

mine the number of state variables. The capacitor and inductor are the two energy-
storing elements in electrical circuits: The physical variables, namely the current
through the inductor and voltages across the capacitor are chosen as the state vari-
ables. The following steps are followed while forming the state space equations for
electrical circuits:
1. Choose all independent inductor currents and the capacitor voltages as the state
variables.
2. The state variables and their first derivatives are expressed in terms of a set of
loop circuits.
3. Write loop equations and eliminate all variables other than the state variables
and their first derivatives.
The following examples illustrate the method of forming state equations for electrical
networks.

Example 7.5
Write the state equations for the network shown in Fig. 7.6.

Solution
1. Let i be the current passing through the inductor L and vc be the voltage across
the capacitor C. These variables are chosen as state variables.

x1 = i
x2 = vc

2. The following loop equation is written:

Fig. 7.6 R-L-C series R L

circuit
i

v(t) vc C
818 7 State Space Modelling and Analysis

di
L + Ri + vc = v(t)
dt
R 1 1
ẋ1 = − x1 − x2 + v(t) (7.17)
L L L
Also

1
vc = i dt
C

Differentiating both sides, we get

1
v̇c = i
C
1
ẋ2 = x1 (7.18)
C
3. Equations (7.17) and (7.18) represent the first derivatives of the chosen states.
Hence
⎡ ⎤
R 1 1
−
⎢ L L⎥ −
ẋ(t) = ⎣ 1 ⎦x(t) + L v(t)
0 0
C

Example 7.6
For the electrical network shown in Fig. 7.7, form the state equation.

Solution
1. Let i be the current passing through the inductor L and vc be the voltage across
the capacitor C. These variables are taken as the state variables. Thus,

Fig. 7.7 Electrical circuit of L ic

Example 7.6
i
i1

v(t) R vc C
7.3 Vector Matrix Differential Equation of Continuous Time System 819

x1 = i
x2 = vc

2. The following loop equation connecting L and C is written:

di
L + vc = v(t)
dt
1 1
ẋ1 = − x2 + v(t) (7.19)
L L
3. Also

ic = i − i1
dvc vc
C =i−
dt R
1 1
ẋ2 = x1 − x2 (7.20)
C RC
4. Combining Eqs. (7.19) and (7.20), we get
⎡ ⎤
1 1
0 −
⎢ ⎥
L ⎦x(t) +
ẋ(t) = ⎣ L v(t)
1 1
− 0
C RC

Example 7.7
For the electrical network shown in Fig. 7.8, form the state equations.

Solution
1. There are three energy-storing elements, namely L , C1 and C2 , and they are
independently connected and there should be three state variables. The following

4 2H
i1 v1 i3 v2

R1 L i5
i2 i4

v(t) C1 0.25F C2 0.5F R2

is
1

Fig. 7.8 Electrical network of Example 7.7

820 7 State Space Modelling and Analysis

state variables are chosen:

x1 = v1
x2 = v2
x3 = i 3

2. The first derivative of these variables is obtained as follows. At v1 node

i1 = i2 + i3
(v − v1 ) dv1
= C1 + i3
R1 dt
dv1 1 1 1
=− v1 − i3 + v
dt R1 C 1 C1 R1 C 1
1 1 1
ẋ1 = − x1 − x3 + v (7.21)
R1 C 1 C1 R1 C 1

3. At note v2 , the following equation is written:

i3 + is = i4 + i5
dv2 v2
= C2 +
dt R2

Substituting the state variables, we get

x2
x3 + i s = C2 ẋ2 +
R2
1 1 1
ẋ2 = − x2 + x3 + is (7.22)
R2 C 2 C2 C2

4. The voltage drop across the inductor L is

di 3
L = v1 − v2
dt
1 1
ẋ3 = x1 − x2 (7.23)
L L
5. Combining Eqs. (7.21), (7.22) and (7.23), we get
⎡ ⎤
1 1 ⎡ 1 ⎤
−
⎢ R1 C 1 0 − ⎥ 0 ⎡ ⎤
⎢ C1 ⎥ ⎢ R1 C 1 ⎥ v(t)
⎢ 1 1 ⎥ ⎢ 1 ⎥
ẋ(t) = ⎢ 0 − ⎥ x(t) + ⎢ 0 ⎥⎣ ⎦
⎢ R2 C 2 C 2 ⎥ ⎣ ⎦ i (t)
⎣ 1 1 ⎦ C2 s
− 0 0 0
L L
7.3 Vector Matrix Differential Equation of Continuous Time System 821

v3
i1 i3

R1 L1 C1 L2
R2

v1(t) v2(t)

C2
v4
i2 (i1 i3)

Fig. 7.9 Electrical network of Example 7.8

Substituting the numerical values, we get

⎡ ⎤ ⎡ ⎤⎡ ⎤
−1 0 −4 1 0 v(t)
ẋ(t) = ⎣ 0 −2 2 ⎦x(t) + ⎣0 2⎦ ⎣ ⎦
0.5 −0.5 0 0 0 i s (t)

Example 7.8
For the electrical network shown in Fig. 7.9, form the vector matrix differential
equation.

Solution
1. There are four energy-storing elements, namely L 1 , L 2 , C1 and C2 . Hence,
four state variables, the current through the inductors and the voltages across the
capacitors are chosen. Thus,

x1 = i 1
x2 = i 3
x3 = v3
x4 = v4

The dimension of A matrix is 4 × 4. Since there are two inputs, namely v1 (t)
and v2 (t), the dimension of B matrix is 4 × 2.
2. The following loop equation is written connecting v1 (t), R1 , L 1 , R2 and C2 :

di 1
v1 (t) = L 1 + i 1 R1 + (i 1 + i 3 )R2 + v4
dt
(R1 + R2 ) R2 1 1
ẋ1 = − x1 − x2 − x4 + v1 (t) (7.24)
L1 L1 L1 L1
822 7 State Space Modelling and Analysis

3. The following loop equation is written connecting v2 (t), L 2 , C1 , R2 and C2 :

di 3
v2 (t) = L 2 + v3 + (i 1 + i 3 )R2 + v4
dt
R2 R2 1 1 1
ẋ2 = − x1 − x2 − x3 − x4 + v(t) (7.25)
L2 L2 L2 L2 L2

4. For the capacitance C1 , the following equation is written:

dv3
C1 = i3
dt
1
ẋ3 = x2 (7.26)
C1

5. For the capacitance C2 , the following equation is written:

dv4
C2 = i1 + i3
dt
1 1
ẋ4 = x1 + x2 (7.27)
C2 C2

6. Combination Eqs. (7.24)–(7.26) and (7.27), the following vector matrix differ-
ential equation is obtained:
⎡ ⎤ ⎡1 ⎤
− (R1L+R 2)
− R2
0 − 1
0
L
⎢ − R2 − R2 − 1 − 1 ⎥
1 1 L 1 L1  
⎢ ⎥ ⎢ 0 1 ⎥ v1 (t)
ẋ(t) = ⎢ L 2 L 2 L 2 L 2 ⎢
⎥x(t) + ⎣ L ⎥
0 0 ⎦ v2 (t)
2
⎣ 0 1
C1
0 0 ⎦
1 1 0 0
C2 C2
0 0

Example 7.9
Develop the state model for the electrical network shown in Fig. 7.10.

(Anna University, April, 2004)

Solution
1. There are two energy-storing elements L and C. The current i 1 through L 1 and
voltage vc across C are chosen as the state variables. Thus

x1 = i 1
x2 = vc

2. The following equation is written connecting R1 , R2 and C:

7.3 Vector Matrix Differential Equation of Continuous Time System 823

vc
L 2H

i1 i2
C 1F

ei R1 3 R2 5 e0

i1 i2

Fig. 7.10 Electrical network of Example 7.9

vc + i 2 R2 = (i 1 − i 2 )R1
R1 vc
i2 = i1 − (7.28)
(R1 + R2 ) (R1 + R2 )

The following equation is written connecting ei , L , C and R2 :

di 1
ei = L + vc + i 2 R2
dt
di 1 R1 R2 vc R2
=L + vc + i 1 −
dt R1 + R2 (R1 + R2 )
 
R1 R2 1 R2 1
ẋ1 = − x1 + −1 x2 + ei (7.29)
L(R1 + R2 ) L (R1 + R2 ) L

3. For the capacitor C, the following equation is written:

dvc
C = i2
dt
i 1 R1 vc
= −
(R1 + R2 ) (R1 + R2 )
R1 1
ẋ2 = − x1 − x2 (7.30)
C(R1 + R2 ) C(R1 + R2 )

4. Combining Eqs. (7.29) and (7.30), we get

⎡  ⎤
(R1 R2 )
− L(R 1 R2
−1 1
1 +R2 ) R1 +R2
ẋ(t) = ⎣ ⎦x(t) + L ei
L
0
R1
C(R1 +R2 )
− C(R11+R2 )

Substituting the numerical value, we get

   
− 15 − 16
3
0.5
ẋ(t) = 16 x(t) + ei
3
8
− 1
8
0
824 7 State Space Modelling and Analysis

Fig. 7.11 First order T.F. u x

realization

1
s

a 10 y
x

5. The output is

e0 = i 2 R2
R1 R2 vc
= i1 −
R1 + R2 R1 + R2
1
= (15x1 − x2 )
8  
1 x
y = e0 = [15 − 1] 1
8 x2

7.4 State Equations from Transfer Function

State equations can be easily obtained from the transfer function of the system.
Consider the first order system with the following transfer function:

10
H (s) =
(s + a)

The system realization is shown in Fig. 7.11.

From Fig. 7.11, the following equations are derived:

ẋ = −ax + u
y = 10x (7.31)

The output of each integrator ( 1s ) is chosen as one state variable. Thus, for an nth
order system, n integrators are required. The following methods of realization are
used to determine the state equation:
1. The direct form or controllable canonical form.
2. The cascade form.
3. The parallel form or diagonal form.
7.4 State Equations from Transfer Function 825

7.4.1 General Case of Representation—Phase Variable or

Controllable Canonical Form

The state space description can be done in several ways. However, the state variables
obtained from the direct form are quite convenient since the state equations can be
immediately written just by inspection of the transfer function. Consider the general
N th order transfer function given as follows:

b0 s N + b1 s N −1 + b2 s N −2 + · · · + b N
H (s) =
s N + a1 s N −1 + a2 s N −2 + · · · + a N
 
b0 + bs1 + bs 22 + · · · + bs NN
=   (7.32)
1 + as1 + as 22 + · · · + as NN

Equation (7.32) is realized in direct form structure and is shown in Fig. 7.12. From
Fig. 7.12, the following equations for the state variable are written:

ẋ1 = x2
ẋ2 = x3
ẋ3 = x4
..
.
ẋ N −1 = x N
ẋ N = −a N x1 − a N −1 x2 , . . . , −a2 x N −1 − a1 x N + u(t) (7.33)
y(t) = b N x1 + b N −1 x2 + · · · + b1 x N + b0 ẋ N
= (b N − b0 a N )x1 + (b N −1 − b0 a N −1 )x2 + · · ·
+(b1 − b0 a1 )x N + b0 u(t) (7.34)

Equations (7.33) and (7.34) can be represented in matrix form given as follows:
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 0 ··· 0 0 x1 0
⎢ 0 0 1 · · · 0 0 ⎥ ⎢ x2 ⎥ ⎢0⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ .. .. .. .. .. .. ⎥ ⎢ .. ⎥ + ⎢ .. ⎥ u(t)
ẋ(t) = ⎢ . . . . . . ⎥⎢ . ⎥
⎥ ⎢ ⎢ ⎥
⎢ ⎥ ⎢.⎥
⎣ 0 0 0 ··· ··· 1 ⎦ ⎣ x N −1 ⎦ ⎣0⎦
−a N −a N −1 −a N −2 · · · −a2 −a1 xN 1
y(t) = [(b N − b0 a N )(b N −1 − b0 a N −1 ) . . . (b1 − b0 a1 )]x + b0 u(t)
= [b̄ N b̄ N −1 . . . b̄1 ]x + b0 u(t) (7.35)

where b̄ N = (b N − b0 a N ). The A matrix given in Eq. (7.35) is said to be in phase

variable canonical form.
826 7 State Space Modelling and Analysis

u(t) x˙ N y(t)
b0

1
s

xN
a1 x˙ N 1 b1

1
s

xN 1
a2 x˙ N 2 b2

1
s

x2
aN 1 x˙1 bN 1

1
s

aN x1 bN

Fig. 7.12 Direct form realization of N th order LTIC system

Step-by-Step Procedure to Determine A, B and C Matrices

1. If the system is described by linear differential equation, convert that in T.F. form.
Find the coefficients of numerator polynomial b0 , b1 , . . . , b N and the coefficients
of the denominator polynomials a0 , a1 , . . . , a N .
2. The elements of the A matrix are written in phase variable canonical form. The
elements of the last row are written in the reverse order with a negative sign as
−a N , −a N −1 , . . . , −a2 , a1 . The elements of the B matrix and identified with 0s
in all the rows and 1 by the last row.
3. The elements of C matrix are identified as given in Eq. (7.35). To remember this
in an easier way, the elements of the first column are obtained from the product
b0 a N being subtracted from b N , the second column from the product b0 a N −1 being
subtracted from b N −1 and so on. This can be easily viewed from Fig. 7.12. For
the state x1 , the right-side branch gain is b N . The left-side branch gain after being
7.4 State Equations from Transfer Function 827

multiplied by b0 is −b0 a N . The sum of these two is (b N − b0 a N ). Similarly, the

second column of C is obtained which corresponds to the state x2 . This is nothing
but (b N −1 − b0 a N −1 ).

Example 7.10
Consider the following differential equation which describes the dynamics of a con-
tinuous time system:

d4 y d3 y d2 y dy du(t)
5 4
+ 2 3
+ 4 2
+7 + 8y = 8 + 7u(t)
dt dt dt dt dt
Form the state space equation.

Solution

d4 y d3 y d2 y dy du(t)
5 + 2 + 4 +7 + 8y = 8 + 7u(t)
dt 4 dt 3 dt 2 dt dt
Taking the Laplace transform on both sides of the above equation, we get

Y (s)
H (s) =
X (s)
(8s + 7)
=  
5 s4 + 2 3
5
s + 45 s 2 + 75 s + 85

where
8 7 2 4 7 8
b0 = 0, b1 = 0, b2 = 0, b3 = , b4 = , a1 = , a2 = , a3 = , a4 =
5 5 5 5 5 5
The state equation is written in phase variable canonical form given as follows:
⎡ ⎤ ⎡ ⎤
0 1 0 0 0
⎢ 0 0 1 0 ⎥ ⎢0⎥
ẋ(t) = ⎢
⎣ 0
⎥ x(t) + ⎢ ⎥ u(t)
0 0 1 ⎦ ⎣0⎦
− 85 − 75 − 45 − 25 1
7
b̄4 = b4 − b0 a4 =
5
8
b̄3 = b3 − b0 a3 =
5
b̄2 = b2 − b0 a2 =0
b̄1 = b1 − b0 a1 =0
1
y = [7 8 0 0] x
5 
C
828 7 State Space Modelling and Analysis

Example 7.11
Consider the following T.F. of a certain continuous time system:

7s 3 + 11s 2 + 14s + 10
H (s) =
s 3 + 8s 2 + 5s + 4

Form the state equations in controllable canonical form.

