SECTION 2: BLOCK DIAGRAMS
& SIGNAL FLOW GRAPHS
ESE 430 – Feedback Control Systems
2 Block Diagrams
K. Webb ESE 430
Block Diagrams
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. Webb ESE 430
Standard Block Diagram Forms
4
The basic input/output relationship for a single block is:
𝑌𝑌 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺 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. Webb ESE 430
Cascade Form
5
Blocks connected in cascade:
𝑋𝑋1 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺1 𝑠𝑠 , 𝑋𝑋2 𝑠𝑠 = 𝑋𝑋1 𝑠𝑠 ⋅ 𝐺𝐺2 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑋𝑋2 𝑠𝑠 ⋅ 𝐺𝐺3 𝑠𝑠 = 𝑋𝑋1 𝑠𝑠 ⋅ 𝐺𝐺2 𝑠𝑠 ⋅ 𝐺𝐺3 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺1 𝑠𝑠 ⋅ 𝐺𝐺2 𝑠𝑠 ⋅ 𝐺𝐺3 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 = 𝐺𝐺1 𝑠𝑠 ⋅ 𝐺𝐺2 𝑠𝑠 ⋅ 𝐺𝐺3 𝑠𝑠
The equivalent transfer function of cascaded blocks is the
product of the individual transfer functions
K. Webb ESE 430
Parallel Form
6
Blocks connected in parallel:
𝑋𝑋1 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺1 𝑠𝑠
𝑋𝑋2 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺2 𝑠𝑠
𝑋𝑋3 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺3 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑋𝑋1 𝑠𝑠 ± 𝑋𝑋2 𝑠𝑠 ± 𝑋𝑋3 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺1 𝑠𝑠 ± 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺2 𝑠𝑠 ± 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺3 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 𝐺𝐺1 𝑠𝑠 ± 𝐺𝐺2 𝑠𝑠 ± 𝐺𝐺3 𝑠𝑠 = 𝑈𝑈 𝑠𝑠 ⋅ 𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 = 𝐺𝐺1 𝑠𝑠 ± 𝐺𝐺2 𝑠𝑠 ± 𝐺𝐺3 𝑠𝑠
The equivalent transfer function is the sum of the individual
transfer functions:
K. Webb ESE 430
Feedback Form
7
Of obvious interest to us, is the feedback form:
𝑌𝑌 𝑠𝑠 = 𝐸𝐸 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 − 𝑋𝑋 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 − 𝑌𝑌 𝑠𝑠 𝐻𝐻 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑌𝑌 𝑠𝑠 1 + 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝐺𝐺 𝑠𝑠
𝑌𝑌 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 ⋅
1 + 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠
The closed-loop transfer function, 𝑇𝑇 𝑠𝑠 , is
𝑌𝑌 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 = =
𝑅𝑅 𝑠𝑠 1 + 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠
K. Webb ESE 430
Feedback Form
8
𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 =
1 + 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠
Note that this is negative feedback, for positive feedback:
𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 =
