Control Systems 0108400

Lect 03 Math Models, LT, Electric

Element, Transfer Function

Dr. M. N. Najem
Summer 2010-2011
May 18, 2011
Mathematical models
• In the previous lecture after obtaining a
schematic, the control engineer makes
simplifying assumptions in order to keep
the ensuing model manageable and still
approximate physical reality.
• Next step is to develop mathematical
models from schematics of physical
systems

Dr. Najem 2
Mathematical models …
• Transfer functions in frequency domain
(classical approach).
• State-space in time domain (modern approach).

• In every case the first step in developing a mathematical

model is to apply the fundamental physical laws of
science an engineering:
– In modeling electric networks: Ohm, KVL, KCL, Nodal,
Equivalent circuit, ..
– in modeling mechnical systems: Newton laws (forces, torques)
• From these equations we obtaing the relationship
between system’s output and input
Dr. Najem 3
Mathematical models …
• Differential equations can describe the
relationship between the input and output
of a system.
• The form of the D.E and its coefficients are
a formulation of the system.
• Although the D.E relates the system to
its input and output, it is not a
satisfying representation from a system
perspective. Look at a LTI nth order D.E:
Dr. Najem 4
 Mathematical models …
d nc(t ) d n1c(t ) dc(t ) d m r (t ) d m1r (t ) dr (t )
n
 an1 n1  ...  a1  a0c(t )  bm m  bm1 m1  ...  b1  b0 r (t )
dt dt dt dt dt dt
• The system parameters (the coefficients), the output, as
well as the input appear through out the D.E.

• Prefer a mathematical representation such that the

input, output, and system are distinct and separate
parts; ( conveniently represent interconnection of several
subsystems). Mathematical function that appears inside the
block called the transfer function, denoted, H(s), or G(s).

Input Output
r(t) System c(t)

Dr. Najem 5
 Review of Laplace transform

Dr. Najem 6
Laplace transform
• Addressed in MATH IV course

L[ f (t )]  F ( s)   f (t )e  st dt
0

1   j

1
L [ F ( s)]  f (t )  F ( s ) e st
ds  f (t )u (t )
2j   j
u (t )  1 t  0
0 t0
• Multiplication f(t) by u(t) yields a time function that is zero
for t < 0.
Dr. Najem 7
Laplace transform
• Refer to Laplace transform in Math IV to
review the various operations and to relate
to Laplace transform tables.
• Inverse of Laplace transform
– Partial-fraction expansion
– Use of various theroems
• Also, refer to solving D.E.s using Laplace
transform.
Dr. Najem 8
LT examples
• Presented and discussed on board

Dr. Najem 9
 Review of resistive electric
networks

Dr. Najem 10
Some elements
Independent sources: Resistor:

Dr. Najem 11
 Superposition example
• Given the following circuit

12V
2kW 4mA
– +

2mA 1kW 2kW

I0

• Calculate I0 using superposition

Dr. Najem 12
Superposition example …
• Solve the problem by leaving one independent source at a time and
opening the other Independent current sources and short circuiting
the other independent voltage sources
Open current source

Short circuit
2kW voltage source

2mA 1kW 2kW

I’0

I’0 = -4/3 mA
Dr. Najem 13
Superposition example …

2kW 4mA

Short circuit

Open current source 1kW 2kW

I’’0

I’’0 = 0
Dr. Najem 14
Superposition example …
Open current source

12V
2kW
– +
Open
current 1kW 2kW
source
I’’’0
Final result:
I’0 = -4/3 mA
I’’’0 = -4 mA I’’0 = 0
I’’’0 = -4 mA

I0 = I’0+ I’’0+ I’’’0 = -16/3 mA

Dr. Najem 15
Use of KVL and KCL

Node1
Loop1:

Loop1
Node1:

Dr. Najem 16
Equivalent circuits
USE THEVENIN TO COMPUTE Vo
Loop Analysis on left button circuit
I 2  2mA
 6V  4kI1  2k ( I1  I 2 )  0
6  2I2 5
I1  mA  mA
6 3
VOC  4k * I1  2k * I 2  20 / 3  4V  32 / 3[V ]

Voc
RTh  2 // 4  2  10 / 3kW Calculate V0
using Voltage
division
17
Nodal analysis and Mesh analysis

Mesh analysis

VS1  R1I1  R2 ( I1  I 2 )  R3 ( I1  I 2 )
 Solve for I1 and I2.
VS1  R1I1  R4 I 2  VS 2

Dr. Najem 18
 Dynamic networks

Dr. Najem 19
Characteristics of dynamic network
• Dynamic Elements  Ohm’s Law: ineffective
• Inductor:

Dr. Najem 20
Characteristics of dynamic network ...

• Dynamic Elements  Ohm’s Law: ineffective

• Capacitor:

Dr. Najem 21
Characteristics of dynamic network ...