Solution

7s 3 + 11s 2 + 14s + 10
H (s) =
s 3 + 8s 2 + 5s + 4

where

b0 = 7, b1 = 11, b2 = 14, b3 = 10, a1 = 8, a2 = 5, a3 = 4

⎡ ⎤ ⎡ ⎤
0 1 0 0
ẋ(t) = ⎣ 0 0 1 ⎦ x(t) + ⎣ 0 ⎦ u(t)
−4 −5 −8 1
b̄3 = (b3 − b0 a3 ) = (10 − 28) = −18
b̄2 = (b2 − b0 a2 ) = (14 − 35) = −21
b̄1 = (b1 − b0 a1 ) = (11 − 56) = −45
y = [−18 − 21 − 45]x + 7u(t)

7.4.2 General Case of Representation-Observable Canonical

Form

Phase variable controllable canonical form of state equation is well suited when the
controller is designed for the desired pole location. However, when an observer is
designed, it is convenient to put the state equation in observable canonical form.
Consider the following T.F. of a continuous time system:

Y (s) b0 s 2 + b1 s + b2
= H (s) = 2 (7.36)
U (s) s + a1 s + a2
b0 + bs1 + bs 22
=
1 + as1 + as 22

Cross-multiplying the above equation, we get

7.4 State Equations from Transfer Function 829

U(s) Y(s)
u(t) b0 x1(t) y(t)

1
s

x1(t)

b1 a1
x2(t)

1
s

x2(t)

b2 a2

Fig. 7.13 Block diagram to represent Eq. (7.37)

  
a1 a2  b1 b2
1+ + 2 Y (s) = b0 + + 2 U (s)
s s s s
 a   
1 a 2 b1 b2
Y (s) = − − 2 Y (s) + b0 + + 2 U (s) (7.37)
s s s s

Equation (7.37) is represented in Fig. 7.13.

From Fig. 7.13, the following state equations are written:

y(t) = x1 (t) + b0 u(t) (7.38)

ẋ1 (t) = −a1 y(t) + x2 (t) + b1 u(t)
= −a1 x1 (t) − a1 b0 u(t) + x2 (t) + b1 u(t)
= −a1 x1 (t) + x2 (t) + (b1 − b0 a1 )u(t) (7.39)
ẋ2 (t) = −a2 y(t) + b2 u(t)
= −a2 x1 (t) − a2 b0 u(t) + b2 u(t)
= −a2 x1 (t) + (b2 − b0 a2 )u(t) (7.40)

Equations (7.39) and (7.40) are combined and expressed in matrix form as
   
−a1 1 (b1 − b0 a1 )
x(t) = x(t) + u(t) (7.41)
−a2 0 (b2 − b0 a2 )
830 7 State Space Modelling and Analysis

The output equation is written using Eq. (7.38) as

y(t) = [1 0]x(t) + b0 u(t) (7.42)

Now consider the following T.F. in general:

Y (s) b0 s n + b1 s n−1 + b2 s n−2 + · · · + bn−1 s + bn

= H (s) = n (7.43)
U (s) s + a1 s n−1 + a2 s n−2 + · · · + an−1 s + an
b0 + bs1 + bs 22 + · · · + bs n−1
n−1
+ bs nn
= (7.44)
1 + as1 + as 22 + · · · + as n−1
n−1
+ as nn

Cross-multiplying the above equation, we get

  
a1 a2 an  b1 b2 bn
1+ + 2 + · · · + n Y (s) = b0 + + 2 + · · · + n U (s)
s s s s s s
 a a2 an 
1
Y (s) = − + + 2 + · · · + n Y (s)
 s s s 
b1 b2 bn
+ b0 + + 2 + · · · + n U (s) (7.45)
s s s

Equation (7.45) is represented in Fig. 7.14. From Fig. 7.14, we write the following
equation:

y(t) = x1 (t) + b0 u(t)

ẋ1 (t) = −a1 y(t) + x2 (t) + b1 u(t)
= −a1 x1 (t) − b0 a1 u(t) + x2 (t) + b1 u(t)
= −a1 x1 (t) + x2 (t) + (b1 − b0 a1 )u(t) (7.46)
ẋ2 (t) = −a2 y(t) + b2 u(t) + x3 (t)
= −a2 x1 (t) − b0 a2 u(t) + b2 u(t)
= −a2 x1 (t) + x3 (t) + (b2 − b0 a2 )u(t) (7.47)
ẋ3 (t) = −a3 y(t) + b3 u(t)
= −a3 x1 (t) − b0 a3 u(t) + b3 u(t)
= −a3 x1 (t) + x4 (t) + (b3 − b0 a3 )u(t) (7.48)

In general, we write

ẋn = −an x1 (t) + (bn − b0 an )u(t) (7.49)

7.4 State Equations from Transfer Function 831

Fig. 7.14 Block diagram to

represent Eq. (7.49)
u(t) b0 x1 y(t)

1
s

x1

b1 a1
x2

1
s

x2

b2 a2
x3

1
s

x3

b3 a3
xn-1

1
s

xn-1

bn-1 an-1
xn

1
s

xn

bn an

From the above Eqs. (7.46)–(7.49), the following vector matrix differential equation
is written:
⎡ ⎤ ⎡ ⎤
−a1 1 0 0 0 · · · 0 b̄1
⎢ −a2 0 1 0 0 · · · 0⎥ ⎢ b̄2 ⎥
⎢ ⎥ ⎢ ⎥
⎢ −a3 0 0 1 0 · · · 0⎥ ⎢ b̄3 ⎥
⎢ ⎥ ⎢ ⎥
ẋ(t) = ⎢ . .. .. .. .. .. .. ⎥ x(t) + ⎢ .. ⎥ u(t)
⎢ .. . . . . . .⎥ ⎥ ⎢ ⎥
⎢ ⎢ . ⎥
⎣−an−1 0 0 0 0 0 1⎦ ⎣b̄n−1 ⎦
−an 0 0 0 0 0 0 b̄n
y(t) = [1 0 0 · · · 0]x(t) + b0 u(t) (7.50)

where
832 7 State Space Modelling and Analysis

b̄1 = b1 − b0 a1
b̄2 = b2 − b0 a2
..
. (7.51)
b̄n−1 = bn−1 − b0 an−1
b̄n = bn − b0 an

Step-by-Step Procedure to Form Observable Canonical Form

1. For the given T.F., from the numerator polynomial determine the coefficients
b0 , b1 , b2 , . . . , bn and from the denominator polynomial determine the coeffi-
cients a1 , a2 , . . . , an . The elements of the A matrix are written in observable
canonical form. The elements of the first column are written in the ascending
order with a negative sign as −a1 , −a2 , . . . , −an , where −a1 represents the coef-
ficient of the first column and the first row while −an is the coefficient of the
first row and the last column. The other elements are represented by 1s and 0s as
shown in Eq. (7.50).
2. The elements of B matrix are determined using Eq. (7.51).
3. The elements of the C matrix are given as

C = [1 0 0 · · · 0]

4. The D matrix has the element b0

Example 7.12
Consider the following T.F. given in Example 7.11:

7s 3 + 11s 2 + 14s + 10
H (s) =
s 3 + 8s 2 + 5s + 4

Form the state equation in observable canonical form. Also find the output equation.

Solution
7s 3 + 11s 2 + 14s + 10
H (s) =
s 3 + 8s 2 + 5s + 4

b0 = 7; b1 = 11; b2 = 14; b3 = 10;

a1 = 8; a2 = 5; a3 = 4
7.4 State Equations from Transfer Function 833

b̄1 = b1 − b0 a1
= (11 − 7 × 8) = −45
b̄2 = b2 − b0 a2
= (14 − 7 × 5) = −21
b̄3 = b3 − b0 a3
= (10 − 7 × 4) = −18

⎡ ⎤ ⎡ ⎤
−a1 1 0 b̄1
ẋ(t) = ⎣−a2 0 1⎦ x(t) + ⎣b̄2 ⎦ u(t)
−a3 0 0 b̄3
y(t) = [1 0 0]x(t) + b0 u(t)

Substituting the numerical values, we get

⎡ ⎤ ⎡ ⎤
−8 1 0 −45
ẋ(t) = ⎣−5 0 1⎦ x(t) + ⎣−21⎦ u(t)
−4 0 0 −18
y(t) = [1 0 0]x(t) + 7u(t)

Example 7.13
Determine the state space model of a continuous time system whose transfer function
is given by

3s 2 + 24s + 44
H (s) =
(s 3 + 12s 2 + 44s + 48)

Use the following methods:

(a) The controllable and observable form.
(b) The cascade form.
(c) The diagonal or Jordan canonical form.
Show that the A matrix is not unique for the given system.

Solution
(a) The Controllable and Observable Form

3s 2 + 24s + 44
H (s) =
(s 3 + 12s 2 + 44s + 48)
834 7 State Space Modelling and Analysis

Fig. 7.15 Controllable u(t) x3 y(t)

canonical form realization of
H (s) =
3s 2 + 24s + 44
(s 3 + 12s 2 + 44s + 48) 1
s

12 3

x2 x3

1
s

44 24
x1 x2

1
s

48 44
x1

Here, b0 = 0; b1 = 3; b2 = 24; b3 = 44; a1 = 12; a2 = 44; a3 = 48. The direct

form realization of H (s) is shown in Fig. 7.15. From Fig. 7.15, the following
equations are written:

ẋ1 = x2
ẋ2 = x3
ẋ3 = −48x1 − 44x2 − 12x3 + u(t)

The state equation for H (s) is therefore written as

⎡ ⎤ ⎡ ⎤
0 1 0 0
ẋ(t) = ⎣ 0 0 1 ⎦ x(t) + ⎣ 0 ⎦ u(t) (7.52)
−48 −44 −12 1
   
A B

Also from Fig. 7.15, the output y(t) is obtained as

y(t) = 44x1 + 24x2 + 3x3

y(t) = [44 24 3 ]x(t)
 
C
7.4 State Equations from Transfer Function 835

The Observable Canonical Form

Since b0 = 0,
⎡ ⎤ ⎡ ⎤
−a1 1 0 b1
ẋ(t) = ⎣−a2 0 1⎦ x(t) + ⎣b2 ⎦ u(t)
−a3 0 0 b3
 
y(t) = 1 0 0 x(t)
⎡ ⎤ ⎡ ⎤
−12 1 0 3
ẋ(t) = ⎣−44 0 1⎦ x(t) + ⎣24⎦ u(t)
−48 0 0 44
 
y(t) = 1 0 0 x(t)

(b) The Cascade Form

3s 2 + 24s + 44
H (s) =
(s 3 + 12s 2 + 44s + 48)

The above equation can be written as

(s + 5.155) (s + 2.845) 1
H (s) = 3
(s + 4) (s + 6) (s + 2)
= H1 (s)H2 (s)H3 (s)

The cascade form realization is shown in Fig. 7.16.

From Fig. 7.16, the following equations are written for the first derivatives of the
states:

u(t) x˙1 x˙ 2 x˙3 y(t)

1 1 1
s s s

4 5.155 6 2.845 2 3
x1 x2 x3
H1(s) H2(s) H3(s)

3(s + 5.155)(s + 2.845)

Fig. 7.16 The cascade realization of H (s) =
(s + 4)(s + 6)(s + 2)
836 7 State Space Modelling and Analysis

ẋ1 = −4x1 + u(t)

ẋ2 = −6x2 + ẋ1 + 5.155x1
= −6x2 − 4x1 + u(t) + 5.155x1
= 1.155x1 − 6x2 + u(t)
ẋ3 = −2x3 + ẋ2 + 2.8455x2
= −2x3 + 1.155ẋ1 − 6x2 + u(t) + 2.845x2
ẋ3 = 1.155x1 − 3.155x2 − 2x3 + u(t)
y(t) = 3x3

The state equations are given as follows:

⎡ ⎤ ⎡ ⎤
−4 0 0 1
ẋ(t) = ⎣ 1.155 −6 0 ⎦ x(t) + ⎣ 1 ⎦ u(t) (7.53)
1.155 −3.155 −2 1
y(t) = [0 0 3] x(t)

(c) The Diagonal or Jordan Canonical Form

3s 2 + 24s + 44
H (s) =
(s 3
+ 12s 2 + 44s + 48)
3s 2 + 24s + 44
=
(s + 2)(s + 4)(s + 6)
1 1 1
= + +
(s + 2) (s + 4) (s + 6)
= H1 (s) + H2 (s) + H3 (s)

The parallel form realization of H (s) is shown in Fig. 7.17.

From Fig. 7.17, the following equations are written for the first derivatives of the
states:

ẋ1 = −2x1 + u(t)

ẋ2 = −4x2 + u(t)

ẋ3 = −6x3 + u(t)
y(t) = x1 + x2 + x3
⎡ ⎤ ⎡ ⎤
−2 0 0 1
ẋ(t) = ⎣ 0 −4 0 ⎦ x(t) + ⎣ 1 ⎦ u(t) (7.54)
0 0 −6 1
y(t) = [1 1 1]x(t)
7.4 State Equations from Transfer Function 837

x˙1
1

1
H1(s) s

2
x1

u(t) x˙2 y(t)

1

1
H2(s) s

4
x2

x˙3
1

1
s
H3(s)
6
x3

Fig. 7.17 The parallel form realization of H (s)

Equations (7.52), (7.53) and (7.54) give the state space description of the sys-
tem T.F. and the system A matrices are not unique. But all of them will
give the same characteristics of the system. In parallel form representation, the
eigenvalues of the system T.F. form the diagonal elements of the A matrix.

Example 7.14
Consider the following T.F. of a certain system:

10(s + 2)
H (s) =
s 2 (s + 1)2 (s + 4)

Determine A, B and C matrices using the Jordan canonical form (Fig. 7.18).
838 7 State Space Modelling and Analysis

x1 x2
1 1
s s
5

35
4

u(t) x˙ 3 x˙4 y(t)

1 1
s s
10
1 1 3
x3 x4
80
9

x˙5

1
s
5
4 36
x5

Fig. 7.18 The parallel form realization of H (s) for Example 7.14

Solution
10(s + 2)
H (s) =
s 2 (s
+ 1)2 (s + 4)
A1 A2 A3 A4 A5
= 2 + + + + (7.55)
s s (s + 1)2 (s + 1) (s + 4)

The residues A1 , A2 , A3 , A4 and A5 are determined by any one method discussed in

previous chapters and are given as follows:

A1 = 5
35
A2 = −
4
10
A3 =
3
7.4 State Equations from Transfer Function 839

80
A4 =
9
5
A5 = −
36
35 1 5 80 1 10 1 5 1
H (s) = − + 2+ + − (7.56)
4 s s 9 (s + 1) 3 (s + 1)2 36 (s + 1)

Equation (7.36) is represented in Fig. 7.15. From Fig. 7.15, the following equations
are in terms of state variables:

ẋ1 = u(t)
ẋ2 = x1
ẋ3 = −x3 + u(t)
ẋ4 = x3 − x4
ẋ5 = −4x5 + u(t)
35 80 10 5
y(t) = − x1 + 5x2 + x3 + x4 − x5
4 9 3 36
The above equations are written in vector matrix differential equation form as
⎡ ⎤ ⎡ ⎤
00 0 0 0 1
⎢1 0 0 0 0 ⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
ẋ(t) = ⎢ ⎥ ⎢ ⎥
⎢0 0 −1 0 0 ⎥ x(t) + ⎢1⎥ u(t)
⎣0 0 1 −1 0 ⎦ ⎣0⎦
0 0 0 0 −4 1
 
35 80 10 5
y(t) = − 5 − x(t)
4 9 3 36

7.5 Transfer Function of Continuous Time System

from State Equations

Consider the state Eqs. (7.1) and (7.2)

ẋ(t) = Ax(t) + Bu(t)

y(t) = C x(t) + Du(t)

Let the initial conditions be zero. Taking the Laplace transform on both sides of the
above equations, we get

[s I − A]X (s) = BU (s)

840 7 State Space Modelling and Analysis

Pre-multiplying both sides by [s I − A]−1 , we get

X (s) = [s I − A]−1 BU (s) (7.57)

Similarly,

Y (s) = C X (s) + DU (s)

= C[s I − A]−1 BU (s) + DU (s)

Y (s)
= H (s) = C[s I − A]−1 B + D (7.58)
U (s)

[s I − A]−1 is called the state transition matrix (STM). In Eq. (7.58),

Adjoint [s I − A]
[s I − A]−1 =
Determinant [s I − A]

Example 7.15
Consider the following state equations:
   
−3 1 1
ẋ = x+ x(t)
−2 0 0
y = [0 1]u(t)

Determine the system function.