1 − 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠
The 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠 factor in the denominator is the loop gain or open-loop
transfer function
The gain from input to output with the feedback path broken is the
forward path gain – here, 𝐺𝐺 𝑠𝑠
In general:
forward path gain
𝑇𝑇 𝑠𝑠 =
1 − loop gain
K. Webb ESE 430
Closed-Loop Transfer Function - Example
9
Calculate the closed-loop transfer function
𝐷𝐷 𝑠𝑠 and 𝐺𝐺 𝑠𝑠 are in cascade
𝐻𝐻1 𝑠𝑠 is in cascade with the feedback system consisting of 𝐷𝐷 𝑠𝑠 ,
𝐺𝐺 𝑠𝑠 , and 𝐻𝐻2 𝑠𝑠
𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 = 𝐻𝐻1 𝑠𝑠 ⋅
1 + 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠 𝐻𝐻2 𝑠𝑠
𝐻𝐻1 𝑠𝑠 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 =
1 + 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠 𝐻𝐻2 𝑠𝑠
K. Webb ESE 430
Unity-Feedback Systems
10
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:
𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 =
1 + 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
K. Webb ESE 430
11 Block Diagram Manipulation
K. Webb ESE 430
Block Diagram Algebra
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. Webb ESE 430
Moving Blocks Back Past a Summing Junction
13
The following two block diagrams are equivalent:
𝑌𝑌 𝑠𝑠 = 𝑈𝑈1 𝑠𝑠 + 𝑈𝑈2 𝑠𝑠 𝐺𝐺 𝑠𝑠 = 𝑈𝑈1 𝑠𝑠 𝐺𝐺 𝑠𝑠 + 𝑈𝑈2 𝑠𝑠 𝐺𝐺 𝑠𝑠
K. Webb ESE 430
Moving Blocks Forward Past a Summing Junction
14
The following two block diagrams are equivalent:
1
𝑌𝑌 𝑠𝑠 = 𝑈𝑈1 𝑠𝑠 𝐺𝐺 𝑠𝑠 + 𝑈𝑈2 𝑠𝑠 = 𝑈𝑈1 𝑠𝑠 + 𝑈𝑈2 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝐺𝐺 𝑠𝑠
K. Webb ESE 430
Moving Blocks Relative to Pickoff Points
15
We can move blocks backward past pickoff points:
And, we can move them forward past pickoff points:
K. Webb ESE 430
Block Diagram Simplification – Example 1
16
Rearrange the following into a unity-feedback system
Move the feedback block, 𝐻𝐻 𝑠𝑠 , forward,
past the summing junction
Add an inverse block on 𝑅𝑅 𝑠𝑠 to
compensate for the move
Closed-loop transfer function:
1
𝐻𝐻 𝑠𝑠 𝐺𝐺 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝐻𝐻 𝑠𝑠
𝑇𝑇 𝑠𝑠 = =
1 + 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠 1 + 𝐺𝐺 𝑠𝑠 𝐻𝐻 𝑠𝑠
K. Webb ESE 430
Block Diagram Simplification – Example 2
17
Find the closed-loop transfer function of the following
system through block-diagram simplification
K. Webb ESE 430
Block Diagram Simplification – Example 2
18
𝐺𝐺1 𝑠𝑠 and 𝐻𝐻1 𝑠𝑠 are in feedback form
𝐺𝐺1 𝑠𝑠
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 =
1 − 𝐺𝐺1 𝑠𝑠 𝐻𝐻1 𝑠𝑠
K. Webb ESE 430
Block Diagram Simplification – Example 2
19
