SECTION 2: BLOCK DIAGRAMS

& SIGNAL FLOW GRAPHS

ESE 430 – Feedback Control Systems
Block
2 Diagrams
K. Webb ESE 430
 Block
3

 In the introductory section we saw examples of block diagrams

to represent systems, e.g.:

 Block diagrams consist of

 Blocks – these represent subsystems – typically modeled by, and labeled
with, a transfer function
 Signals – inputs and outputs of blocks – signal direction indicated by
arrows – could be voltage, velocity, force, etc.
 Summing junctions – points were signals are algebraically summed –
subtraction indicated by a negative sign near where the signal joins the
summing junction
K. ESE
 Standard Block Diagram
4

 The basic input/output relationship for a single block is:

𝑌𝑌 𝑠𝑠 = ⋅ 𝐺𝐺
𝑈𝑈 𝑠𝑠 𝑠𝑠

 Block diagram blocks can be connected in three basic forms:

 Cascade
 Parallel
 Feedback

 We’ll next look at each of these forms and derive a

single- block equivalent for each
K. ESE
 Cascade
5

 Blocks connected in cascade:

𝑋𝑋𝑠𝑠 = ⋅ = ⋅ 𝑠𝑠
𝑈𝑈 𝑠𝑠 𝐺𝐺1𝑠𝑠 𝑋𝑋1 𝑠𝑠 𝐺𝐺2
,
1

𝑌𝑌𝑠𝑠 = 𝑋𝑋2 𝑠𝑠
𝑠𝑠 ⋅ 𝑠𝑠 ⋅ 𝑠𝑠 ⋅ 𝑠𝑠
𝑋𝑋 𝐺𝐺 𝐺𝐺 𝐺𝐺 ⋅ 𝐺𝐺𝑒𝑒𝑒𝑒
𝑠𝑠
𝑌𝑌𝑠𝑠 2 3 = 𝑠𝑠 2 3
𝑋𝑋1
𝐺𝐺𝑒 = 𝑈𝑈
𝑠𝑠 ⋅ 𝐺𝐺 𝑠𝑠
1 ⋅ 𝐺𝐺𝑠𝑠
3 = 𝑈𝑈 𝑠𝑠
⋅ 𝐺𝐺𝑠𝑠
𝑒𝑒𝑒
𝑠𝑠 = 𝑠𝑠 ⋅ 2
𝑠𝑠
𝐺𝐺1 𝐺𝐺2 𝑠𝑠 ⋅
𝐺𝐺3

 The equivalent transfer function of cascaded blocks is the

K. ESE
 Cascade
5
product of the individual transfer functions

K. ESE
 Parallel
6

 Blocks connected in
parallel: 𝑋 = ⋅
𝑋1 𝑠𝑠 𝑈𝑈 𝑠𝑠 𝐺𝐺1𝑠𝑠
𝑋 𝑠𝑠 = 𝑠𝑠 ⋅ 𝑠𝑠
𝑋2 𝑈𝑈 𝐺𝐺2
𝑋 𝑠𝑠 = 𝑠𝑠 ⋅ 𝑠𝑠𝑠𝑠 ± 𝑋𝑋𝑠𝑠
3
𝑋3 𝑈𝑈 𝐺𝐺3
𝑌𝑌𝑠𝑠 = 𝑠𝑠 ±
𝑋𝑋1 𝑋𝑋2

𝑌𝑌 = ⋅ ± ⋅ ± 𝑈𝑈
𝑠𝑠 ⋅ 𝐺𝐺𝑠𝑠
3
𝑈𝑈 𝑠𝑠 𝐺𝐺1 𝑠𝑠 𝑈𝑈 𝑠𝑠 𝐺𝐺2𝑠𝑠
𝑌𝑌 𝑠𝑠 𝑠𝑠 = 𝑠𝑠 ⋅ 𝐺𝐺𝑒𝑒𝑒𝑒
𝑠𝑠
𝑠𝑠 = 𝑠𝑠 𝐺𝐺 𝑠𝑠 ± 𝑠𝑠 ± 𝑈𝑈
𝑈𝑈 1 𝐺𝐺2 𝐺𝐺3
𝑠𝑠 𝑠𝑠 𝑠𝑠
K. ESE
 Parallel
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 = ± ± 𝐺𝐺3
6