Solution
   
−3 1 1
ẋ = x+ u
−2 0 0
y= [0 1]x
 
(s + 3) −1
(s I − A) =
2 s
 
1 s 1
(s I − A)−1 =
(s + 1)(s + 2) −2 (s + 3)
  
1 s 1 1
(s I − A)−1 B =
(s + 1)(s + 2) −2 (s + 3) 0
 
1 s
=
(s + 1)(s + 2) −2
7.5 Transfer Function of Continuous Time System from State Equations 841

Y (s)
H (s) =
U (s)
= C[s I − A]−1 B
 
1 s
= [0 1]
(s + 1)(s + 2) −2

−2
H (s) =
(s + 1)(s + 2)

7.6 Solution of State Equations

The state equations of a linear time invariant system are solved in both time and
frequency domains. In the frequency domain, the Laplace transform method is used.
These two methods are discussed ahead.

7.6.1 Laplace Transform Solution of State Equations

Consider the vector matrix differential equation of (7.1)

ẋ = Ax + Bu(t)

Taking the Laplace transform on both sides of the above equation, we get

s X (s) − x(0) = AX (s) + BU (s)

(s I − A)X (s) = x(0) + BU (s) (7.59)

Pre-multiplying both sides of Eq. (7.59) by [s I − A]−1 , we get

X (s) = [s I − A]−1 [x(0) + BU (s)]

= φ(s)[x(0) + BU (s)] (7.60)

where

φ(s) = [s I − A]−1 (7.61)

X (s) = φ(s)x(0) + φ(s)BU (s)

Taking the inverse Laplace transform, we get

x(t) = L −1 [φ(s)x(0)] + L −1 [φ(s)BU (s)] (7.62)

842 7 State Space Modelling and Analysis

φ(s) defined in Eq. (7.61) is the STM. In Eq. (7.62), L −1 [φ(s)x(0)] gives the zero
input response and L −1 [φ(s)BU (s)] gives the zero state response.

Example 7.16
A certain system is described by the following state equation:

ẋ = Ax + Bu(t)

where
   
−3 1 1
A= ; B= ; u(t) = u(t)
−2 0 0

The initial conditions are x1 (0) = 1 and x2 (0) = −1. Find STM and hence x(t). Also
find y(t) if C = [0 1].

Solution
   
−3 1 1
ẋ = x+ u(t)
−2 0 0
 
(s + 3) −1
(s I − A) =
2 s

The STM is
 
1 s 1
φ(s) = [s I − A]−1 =
(s + 1)(s + 2) −2 s + 3
X (s) = [s I − A]−1 x(0) + [s I − A]−1 BU (s)
  
−1 1 s 1 1
(z I − A) x(0) =
(s + 1)(s + 2) −2 s + 3 −1
 
1 (s − 1)
=
(s + 1)(s + 2) −(s + 5)
⎡ (s − 1) ⎤
⎢ 1)(s + 2) ⎥
= ⎣ (s +(s + 5) ⎦
−
(s + 1)(s + 2)

Given u(t) = u(t) and U (s) = 1s ,

  
−1 1 s 1 1 1
[s I − A] BU (s) =
(s + 1)(s + 2) −2 s + 3 0 s
7.6 Solution of State Equations 843
 
1 s
=
s(s + 1)(s + 2) −2
⎡ 1 ⎤
⎢ + 2) ⎥
= ⎣ (s + 1)(s 2 ⎦
−
s(s + 1)(s + 2)
x(t) = L {[s I − A]−1 x(0) + [s I − A]−1 BU (s)}
−1
⎡ (s − 1) ⎤ ⎡ 1 ⎤
⎢ + 2) ⎥ + L −1 ⎢ (s + 1)(s + 2) ⎥
= L −1 ⎣ (s +(s1)(s+ 5) ⎦ ⎣ −2 ⎦
(s + 1)(s + 2) s(s + 1)(s + 2)
⎡ 2 3 ⎤
− +
⎢ 1) (s + 2) ⎥
= L −1 ⎣ (s + 4 3 ⎦
− +
(s + 1) (s + 2)
⎡ 1 1 ⎤
−
⎢ ⎥
+ L −1 ⎣ 1 (s + 1)2 (s + 2)1 ⎦
− + −
s (s + 1) (s + 2)
 
2e−2t − e−t
x(t) =
2e−2t − 2e−t − 1

x1 (t) = (2e−2t − e−t )u(t)

x2 (t) = (2e−2t − 2e−t − 1)u(t)
y(t) = C x(t) = 2e−2t − 2et − 1

7.6.2 Time Domain Solution to State Equations

Consider the state equation

ẋ = Ax + Bu(t) (7.63)

Pre-multiplying Eq. (7.47) by e−At on both sides, we get

e−At ẋ = e−At Ax + e−At Bu(t) (7.64)

(e−At ẋ − e−At Ax) = e−At Bu(t)
d −At
[e x] = e−At Bu(t)
dt
By integrating both sides of the above equation from 0 to t, we get
844 7 State Space Modelling and Analysis
 t
e−At x − x(0) = e−Aτ Bu(τ )dτ
0

 t
x(t) = e At x(0) + e A(t−τ ) Bu(τ )dτ (7.65)
0

In Eq. (7.65), e At is the STM; Eq. (7.65) can be generalized to any initial value t0 and
hence it can be modified as
 t
A(t−t0 )
x(t) = e x(t0 ) + e A(t−τ ) Bu(τ )dτ
t0
= Free response + Forced response (7.66)

7.6.3 Determination of e At —The Cayley–Hamilton Theorem

To determine x(t) which is the solution of the vector matrix differential Eq. (7.65),
it is necessary to determine the STM e At . This can be obtained using the Cayley–
Hamilton theorem. According to this theorem, an n × n square matrix A satisfies its
own characteristic equation |λI − A| = 0 where λs are the eigenvalues of system
matrix A. Let


n−1
F(A) = α k Ak
k=0


n−1
F(λ) = α k λk
k=0


n−1
e At = α k Ak (7.67)
k=0

Equation (7.67) should satisfy for all the eigenvalues of A.

7.6.3.1 Determination of e At for Distinct Eigenvalues of A

If the eigenvalues of A are distinct, then the following procedure is followed to

evaluate e At :
1. Determine the eigenvalues of the matrix A from |λI − A| = 0.
2. For distinct eigenvalues, the following equations are written:
7.6 Solution of State Equations 845

eλ1 t = α0 + α1 λ1 + α2 λ21 + · · · + αn−1 λn−1

1
eλ2 t = α0 + α1 λ2 + α2 λ22 + · · · + αn−1 λn−1
2
..
.
eλn t = α0 + α1 λn + α2 λ2n + · · · + αn−1 λn−1
n (7.68)

Thus, we will have n simultaneous equations if there are n distinct eigenvalues.

By solving these simultaneous equations, α0 , α1 , α2 , . . . , αn can be determined.
3. Using the following equation, e At can be evaluated:

e At = α0 I + α1 A + α2 A2 + · · · + αn−1 An−1 (7.69)

where I = n × n identity matrix.

7.6.3.2 Determination of e At for Multiple Eigenvalues of A

Let us assume that λ = λi has multiplicity of m. If all the other eigenvalues are
distinct, then the number of distinct eigenvalues are (n − m + 1). Corresponding
to these eigenvalues, we will have (n − m + 1) independent equations. For the rest
(m − 1), we use Cauchy’s residue theorem.
  n−1 
d i f (λ)  di 
= αk λ k
(7.70)
dλi λ=λi dλi k=0
λ=λi

where i = 1, 2, 3, . . . , (m − 1). The following examples illustrate the method of

evaluating STM and the solution for x(t).

Example 7.17
Consider the following vector matrix differential equation:

   
−3 1 1
ẋ = x+ u(t)
−2 0 0

with the initial conditions

 
1
x(0) =
−1

Find (a) e At and (b) x(t) for unit step input signal.
846 7 State Space Modelling and Analysis

Solution
 
−3 1
A=
−2 0
|λI − A| = λ2 + 3λ + 2
= (λ + 1)(λ + 2)
λ1 = −1
λ2 = −2

Using Eq. (7.68), we get

e−t = α0 − α1
e−2t = α0 − 2α1

Solving the above simultaneous equations for α0 and α1 , we get

α0 = 2e−t − e−2t
α1 = e−t − e−2t
e At = α0 I + α1 A
 −t   
(2e − e−2t ) 0 −t −2t −3 1
= + (e − e )
0 (2e−t − e−2t ) −2 0
 −t −2t −t −2t

(−e + 2e ) (e − e )
=
2(−e−t + e−2t ) (2e−t − e−2t )
 t
x(t) = e At x(0) + e A(t−τ ) Bu(τ )dτ
0

Free Response x F R (t)

  
(−e−t + 2e−2t ) (e−t − e−2t ) 1
x F R (t) = e At x(0) =
2(−e−t + e−2t ) (2e−t − e−2t ) −1
 −2t 
(3e − 2e−t )
=
(3e−2t − 4e−t )

Forced Response x F O (t)

 t
x F O (t) = e A(t−τ ) Bu(τ )dτ
0
 t  
−e−(t−τ ) + 2e−2(t−τ ) e−(t−τ ) − e−2(t−τ ) 1
= dτ
0 2(−e−(t−τ ) + e−2(t−τ ) ) 2e−(t−τ ) − e−2(t−τ ) 0
 t  
−e−(t−τ ) + 2e−2(t−τ )
= dτ
0 2(−e−(t−τ ) + e−2(t−τ ) )
7.6 Solution of State Equations 847
 t 
−e−(t−τ ) + 2e−2(t−τ )
=
2(−e−(t−τ ) + e−2(t−τ ) )
  0
e−t − e−2t
=
−1 + 2e−t − e−2t

Hence, the total response is

x(t) = x F R (t) + x F O (t)

 −2t   
3e − 2e−t e−t − e−2t
= +
3e−2t − 4e−t 2e−t − e−2t − 1

 
(2e−2t − e−t )
x(t) =
(2e−2t − 2e−t − 1)

The result is the same as derived in Example 7.16.

Example 7.18
Determine e At for the following A matrix:
⎡ ⎤
4 1 −2
A = ⎣1 0 2 ⎦
1 −1 3

Solution
⎡ ⎤
4 1 −2
A = ⎣1 0 2 ⎦
1 −1 3
F(λ) = |λI − A|
= λ3 − 7λ2 − 9 + 15λ
= (λ − 1)(λ − 3)2

This is the case of repeated eigenvalues.

848 7 State Space Modelling and Analysis

λ1 = 1
λ2 = 3
λ3 = 3
et = α0 + α1 + α2
e3t = α0 + 3α1 + 9α2
d λt d
e = (α0 + α1 λ + α2 λ2 )
dλ dλ
For λ = 3

te3t = α1 + 6α2

Solving the above three simultaneous equations, we get

1 t
α0 = (9e + 6te3t − 5e3t )
4
1
α1 = (−6et − 8te3t + 6e3t )
4
1
α2 = (et + 2te3t − e3t )
4
e At = α0 I + α1 A + α2 A2
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
100 4 1 −2 15 6 −12
= α0 ⎣ 0 1 0 ⎦ + α1 ⎣ 1 0 2 ⎦ + α2 ⎣ 6 −1 4 ⎦
001 1 −1 3 6 −2 5

⎡ ⎤
(−te3t + e3t ) (te3t ) (−2te3t )
e At =⎣ (te3t ) (2et + te3t − e3t ) −2(et + te3t − e3t ) ⎦
(te )
3t
(et + te3t − e3t ) (−et − 2te3t + 2e3t )

Example 7.19
A system is represented by the state equation

   
0 1 0
ẋ(t) = x(t) + u(t)
−4 −5 1

Find the STM.

7.6 Solution of State Equations 849

Solution The Laplace transform of STM is given by

φ(s) = [s I − A]−1
Adjoint of [s I − A]
=
|s I − A|
 
s −1
[s I − A] =
4 s+5
 
s+5 1
Adjoint of [s I − A] =
−4 s
|s I − A| = s(s + 5) + 4
= (s + 1)(s + 4)
φ(s) = [s I − A]−1
 
1 s+5 1
=
(s + 1)(s + 4) −4 s
 
(s+5) 1
= (s+1)(s+4) (s+1)(s+4)
−4 s
(s+1)(s+4) (s+1)(s+4)
 A4 
A1
+ A2 A3
+
= s+1 s+4 s+1 s+4
A5
s+1
+ A6 A7
s+4 s+1
+ A8
s+4
A1 (s + 4) + A2 (s + 1) = s + 5

4
For s = −1, A1 =
3
1
For s = −4, A2 = −
3
A3 (s + 4) + A4 (s + 1) = 1
1
For s = −1, A3 =
3
1
For s = −4, A4 = −
3
A5 (s + 4) + A6 (s + 1) = −4

4
For s = −1, A5 = −
3
4
For s = −4, A6 =
3
A7 (s + 4) + A8 (s + 1) = s
1
For s = −1, A7 = −
3
4
For s = −4, A8 =
3
850 7 State Space Modelling and Analysis

Substituting these values in φ(s), we get

 4   
1
3 s+1
− 13 s+4
1 1
3
1
s+1
− s+4
1
φ(s) =    
− 43 s+1
1
− s+4 1
− 13 s+11
− s+4
4

Taking the inverse Laplace transform, the following STM is obtained:

 4  
e−t − 13 e−4t 1 −t
(e − e−4t )
φ(t) = 3 3
− 43 (e−t − e−4t ) − 13 (e−t − 4e−4t )

Example 7.20
Determine the transform matrix for the state model given as follows:
   
−1 −1 1
ẋ = x+ u
3 −5 1

(Anna University, November, 2010)

Solution
 
−1 −1
A=
3 −5
 
s+1 1
[s I − A] =
−3 s + 5
|s I − A| = (s + 1)(s + 5) + 3
= s 2 + 6s + 8
= (s + 2)(s + 4)

The eigenvalues are

λ1 = −2; λ2 = −4

φ(s) = [s I − A]−1
 
1 s+1 1
=
(s + 2)(s + 4) −3 s + 5
 
(s+1) 1
(s+2)(s+4) (s+2)(s+4)
φ(s) = −3 (s+5)
(s+2)(s+4) (s+2)(s+4)
1    1 
3
2 s+2
− 1
s+4
1
2
− s+2 + 1
s+4
=    1 
3 1
2 s+2
− 1
s+4
1
2
− s+2 + 3
s+4
−1
φ(t) = L φ(s)
7.6 Solution of State Equations 851

The STM is
 3   1 −2t 1 −4t 
e−2t − 21 e−4t − e + e
φ(t) = e At
=  23 −2t 3 −4t
  21 −2t 23 −4t 
2
e − 2e −2e + 2e

Example 7.21
Determine the time response for a system given as follows:

ẋ1 = −x1
ẋ2 = x1 − x2 + u(t)
 
1
x(0) =
0

(Anna University, November, 2010)

Solution
   
−1 0 0
ẋ = x+ u
1 −1 1
 
s+1 0
[s I − A] =
−1 s + 1
|s I − A| = (s + 1)2

The eigenvalues are λ1 = −1 and λ2 = −1.

eλ1 t = α0 + λ1 α1
e−t = α0 − α1
d λt
e = α1
dλ

−te−t = α1
α0 = α1 + e−t
= e−t − te−t
φ(t) = e At
= α0 I + α1 A
 −t   
e − te−t 0 −t −1 0
= + (−te )
0 e−t − te−t 1 −1
852 7 State Space Modelling and Analysis
 
e−t 0
=
−te−t e−t
 t 
e−(t−τ ) 0
x(t) = e At x(0) + Bu(τ )dτ
0 −(t − τ )e−(t−τ ) e−(t−τ )
 −t    t   
e 0 1 e−(t−τ ) 0 0
= + dτ
−te−t e−t 0 0 −(t − τ )e −(t−τ ) −(t−τ )
e 1
 −t   t  
e 0
= + dτ
−te−t 0 e−(t−τ )
 −t   t
e 0
= + −t τ
−te−t e e 0
 −t   
e 0
= +
−te−t (1 − e−t )
 
e−t
=
1 − e−t (1 + t)

x1 = e−t
x2 = 1 − e−t (1 + t)

Example 7.22
For a system represented by the state equation

ẋ(t) = Ax(t)
 −t   
e 1
x(t) = for x(0) =
−e−t −1
 −3t   
e 1
x(t) = for x(0) =
−3e−3t −3

Determine the state transition matrix φ(t) and the system matrix A.