Move 𝐺𝐺2 𝑠𝑠 backward past the pickoff point
Block from previous step, 𝐺𝐺2 𝑠𝑠 , and 𝐻𝐻2 𝑠𝑠 become a
feedback system that can be simplified
K. Webb ESE 430
Block Diagram Simplification – Example 2
20
Simplify the feedback subsystem
Note that we’ve dropped the function of 𝑠𝑠 notation, 𝑠𝑠 , for clarity
𝐺𝐺1 𝐺𝐺2
1 − 𝐺𝐺1 𝐻𝐻1 𝐺𝐺1 𝐺𝐺2
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 = =
𝐺𝐺1 𝐺𝐺2 𝐻𝐻2 1 − 𝐺𝐺1 𝐻𝐻1 + 𝐺𝐺1 𝐺𝐺2 𝐻𝐻2
1+
1 − 𝐺𝐺1 𝐻𝐻1
K. Webb ESE 430
Block Diagram Simplification – Example 2
21
Simplify the two parallel subsystems
𝐺𝐺4
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 = 𝐺𝐺3 +
𝐺𝐺2
K. Webb ESE 430
Block Diagram Simplification – Example 2
22
Now left with two cascaded subsystems
Transfer functions multiply
𝐺𝐺1 𝐺𝐺2 𝐺𝐺3 + 𝐺𝐺1 𝐺𝐺4
𝐺𝐺𝑒𝑒𝑒𝑒 𝑠𝑠 =
1 − 𝐺𝐺1 𝐻𝐻1 + 𝐺𝐺1 𝐺𝐺2 𝐻𝐻2
K. Webb ESE 430
Block Diagram Simplification – Example 2
23
The equivalent, close-loop transfer function is
𝐺𝐺1 𝐺𝐺2 𝐺𝐺3 + 𝐺𝐺1 𝐺𝐺4
𝑇𝑇 𝑠𝑠 =
1 − 𝐺𝐺1 𝐻𝐻1 + 𝐺𝐺1 𝐺𝐺2 𝐻𝐻2
K. Webb ESE 430
24 Multiple-Input Systems
K. Webb ESE 430
Multiple Input Systems
25
Systems often have more than one input
E.g., reference, 𝑅𝑅 𝑠𝑠 , and disturbance, 𝑊𝑊 𝑠𝑠
Two transfer functions:
From reference to output
𝑇𝑇 𝑠𝑠 = 𝑌𝑌 𝑠𝑠 ⁄𝑅𝑅 𝑠𝑠
From disturbance to output
𝑇𝑇𝑤𝑤 𝑠𝑠 = 𝑌𝑌 𝑠𝑠 /𝑊𝑊 𝑠𝑠
K. Webb ESE 430
Transfer Function – Reference
26
Find transfer function from 𝑅𝑅 𝑠𝑠 to 𝑌𝑌 𝑠𝑠
A linear system – superposition applies
Set 𝑊𝑊 𝑠𝑠 = 0
𝑌𝑌 𝑠𝑠 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 = =
𝑅𝑅 𝑠𝑠 1 + 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
K. Webb ESE 430
Transfer Function – Reference
27
Next, find transfer function from 𝑊𝑊 𝑠𝑠 to 𝑌𝑌 𝑠𝑠
Set 𝑅𝑅 𝑠𝑠 = 0
System now becomes:
𝑌𝑌 𝑠𝑠 𝐺𝐺𝑤𝑤 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇𝑤𝑤 𝑠𝑠 = =
𝑊𝑊 𝑠𝑠 1 + 𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
K. Webb ESE 430
Multiple Input Systems
28
Two inputs, two transfer functions
𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠 𝐺𝐺𝑤𝑤 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝑇𝑇 𝑠𝑠 =
1+𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
and 𝑇𝑇𝑤𝑤 𝑠𝑠 =
1+𝐷𝐷 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝐷𝐷 𝑠𝑠 is the controller transfer function
Ultimately,
we’ll determine this
We have control over both 𝑇𝑇 𝑠𝑠 and 𝑇𝑇𝑤𝑤 𝑠𝑠
What do we want these to be?
Design 𝑇𝑇 𝑠𝑠 for desired performance
Design 𝑇𝑇𝑤𝑤 𝑠𝑠 for disturbance rejection
K. Webb ESE 430
29 Signal Flow Graphs
K. Webb ESE 430
Signal Flow Graphs
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. Webb ESE 430
Block Diagram → Signal Flow Graph
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. Webb ESE 430