𝐺𝐺1 𝐺𝐺2

 The equivalent transfer function is the sum of the

individual transfer functions:

K. ESE
 Feedback
7

 Of obvious interest to us, is the feedback form:

𝑌𝑌 = 𝐺𝐺𝑠𝑠
𝐸𝐸 𝑠𝑠
𝑌𝑌 𝑠𝑠 𝑠𝑠 − 𝐺 𝑠𝑠
= 𝑋𝑋 𝑠𝑠 𝐺 𝐺𝐺
𝑌𝑌 𝑠𝑠 𝑠𝑠
− 𝑠𝑠𝐻 𝑠𝑠
𝑅𝑅 𝑠𝑠
𝑌𝑌 𝐻
𝑠𝑠 =

𝑅𝑅
𝑠𝑠
𝑌 𝑌 𝑅
1
𝑌 𝑠𝑠 =
�
𝑠𝑠
�
�
+
𝑌 �
�
𝑠𝑠 𝑠𝑠
�
K. ESE
 Feedback
𝑠𝑠 𝑠𝑠
𝐻𝐻 𝑠𝑠𝐺𝐺 = 𝐺𝐺
𝑠𝑠
8

𝑅𝑅
⋅
𝑠𝑠
1 + 𝐺𝐺 𝐻𝐻

 The closed-loop transfer function, 𝑇𝑇

𝑠𝑠, is

𝑇𝑇 𝑌 𝐺 𝑠𝑠
𝑌 𝑠𝑠 𝐺
=
𝑠𝑠 𝑅 = 𝐻𝐻𝑠𝑠
𝑠𝑠 1 + 𝑠𝑠
𝑅 𝐺𝐺

K. ESE
 Feedback
9

𝑇𝑇 𝐺 𝑠𝑠
𝐺
𝐻𝐻𝑠𝑠
𝑠𝑠 = 1 +
𝑠𝑠
𝐺𝐺

Note that this is negative feedback, for positive feedback:

𝐺𝐺𝑠𝑠


𝑇𝑇 𝑠𝑠 =
1 − 𝑠𝑠𝐻𝐻𝑠𝑠
𝐺𝐺
𝐻
𝐺𝐺 𝑠𝑠𝐻 𝑠𝑠
 The factor in the denominator is the loop gain or open-loop

transfer function

forward path gain – here, 𝐺𝐺 𝑠𝑠

 The gain from input to output with the feedback path broken is the
K. ESE
 Feedback
10
In general:
forward path gain


𝑇𝑇
𝑠𝑠 = 1 − loop gain

K. ESE
 Closed-Loop Transfer Function -
11

 Calculate the closed-loop transfer function

𝐷𝐷 𝑠𝑠 and
𝐺𝐺 𝑠𝑠
 are in cascade

𝐻𝐻1 𝑠𝑠 𝑠𝑠is in
cascade with the feedback system consisting of 𝐷𝐷


𝐺𝐺 𝑠𝑠 , and 𝐻𝐻
𝑠𝑠2
,

𝑇𝑇 𝐷𝐷 𝐷
𝑠𝑠 = 𝑠𝑠 ⋅ 1 +
𝑠𝑠 �
�
𝐻𝐻1
K. ESE
 Closed-Loop Transfer Function -
𝑠𝑠
𝐺𝐺 𝐺𝐺 𝑠𝑠 𝐻 𝑠𝑠𝐻2 𝑠𝑠
12

𝑇𝑇 𝐷 𝐺𝐺𝑠𝑠
= 𝐷 𝑠𝑠
𝑠𝑠𝐻𝐻𝑠𝑠
2
𝐻𝐻1 𝑠𝑠
𝑠𝑠𝐺
𝑠𝑠 1 + 𝐺
𝐷𝐷

K. ESE
 Unity-Feedback Systems
13

 We’re often interested in unity-feedback systems

 Feedback path gain is unity

 Can always reconfigure a system to unity-feedback form
 Closed-loop transfer function is:

𝑇𝑇 𝐷𝐷𝑠𝑠𝐺 𝑠𝑠
𝑠𝑠 = 1 + 𝐺 𝐺𝐺
𝑠𝑠 𝑠𝑠
𝐷𝐷
K. ESE
Block
1 Diagram Manipulation
1
K. Webb ESE 430
 Block Diagram
12