(Anna University, November, 2010)

Solution Let the STM e At be expressed as
 
a11 a12
e At
=
a21 a22
x(t) = e At x(0)
    
e−t a11 a12 1
=
−e−t a21 a22 −1
7.6 Solution of State Equations 853

e−t = a11 − a12

−e−t = a21 − a22 (7.71)
    
e−3t a11 a12 1
=
−3e−3t a21 a22 −3
e−3t = a11 − 3a12
−3e−3t = a21 − 3a22 (7.72)

From Eqs. (7.71) and (7.72), the following simultaneous equations are formed:

e−t = a11 − a12

e−3t = a11 − 3a12
____________________________
1 −t
a12 = [e − e−3t ]
2
1 −t
a11 = [3e − e−3t ] (7.73)
2
−e−t = a21 − a22
−3e−3t = a21 − 3a22
____________________________
3 −3t 1 −t
a22 = e − e
2 2
3 −3t
a21 = [e − e−t ] (7.74)
2
From Eqs. (7.73) and (7.74), the state transition matrix is written as
 3  
e−t − 21 e−3t 1 −t
(e − e−3t )
e At
= φ(t) = 2
3 −3t
2 (7.75)
2
(e − e−t ) 1
2
(3e−3t − e−t )

Taking the Laplace transform, we get

1 1 
( 3 − 1
) (
1 1
− )
φ(s) = 2 s+1 s+3 2 s+1 s+3
(
3 1
2 s+3
− 1
s+1
) (
1 3
2 s+3
− 1
s+1
)
 
1 (s + 4) 1
= (7.76)
(s + 1)(s + 3) −3 s

From Eq. (7.76), the determinant is

(s + 1)(s + 3) = 0

The eigenvalues are λ1 = −1 and λ2 = −3. But

854 7 State Space Modelling and Analysis

Adjoint of φ(s)
[φ(s)]−1 =
Determinant of φ(s)

Since

φ(s) = [s I − A]−1
s I − A = [φ(s)]−1
 
s+4 1
Adjoint of φ(s) = Adjoint of
−3 s
 
s −1
=
3 s+4

Thus
 
s −1
[s I − A] =
3 s+4
   
s0 s −1
A= −
0s 3 s+4
 
0 1
A=
−3 −4

7.7 Controllability of Linear Continuous Time System

The controllability concept was introduced by R. Kalman of the USA in the year
1960. This concept is related to linear algebraic equations whose solution exists iff the
rank of the system A matrix is full. However, to achieve this there is no restriction in
the choice of the input. Therefore, the concept of controllability is defined as follows.
A linear time invariant continuous time system is said to be completely state
controllable at time t = t0 if it is possible to transfer the initial state x(t0 ) to any
final state x(t f ) by applying any control vector u(t) which has no constraints
in its choice in a finite time interval t0 ≤ t ≤ t f .

7.7.1 State Controllability Condition

For a given linear time invariant continuous time system, the controllability condition
is derived as follows.
Consider the vector matrix differential equation of the linear system given by (7.1)

ẋ(t) = Ax(t) + Bu(t)

7.7 Controllability of Linear Continuous Time System 855

y(t) = C x(t) (7.77)

where the symbols and notations have the same meaning and dimensions as discussed
before. The solution of Eq. (7.77) is given by Eq. (7.65) as
 t
x(t) = e x(0) +
At
e A(t−τ ) Bu(τ )dτ (7.78)
0

Without loss of generality, we can choose the origin of the state space as the final
state at t = t f . Equation (7.78) is written as
 t
0 = e At x(0) + e A(t−τ ) Bu(τ )dτ
0
 tf
−Aτ
x(0) = − e Bu(τ )dτ (7.79)
0

Using the Cayley–Hamilon theorem, e−Aτ can be written as


n−1
e−Aτ = αk (τ )Ak (7.80)
k=0

Substituting Eq. (7.80) in (7.79), we get


n−1  tf
x(0) = − k
A B αk (τ )u(τ )dτ (7.81)
k=0 0

Let
 tf
αk (τ )u(τ )dτ = βk (7.82)
0

Equation (7.82) can be estimated without putting any restriction in the choice of the
input u(τ ). Substituting Eq. (7.82) in (7.81), we get


n−1
x(0) = − Ak Bβk
k=0
⎡ ⎤
β0
⎢ β1 ⎥
⎢ ⎥
⎢ β2 ⎥
= −[B AB A B · · · A 2 n−1
B] ⎢ ⎥ (7.83)
⎢ .. ⎥
⎣ . ⎦
βn−1
856 7 State Space Modelling and Analysis

It is well known from linear algebra that in order to have a solution of Eq. (7.83), it
is necessary that the inverse of the following matrix exists:

Q = [B AB A2 B · · · An−1 B] (7.84)

Further, we know that the linear algebraic equations with n unknowns, Eq. (7.83)
has a unique solution iff the matrix Q has rank n. The matrix Q in Eq. (7.84) is called
the controllability matrix. With the controllability matrix having full rank, it gives
the sufficient condition for the system to be completely state controllable.

7.7.2 Output Controllability Condition

A linear time invariant continuous time system is said to be completely output con-
trollable at time t = t0 if it is possible to transfer the initial output y(t0 ) to any final
output y(t f ) by applying any control vector u(t) which has no constraints in its
choice in a finite time interval t0 ≤ t ≤ t f .
In Eq. (7.7), y(t) is the output vector which has dimension m × 1 where m is less
than n. Let the final output at t = t f be chosen as the origin of the state space. The
output y(t) can be obtained from Eq. (7.78) and can be written as
  tf 
y(t f ) = 0 = Ce At f x(0) + e−Aτ Bu(τ )dτ (7.85)
0

By following the procedure for obtaining the condition for complete state controlla-
bility, we can obtain the following condition for output controllability. For complete
output controllability, the rank of the composite matrix P whose dimension is m × nr
should have rank m. The matrix P is called the output matrix and is given by

P = [C B C AB C A2 B · · · C An−1 B] (7.86)

Example 7.23
Consider the following vector matrix differential equation:
⎡ ⎤ ⎡ ⎤
0 1 0 0
ẋ = ⎣0 −1 1 ⎦ x + ⎣ 0 ⎦ u
0 −1 10 10

Test whether the system is controllable.

7.7 Controllability of Linear Continuous Time System 857

Solution For the given system

⎡ ⎤ ⎡ ⎤
0 1 0 0
A = ⎣0 −1 1 ⎦ and B = ⎣ 0 ⎦
0 −1 10 10

⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 0 0 0
AB = ⎣0 −1 1 ⎦ ⎣ 0 ⎦ = ⎣ 10 ⎦
0 −1 10 10 100
⎡ ⎤⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 0 0 1 0 0 10
A2 B = ⎣0 −1 1 ⎦ ⎣0 −1 1 ⎦ ⎣ 0 ⎦ = ⎣ 90 ⎦
0 −1 10 0 −1 10 10 990

The controllability matrix is given by

⎡ ⎤
0 0 10
Q = [B AB A2 B] = ⎣ 0 10 90 ⎦
10 100 990

The determinant of the controllability matrix is

 
 0 0 10 
 
|Q| =  0 10 90 
10 100 990
     
10 90 0 90 0 10
= 0× −0× + 10
100 990 10 990 10 100
= 0 − 0 + (−1000) = −1000

The determinant does not vanish. The rank of the controllability matrix is full. Hence,
the system is completely state controllable. Further, just by observation of the con-
trollability matrix Q, the three columns are seen to be linearly independent and thus
conclude that the rank of Q is 3 and the system is completely state controllable.

Example 7.24
Consider the system described by the following state equation. Determine whether
the system is completely state controllable
⎡ ⎤ ⎡ ⎤
0 1 −3 0 −2
ẋ = ⎣ 2 −5 4 ⎦ x + ⎣1 −4⎦ u
−8 6 7 5 −3
858 7 State Space Modelling and Analysis

Solution
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 −3 0 −2 −14 5
AB = ⎣ 2 −5 4 ⎦ ⎣1 −4⎦ = ⎣ 15 4 ⎦
−8 6 7 5 −3 41 −29
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 −3 −14 5 −108 91
A2 B = ⎣ 2 −5 4 ⎦ ⎣ 15 4 ⎦ = ⎣ 61 −126⎦
−8 6 7 41 −29 265 −219

The controllability matrix is

⎡ ⎤
0 −2 −14 5 −108 91
Q = [B AB A2 B] = ⎣1 −4 15 4 61 −126⎦
5 −3 41 −29 265 −219

Considering any 3 × 3 square matrix, we may choose, for example,

⎡ ⎤
0 −2 −14
Q = ⎣1 −4 15 ⎦
5 −3 41

The three columns are linearly independent. In other words, one column cannot be
expressed as the linear combination of the other, and hence the rank of Q is 3 and
the system is completely state controllable. Thus, in this case it is not necessary to
calculate A2 B. Now the above fact is proved by determining the determinant of Q.
⎡ ⎤
0 −2 −14
Q = ⎣1 −4 15 ⎦ = −306
5 −3 41

The determinant Q does not vanish. Hence, the rank of the controllability matrix is
3. The system is completely state controllable.

Example 7.25
Consider the system described by the following state and output equations:
   
−2 0 1
ẋ = x+ u
0 −1 1
y = [1 1]x

Test whether the system is completely state controllable and completely output con-
trollable.
7.7 Controllability of Linear Continuous Time System 859

Solution Test for State Controllability

 
−2 0
A=
0 −1
 
1
B=
1
    
−2 0 1 −2
AB = =
0 −1 1 −1

The controllability matrix is

 
1 −2
Q = [B AB] =
1 −1

The columns of Q are linearly independent and hence the rank of Q is 2. The
system is completely state controllable. The determinant of Q is obtained as
|Q| = −1 + 2 = 1.
Test for Output Controllability
The output controllability matrix

P = [C B C AB]
 
1
C B = [1 1] =2
1
 
−2
C AB = [1 1] = −3
−1
P = [2 − 3]

Rank of P is 1. The system is output controllable.

Example 7.26
Consider the system described by the following state equation:
   
−1 0 0
ẋ = x+ u
0 −1 1

Test the state controllability of the system.

Solution
 
−1 0
A=
0 −1
860 7 State Space Modelling and Analysis
 
0
B=
1
    
−1 0 0 0
AB = =
0 −1 1 −1

The controllability matrix Q is

 
0 0
Q = [B AB] =
1 −1

The second column of Q can be obtained by multiplying the first column by a factor
−1. Hence, the matrix Q is linearly dependent. Hence, it is singular matrix and the
system is not completely state controllable. |Q| = 0.

7.7.3 Controllability and Observability Tests in the Frequency

Domain

In the previous sections, the controllability concept was introduced in the time domain
for the state space model. The controllability condition can be obtained in the fre-
quency domain as explained below.
Let H (s) be the transfer function of a single-input single-output system.

H (s) = C[s I − A]−1 B

H (s) is the ratio of two polynomials which correspond to the poles and zeros of
H (s). A very powerful and simple theorem has been developed that gives the single
condition for controllability. According to this theorem,
A single-input single-output system is said to be completely state (output)
controllable if no pole–zero cancellation occurs in the transfer function H (s).
Consider the following T.F:

(s + 2)
H (s) =
(s + 2)(s + 3)

By pole–zero cancellation, the above T.F. becomes

1
H (s) =
(s + 3)

In the above equation, one degree of freedom is lost. The original second order is
reduced to a first order system. Thus, it is not possible to take one of the two states to
the desired location by applying any input. The system now becomes not completely
7.7 Controllability of Linear Continuous Time System 861

controllable. For the example cited above, the rank of the controllability matrix is
obtained as one and not two.

7.8 Observability of Linear Continuous Time System

Consider the following vector matrix differential equation of an autonomous system:

ẋ = Ax
y = Cx (7.87)

where the symbols have the usual meaning and dimension.

A linear time invariant continuous time system is said to be completely
observable if every initial state x(t0 ) can be determined by measuring y(t)
in a finite interval of time t0 ≤ t ≤ t f .
The controllability gives the necessary and sufficient condition for designing a
feedback controller so that all the eigenvalues of the given system are placed in
the desired locations in the s-plane as one wishes. If any state is not available for
measurement, then one has to construct an observer by means of which the unmea-
surable states are reconstructed by measuring the known states. For the design of the
observer, the system should be observable. Similar to the controllability condition,
we establish below the observability condition.

7.8.1 Observability Condition

For a free system, the output as given by Eq. (7.65) is

y(t) = Ce At x(0)

Using the Cayley–Hamilton theorem, we get


n−1
e At = αk Ak (7.88)
k=0

Substituting Eq. (7.88) in (7.87), we get


n−1
y(t) = αk C Ak x(0)
k=0
= α0 C x(0) + α1 C Ax(0) + · · · + αn−1 C An−1 x(0) (7.89)
862 7 State Space Modelling and Analysis

By measuring the output y(t), in a finite time interval t0 ≤ t ≤ t f , x(0) can be

determined. Equation (7.89) can be written as
⎡ ⎤
C
⎢ ⎥ CA
y(t) ⎢ ⎥
n−1 k = ⎢ ⎥ x(0)
.. (7.90)
k=0 α
⎣ ⎦ .
n−1
CA

For the solution of Eq. (7.90), for x(0), the inverse of the following matrix should
exist:
⎡ ⎤
C
⎢ CA ⎥
⎢ ⎥
S=⎢ . ⎥ (7.91)
⎣ .. ⎦
C An−1

For the existence of inverse of S, the rank of the matrix should be n which can be
obtained using the theory of linear algebra. Therefore, the observability condition is
stated as follows.
For a continuous time system to be completely observable, the necessary and
sufficient condition is that the n × nm matrix

S = [C T A T C T · · · (A T )n−1 C T ] (7.92)

should have rank n which implies that the matrix S has n linearly independent column
vectors.

Example 7.27
Determine the controllability and observability of the following state model:
⎡ ⎤ ⎡ ⎤
1 1 0 0
ẋ = ⎣0 −2 1 ⎦ x + ⎣ 1 ⎦ u
0 0 −1 −2
y = [1 0 0]x

(Anna University, December, 2009)

Solution Controllability Test
⎡ ⎤
1 1 0
A = ⎣0 −2 1 ⎦
0 0 −1
7.8 Observability of Linear Continuous Time System 863
⎡ ⎤
0
B =⎣ 1 ⎦
−2
 
C = 100
⎡ ⎤⎡ ⎤ ⎡ ⎤
1 1 0 0 1
AB = ⎣0 −2 1 ⎦ ⎣ 1 ⎦ = ⎣−4⎦
0 0 −1 −2 2
⎡ ⎤⎡ ⎤ ⎡ ⎤
1 1 0 1 −3
A2 B = ⎣0 −2 1 ⎦ ⎣−4⎦ = ⎣ 10 ⎦
0 0 −1 2 −2

The controllability matrix Q is given by

⎡ ⎤
0 1 −3
Q = ⎣ 1 −4 10 ⎦
−2 2 −2

By observation of Q, no column of Q can be expressed as the linear combination of

the other columns. Hence, Q has full rank and hence the system is completely state
controllable. This can also be verified by determining the determinant of Q
 
 0 1 −3 
 
|Q| =  1 −4 10  = −1(−2 + 20) − 3(2 − 8) = 0
 −2 2 −2 

The determinant is zero and hence the rank of the controllability matrix is not full.
The system is not controllable.
Observability Test
⎡ ⎤
1
CT = ⎣0⎦
0
⎡ ⎤
1 0 0
AT = ⎣1 −2 0⎦
0 1 −1
⎡ ⎤⎡ ⎤ ⎡ ⎤
1 0 0 1 0 0 1 0 0
(A T )2 = ⎣1 −2 0 ⎦ ⎣1 −2 0 ⎦ = ⎣−1 4 0⎦
0 1 −1 0 1 −1 1 −3 1
⎡ ⎤⎡ ⎤ ⎡ ⎤
1 0 0 1 1
AT C T = ⎣1 −2 0 ⎦ ⎣0⎦ = ⎣1⎦
0 1 −1 0 0
864 7 State Space Modelling and Analysis
⎡ ⎤⎡ ⎤ ⎡ ⎤
1 0 0 1 1
(A T )2 C T = ⎣−1 4 0⎦ ⎣0⎦ = ⎣−1⎦
1 −3 1 0 1

The observability matrix is given by

⎡ ⎤
 T T T T 2 T 11 1
S = C A C (A ) C = ⎣0 1 −1⎦
00 1
|S| = 1

Observation of the observability matrix S shows that all the columns are linearly
independent. Further, the determinant of S is determined as |S| = 1. Hence, the rank
of S is full (3). The system is observable. The given system is not controllable but
observable.