• Example:

di(t ) 1 t
v(t )  L   i(t )d  Ri (t )
dt C 0

If
then
v(t)

Dr. Najem 22
Dynamic relationships into
Algebraic operation  LT

• Using Laplace Transform on previous circuit

1  I ( s ) i ( 1)
(0) 
Vs ( s)  L[ (t )]  L( sI ( s)  i(0 ))  

   RI ( s)
C s s 
1
 ( Ls) I ( s)  I ( s)  RI ( s)
Cs
• Then generalize resistor R into impedance Z
1
Z1 ( s)  Ls Z 2 (s) 
Cs
Dr. Najem 23
Dynamic relationships into
Algebraic operation  LT …

 Vs ( s)  Z1 ( s) I ( s)  Z 2 ( s) I ( s)  RI ( s)
Vs ( s)
 I (s) 
Z1 ( s)  Z 2 ( s)  R

Dr. Najem 24
Dynamic elements  Laplace
transform models
• Capacitor LT model

Dr. Najem 25
Dynamic elements  Laplace
transform models
• Inductor LT model

Dr. Najem 26
Resistor LT model
• V(s) = RI(s)

Sources as LT models

Dr. Najem 27
Laplace and analysis methods
• Laplace application to analysis techniques

Generalized Ohms Law 

superposition 
 These will be
applied as in
KVL and KCL  Resistive Networks
Equivalent circuit 

Nodal analysis and mesh analysis

Key: Laplace transform models of (dynamic) elements.

Dr. Najem 28
 Transfer Functions

Dr. Najem 29
System representations

• Mathematical form of representing

(modeling) the system.
• In this course we are concerned with the
following system representation methods:
– Differential equations
– Transfer functions
– State-space

Dr. Najem 30
Transfer Functions
• System analysis: emphasize is on relation between
input and output (using blocks)
• Circuit analysis: detailed analysis (voltages, branch
currents)
• System analysis: how the system processes the input to
form the output, or how the system transforms the input
into output.
• Output: variable to be controlled
• Input: variable used and to be adjusted to change or
influence the output

Input Output
Dynamic system c(t)
r(t)
Dr. Najem 31
Transfer Function H(s) or G(s)
• Gives Quantitative Description of ‘ how
the system processes the input to form the
output’.

Dr. Najem 32
 input to H(s) or G(s)
• Input is  (t) impulse fucntion
• The resultant output y(t) is called: the unit
impulse response.
• In this case: X(s) = L [ (t)] = 1 and
Y(s) = Laplace Transform of the unit impulse
response

Therefore, H(s) = Y(s)/X(s) = Y(s)

Therefore, the transfer function of a system is

the Laplace transform of the unit impulse
response of the system
Dr. Najem 33
Facts on Transfer Functions
• Independent of input, a property of the system structure
and parameters.
• Obtained with zero initial conditions.
• Generates a mathematical model of the system and
algebraically relates the representation of the output to
the representation of the input,.
• Applies only on LTI systems.
• Rational Function of s (Linear, lumped, fixed)
• H(s): Transfer function can be easily converted into
frequency response function of the system H(jw) or H(
j2f ). Simply replace s = jw.
• |H( j2f )| or |H( j )|: amplitude response function and
H(j2f) or H( j ): Phase response function
Dr. Najem 34
Properties of Transfer Function
• Properties of Transfer Function for Linear, Lumped
stable systems.

bm s m  bm 1s m 1  ...  b0 N ( s )
H ( s)  n 1

an s  an 1s  ...  a0 D( s )
n

• Corresponding differential equations:

d ( n ) y (t ) d ( n 1) y (t )
an n
 a n 1 n 1
 ...  a 0 y (t )
dt dt
d ( m ) x(t ) d ( m 1) x(t )
 bm m
 bm 1 m 1
 ...  b0 x(t )
dt dt
coef. ai, bj: all real! Why? Results from real system components.
Dr. Najem 35
Properties of Transfer Function …
• Differential equations do not separate the representation
of input, output, and system;  transfer functions do.
• Replacing D.Es. with algebraic equations simplifies the
representation of individual subsystems and simplifies
modeling interconnected subsystems.
• Roots of N(s), D(s): are real, or complex conjugate pairs.

• Zeros of the transfer function:= roots of N(s)

• Poles of the transfer function:= roots of D(s)

D( s )  ( s  3s  1)  D( s )  ( s  2)( s  1)
2

poles : 2,1
Dr. Najem 36
H(s) of BIBO stable system
H(s) = G(s) = N(s) / D(s)
• Degree of N(s) must be  Degree of D(s).
– If degree N(s) > Degree D(s) then divide:

N ' ( s)
 H ( s)  ck s  ...  c1s  c0 
k
D( s )
• but, under a bounded-input x(t) = u(t) (unit
step function) => L[u(t)] = X(s) = 1/s, then:

Dr. Najem 37
H(s) of BIBO stable system …
k 1 c0 N ' ( s )
 Y ( s )  ck s  ...  c1  
s sD( s )
 y (t )  ...  c1 (t )  ...

(
 (t ) not bounded!
so y(t) unbounded)

Dr. Najem 38
 H(s) of BIBO stable system …
• Poles: must lie in the left half of the s-
plan.
D( s)  ( s   1 )(s   2 )...(s   n )
then: Re( j )  0

WHY? …..

• Question: Any restriction on zeros?

• Answer: No (for BIBO stable system)
Dr. Najem 39
Next 2 lectures modeling of:
– Electric Networks Transfer Functions
– Translational Mechanical Systems Transfer
Functions
– Rotational Mechanical Systems Transfer
Functions
– Transfer Functions for Systems with Gears
– Electromechanical Systems Transfer
Functions
– Electric/Mechanical Circuit Analogs
– You do linearization on your own
Dr. Najem 40