 Often want to simplify block diagrams into

simpler, recognizable forms
 To determine the equivalent transfer function

 Simplify to instances of the three standard

forms, then simplify those forms
 Move blocks around relative to summing
junctions and pickoff points – simplify to a
standard form
 Move blocks forward/backward past summing junctions
 Move blocks forward/backward past pickoff points

K. ESE
 Moving Blocks Back Past a Summing
13

 The following two block diagrams are equivalent:

𝑌𝑌 𝑠𝑠 = 𝑠𝑠 + 𝑠𝑠𝐺𝐺 𝑠𝑠 = 𝑠𝑠𝐺𝐺 𝑠𝑠 + 𝑠𝑠𝐺𝐺𝑠𝑠

𝑈𝑈2 𝑈𝑈1 𝑈𝑈2
𝑈𝑈1
K. ESE
 Moving Blocks Forward Past a Summing
14

 The following two block diagrams are equivalent:

1
𝑌𝑌 𝑠𝑠 = 𝑠𝑠𝐺𝐺 𝑠𝑠 + 𝑠𝑠 = 𝑠𝑠 + 𝐺 𝑠𝑠
𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑈𝑈1 𝑈𝑈2 𝑈𝑈2 𝐺
𝑈𝑈1
K. ESE
 Moving Blocks Relative to Pickoff
15

 We can move blocks backward past pickoff points:

 And, we can move them forward past pickoff points:

K. ESE
 Block Diagram Simplification – Example
16

 Rearrange the following into a unity-feedback system

Move the feedback block, 𝑠𝑠,

𝐻𝐻


past the summing junction forward, to

 Add an inverse block on 𝑅𝑅 𝑠𝑠
compensate for the move

1
Closed-loop transfer function:
𝐺𝐺


𝐻𝐻 𝑠𝑠 𝑠𝑠
𝑠
K. ESE
 Block Diagram Simplification – Example
𝑇𝑇 𝐻 𝐻 𝐺 𝑠𝑠
𝑠𝑠 = 1 + 𝐺
16

𝑠𝑠𝐻 = 𝐻𝐻𝑠𝑠
𝐺𝐺 𝐻 𝑠𝑠 1 + 𝑠𝑠
𝐺𝐺

K. ESE
 Block Diagram Simplification – Example
17

 Find the closed-loop transfer function of the

following system through block-diagram
simplification

K. ESE
 Block Diagram Simplification – Example
𝐺 𝑠𝑠 and 𝑠𝑠 are in feedback form
18

𝐺1 𝐻𝐻1


𝐺𝐺𝑒𝑒𝑒𝑒 𝐺𝐺1
K. = ESE
 Block Diagram Simplification – Example
𝑠𝑠
1 − 𝐺𝐺1 𝑠𝑠 𝑠𝑠
19

𝐻𝐻1 𝑠𝑠

K. ESE
 Block Diagram Simplification – Example

𝑠𝑠 backward past the pickoff point

20

𝐺𝐺2
 Move

𝐺𝐺2 𝐻𝐻2
 Block from previous step, , and become a

K. ESE
 Block Diagram Simplification – Example
21 𝑠𝑠 𝑠𝑠
feedback system that can be simplified

K. ESE
 Block Diagram Simplification – Example
22

Note that we’ve dropped the function of 𝑠𝑠

Simplify the feedback subsystem
𝑠𝑠, for clarity


notation,

𝐺𝐺1𝐺𝐺2 𝐺𝐺1𝐺𝐺2𝐻𝐻2
1 + 1 − 𝐺𝐺 𝐻𝐻
1− 1 1
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 =
𝐺𝐺1𝐻𝐻1
K. ESE
 Block Diagram Simplification – Example
𝐺𝐺1𝐺𝐺2
23