Example 7.28
Determine the observability of the system given as follows:
⎡ ⎤ ⎡ ⎤
0 1 0 0
ẋ = ⎣0 0 1 ⎦ x + ⎣0⎦ u
0 −2 −3 1
 
y= 341 x

(Anna University, December, 2010)

Solution
⎡ ⎤
0 1 0
A = ⎣0 0 1 ⎦
0 −2 −3
⎡ ⎤
0
B = ⎣0⎦
1
 
C = 341
⎡ ⎤
3
CT = ⎣4⎦
1
⎡ ⎤
00 0
AT = ⎣1 0 −2⎦
0 1 −3
7.8 Observability of Linear Continuous Time System 865
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 0 0 00 0 0 0 0
(A T )2 = ⎣1 0 −2⎦ ⎣1 0 −2⎦ = ⎣0 −2 6⎦
0 1 −3 0 1 −3 1 −3 7
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 0 0 3 0
A T C T = ⎣1 0 −2⎦ ⎣4⎦ = ⎣1⎦
0 1 −3 1 1
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 0 0 3 0
(A T )2 C T = ⎣0 −2 6⎦ ⎣4⎦ = ⎣−2⎦
1 −3 7 1 −2

The observability matrix is given by

⎡ ⎤
 T T T T 2 T 30 0
S = C A C (A ) C = ⎣4 1 −2⎦
1 1 −2

By observing the controllability matrix S, it is seen that column 3 can be expressed

by multiplying column 2 by −2. Hence, the rank of the matrix S is not full and the
system is not observable. The determinant of S is determined as

|S| = 3(−2 + 2) = 0

7.9 State Equation of Discrete Time System

The discrete time system is described by the difference equation. A continuous time
system which is described by an n th order differential equation is converted into n first
order equations and the descriptions for n state variables are given. In an analogous
manner to continuous time system, the discrete time system which is described by
N th order difference equation is converted into N first order difference equations, and
the description for N state variables are given in the form of vector matrix difference
equation.
A linear time invariant discrete time system can be represented by the following
vector matrix difference equation:

x[k + 1] = Ax[k] + Bu[k]

y[k] = C x[k] + Du[k] (7.93)

where the symbols of Eq. (7.93) are defined as

x[k] = N × 1 (State Vector)

y[k] = p × 1 (Output Vector)
u[k] = r × 1 (Input Vector)
866 7 State Space Modelling and Analysis

u[k] x[k 1] Delay x[k] y[k]

B C
z 1

Fig. 7.19 Block diagram representation of state equation of discrete time system

A = N × N (State Matrix)
B = N × r (Input Matrix)
C = p × N (Output Matrix)
D = p × r (Direct Transmission Matrix)

The block diagram representation of Eq. (7.93) is shown in Fig. 7.19.

For example, consider the following second order difference equation:

y[k + 2] + a1 y[k + 1] + a2 y[k] = bu[k]

Let us choose

y[k] = x1 [k]
y[k + 1] = x1 [k + 1]
= x2 [k]
y[k + 2] = x2 [k + 1]

Substituting the above equations in the given second order difference equation, we
get

x2 [k + 1] + a1 x2 [k] + a2 x1 [k] = bu[k]

x2 [k + 1] = −a2 x1 [k] − a1 x2 [k] + bu[k]

From the above equations, the following vector matrix difference equation is
obtained:
   
0 1 0
x[k + 1] = x[k] + u[k]
−a2 −a1 b
y[k] = [1 0]x[k]
7.9 State Equation of Discrete Time System 867

7.9.1 Discrete Time State Equation in Controllable

Canonical Form

Let H (z) be the discrete time system transfer function which can be expressed as
follows:

Y [z] b0 z N + b1 z N −1 + · · · + b N −1 z + b N
= H [z] = N
U [z] z + a1 z N −1 + · · · + a N −1 z + a N
b0 + b1 z −1 + · · · + b N −1 z N −1 + b N z N
= (7.94)
1 + a1 z −1 + · · · + a N −1 z N −1 + a N z N

Let H [z] = H1 (z)H2 (z), where

H1 [z] = (b0 + b1 z −1 + · · · + b N −1 z N −1 + b N z N )
1
H2 [z] =
(1 + a1 z + · · · + a N −1 z N −1 + a N z N )
−1

Y [z] = H1 [z]W [z] (7.95)

where

W [z] = H2 [z]U [z] (7.96)

Equations (7.95) and (7.96) are represented in Fig. 7.20. From Fig. 7.20, the following
equations are written:

x1 [n + 1] = x2 [n]
x2 [n + 1] = x3 [n]
..
.
x N −1 [n + 1] = x N [n]
x N [n + 1] = −a N x1 [n] − a N −1 x2 [n] − . . . − a1 x N [n] + u[n] (7.97)
y[n] = b N x1 [n] + b N −1 x2 [n] + . . . + b1 x N [n] + b0 u[n]

Substituting for x N +1 [n] from Eq. (7.97), we get

y[n] = (b N − b0 a N )x1 [n] + (b N −1 − b0 a N −1 )x2 [n] + · · ·

+(b1 − b0 a1 )x N [n] + b0 u[n]
= b̄ N x1 [n] + b̄ N −1 x2 [n] + · · · + b̄1 x N + b0 u[n] (7.98)
868 7 State Space Modelling and Analysis

W[z]
u[n] xN(n 1) y[n]
b0

z 1

xN[n]
a1 b1

z 1

xN 1
a2 b2

z 1

xm 1[n]
am x2(n 1) bm

z 1

x2[n]
aN 1 b
x1(n 1) N 1

z 1

aN x1[n] bN

Fig. 7.20 Direct form II realization of N th order discrete time system

where b̄i = (bi − b0 ai ). Equation (7.97) represents the state equations and Eq. (7.98)
represents the output equation. The above equations are written in matrix form as
⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
x1 [n + 1] 0 1 0 0 0 x1 [n] 0
⎢ x2 [n + 1] ⎥ ⎢ 0 0 1 0 ⎥ ⎢
0 ⎥ ⎢ x2 [n] ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎢0⎥
⎢ x [n + 1] ⎥ ⎢ 0 0 ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 3 ⎥ ⎢ 0 0 1 ⎥ ⎢ x3 [n] ⎥ ⎢ 0 ⎥
⎢ . ⎥=⎢ . . . . . ⎥ ⎢ . ⎥+⎢ . ⎥
⎢ . ⎥ ⎢ ⎥ ⎢ ⎥ ⎢.⎥
⎢ . ⎥ ⎢ .. .
.
.
.
.
.
. ⎥⎢
. .
. ⎥ ⎢.⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣ x N −1 [n] ⎦ ⎣ 0 0 0 0 1 ⎦ ⎣ x N −1 [n] ⎦ ⎣ 0 ⎦
x N [n] −a N −a N −1 −a N −2 −a2 −a1 x N [n] 1
       
7.9 State Equation of Discrete Time System 869

x[n + 1] A x[n] B (7.99)

y[n] = [b̄ N b̄ N −1 . . . b̄1 ] + b0 u[n]
  
C D (7.100)

The general form of state and output equations is therefore written as

x[n + 1] = Ax[n] + Bu[n]

y[n] = C x[n] + Du[n] (7.101)

The state equations for a discrete system can be obtained by several methods.
However, we represent here by controllable canonical form, observable canonical
form and diagonal form. The following analogy between continuous and discrete
time systems is to be noted:
1. In a continuous time system, the output of each integrator is identified as a state.
In a discrete time system, the output of each delay element is identified as the
state.
2. In a continuous time system, the input of each integrator is identified as a first
derivative from which the first order differential equation is formed. In a discrete
time system, the input to each delayed element is identified to form the first order
difference equation.
The following examples illustrate the method of forming state equations.

Example 7.29
Form the state equations of canonical form for the following T.F. of a discrete time
system:

5z 4 + 7z 3 + 8z 2 + 2z + 10
H [z] =
z 4 + 6z 3 + 7z 2 + 4z + 9

Solution

5z 4 + 7z 3 + 8z 2 + 2z + 10 5 + 7z −1 + 8z −2 + 2z −3 + 10z −4
H [z] = =
z 4 + 6z 3 + 7z 2 + 4z + 9 1 + 6z −1 + 7z −2 + 4z −3 + 9z −4

where

b0 = 5; b1 = 7; b2 = 8; b3 = 2; b4 = 10;
a1 = 6; a2 = 7; a3 = 4; a4 = 9
870 7 State Space Modelling and Analysis
⎡ ⎤ ⎡ ⎤
0 1 0 0 0
⎢ 0 0 1 0 ⎥ ⎢0⎥
x[n + 1] = ⎢
⎣ 0
⎥ x[n] + ⎢ ⎥ u[n]
0 0 1 ⎦ ⎣0⎦
−9 −4 −7 −6 1
b̄4 = b4 − b0 a4 = 10 − 5 × 9 = −35
b̄3 = b3 − b0 a3 = 2 − 5 × 4 = −18
b̄2 = b2 − b0 a2 = 8 − 5 × 7 = −27
b̄1 = b1 − b0 a1 = 7 − 5 × 6 = −23
y[n] = [b̄4 b̄3 b̄2 b̄1 ]u[n] + b0 u[n]

y[n] = [−35 − 18 − 27 − 23]x[n] + 5u[n]

Example 7.30
Consider the following difference equation:

4y[n − 3] + 6y[n − 2] − 5y[n − 1] + y[n] = u[n] + 5x[n − 1]

Form state equations.

Solution

Y [z] (1 + 5z −1 )
= H [z] =
U [z] (1 − 5z −1 + 6z −2 + 4z −3 )

where b0 = 1; b1 = 5; b2 = 0; b3 = 0; a1 = −5; a2 = 6; a3 = 4
⎡ ⎤ ⎡ ⎤
0 1 0 0
x[n + 1] = ⎣ 0 0 1 ⎦ x[n] + ⎣ 0 ⎦ u[n]
−4 −6 5 1
b̄3 = b3 − b0 a3 = −4
b̄2 = b2 − b0 a2 = −6
b̄1 = b1 − b0 a1 = 5 + 5 = 10

y[n] = [−4 − 6 10]x[n] + u[n]

7.9.2 Observable Canonical Form Model

Consider the following system function:

7.9 State Equation of Discrete Time System 871

Y [z] b0 + b1 z −1 + b2 z −2
= H [z] =
U [z] 1 + a1 z −1 + a2 z −2

Cross-multiplying the above equation, we get

[1 + a1 z −1 + a2 z −2 ]Y [z] = [b0 + b1 z −1 + b2 z −2 ]U [z]

Y [z] = −a1 z −1 Y [z] − a2 z −2 Y [z] + b0 U [z]
+b1 z −1 U [z] + b2 z −2 U [z] (7.102)

Equation (7.102) is represented in Fig. 7.21, where the states and the delay elements
are shown.
From Fig. 7.21, the following equations are written in terms of states:

y[n] = x1 [n] + b0 u[n]

x1 [n + 1] = −a1 y[n] + x2 y[n] + b1 u[n]
= −a1 x1 [n] + x2 [n] + (b1 − b0 a1 )u[n] (7.103)
x2 [n + 1] = −a2 y[n] + b2 u[n]
= −a2 x1 [n] + (b2 − b0 a1 )u[n] (7.104)

Equations (7.103) and (7.104) can be combined and expressed in matrix form as
   
−a1 1 (b1 − b0 a1 )
x[n + 1] = x[n] + u[n]
−a2 0 (b2 − b0 a2 )
y[n] = [1 0]x[n] + b0 u[n]

u[n] y[n]
U(z) b0 x1[n] Y(z)

z 1

x1[n 1]

b1 a1
x2[n]

z 1

x2[n 1]

b2 a2

Fig. 7.21 Canonical form structure

872 7 State Space Modelling and Analysis

In general, for an N th order difference equation,

⎡ ⎤ ⎡ ⎤
−a1 1 0 0 · · · 0 (b1 − b0 a1 )
⎢ −a2 0 1 0 · · · 0 ⎥ ⎢ (b2 − b0 a2 ) ⎥
⎢ ⎥ ⎢ ⎥
⎢ −a3 0 0 1 · · · 0 ⎥ ⎢ (b3 − b0 a3 ) ⎥
⎢ ⎥ ⎢ ⎥
x[n + 1] = ⎢ . .. .. .. .. .. ⎥ x[n] + ⎢ .. ⎥ u[n](7.105)
⎢ .. . . . . .⎥ ⎥ ⎢ . ⎥
⎢ ⎢ ⎥
⎣ −a N −1 0 0 0 · · · 1 ⎦ ⎣ (b N −1 − b0 a N −1 ) ⎦
−a N 0 0 0 · · · 0 (b N − b0 a N )

where b̄i is defined as b̄i = (bi − b0 ai )

y[n] = [1 0 0 · · · 0]x[n] + b0 u[n] (7.106)

Example 7.31
Form the state equations of observable canonical form for the following T.F. of a
discrete time system:

5z 4 + 7z 3 + 8z 2 + 2z + 10
H [z] =
z 4 + 6z 3 + 7z 2 + 4z + 9

Solution

5z 4 + 7z 3 + 8z 2 + 2z + 10
H [z] =
z 4 + 6z 3 + 7z 2 + 4z + 9
5 + 7z −1 + 8z −2 + 2z −3 + 10z −4
=
1 + 6z −1 + 7z −2 + 4z −3 + 9z −4

where b0 = 5; b1 = 7; b2 = 8; b3 = 2; b4 = 10; a1 = 6; a2 = 7; a3 = 4; a4 = 9

b̄1 = b1 − b0 a1 = 7 − 5 × 6 = −23
b̄2 = b2 − b0 a2 = 8 − 5 × 7 = −27
b̄3 = b3 − b0 a3 = 2 − 5 × 4 = −18
b̄4 = b4 − b0 a4 = 10 − 5 × 9 = −35
⎡ ⎤ ⎡ ⎤
−6 1 0 0 −23
⎢ −7 0 1 0 ⎥ ⎢ −27 ⎥
x[n + 1] = ⎢ ⎥ ⎢ ⎥
⎣ −4 0 0 1 ⎦ x[n] + ⎣ −18 ⎦ u[n]
−9 0 0 0 −35

y[n] = [1 0 0 0]x[n] + 5u[n]

7.9 State Equation of Discrete Time System 873

b0

x1[n 1]

z 1
x[n]

1 A1
x1[n]

y[n]
x2[n 1]

z 1

2 A2
x2[n]

xN[n 1]

z 1

AN
N xN[n]

Fig. 7.22 Diagonal form model

7.9.3 Diagonal Form (Parallel Form) Model

Consider the system transfer function given as follows:

b0 z N + b1 z N +1 + · · · + b N −1 z + b N
H [z] = (7.107)
z N + a1 z N +1 + · · · + a N −1 z + a N
b0 + b1 z −1 + · · · + b N −1 z −(N −1) + b N z −N
=
1 + a1 z −1 + · · · + a N −1 z −(N −1) + a N z −N
A1 AN
= b0 + + ··· + (7.108)
(z − λ1 ) (z − λ N )

where λ1 , λ2 , . . . , λn are the distinct eigenvalues of H [z]. Equation (7.108) is rep-

resented in Fig. 7.22.
From Fig. 7.22, the following state equations are written:

x1 [n + 1] = λ1 x1 [n] + u[n]
874 7 State Space Modelling and Analysis

x2 [n + 1] = λ2 x2 [n] + u[n]
..
.
x N [n + 1] = λ N x N [n] + u[n] (7.109)

y[n] = A1 x1 [n] + A2 x2 [n] + · · · + A N x N [n] + b0 u[n] (7.110)

Equations (7.109) and (7.110) can be represented in matrix form as

⎡ ⎤ ⎡ ⎤
λ1 0 ··· 0
0 1
⎢0 λ2 ··· 0 ⎥
0 ⎢1⎥
⎢ ⎥ ⎢ ⎥
⎢ ··· 0 ⎥
λ3 ⎢ ⎥
x[n + 1] = ⎢ 0 0 ⎥ x[n] + ⎢ 1 ⎥ u[n] (7.111)
⎢ .. .. .. .. ⎥ ⎢ .. ⎥
⎣ . .··· . ⎦
. ⎣.⎦
0 0 0 · · · λN 1

y[n] = [A1 A2 A3 · · · A N ]x[n] + b0 u[n] (7.112)

Example 7.32
A certain discrete time system has the following T.F.:

z 3 + 10z 2 + 32z + 29
H [z] =
z 3 + 9z 2 + 26z + 24

Form the state equation with its A matrix in diagonal form.