Signal Flow Graph – Example 1
32
Convert to a signal flow graph
Label any unlabeled signals
Place a node for each signal
K. Webb ESE 430
Signal Flow Graph – Example 1
33
Connect nodes with branches, each representing a system block
Note the -1 to provide negative feedback of 𝑋𝑋2 𝑠𝑠
K. Webb ESE 430
Signal Flow Graph – Example 1
34
Nodes with a single input and single output can be
eliminated, if desired
This makes sense for 𝑋𝑋1 𝑠𝑠 and 𝑋𝑋2 𝑠𝑠
Leave 𝑈𝑈 𝑠𝑠 to indicate separation between controller and plant
K. Webb ESE 430
Signal Flow Graph – Example 2
35
Revisit the block diagram from earlier
Convert to a signal flow graph
Label all signals, then place a node for each
K. Webb ESE 430
Signal Flow Graph – Example 2
36
Connect nodes with branches
K. Webb ESE 430
Signal Flow Graph – Example 2
37
Simplify – eliminate 𝑋𝑋5 𝑠𝑠 , 𝑋𝑋6 𝑠𝑠 , 𝑋𝑋7 𝑠𝑠 , and 𝑋𝑋8 𝑠𝑠
K. Webb ESE 430
Signal Flow Graphs vs. Block Diagrams
38
Signal flow graphs and block diagrams are
alternative, though equivalent, tools for graphical
representation of interconnected systems
A generalization (not a rule)
Signal flow graphs – more often used when dealing
with state-space system models
Block diagrams – more often used when dealing with
transfer function system models
K. Webb ESE 430
39 Mason’s Rule
K. Webb ESE 430
Mason’s Rule
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. Webb ESE 430
Loop Gain
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
Here, there are three loops with the following gains:
1. −𝐺𝐺1 𝐻𝐻3
2. 𝐺𝐺2 𝐻𝐻1
3. −𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
K. Webb ESE 430
Forward Path Gain
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:
1. 𝐺𝐺1 𝐺𝐺2 𝐺𝐺3 𝐺𝐺4
2. 𝐺𝐺1 𝐺𝐺2 𝐺𝐺5
K. Webb ESE 430
Non-Touching Loops
43
Non-touching loops – loops that do not have any
nodes in common
Here,
1. −𝐺𝐺1 𝐻𝐻3 does not touch 𝐺𝐺2 𝐻𝐻1
2. −𝐺𝐺1 𝐻𝐻3 does not touch −𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
K. Webb ESE 430
Non-Touching Loop Gains
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
1. −𝐺𝐺1 𝐻𝐻3 ⋅ 𝐺𝐺2 𝐻𝐻1
2. −𝐺𝐺1 𝐻𝐻3 ⋅ −𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
K. Webb ESE 430
Mason’s Rule
45
𝑃𝑃
𝑌𝑌 𝑠𝑠 1
𝑇𝑇 𝑠𝑠 = = � 𝑇𝑇𝑘𝑘 Δ𝑘𝑘
𝑅𝑅 𝑠𝑠 Δ
𝑘𝑘=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. Webb ESE 430
Mason’s Rule - Example
46
# of forward paths: Σ(NTLGs taken two-at-a-time):
𝑃𝑃 = 2 −𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐻𝐻1 + 𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
Forward path gains: Δ:
𝑇𝑇1 = 𝐺𝐺1 𝐺𝐺2 𝐺𝐺3 𝐺𝐺4 Δ = 1 − −𝐺𝐺1 𝐻𝐻3 + 𝐺𝐺2 𝐻𝐻1 − 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