= 1 − 𝐺𝐺 𝐻𝐻 +
1 1
𝐺𝐺1𝐺𝐺2𝐻𝐻2

K. ESE
 Block Diagram Simplification – Example
24

 Simplify the two parallel subsystems

𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 = 𝐺𝐺4
+ 𝐺𝐺
𝐺𝐺3 2

K. ESE
 Block Diagram Simplification – Example
25

 Now left with two cascaded subsystems

 Transfer functions multiply

𝐺𝐺𝑒𝑒𝑒𝑒 𝐺𝐺1𝐺𝐺2𝐺𝐺3 + 𝐺𝐺1𝐺𝐺4

𝑠𝑠 =
1 − 𝐺𝐺1𝐻𝐻1 + 𝐺𝐺1𝐺𝐺2𝐻𝐻2

K. ESE
 Block Diagram Simplification – Example
26

 The equivalent, close-loop transfer function is

𝑇𝑇 𝐺𝐺1𝐺𝐺2𝐺𝐺3 + 𝐺𝐺1𝐺𝐺4
𝑠𝑠 =
1 − 𝐺𝐺1𝐻𝐻1 + 𝐺𝐺1𝐺𝐺2𝐻𝐻2
K. ESE
Multiple-Input
2 Systems
4
K. Webb ESE 430
 Multiple Input
25

E.g., reference, 𝑅𝑅
𝑠𝑠, and disturbance, 𝑊𝑊
Systems often have more than one input
𝑠𝑠


 Two transfer functions: 𝑌𝑌

From reference to output
𝑇𝑇𝑠𝑠 =


From disturbance to output 𝑌𝑌 𝑠𝑠

𝑇𝑇𝑤𝑤


𝑠𝑠
=
K. ESE
 Multiple Input
25

⁄𝑅𝑅 𝑠𝑠

/𝑊𝑊 𝑠𝑠 𝑠𝑠

K. ESE
 Transfer Function –
Find transfer function from 𝑅𝑅
𝑠𝑠 to 𝑌𝑌
𝑠𝑠
26

 Set 𝑊𝑊 𝑠𝑠 = 0
A linear system – superposition applies

𝑇𝑇 𝑌 𝑠𝑠 𝐷𝐷𝑠𝑠𝐺 𝑠𝑠
𝑌 =
= 𝑠𝑠 1 + 𝐺 𝐺𝐺𝑠𝑠
𝑠𝑠
𝑠𝑠 𝑅 𝐷𝐷
𝑅
K. ESE
 Transfer Function –
Next, find transfer function from 𝑊𝑊
𝑠𝑠 to 𝑌𝑌
𝑠𝑠
27

𝑅𝑅𝑠𝑠 = 0


 Set
 System now becomes:

𝑇𝑇𝑤𝑤 𝑌 𝑊
𝑠𝑠 𝐺𝐺𝑤𝑤
𝑌 𝑠𝑠 = 1 + 𝐷𝐷
𝑠𝑠 = 𝑊 𝑠𝑠
K. ESE
 Transfer Function –
𝐺𝐺 𝑠𝑠
28

𝐺𝐺 𝑠𝑠 𝑠𝑠

K. ESE
 Multiple Input
29

 Two inputs, two transfer functions

𝑇𝑇 𝑠𝑠 = 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑠𝑠 = 𝐺𝐺𝑤𝑤 𝑠𝑠 𝐺𝐺 𝑠𝑠
1+𝐷𝐷 𝑠𝑠 1+𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
and
𝐺𝐺 𝑠𝑠

𝑇𝑇𝑤𝑤

 𝐷𝐷
𝑠𝑠 is the controller transfer function

 We have control over both 𝑇𝑇𝑠𝑠 and 𝑇𝑇𝑤𝑤

 Ultimately,
we’ll determine this
𝑠𝑠
 What do we want these to
be?
K. ESE
 Multiple Input
𝑠𝑠 for desired performance
30

𝑇𝑇
 Design

𝑠𝑠
for disturbance rejection

𝑇𝑇𝑤𝑤
 Design

K. ESE
Signal
2 Flow Graphs
9
K. Webb ESE 430
 Signal Flow
30

 An alternative to block diagrams for graphically describing systems

 Signal flow graphs consist of:

 Nodes –represent signals
 Branches –represent system blocks
 Branches labeled with system transfer functions
 Nodes (sometimes) labeled with signal names
 Arrows indicate signal flow direction
 Implicit summation at nodes
 Always a positive sum
 Negative signs associated with branch transfer functions
K. ESE
 Block Diagram → Signal Flow
31