Solution Dividing the numerator polynomial with denominator polynomial, we get

1

z + 9z + 26z + 24 z 3 + 10z 2 + 32z + 29
3 2

z 3 + 9z 2 + 26z + 24
z 2 + 6z + 5

(z 2 + 6z + 5)
H [z] = 1 +
(z 3 + 9z 2 + 26z + 24)
(z + 6z + 5) = (z + 1)(z + 5)
2

(z + 9z + 26z + 24) = (z + 2)(z + 3)(z + 4)

3 2
7.9 State Equation of Discrete Time System 875

Im

A3 A2 A1
Re
5 4 3 2 1 0

Fig. 7.23 Pole–zero diagram

(z + 1)(z + 5)
H [z] = 1 +
(z + 2)(z + 3)(z + 4)
(z + 1)(z + 5) A1 A2 A3
= + +
(z + 2)(z + 3)(z + 4) (z + 2) (z + 3) (z + 4)

The pole–zero locations are shown in Fig. 7.23.

−3 × 1 3
A1 = =−
2×1 2
2×2
A2 = =4
1×1
−3 × 1 3
A3 = =−
2×1 2

3 1 4 3 1
H [z] = 1 − + −
2 (z + 2) (z + 3) 2 (z + 4)

The eigenvalues are λ1 = −2, λ2 = −3 and λ3 = −4

⎡ ⎤ ⎡ ⎤
−2 0 0 1
x[n + 1] = ⎣ 0 −3 0 ⎦ x[n] + ⎣ 1 ⎦ u[n]
0 0 −4 1

 
3 3
y[n] = − 4 − x[n] + u[n]
2 2
876 7 State Space Modelling and Analysis

Example 7.33
Find the state variable matrices A, B, C and D for the equation

y[n] − 3y[n − 1] − 2y[n − 2] = u[n] + 5x[n − 1] + 6x[n − 2]

(Anna University, November, 2007)

Solution Taking z-transform on both sides of the equation, we get

Y [z] (1 + 5z −1 + 6z −2 )
= H [z] =
U [z] (1 − 3z −1 − 2z −2 )

where b0 = 1; b1 = 5; b2 = 6; a1 = −3; a2 = −2

b̄2 = b2 − b0 a2 = 6 + 2 = 8
b̄1 = b1 − b0 a1 = 5 + 3 = 8
D = b0 = 1

   
01 0
x[n + 1] = x[n] + u[n]
23 1
  
A B
y[n] = [8 8] x[n] + 
1 u[n]
 
C D

Example 7.34
A continuous time system has the state variable description given as follows.
   
2 −1 1
A= ; B= ; C = [3 1]; D = [2]
1 0 0

Determine the transfer function.

Solution
 
(z − 2) 1
(z I − A) =
−1 z
Adjoint
(z I − A)−1 =
Determinant  
1 z 1
= 2
(z − 2z + 1) 1 (2 − z)
7.9 State Equation of Discrete Time System 877
  
1 z 1 1
(z I − A)−1 B =
(z 2 − 2z + 1) 1 (2 − z) 0
 
1 z
=
(z 2 − 2z + 1) 1
 
1 z
C(z I − A)−1 B = [3 1]
(z 2 − 2z + 1) 1
(3z + 1)
=
(z − 2z + 1)
2

(3z + 1)
C(z I − A)−1 B + D = +2
(z − 2z + 1)
2

(2z 2 − z + 3)
H (z) =
(z 2 − 2z + 1)

7.9.4 Solution of State Equation

The solution of state Eq. (7.93) of discrete time system can be obtained by the
following methods:
1. By the classical method.
2. The z-transform method.

7.9.4.1 Solution of State Equation by Classical Method—The Recursive

Method

Consider the homogeneous equation of (7.93) which is written as

x[k + 1] = Ax[k]
k = 0, 1, 2, . . . (7.113)

The solution of the above equation may be obtained as

x[k] = Aφ[k]

where φ[k] is called the fundamental or state transition matrix which is unique for
the given system and satisfies the equation

[k + 1] = Aφ[k] (7.114)

878 7 State Space Modelling and Analysis

where

[0] = I
φ[k] = n × n matrix

The solution of (7.113) may be written as

x[k] = φ[k]x[0] (7.115)

For k = 0
[1] = A

For k = 1

[2] = A[1]
= A2

In general

φ[k] = Ak for k = 0, 1, 2, . . .

x[k] = Ak x[0] (7.116)

For a non-homogenous Eq. (7.113), the solution is obtained as


k
x[k] = Ak x[0] + Ak− j Bu( j − 1) (7.117)
j=1

7.9.4.2 Solution by using z-Transform Method

x[k + 1] = Ax[k] + Bu[k] (7.118)

Taking z-transform on both sides of the above equation, we get

Z x[z] − zx[0] = Ax[z] + Bu[z]

[z I − A]X [z] = zx[0] + BU [z]

X [z] = [z I − A]−1 zx[0] + [z I − A]−1 BU [z] (7.119)

7.9 State Equation of Discrete Time System 879

Taking inverse z-transform for Eq. (7.119), x[k] is obtained.

Example 7.35
Consider the following vector matrix difference equation:
   
−3 1 1
x[k + 1] = x[k] + u[k]
−2 0 0
 
1
x[0] =
−1

The input is a unit step sequence. Derive an expression for x[k].

Solution
 
−3 1
A=
−2 0
 
1
B=
0
 
1
x[0] =
−1

From Eq. (7.118), we get

X (z) = [z I − A]−1 zx[0] + [z I − A]−1 BU [z]

Let

X 0 [z] = [z I − A]−1 zx[0]

X f [z] = [z I − A]−1 BU (z)

For unit step sequence

z
U (z) =
z−1

 
(z + 3) −1
[z I − A] =
2 z
 
−1 z 1 1
[z I − A] =
−2 z + 3 (z + 1)(z + 2)
880 7 State Space Modelling and Analysis
  
z z 1 1
X 0 (z) =
(z + 1)(z + 2) −2 z + 3 −1
 
X 0 [z] 1 (z − 1)
=
z (z + 1)(z + 2) −(z + 5)
 
3
(z+2)
− (z+1)
2
= 3
(z+2)
− (z+1)
4

 
3z
− 2z
X 0 [z] = (z+2) (z+1)
3z
(z+2)
− (z+1)
4z

Taking the inverse z-transform, we get

 
3(−2)k − 2(−1)k
x0 [k] =
3(−2)k − 4(−1)k
  
−1 1 z 1 1
[z I − A] B =
(z + 1)(z + 2) −2 z + 3 0
 
1 z
=
(z + 1)(z + 2) −2
X f [z] = [z I − A]−1 BU [z]
 
z z
=
(z − 1)(z + 1)(z + 2) −2
 
X f [z] 1 z
=
z (z − 1)(z + 1)(z + 2) −2
 
1
− 23 (z+2)1
+ 6(z−1)
1
= 2(z+1)
1
(z+1)
− 3(z+2)
2
− 3(z−1)
1

 
1 z
− 23 (z+2)
z
+ 6(z−1)
z
X f [z] = 2 z+1
z
(z+1)
− 23 z+2z
− z(z−1)
z

Taking the inverse z-transform, we get

1 
(−1)k − 23 (−2)k + 16
x f [k] = 2
(−1)k − 23 (−2)k + 13

The total response is given by

x[k] = x0 [k] + x f [k]

  1 
3(−2)k − 2(−1)k (−1)k − 23 (−2)k + 16
= + 2
3(−2)k − 4(−1)k (−1)k − 23 (−2)k − 13
7 
(−2)k − 23 (−1)k + 16
x[k] = 27
2
(−2)k − 3(−1)k − 13
7.9 State Equation of Discrete Time System 881

7 3 1
x1 [k] = (−2)k − (−1)k +
2 2 6
7 1
x2 [k] = (−2)k − 3(−1)k −
2 3

7.10 Controllability and Observability of Discrete Time

System

In order to understand what is going on inside the system just by observation, the
given dynamic system should be observable. The dual concepts of controllability
and observability are related to linear systems of algebraic equations. The necessary
and sufficient conditions for the existence of these two concepts are derived below.

7.10.1 Controllability Condition for Discrete Time System

Consider the vector matrix difference equation which is given as follows:

x[k + 1] = Ax[k] + BU [k] (7.120)

The linear discrete time system Eq. (7.120) is said to be absolutely state controllable if
it is possible to transfer any initial state x[0] = x0 to any final state x f in a finite time.
This implies whether it is possible to find a control sequence u[0], u[1] · · · u[n − 1]
such that x[k] = x f . Equation (7.120) for various input sequences is given as

x[1] = Ax[0] + Bu[0]

x[2] = Ax[1] + Bu[1] = A2 x[0] + ABu[0] + Bu[1]
..
.
x[n] = An x[0] + An−1 Bu[0] + · · · + Bu[n − 1] (7.121)
882 7 State Space Modelling and Analysis
⎡ ⎤
u[n − 1]
⎢u[n − 2]⎥
  ⎢ ⎥
⎢ .. ⎥
x[n] − An x[0] = B AB A2 B · · · An−1 B ⎢ . ⎥ (7.122)
⎢ ⎥
⎣ u[1] ⎦
u[0]
⎡ ⎤
u[n − 1]
⎢u[n − 2]⎥
⎢ ⎥
⎢ .. ⎥
⎢ . ⎥ = Q −1 [x[n] − An x[0]] (7.123)
⎢ ⎥
⎣ u[1] ⎦
u[0]

where Q = [B AB A2 B · · · An−1 B] is a n × n square matrix. For any x[n] = x f ,

the input sequence transfers the initial state x0 to any final state x f in n steps. This
requires the existence of Q which is called the controllability matrix. In other
words, for the discrete system to be completely state controllable, the controllability
matrix Q should have full rank.

7.10.2 Observability Condition for Discrete Time System

Consider the following state and output equations of a linear time invariant discrete
time autonomous system:

x[k + 1] = Ax[k]
y[k] = C x[k] (7.124)

where the symbols have the usual meaning and the matrices A and C have appropriate
dimensions. The discrete time system is said to be observable if it is possible to
establish the unknown initial state x[0] by measuring the output y[k]. Since x[0] is
an n dimensional state vector, by taking n measurements it is possible to establish
x[0]. For k = 0, 1, 2, . . . , (n − 1), the output sequence is expressed as

y[0] = C x[0]
y[1] = C x[1] = C Ax[0]
y[2] = C x[2] = C A2 x[0]
..
.
y[n − 1] = C x[n − 1] = C An−1 x[0] (7.125)
7.10 Controllability and Observability of Discrete Time System 883
⎡ ⎤ ⎡ ⎤
y[0] C
⎢ y[1] ⎥ ⎢ CA⎥
⎢ ⎥ ⎢ ⎥
⎢ y[2] ⎥ ⎢ ⎥
C A2
⎢ ⎥= ⎢ ⎥ x[0]
⎢ .. ⎥ ⎢ ⎥
..
⎣ . ⎦ ⎣ ⎦.
y[n − 1] C An−1
= Sx[0] (7.126)

where
⎡ ⎤
C
⎢ ⎥
CA
⎢ ⎥
⎢ ⎥
C A2
S=⎢ ⎥ (7.127)
⎢ ⎥ ..
⎣ ⎦ .
n−1
CA
x[0] = S −1 y[k] (7.128)

S is called the observability matrix. From Eq. (7.128), x[0] can be obtained from
y[k] if the inverse of the observability matrix S is non-singular. In other words,

S = [C T A T C T (A T )2 C T · · · (A T )n−1 C T ] (7.129)

should have rank n.

Example 7.36
Consider the discrete time system described by the following state equation:
   
0 6 3 6
x[k + 1] = x+ u
−1 −5 −1 −2

Test whether the system is state controllable.

Solution
 
0 6
A=
−1 −5
 
3 6
B=
−1 −2
    
0 6 3 6 −6 −12
AB = =
−1 −5 −1 −2 2 4

The controllability matrix Q is

884 7 State Space Modelling and Analysis
 
Q = B AB
 
3 6 −6 −12
=
−1 −2 2 4

By observing, it is seen column two is obtained by multiplying column one by a

factor 2. Similarly, column three is obtained by multiplying column two by −1 and
column four is obtained by multiplying column three by 2. Thus, the column vectors
are linearly dependent. No square matrix can be formed which will give full rank
which is 2. Hence, the given system is not completely state controllable.

Example 7.37
Consider the system described by the following vector matrix difference equation:
⎡ ⎤ ⎡ ⎤
0 1 0 1
x[k + 1] = ⎣ 3 0 2 ⎦ x[k] + ⎣4⎦ u[k]
−12 −7 −6 6

Test whether the given system is completely state controllable.

Solution
⎡ ⎤
0 1 0
A =⎣ 3 0 2⎦
−12 −7 −6
⎡ ⎤
1
B = ⎣4 ⎦
6
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 0 1 4
AB =⎣ 3 0 2 ⎦ ⎣4⎦ = ⎣ 15 ⎦
−12 −7 −6 6 −76
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 1 0 4 15
A2 B =⎣ 3 0 2 ⎦ ⎣ 15 ⎦ = ⎣−140⎦
−12 −7 −6 −76 303

The controllability matrix is given by

⎡ ⎤
 1 4
 15
Q = B AB A2 B = ⎣4 15 −140⎦
6 −76 303

By observation none of the column vectors of Q can be expressed as a linear combina-

tion of other column vectors. The column vectors are linearly independent and hence
the matrix Q has full rank. Therefore, the system is completely state controllable.
7.10 Controllability and Observability of Discrete Time System 885

The rank of the controllability matrix is determined as

|Q| = 1(15 × 303 − 140 × 76) − 4(4 × 303 + 6 × 140)

+15(4 × (−76) − 6 × 15)
= −6095 − 8208 − 5910 = −20213

The determinant of Q does not vanish. Hence, the rank of Q is 3. The system is
completely state controllable.

Example 7.38
Verify whether the system described by the following state equation of discrete time
system is observable:
   
−5 4 1
x[k + 1] = x[k] + u[k]
−6 5 1
y[k] = [−2 3]

Solution
 
−2
C =T
3
 
−5 −6
A =
T
4 5
    
−5 −6 −2 −8
A C =
T T
=
4 5 3 7

The observability matrix

 
  −2 −8
S = C T AT C T =
3 7
|S| = −14 + 24 = 10

The determinant of S is not zero. Hence, the determinant of the observability matrix
S exists and the system is observable.
886 7 State Space Modelling and Analysis

7.11 Sampled Data System

7.11.1 Introduction

Sampled data system has now become very popular with the introduction of inex-
pensive microprocessors and mini-computers. Sampled data systems or otherwise
called digital control systems arise whenever the measurements necessary for control
are obtained in an intermittent form. Also when the complexity of the system to be
controlled increases and for large scale systems, it is now customary to intentionally
introduce a digital computer in the system. The digital computer sends out a control
signal to each plant only periodically. In an analog system, the plant or the process to
be controlled is described by means of differential equations. On the other hand, in
the discrete time system (Digital control system) the signals vary at discrete intervals
of time. Such systems are governed by difference equations. In such systems, the
components connected respond only to discrete signals and the variables are discrete
in nature. If both analog and digital components are connected in a system, then the
system is called a sampled data system. Here, the variables appear in discrete form
at one point and in analog form at some other point.