𝑇𝑇2 = 𝐺𝐺1 𝐺𝐺2 𝐺𝐺5 + −𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐻𝐻1 + 𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
Σ(loop gains):
−𝐺𝐺1 𝐻𝐻3 + 𝐺𝐺2 𝐻𝐻1 − 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
K. Webb ESE 430
Mason’s Rule – Example - Δ𝑘𝑘
47
Simplest way to find Δ𝑘𝑘 terms is to calculate Δ with the 𝑘𝑘 𝑡𝑡𝑡
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. Webb ESE 430
Mason’s Rule – Example - Δ𝑘𝑘
48
𝑘𝑘 = 2:
Similarly, removing forward path 2 leaves no loops, so
Δ2 = 1 − 0
Δ2 = 1
K. Webb ESE 430
Mason’s Rule - Example
49
For our example: 𝑃𝑃
𝑌𝑌 𝑠𝑠 1
𝑃𝑃 = 2 𝑇𝑇 𝑠𝑠 = = � 𝑇𝑇𝑘𝑘 Δ𝑘𝑘
𝑇𝑇1 = 𝐺𝐺1 𝐺𝐺2 𝐺𝐺3 𝐺𝐺4
𝑅𝑅 𝑠𝑠 Δ
𝑘𝑘=1
𝑇𝑇2 = 𝐺𝐺1 𝐺𝐺2 𝐺𝐺5
Δ = 1 + 𝐺𝐺1 𝐻𝐻3 − 𝐺𝐺2 𝐻𝐻1 + 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2 − 𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐻𝐻1 + 𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
Δ1 = 1
Δ2 = 1
The closed-loop transfer function:
𝑇𝑇1 Δ1 + 𝑇𝑇2 Δ2
𝑇𝑇 𝑠𝑠 =
Δ
𝐺𝐺1 𝐺𝐺2 𝐺𝐺3 𝐺𝐺4 + 𝐺𝐺1 𝐺𝐺2 𝐺𝐺5
𝑇𝑇 𝑠𝑠 =
1 + 𝐺𝐺1 𝐻𝐻3 − 𝐺𝐺2 𝐻𝐻1 + 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2 − 𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐻𝐻1 + 𝐺𝐺1 𝐻𝐻3 𝐺𝐺2 𝐺𝐺3 𝐻𝐻2
K. Webb ESE 430
50 Preview of Controller Design
K. Webb ESE 430
Controller Design – Preview
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
𝑟𝑟𝑟𝑟𝑟𝑟
A single real pole at 𝑠𝑠 = −2
𝑠𝑠𝑠𝑠𝑠𝑠
K. Webb ESE 430
Open-Loop Step Response
52
This system
exhibits the
expected first-
order step
response
No overshoot or
ringing
K. Webb ESE 430
Add Feedback
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
𝑇𝑇 𝑠𝑠 = 𝑠𝑠 + 2 =
1
1 + 𝐷𝐷 𝑠𝑠 𝑠𝑠 + 2 𝑠𝑠 + 2 + 𝐷𝐷 𝑠𝑠
Clearly the addition of feedback and the controller
changes the transfer function – but how?
Let’s consider a couple of example cases for 𝐷𝐷 𝑠𝑠
K. Webb ESE 430
Add Feedback
54
First, consider a simple gain block for the controller
Error signal, 𝐸𝐸 𝑠𝑠 , amplified by a constant gain, 𝐾𝐾𝐶𝐶
A proportional controller, with gain 𝐾𝐾𝐶𝐶
Now, the closed-loop transfer function is:
𝐾𝐾𝐶𝐶
𝑠𝑠 + 2 𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = =
𝐾𝐾𝐶𝐶
1 + 𝑠𝑠 + 2 𝑠𝑠 + 2 + 𝐾𝐾𝐶𝐶
A single real pole at 𝑠𝑠 = − 2 + 𝐾𝐾𝐶𝐶
Pole moved to a higher frequency
A faster response
K. Webb ESE 430
Open-Loop Step Response
55
As feedback gain
increases:
Pole moves to a
higher frequency
Response gets
faster
K. Webb ESE 430
First-Order Controller
56
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 + 𝑏𝑏 𝑠𝑠 + 2 𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = = 2
𝐾𝐾𝐶𝐶 1 𝑠𝑠 + 2 + 𝑏𝑏 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶
1+
𝑠𝑠 + 𝑏𝑏 𝑠𝑠 + 2
The closed-loop system is now second-order
One pole from the plant
One pole from the controller
K. Webb ESE 430
First-Order Controller
57
𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = 2
𝑠𝑠 + 2 + 𝑏𝑏 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶
Two closed-loop poles:
𝑏𝑏 + 2 𝑏𝑏 2 − 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. Webb ESE 430
Design to Specifications
58
The second-order closed-loop transfer function
𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = 2
𝑠𝑠 + 2 + 𝑏𝑏 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶
can be expressed as
𝐾𝐾𝐶𝐶 𝐾𝐾𝐶𝐶
𝑇𝑇 𝑠𝑠 = 2 2 = 2
𝑠𝑠 + 2𝜁𝜁𝜔𝜔𝑛𝑛 𝑠𝑠 + 𝜔𝜔𝑛𝑛 𝑠𝑠 + 2𝜎𝜎𝑠𝑠 + 𝜔𝜔𝑛𝑛2
Let’s say we want a closed-loop response that satisfies the
following specifications:
%𝑂𝑂𝑂𝑂 ≤ 5%
𝑡𝑡𝑠𝑠 ≤ 600 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
Use %𝑂𝑂𝑂𝑂 and 𝑡𝑡𝑠𝑠 specs to determine values of 𝜁𝜁 and 𝜎𝜎
Then use 𝜁𝜁 and 𝜎𝜎 to determine 𝐾𝐾𝐶𝐶 and 𝑏𝑏
K. Webb ESE 430
Determine 𝜁𝜁 from Specifications
59
Overshoot and damping ratio, 𝜁𝜁, are related as
follows:
− ln 𝑂𝑂𝑂𝑂
𝜁𝜁 =
𝜋𝜋 2 + ln2 𝑂𝑂𝑂𝑂
The requirement is %𝑂𝑂𝑂𝑂 ≤ 5%, so
− ln 0.05
𝜁𝜁 ≥ = 0.69
𝜋𝜋 2 + ln2 0.05
Allowing some margin, set 𝜁𝜁 = 0.75
K. Webb ESE 430
Determine 𝜎𝜎 from Specifications
60
Settling time (±1%) can be approximated from 𝜎𝜎 as
4.6
𝑡𝑡𝑠𝑠 ≈
𝜎𝜎
The requirement is 𝑡𝑡𝑠𝑠 ≤ 600 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
Allowing for some margin, design for 𝑡𝑡𝑠𝑠 = 500 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
4.6 4.6
𝑡𝑡𝑠𝑠 ≈ = 500 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 → 𝜎𝜎 =
𝜎𝜎 500 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
which gives
𝑟𝑟𝑟𝑟𝑟𝑟
𝜎𝜎 = 9.2
𝑠𝑠𝑠𝑠𝑠𝑠
We can then calculate the natural frequency from 𝜁𝜁 and 𝜎𝜎
𝜎𝜎 9.2 𝑟𝑟𝑟𝑟𝑟𝑟
𝜔𝜔𝑛𝑛 = = = 12.27
𝜁𝜁 0.75 𝑠𝑠𝑠𝑠𝑠𝑠
K. Webb ESE 430
Determine Controller Parameters from 𝜎𝜎 and 𝜔𝜔𝑛𝑛
61
The characteristic polynomial is
𝑠𝑠 2 + 2 + 𝑏𝑏 𝑠𝑠 + 2𝑏𝑏 + 𝐾𝐾𝐶𝐶 = 𝑠𝑠 2 + 2𝜎𝜎𝜎𝜎 + 𝜔𝜔𝑛𝑛2
Equating coefficients to solve for 𝑏𝑏:
2 + 𝑏𝑏 = 2𝜎𝜎 = 18.4
𝑏𝑏 = 16.4
and 𝐾𝐾𝑐𝑐 :
2𝑏𝑏 + 𝐾𝐾𝐶𝐶 = 𝜔𝜔𝑛𝑛2 = 12.27 2 = 150.5
𝐾𝐾𝐶𝐶 = 150.5 − 2 ⋅ 16.4 = 117.7 → 118
𝐾𝐾𝑐𝑐 = 118
The controller transfer function is
118
𝐷𝐷 𝑠𝑠 =
𝑠𝑠 + 16.4
K. Webb ESE 430
Closed-Loop Poles
62
Closed-loop system
is now second order
Controller designed
to place the two
closed-loop poles at
desirable locations:
𝑠𝑠1,2 = −9.2 ± 𝑗𝑗𝑗.13 Controller Plant
pole pole
𝜁𝜁 = 0.75
𝜔𝜔𝑛𝑛 = 12.3
K. Webb ESE 430
Closed-Loop Step Response
63
Closed-loop step
response satisfies
the specifications
Approximations
were used
If requirements not
met - iterate
K. Webb ESE 430