 To convert from a block diagram to a signal

flow graph:
1. Identify and label all signals on the block diagram
2. Place a node for each signal
3. Connect nodes with branches in place of the blocks
 Maintain correct direction
 Label branches with corresponding transfer functions
 Negate transfer functions as necessary to provide
negative feedback
4. If desired, simplify where possible

K. ESE
 Signal Flow Graph – Example
32

 Convert to a signal flow graph

 Label any unlabeled signals

 Place a node for each signal

K. ESE
 Signal Flow Graph – Example
33

 Connect nodes with branches, each representing a system block

 Note the -1 to provide negative feedback of 𝑋𝑋𝑠𝑠

2

K. ESE
 Signal Flow Graph – Example
34

 Nodes with a single input and single output can

be eliminated, if desired
This makes sense for 𝑋𝑋𝑠𝑠1 and 𝑋𝑋 𝑠𝑠
2
Leave 𝑈𝑈𝑠𝑠 to indicate separation between controller and plant


K. ESE
 Signal Flow Graph – Example
35

 Revisit the block diagram from earlier

 Convert to a signal flow graph

 Label all signals, then place a node for each

K. ESE
 Signal Flow Graph – Example
36

 Connect nodes with branches

K. ESE
 Signal Flow Graph – Example
37

𝑠𝑠, 𝑠𝑠, 𝑠𝑠, and 𝑋𝑋

𝑠𝑠8
𝑋𝑋5 𝑋𝑋 𝑋𝑋
 Simplify – eliminate

6 7

K. ESE
 Signal Flow Graph – Example
38

K. ESE
 Signal Flow Graphs vs. Block
39

 Signal flow graphs and block diagrams are

alternative, though equivalent, tools for
graphical representation of interconnected
systems
 A generalization (not a rule)
 Signalflow graphs – more often used when dealing
with state-space system models
 Blockdiagrams – more often used when dealing with
transfer function system models

K. ESE
Mason’s
3 Rule
9
K. Webb ESE 430
 Mason’s
40

 We’ve seen how to reduce a complicated

block diagram to a single input-to-output
transfer function
 Many successive simplifications

 Mason’s rule provides a formula to calculate

the same overall transfer function
 Singleapplication of the formula
 Can get complicated

 Before presenting the Mason’s rule formula,

we need to define some terminology

K. ESE
 Loop
41

 Loop gain – total gain (product of individual gains)

around any path in the signal flow graph
 Beginning and ending at the same node
 Not passing through any node more than once

−𝐺𝐺1𝐻𝐻3
 Here, there are three loops with the following gains:

𝐺𝐺2𝐻𝐻1
1.

−𝐺𝐺2𝐺𝐺3𝐻𝐻2
2.
3.

K. ESE
 Forward Path
42

 Forward path gain – gain along any path from the

input to the output
 Not passing through any node more than once
 Here, there are two forward paths with the
following gains:
𝐺𝐺1𝐺𝐺2𝐺𝐺3𝐺𝐺4
𝐺𝐺1𝐺𝐺2𝐺𝐺5
1.

2.

K. ESE
 Non-Touching
43

 Non-touching loops – loops that do not have

any nodes in common
Here,
−𝐺𝐺1𝐻𝐻3 does not touch 𝐺𝐺2𝐻𝐻1


−𝐺𝐺1𝐻𝐻3 does not touch −𝐺𝐺2𝐺𝐺3𝐻𝐻2

1.

2.

K. ESE
 Non-Touching Loop
44

 Non-touching loop gains – the product of loop

gains from non-touching loops, taken two, three,
four, or more at a time
Here, there are only two pairs of non-touching loops
−𝐺𝐺1𝐻𝐻3 ⋅ 𝐺𝐺2𝐻𝐻1


−𝐺𝐺1𝐻𝐻3 ⋅ −𝐺𝐺2𝐺𝐺3𝐻𝐻2
1.

2.