7.11.2 Advantages of Sampled Data Control Systems

1. By changing the software, it is easy to modify the control law in digital control
systems whereas it is not so in analog systems.
2. Sampled data control systems use control elements which require low energy
signals. Thus, large power can be controlled with low power signals.
3. The drift associated with analog components is more and hence the accuracy is
less. The drift is minimum in digital control systems and hence the accuracy is
more here.
4. The use of differential controllers in both analog and digital control systems has
become popular now. The use of a differential controller in continuous systems not
only enhances the noise but also produces additional noise. Further, in a sampled
data control system, there is no need for amplification of the analog signal, and
this reduces the noise in the system.
5. In sampled data control systems, the time intervals between successive signals
are used to control many other separate plants and thus the overall cost is reduced.
6. In a sampled data control system, the use of digital computers helps in data
acquisition and monitoring of all the variables in the system. The digital computer
ensures the safety of the plant. The startup and shutdown procedures are made
simple in computer controlled systems.
7. It is easy to implement optimal and adaptive control in sampled data time system.
7.11 Sampled Data System 887

7.11.3 Disadvantages

1. The analysis and design of a sampled data control system are more complex and
tedious as compared to continuous control systems.
2. The use of A/D, D/A converters and digital computer in sampled data time system
introduces a delay in sending the signal, and thus it is difficult to achieve the
performance objectives as predicted by theoretical calculation which assume no
delay.
3. Generally, the sampled data system is less stable compared to the continuous
system.
4. In sampled data control system, while reconstructing the continuous signal from
a discrete signal, there is a loss of signal information.
5. In sampled data system, it is very difficult to get information about the system
behaviour in between sampling intervals.

7.11.4 A Sampled Data Closed Loop Control System

The block diagram of a typical sampled data closed loop control system is shown in
Fig. 7.24. The input r (t), output c(t) and error e(t) are continuous signals. The error
signal e(t) is converted into a sequence of numbers m(kT ) by means of an analog
to digital (A/D) converter. For this, it is necessary to sample the analog signal e(t)
to e(kT ) at instants kT where k = 0, 1, 2, . . ., and T is the time interval between
any sampling (T is called the sampling period). The A/D converter samples e(t)
into e(kT ) and sends a sequence of numbers m(kT ) suitable for use by the digital
computer. Thus, the A/D converter first converts the continuous signal e(t) into
e(kT ). This process is called digitization. The discrete signal e(kT ) is converted into
m(kT ). This process is known as quantization. The output of the digital computer
appears as u(kT ). Since the plant requires continuous signal u(t), it is necessary
to convert the discrete signal u(kT ) to u(t). This is done by means of a digital to
analog (D/A) converter which is followed by a data extrapolator or otherwise called
a hold circuit. In the computer hardware, there is a clock which sends a pulse every T
seconds to A/D and D/A converters. The converter sends a number to the computer.
In a sampled data system, the A/D converter, digital computer, D/A converter and
the clock are all represented by a sampling switch followed by a hold circuit.

7.11.5 Sampling Process

The process of converting a continuous signal into a discrete signal is known as

sampling. This process may occur in one or more places in the system. In a block
diagram, it is represented by a switch.
888 7 State Space Modelling and Analysis

m(kT) u(kT) u(t)

r(t) e(t) Digital c(t)
A/D D/A Plant
Converter Computer Converter

Clock

Fig. 7.24 A sampled data closed loop control system

The input signal x(t) is a continuous signal. The sampler output x ∗ (t) is a train
of pulses. The continuous signal x(t) and the output of the sampler x ∗ (t) may be
related by
x ∗ (t) = δT (t)x(t) (7.130)

where δT (t) represents a train of unit impulses. The sampler output is equal to the
product of continuous signal x(t) and the train of unit impulses. In other words,
sampler may be considered as a modulator with the input x(t) as the modulating
signal and the train of unit impulses as the carrier. The output x ∗ (t) is the modulated
signal. The sampling process is shown in Fig. 7.25. Now the train of impulses may
be represented by the following equation:
∞

δT (t) = δT (t − kT ) (7.131)
−∞

where δT (t − kT ) is the unit impulse function occurring at t = kT where k =

0, 1, 2, 3, . . . and T is the sampling period. If the continuous signal is sampled in a
periodic manner, the sampled signal may be represented by
∞

x ∗ (t) = x(t)δT (t − kT ) (7.132)
−∞
∞

= x(kT )δT (t − kT ) (7.133)
−∞

7.11.6 Sampled Data System Variables

The different variables appearing in a sampled data system can be described in terms
of time and amplitude. They are grouped as continuous amplitude and continuous (C-
C), continuous amplitude and discrete time (C-D), discrete amplitude and continuous
time (D-C) and discrete amplitude and discrete time (D-D). The commonly sampled
7.11 Sampled Data System 889

(a)
(t)

x(t) x*(t) g(t) x(t) x*(t) g(t)

Modulator
T
Sampler
(b) (c)
x(t)

(t)

t T 2T 3T 4T t

(d)

x*(t)

T 2T 3T 4T 5T t

Fig. 7.25 Representation of the sampling process

data system is represented in the block diagram form in Fig. 7.26a, and the signals
at different points are presented in Fig. 7.26b. The continuous amplitude, continuous
time signal (C-C) (Fig. 7.26b) is applied as the input to the sampler as e(t), and the
output of the sampler appears as e∗ (t) which has continuous amplitude but in discrete
time as shown in Fig. 7.26c. This signal appears as the input to a hold circuit. The
output of the hold circuit has continuous amplitude and continuous time (C-C) as
shown in Fig. 7.26d. It is to be noted here that e∗ (t) can be an infinite number of
values within some predetermined range of amplitude.
Figure 7.27a represents the block of an open loop sampled data control system.
The continuous signal e(t) (C-C) is sampled as e∗ (t) in continuous amplitude and
discrete time (C-D). The A/D converter quantizes e∗ (t) into e(kT ) as a (D-D) signal
(Fig. 7.27d) and is applied to the digital computer. It is to be noted here that e∗ (t) is
quantized such that it can be only one of a finite number of values within a specific
value range. The digital computer itself manipulates the quantized value (a base 2
number) into another discrete amplitude–discrete time signal f (kT ) (Fig. 7.27e).
890 7 State Space Modelling and Analysis

(a)

e(t) x*(t) m(t)

HOLD
C-C T C-D C-C

(b) (c)
e(t) C-C
C-D
e*(t)

0 T 2T t 0 T 2T t
(d)
C-C

m(t)

0 T 2T t

Fig. 7.26 Input–output of a hold circuit

Finally, the D/A converter or the hold device transforms f (kT ) into m(t) which has
continuous amplitude in continuous time (Fig. 7.27f).

7.11.7 Hold Devices

The essential element of a discrete system is the sampler. It admits the continuous
signal T sec. After being quantized, it is applied to the digital computer. The output
of the digital computer is a sampled signal. The sampled signal has to be converted
into a continuous signal before it is applied to the controller. This is done by a data
extrapolator or a data hold circuit. This device reconstructs a sampled signal into a
continuous time signal based on the knowledge of the past samples. It is usually a
part of the D/A converter. The hold devices are classified according to the number of
prior samples utilized for predicting the sampled function during waiting intervals.
It is evident from Fig. 7.28 that the value of the reconstructed function using a
zero order hold circuit during any waiting period is equal to the value of the sampled
7.11 Sampled Data System 891

(a)

e(t) e*(L) eD(kT) Digital f(kT) D/A (sample m(t)

A/D and HOLD)
C-C T C-D D-D Controller D-D D-C

(b) (c)
e(t) C-C
C-D
e*(t)

0 T 2T t 0 T 2T t

(d) (e)

c1
D-D D-D
eD(kT) p3 f(kT) c2
p2 c3
p1

0 T 2T t 0 T 2T t

(f)
D-C
m(t)

0 T 2T t

Fig. 7.27 Representation of open loop sampled data system

function at the beginning of the interval. This leads to the staircase approximation
of a continuous time function.
892 7 State Space Modelling and Analysis

f0(t)

0 T t

Fig. 7.28 Characteristic of zero prder hold device

(a) (b)

f(t) f*(t)

0 t 0 t

(c)
Reconstructed
fh(t) signal

0 t

Fig. 7.29 Signal reconstruction using zero order hold circuit

7.11.8 Signal Reconstruction Using Zero Order Hold Device

Figure 7.29 shows the reconstruction of continuous signal using zero order hold
(ZOH). Figure 7.29a shows the continuous signal. Figure 7.29b shows the signal
after the sampler. Figure 7.29c shows the output of the zero order hold circuit. The
zero order hold circuit keeps its output level equal to the magnitude of an input pulse
and then resets itself when a new pulse arrives. On the arrival of the new pulse, the
signal jumps to the new height and holds the output at the level. Thus, the output of
the ZOH leads to the staircase approximation as shown in Fig. 7.29c.
7.11 Sampled Data System 893

7.11.9 Transfer Function of a ZOH

Figure 7.28 shows the characteristics of a ZOH. It is a composition of positive unit

step and a negative unit step of delay T . The transfer function of a unit positive step
is 1/s at that of a negative unit step with a time delay − 1s e−sT . Hence, the transfer
function of a ZOH is
1
G oh = [1 − e−sT ] (7.134)
s

7.11.10 The Sampling Theorem

The signal x(t), δ(t) and g(t) shown in Fig. 7.25 are connected by the following
equation:

g(t) = x(t)δT (t) (7.135)

where
∞

δT (t) = δ(t − nTs ) (7.136)
n=−∞

δTs (t) is called sampling function, Ts is the sampling period and ωs = 2π

Ts
is the
fundamental sampling frequency. This is also called radian frequency and is related
to cyclic frequency as

2π
ωs = = 2π f s
Ts

where f s = T1s is the cyclic frequency. The sampled signal g(t) shown in Fig. 7.25(e)
consists of impulses spaced every Ts seconds, which is the sampling interval. δT (t)
can be expressed as a trigonometric Fourier series as

1
δT (t) = [1 + 2 cos ωs t + 2 cos 2ωs t + 2 cos 3ωs t + · · · ] (7.137)
Ts
g(t) = x(t)δT (t)
1
= [x(t) + 2x(t) cos ωs t + 2x(t) cos 2ωs t + · · · ] (7.138)
Ts

From the knowledge of the Fourier transform of continuous time signal, the following
equations are written:
894 7 State Space Modelling and Analysis

FT
x(t) ←→ X (ω)
FT
2x(t) cos ωs t ←→ X (ω − ωs ) + X (ω + ωs )
FT
2x(t) cos 2ωs t ←→ X (ω − 2ωs ) + X (ω + 2ωs )
FT
g(t) ←→ G(ω) (7.139)

Substituting Eq. (7.139) in Eq. (7.138), we get

1
G(ω) = [X (ω) + X (ω − ωs ) + X (ω + ωs )
Ts
+X (ω − 2ωs ) + X (ω − 2ωs ) + · · · ] (7.140)

∞
1 
G(ω) = X (ω − nωs ) (7.141)
Ts n=−∞

In Eq. (7.140), the first term in the bracket is X (ω). The second term is
X (ω − ωs ) + X (ω + ωs ). This represents the spectrum of X ( jω) shifted to ωs and
−ωs . Similarly, the third term X (ω − 2ωs ) + X ((ω + 2ωs )) which represents the
spectrum X (ω) is shifted by 2ωs and −2ωs and so on. The frequency spectrum
X (ω) and δ(ω) are represented in Fig. 7.30a and b, respectively. In Fig. 7.30a, ωm
is the maximum frequency content of the continuous time signal. Figure 7.30c
represents the frequency spectrum G(ω) of the continuous sampled signal for
ωs > 2ωm . Spectrum of the sampled signal G(ω) for ωs < 2ωm is shown in Fig. 7.30d.
To reconstruct the continuous time signal x(t) from sampled signal g(t), we
should be able to recover X (ω) from G(ω). This recovery is possible if there is no
overlap between successive G(ω). From Fig. 7.30c, this is possible if ωs > 2ωm or

fs > 2 fm (7.142)

The minimum sampling rate f s = 2 f m is called the Nyquist rate of x(t) and the cor-
responding time interval Ts = 1fs = 2 1fm is called Nyquist interval of x(t). Samples
of a signal taken at its Nyquist rate are the Nyquist samples.
From Eq. (7.142), the Shannon sampling theorem or simply sampling theorem is
stated as follows.
A band-limited signal of finite energy which has no frequency component
higher than f m can be completely described and recovered back if the sampling
frequency is twice the highest frequency of the given signal.
The proof of the theorem is given in Eq. (7.141) and Fig. 7.30c. If f s < 2 f m ,
Eq. (7.141) is represented in Fig. 7.30d and here overlapping between successive
samples occurs, and therefore it is not possible to recover x(t) from the frequency
spectrum G(ω) when passed through low-pass filter.
7.12 MATAB Program 895

(a) X( )

m 0 m

(b) T( )

2
T

3 s 2 s s 0 s 2 s 3 s

(c) G( ) s> 2 m
1
T

2 s s m 0 m s 2 s
( s m)

(d) G( ) s< 2 m

1
T

2 s s 0 s m m s 2 s
aliasing

Fig. 7.30 Continuous signal, impulse train sampled signal spectrum

7.12 MATAB Program

7.12.1 Conversion of State Space Model to Transfer Function

Model and Vice Versa

Example 7.39
Determine the transfer function and the eigenvalues of the system represented in
state space using MATLAB.
896 7 State Space Modelling and Analysis
⎡ ⎤ ⎡ ⎤
4 1 −2 1
d x(t)
= ⎣1 0 2 ⎦ x(t) + ⎣2⎦ u(t)
dt 1 −1 3 3
y(t) = [2 − 6 5]x(t)

PROGRAM

clc;
A = [41 − 2; 102; 1 − 13];
B = [1; 2; 3];
C = [2 − 65];
D = 0;
[num,den]=ss2tf(A,B,C,D)
sys=tf(num,den)
EigenValues=roots(den)

OUTPUT

num = 0 5.0000 − 37.0000 60.0000

den = 1.0000 − 7.0000 15.0000 − 9.0000
Transfer function:
5s 2 − 37s + 60/s 3 − 7s 2 + 15s − 9
Eigen Values = 3.0000
3.0000
1.0000

Example 7.40
Write a Program to obtain the transfer function of the system defined by the following
state space equations:
⎡ ⎤ ⎡ ⎤
0 1 0 0
d x(t) ⎣
= 0 0 1 ⎦ x(t) + ⎣0⎦ u(t)
dt −5 −10 −3 1
y(t) = [4 5 1]x(t)
7.12 MATAB Program 897

PROGRAM

clc;
A = [010; 001; −5 − 10 − 3];
B = [0; 0; 1];
C = [451];
D = 0;
[num,den]=ss2tf(A,B,C,D) sys=tf(num,den)

OUPUT

num = 0 1.0000 5.0000 4.0000

den = 1.0000 3.0000 10.0000 5.0000
Transfer function:
s 2 + 5s + 4/s 3 + 3s 2 + 10s + 5

7.12.2 Conversion of Transfer Function to State Model

Example 7.41
Write a Program to obtain the state space equations for the transfer function given
as follows:
5s + 2
T (s) =
s 3 + 7s 2 + 3s + 5

PROGRAM

clc;
num=[0 0 5 2];
den=[1 7 3 5];
[A,B,C,D]=tf2ss(num,den)

OUTPUT

⎡ ⎤
−7 −3 −5
A=⎣1 0 0 ⎦
0 1 0
⎡ ⎤
1
B = ⎣0⎦
0
898 7 State Space Modelling and Analysis

C = [0 5 2]
D = [0]

7.12.3 To Write a Program to Obtain the STM

Example 7.42
Write a Program to find the state transition matrix for the following A matrix:
 
−3 1
A=
−2 0

PROGRAM

clc;
A = [−31; −20];
T=sym(‘t’)
STM=expm(A*t)

OUTPUT

STM =[2/exp(2*t) - 1/exp(t), 1/exp(t) - 1/exp(2*t)]

[2/exp(2*t) - 2/exp(t), 2/exp(t) - 1/exp(2*t)]

7.12.4 To Determine Controllability and Observability of the

System

Example 7.43
A certain control system is described by the following vector matrix differential
equation:
⎡ ⎤ ⎡ ⎤
1 2 1 1
dx
= ⎣−1 −4 −3⎦ x(t) + ⎣4⎦ u(t)
dt 1 2 3 6
y(t) = [1 1 2]x(t)
7.12 MATAB Program 899

Determine whether the above system is completely state controllable, completely

output controllable and observable.