K. ESE
 Mason’s
𝑌𝑌
45

𝑃𝑃

= 𝑠𝑠 1
𝑇𝑇 𝑠𝑠
𝑅 𝑠𝑠 = � 𝑇𝑇𝑘𝑘 Δ𝑘𝑘
𝑅 Δ𝑘𝑘=1
where
𝑃𝑃 = # of forward paths
𝑇𝑇𝑘𝑘 = gain of the 𝑘𝑘𝑡𝑡𝑡 forward path
Δ = 1 − Σ(loop gains)
+Σ(non-touching loop gains taken two-at-a-time)
−Σ(non-touching loop gains taken three-at-a-time)
+Σ(non-touching loop gains taken four-at-a-time)
−Σ …
Δ𝑘𝑘 = Δ − Σ(loop gain terms in Δ that touch the 𝑘𝑘𝑡𝑡𝑡 forward path)

K. ESE
 Mason’s Rule -
46

# of forward paths: 𝐺𝐺2𝐺𝐺3𝐻𝐻2

𝑃𝑃 = 2


Forward path gains:

𝑇𝑇1 = 𝐺𝐺1𝐺𝐺2𝐺𝐺3𝐺𝐺4


𝑇𝑇2 = 𝐺𝐺1𝐺𝐺2𝐺𝐺5
Σ(loop gains):
−𝐺𝐺1𝐻𝐻3 + 𝐺𝐺2𝐻𝐻1 −


K. ESE
 Mason’s Rule -
Σ(NTLGs taken two-at-
46

a-time):
+ −𝐺𝐺1𝐻𝐻3𝐺𝐺𝐺𝐺
2𝐻𝐻1𝐻𝐻
1 3𝐺𝐺2𝐺𝐺3𝐻𝐻

Δ:
Δ=1−


−𝐺𝐺1𝐻𝐻3 + 𝐺𝐺2𝐻𝐻1 − 𝐺
+ −𝐺𝐺1𝐻𝐻3𝐺𝐺2𝐻𝐻1 + 𝐺𝐺1𝐻

K. ESE
 Mason’s Rule – Example -
Simplest way to find Δ𝑘𝑘 terms is to calculate Δ with the 𝑘𝑘𝑡𝑡𝑡
47


path removed – must remove nodes as well
𝑘𝑘 = 1:

With forward path 1 removed, there are no loops, so

Δ1 = 1 − 0


Δ1 = 1

K. ESE
 Mason’s Rule – Example -
48

 𝑘𝑘 = 2:

Similarly, removing forward path 2 leaves no loops, so

Δ2 = 1 − 0


Δ2 = 1

K. ESE
 Mason’s Rule -
49

𝑃𝑃
𝑌𝑌 𝑠𝑠1
For our example:
𝑃𝑃 = 2


𝑇𝑇1 = 𝐺𝐺1𝐺𝐺2𝐺𝐺3𝐺𝐺4 𝑇𝑇 𝑠𝑠==� 𝑇𝑇𝑘𝑘Δ𝑘𝑘

𝑇𝑇2 = 𝐺𝐺1𝐺𝐺2𝐺𝐺5 𝑅𝑅 𝑠𝑠Δ
𝑘𝑘=1
Δ = 1 + 𝐺𝐺1𝐻𝐻3 − 𝐺𝐺2𝐻𝐻1 + 𝐺𝐺2𝐺𝐺3𝐻𝐻2 − 𝐺𝐺1𝐻𝐻3𝐺𝐺2𝐻𝐻1 +
𝐺𝐺1𝐻𝐻3𝐺𝐺2𝐺𝐺3𝐻𝐻2 Δ1 = 1
Δ2 = 1

 The closed-loop transfer function:

𝑇𝑇 𝑇𝑇1Δ1 + 𝑇𝑇2Δ2
𝑠𝑠 = Δ
𝑇𝑇 𝐺𝐺1𝐺𝐺2𝐺𝐺3𝐺𝐺4 + 𝐺𝐺1𝐺𝐺2𝐺𝐺5
𝑠𝑠
= 1 + 𝐺𝐺1𝐻𝐻3 − 𝐺𝐺2𝐻𝐻1 + 𝐺𝐺2𝐺𝐺3𝐻𝐻2 − 𝐺𝐺1𝐻𝐻3𝐺𝐺2𝐻𝐻1 +
K. ESE
 Mason’s Rule -
𝐺𝐺1𝐻
50