PROGRAM

clc;
A = [121; −1 − 4 − 3; −123];
B = [1; 4; 6];
C = [112];
D=0;

disp(‘Rank of the Matrix’)

Rankc=rank([B A*B A2 *B]) % To check the controllability
Ranko= rank([C’ A’*C’ A’2 *C’]) % To check the observability
Rankoc= rank([C*B C*A*B C*A2 *B]) % To check the output Controllability

OUTPUT

Rankc =3
Ranko = 3
Rankoc = 1

From the above, the system is completely state controllable and observable.
The system is not output controllable since the rank of the matrix is not three.

SUMMARY

1. An N th order system (continuous as well as discrete) can be described in terms of

N variables with N first order equations. These variables are called state variables.
2. State variable representation is not unique. However, the solution of state equation
is unique.
3. The state equations are written in a specific format. Such an equation is called
vector matrix differential/difference equation.
4. The state variables give internal and external descriptions of the system. Thus,
physical variables can be chosen as state variables, and their behaviour can be
readily studied.
5. State equations are written from system structure or from block diagrams.
6. State equations are solved by time domain or frequency domain methods.
900 7 State Space Modelling and Analysis

EXERCISE
Short Answer Type Questions
1. Define the state of a system.
The state of a system is defined as the minimum number of initial conditions
that must be specified at any initial time t0 so that the complete behaviour of the
system at any time t > t0 is determined if the input u(t) is known.
2. What do you understand by state vector?
If N state variables are required to completely describe the behaviour of the
system, then these N variables are the N components of a vector q. Such a
vector is called a state vector.
3. What is state space?
The N dimensional space whose coordinate axes consist of x1 axis, x2 axis,
. . . , x N axis is called state space.
4. What are state space equations?
Input variables, output variables and state variables are involved in the modelling
of dynamic systems. The dynamic equations involving these three variables are
called state space equations.
5. What is a vector matrix differential/difference equation?
State variable equations are expressed in the time domain by using compact
vector matrix notations. These equations when written for CT systems are called
vector matrix differential equations and when written for DT systems are called
vector matrix difference equations.
6. What is state transition matrix (STM)?
The matrix which is unique for a given system that transforms any initial state
q(t0 ) to any final state q(t f ) is called the state transition matrix. It contains all
the information about the system dynamics at all times. For a continuous time
system, the STM φ(t) = e At .
7. What is the Cayley–Hamilton theorem?
According to the Cayley–Hamilton theorem, the matrix A satisfies its own char-
acteristic equation. This property is used to evaluate the STM e At .
8. What physical variables are chosen as state variables in an electrical circuit
and mechanical systems?
The current through the inductor and the voltage across the capacitor are chosen
as state variables in electrical circuits. The displacement and velocity of energy-
storing elements such as mass (inertia) and spring are chosen as state variables
in mechanical systems.
9. What are the advantages of the state space model over that of the transfer
function model?
(a) The state space model is applicable to linear, non-linear and time varying
systems, whereas the T.F. model is applicable only to linear systems.
7.12 MATAB Program 901

(b) The T.F. model requires initial conditions to be zero whereas for the state
space model, the initial conditions need not be zero.
(c) System design modelled by T.F. is based on trial and error and it will not in
general lead to optimal control. The famous optimal control theory which
follows a systematic design procedure uses the state space model of the
system.
10. For a particular system, the A matrix is represented in more than one form.
What is the nature of characteristic equation?
In the state space model, even though the system A matrix is not unique, the
characteristic equation is the same and is unique.
Long Answer Type Questions
1. Consider the mechanical system shown in Fig. 7.31. Form the state space
equation.
⎡ ⎤
− RL 0 0 ⎡1⎤
⎢ 0 0 1 ⎥ L
ẋ(t) = ⎢
⎣ Ks K eq
⎥ ⎣ ⎦
Beq ⎦ x(t) + 0 e(t)
− − 0
Meq Meq Meq

where
l1 l2
Meq = M1 + M2
l2 l1
l1 l2
Beq = B1 + B2
l2 l1
l1 l2
K eq = K1 + K2
l2 l1

2. Consider the electrical circuit shown in Fig. 7.32. Form the state space equa-
tion.
⎡ ⎤ ⎡ ⎤
11 1 5
⎢− ⎥ ⎢ ⎥
ẋ(t) = ⎣ 34 41 ⎦ x(t) + ⎣ 41 ⎦ v(t)
− −
8 8 8
3. Consider the following T.F. of a certain continuous time system. Form the
state space equations in canonical form I model.

7z 2 + 8z + 4
H(z) =
z 3 + 5z 2 + 9z + 10
902 7 State Space Modelling and Analysis
⎡ ⎤ ⎡ ⎤
−5 1 0 7
ẋ(t) = ⎣ −9 0 1 ⎦ x(t) + ⎣ 8 ⎦ u(t)
−10 0 0 4

y(t) = [1 0 0]x(t)

4. Consider the following T.F. of continuous time system. Form the state space
equation in canonical form II model.

5z 3 + 6z 2 + 2z + 10
H(z) =
z 3 + 7z 2 + 4z + 5
⎡ ⎤ ⎡ ⎤
0 1 0 0
ẋ(t) = ⎣ 0 0 1 ⎦ x(t) + ⎣ 0 ⎦ u(t)
−5 −4 −7 1

y(t) = [−15 − 18 − 29]x(t) + 5u(t)

5. Consider the following T.F. of continuous time system. Form the state space
equation in diagonal form model.

(z 2 + 6z + 8)
H(z) =
(z 3 + 27z 2 + 230z + 600)
⎡ ⎤ ⎡ ⎤
−5 0 0 1
ẋ(t) = ⎣ 0 −10 0 ⎦ x(t) + ⎣ 1 ⎦ u(t)
0 0 −12 1
 
3 24 40
y(t) = − x(t)
35 5 7

6. Consider the following differential equation:

d3 y d2 y dy dx
− 5 +6 + 7y = 2 + 5u(t)
d t3 d t2 dt dt
Form the state space equation in canonical form II model.
⎡ ⎤ ⎡ ⎤
0 1 0 0
ẋ(t) = ⎣ 0 0 1 ⎦ x(t) + ⎣ 0 ⎦ u(t)
−7 −6 5 1

y(t) = [5 2 0] x(t)
7.12 MATAB Program 903

7. Consider the following T.F. of a certain discrete time system:

4z 2 − 5z + 10
H(z) =
z 3 + 2z 2 − 7z + 9

Form the state variable equation in canonical form II model.

⎡ ⎤ ⎡ ⎤
0 1 0 0
x[n + 1] = ⎣ 0 0 1 ⎦ x[n] + ⎣ 0 ⎦ u[n]
−9 7 −2 1

y[n] = [10 − 5 4] x[n]

8. Consider the following T.F. of a certain discrete system given below. Form
the state space equation of canonical form I model.

2z 2 + 6z + 9
H(z) =
z3 + 8z 2 + 7z + 16
⎡ ⎤ ⎡ ⎤
−8 1 0 2
x[n + 1] = ⎣ −7 0 1 ⎦ x[n] + 6 ⎦ u[n]
⎣
−16 0 0 9

y[n] = [1 0 0] x[n]

9. A certain discrete time system is described by the following difference equa-

tion:
1 1 1
y[n] − y[n − 1] − y[n − 2] + y[n − 3]
12 6 48
3 1
= x[n − 1] − x[n − 2] + x[n − 3]
8 32
Determine the system T.F. Form the state space equation in the diagonal
form. What are the eigenvalues of the system?
⎡1 ⎤ ⎡ ⎤
2
0 0 1
x[n + 1] = ⎣ 0 − 16 0 ⎦ x[n] + ⎣ 1 ⎦ u[n]
0 0 − 41 1
 
3 35
y[n] = − 3 x[n]
16 16
904 7 State Space Modelling and Analysis

l1 l2

i(t)
M1, L, R
M2
e(t)
f(t) Ksi(t)

K2 B2

Fig. 7.31 Mechanical system with lower arrangement

R1 1 R3 3

R2
v(t) C 2F
L 1H

Fig. 7.32 Electrical circuit for Question 2

z −1 − 38 z −2 + 1 −3
z
H [z] = 1 −1 1 −2
32
1 −3
1− 12
z − 6
z − 48 z

The eigenvalues are λ1 = 21 , λ2 = − 16 , λ3 = − 14 .

10. Find the state equation of a continuous time LTI system described by

ÿ(t) + 3 ẏ(t) + 2 y(t) = u(t)

(Anna University, May, 2007)

   
0 1 0
ẋ(t) = x(t) + u(t)
−2 −3 1
Index

A Compensator design in the frequency

A.C. two phase servomotor, 66 domains, 727
A/D converter, 16 Lag compensator, 749
Absolute stability, 519 Lag lead compensator, 765
Angular acceleration, 47 Lead compensator, 727
Angular displacement, 46 Compensator design using root locus, 777
Angular velocity, 47 Lag compensator, 787
Asymptotic stability, 503 Lag–lead compensator, 796
Lead compensator, 777
Complex s-plane, 20
B Computer controlled system, 16
Back emf constant, 54 Constant M circles, 451
Bandwidth, 354, 356 Constant N circles, 453
BIBO stability, 503 Control system, 4
Block diagram, 26 Controllability and observability of DT sys-
Block diagram connection, 87 tem, 881
cascade form, 88 controllability condition, 881
feedback form, 88 observability condition, 882
parallel form, 88 Controllability of continuous time system,
Block diagram for multiple inputs, 103 854
Block diagram reduction, 86 output controllability, 856
parts, 87 state controllability, 854
rule, 88
Controllable canonical form, 825
Bode asymptotic phase plots, 366
Controllers, 268
Bode plots, 364
Cutoff frequency, 356
advantages, 365
Cutoff rate, 356
disadvantages, 365

C D
Cascade form, 835 D.C. generator, 64
Cauchy’s theorem, 552 D.C. motor, 60
Cayley–Hamilton theorem, 844 armature controlled, 63
Characteristic roots, 505 field controlled, 60
Clock, 16 D.C. Tachogenerator, 65
Closed loop system, 7 D/A converter, 16

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer 905
Nature Switzerland AG 2022
S. Palani, Automatic Control Systems,
https://doi.org/10.1007/978-3-030-93445-3
906 Index

Derivative controller, 281 L

Diagonal form, 836 Lever, 33
Digital Computer, 16 Linear acceleration, 33
Linear displacement, 33
Linear velocity, 33
E Liquid level system, 13
Eigen values, 844 Log magnitude versus phase plot, 407
Electrical analogy, 71
force Current, 73
force Voltage, 72 M
Electromechanical system, 54 Mapping contour concept, 548
Mapping point concept, 548
Marginal stability, 503
Mason’s gain formula, 113
F Mass, 33
Feedback, 6 Mechanical systems, 32
Feedback system, 7 rotational, 32
First order system, 165 translational, 46
impulse responses, 165 Minimum phase transfer function, 23
modelling, 165
rise time, 168
settling time, 169 N
step response, 166 Nichols chart, 454
time constant, 167 advantages, 455
time delay, 169 Nichols plot, 407
time domain specifications, 167 advantages, 407
Force, 33 disadvantages, 407
Frequency domain specifications, 409 Non-minimum phase T.F., 24
Frequency response method Nyquist plot, 332
advantages, 328 Nyquist stability criterion, 544, 553
advantages, 544

G
Gain cross-over, 414 O
Gain cross-over frequency, 414 Observability of continuous time system,
Gain margin, 412 861
Gear, 46 Observable canonical form, 828
Generalized error constants, 250 Open loop system, 6
Order of the system, 21
Output, 4
H
Hold circuit, 889–892
P
Parallel compensators, 724
Parallel form, 836
I Parameter, 4
Impulse response function, 504 PD controller, 286
Inertial, 46 Phase cross-over, 418
Input, 4 Phase cross-over frequency, 417
Integral controller, 280 Phase margin, 412
Phase margin and gain margin, 424
Bode plot, 424
J Phase margin in polar plot, 414
Jordan canonical form, 836 Phase variable canonical form, 825
Index 907

PI controller, 274 damping ratio, 173

PID controller, 274, 291 impulse response, 174
PID controller parameter selection, 292 modelling, 172
Plant, 3, 16 natural frequency of oscillation, 173
Polar plot, 351 over damped, 176, 182
Polar plot with time delay, 349 step response, 179
Poles and zeros, 19 under damped, 176, 179
Position control system, 15 Series compensators, 724
Principle of argument, The, 552 Series-parallel compensators, 724
Process, 3 Servomechanism, 5, 15
Properties of Root Locus, 636 SFG applied to block diagrams, 129
angle of arrival, 640 SFG for electrical network, 136
angle of departure, 639 Signal flow graph, 110
angles of arrival and departure, 641 Signal flow graph algebra, 110
breakaway and break-in points, 641 addition rule, 110
centroid, 638 multiplication rule, 111
ending point, The, 636 transmission rule, 111
intersection with the imaginary axis, 642 Signal flow graph terminologies, 112
number of root loci, 637 branch, 112
real axis root loci, 638 forward path, 112
starting point, The, 636 loop, 112
symmetry of root loci, 637 loop gain, 112
Proportional controller, 276 mixed nodes, 112
node, 112
non-touching loops, 113
R
path, 112
Rate controller, 269
path gain, 112
Regulator system, 5, 10
self loop, 113
Relative stability, 504, 542
sink, 112
Reset controller, 280
source, 112
Resonant frequency, 353
Signal reconstruction, 892
Resonant peak, 353
Root locus concept, 633 Sinusoidal transfer function, 22
advantages, 632 Solenoid, 54
magnitude criterion, 634 Solution of state equation, 841
phase angle criterion, 634 Solution of state equation of DT system, 877
Root locus gain margin and phase margin, z-transform method, 878
666 recursive method, 877
Routh’s array, 520 Spring, 33
Routh–Hurwitz criterion, 517 Stability concept, 502
disadvantage, 518 State, 808
State equation of CT system, 808
State equation of DT system, 865
S controllable canonical form, 867
Sampled data system, 886 observable canonical form, 870
advantages and disadvantages, 886 State transition matrix (STM), 841
closed loop sampled data system, 887 Static error constants, 230
Sampling process, 887 acceleration error constants, 237
Sampling theorem, 893 position error constants, 231
Second Order system, 179 velocity error constants, 234
importance Of Impulse Response, 178 Steady state error, 229
Second order system, 172 Steady state error for disturbances, 248
critically damped, 176, 181 Steady state error for non-Unity feedback,
damping factor, 173 244
908 Index

Steering Control system, 10 Torque, 46

Synchros, 70 Torque–Current analogy, 81
control transformer, 71 Torque–Voltage analogy, 81
generator, 70 Torsional spring, 46
System, 3 Transfer function, 17
System classification, 17 improper, 18
proper, 18
strictly proper, 18
T Transient response and frequency response
T.F. form Bode plot, 397 correlation, 352
Tachometer feedback controller, 269 Type of the System, 230
Temperature control systems, 14
Three term controllers, 274
Time domain specification of second order V
system, 183 Variables, 4
number of oscillation, 190 Vector matrix differential equation, 809
peak order-shoot, 185 Viscous friction, 33, 46
peak time, 187
period of oscillation, 190
rise time, 189 Z
settling time, 187 Zero input response, 503
time constant, 189 Zero input stability, 503
time delay, 190 Zero state response, 503