𝐻3𝐺𝐺2
𝐺𝐺3𝐻
𝐻2

K. ESE
Preview
5 of Controller Design
0
K. Webb ESE 430
 Controller Design –
51

 We now have the tools necessary to determine

the transfer function of closed-loop feedback
systems
 Let’s take a closer look at how feedback can help
us achieve a desired response
 Just a preview – this is the objective of the whole course

 Consider a simple first-order system

 A single real pole at 𝑠𝑠 = −2

K. ESE
 Controller Design –
𝑟𝑟𝑟𝑟𝑟𝑟
51

𝑠𝑠𝑒𝑒𝑠𝑠

K. ESE
 Open-Loop Step
52

 This system
exhibits the
expected
first- order
step response
 Noovershoot or
ringing

K. ESE
 Add
53

 Now let’s enclose the system in a feedback loop

 Add controller block with transfer function 𝐷𝐷

𝑠𝑠
 Closed-loop transfer function becomes:
1 𝐷𝐷 𝑠𝑠
𝐷𝐷 𝑠𝑠 𝑠 𝑠 +
𝑇𝑇 𝑠𝑠 =
𝑠𝑠 + 2 + 𝐷𝐷 𝑠𝑠
=
1 + 𝐷𝐷 𝑠𝑠
𝑠𝑠1

K. ESE
 Add
54
 Clearly the addition of feedback and the
controller changes the transfer function – but

Let’s consider a couple of example cases for 𝐷𝐷

how?
 𝑠𝑠

K. ESE
 Add
55

 First, consider a simple gain block for the controller

, amplified by a constant gain, 𝐾𝐾𝐶𝐶

𝐸𝐸 𝑠𝑠
 Error signal,

 A proportional controller, with gain 𝐾𝐾𝐶𝐶

 Now, the closed-loop transfer function is:
𝐾𝐾𝐶𝐶
𝑠𝑠+2 𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = =
𝐾𝐾𝐶𝐶 𝑠𝑠 + 2 + 𝐾𝐾𝐶𝐶
1 𝑠𝑠 +
2

 A single real pole at 𝑠𝑠 = −

2 + 𝐾𝐾𝐶𝐶
 Pole moved to a higher frequency
K. ESE
 Add
56
 A faster response

K. ESE
 Open-Loop Step
57

 As feedback
gain increases:
 Pole moves to a
higher frequency
 Response gets
faster

K. ESE
 First-Order
58

 Next, allow the controller to have some dynamics of its own

Now the controller is a first-order block with gain 𝐾𝐾𝐶𝐶 and a pole at
𝑠𝑠 = −𝑏𝑏


 This yields the following closed-loop transfer function:

𝐾𝐾𝐶𝐶 1
𝑠𝑠+𝑏 𝑏 𝑠𝑠+2 𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = 𝐾𝐾𝐶𝐶 1 =
1+ 𝑠𝑠2 +2 + 𝑏𝑏 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶
𝑠𝑠 + 𝑏𝑏𝑠𝑠 + 2

 The closed-loop system is now second-order

 One pole from the plant
 One pole from the controller

K. ESE
 First-Order
59

𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 =
𝑠𝑠2 +2 + 𝑏𝑏 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾

 Two closed-loop poles:

𝑏𝑏 + 2 𝑏𝑏2 − 4𝑏𝑏 + 4 − 4𝐾𝐾𝐶𝐶

𝑠𝑠1,2 = − ±
2 2
 Pole locations determined by 𝑏𝑏 and 𝐾𝐾𝐶𝐶
 Controller parameters – we have control over these
 Design the controller to place the poles where we want them
 So, where do we want them?
 Design to performance specifications
 Risetime, overshoot, settling time, etc.
K. ESE
 Design to
60

 The second-order closed-loop transfer function

𝑇𝑇
𝐾𝐾𝐶𝐶
𝑠𝑠 = 𝑠𝑠2 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶
2 + 𝑏𝑏
+
can be expressed
as
𝑇𝑇 𝐾𝐾𝐶𝐶 𝐾𝐾𝐶𝐶
𝑠𝑠 = = 𝑠𝑠2 + 2𝜎𝜎𝑠𝑠 + 𝜔𝜔2
𝑠𝑠2 + 2𝜁𝜁𝜔𝜔𝑛𝑛𝑠𝑠
+ 𝜔𝜔2
� �

 Let’s say we want a closed-loop response that satisfies

the following specifications:
%𝑂𝑂𝑂𝑂 ≤ 5%
 𝑡𝑡𝑠𝑠 ≤ 600 𝑚𝑚𝑠𝑠𝑚𝑚𝑚𝑚


K. ESE
 Design to
Use %𝑂𝑂𝑂𝑂 and 𝑡𝑡𝑠𝑠 specs to determine values of 𝜁𝜁 and 𝜎𝜎
61

Then use 𝜁𝜁 and 𝜎𝜎 to determine 𝐾𝐾𝐶𝐶 and 𝑏𝑏




K. ESE
 Determine 𝜁𝜁 from
Overshoot and damping ratio, 𝜁𝜁, are related
62


as follows:
𝜁𝜁 = − ln 𝑂𝑂𝑂𝑂
𝜋𝜋2 + ln2 𝑂𝑂𝑂𝑂

 The requirement is %𝑂𝑂𝑂𝑂 ≤ 5%, so

𝜁𝜁 ≥ − ln
0.05 = 0.69
𝜋𝜋2 + ln2 0.05

 Allowing some margin, set 𝜁𝜁 = 0.75

K. ESE
 Determine 𝜎𝜎 from
Settling time (±1%) can be approximated from 𝜎𝜎 as
63

4.6


𝑡𝑡𝑠𝑠 ≈
𝜎𝜎
The requirement is 𝑡𝑡𝑠𝑠 ≤ 600 𝑚𝑚𝑠𝑠𝑚𝑚𝑚𝑚
Allowing for some margin, design for 𝑡𝑡𝑠𝑠 = 500 𝑚𝑚𝑠𝑠𝑚𝑚𝑚𝑚


4.6 4.


𝑡𝑡𝑠𝑠 = 500 𝑚𝑚𝑠𝑠𝑚𝑚𝑚𝑚 →

𝜎𝜎 = 6
≈ 𝜎𝜎
500
𝑚𝑚𝑠𝑠𝑚𝑚𝑚𝑚
which gives
𝑟𝑟𝑟𝑟𝑟𝑟

K. ESE
 Determine 𝜎𝜎 from
𝜎𝜎 = 9.2
𝑠𝑠𝑚𝑚𝑚𝑚
64

We can then calculate the natural frequency from 𝜁𝜁 and 𝜎𝜎

𝜎𝜎 9. 𝑟𝑟𝑟𝑟𝑟𝑟


2
𝜔𝜔𝑛𝑛 = = = 12.27
𝜁𝜁 0.75 𝑠𝑠𝑚𝑚𝑚𝑚

K. ESE
 Determine Controller Parameters from 𝜎𝜎 and
65

 The characteristic polynomial is

𝑠𝑠2 + 2 + 𝑏𝑏
𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶 = 𝑠𝑠2 + 2𝜎𝜎𝑠𝑠 + 𝜔𝜔2
�

 Equating coefficients to solve for 𝑏𝑏:

2 + 𝑏𝑏 = 2𝜎𝜎 = 18.4
𝑏𝑏 = 16.4

𝐾𝐾𝑠𝑠 :
and
2𝑏𝑏 + 𝐾𝐾𝐶𝐶 =� 12.27 = 150.5
2

𝜔𝜔2 =
𝐾𝐾𝐶𝐶 = 150.5 − 2 ⋅ 16.4 = 117.7 → 118
𝐾𝐾𝑠𝑠 = 118

The controller transfer function is

118


K.
𝑠𝑠 ESE
 Determine Controller Parameters from 𝜎𝜎 and
𝐷𝐷 =
𝑠𝑠 + 16.4
66

K. ESE
 Closed-Loop
67

 Closed-loop system
is now second
order

 Controller designed
to place the two
closed-loop poles
at desirable
Controller Plant
locations:
 𝑠𝑠1 ,2 = −9.2 ± 𝑗𝑗𝑗.13
pole pole

 𝜁𝜁 = 0.75
K. ESE
 Closed-Loop
𝜔𝜔𝑛𝑛 = 12.3
68

K. ESE
 Closed-Loop Step
69

 Closed-loop step
response
satisfies the
specifications
 Approximations
were used
 If requirements not
met - iterate

K. ESE