Signal flow graphs consist of:
๐ Nodes โrepresent signals
๐ Branches โrepresent system blocks
Branches labeled with system transfer functions
1. Control Signal Flow Graphs
๏จ An alternative to block diagrams for graphically describing systems
๏จ Signal flow graphs consist of:
๐ Nodes โrepresent signals
๐ Branches โrepresent system blocks
๏จ Branches labeled with system transfer functions
๏จ Nodes (sometimes) labeled with signal names
๏จ Arrows indicate signal flow direction
๏จ Implicit summation at nodes
๐ Always a positive sum
๐ Negative signs associated with branch transfer functions
2. Block Diagram โ Signal Flow Graph
๏จ Toconvert 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
๏ฎ Maintain correct direction
๏ฎ Label branches with corresponding transfer functions
๏ฎ Negate transfer functions as necessary to provide negative
feedback
4. If desired, simplify where possible
3. Signal Flow Graph โ Example 1
๏จ Convert to a signal flow graph
๏จ Label any unlabeled signals
๏จ Place a node for each signal
4. Signal Flow Graph โ Example 1
๏จ Connect nodes with branches, each representing a system block
๏จ Note the -1 to provide negative feedback of X2 ๐
5. Signal Flow Graph โ Example 1
๏จ Nodes with a single input and single output can be
eliminated, if desired
๐ This makes sense for X1 ๐ and X2 ๐
๐ Leave ๐ ๐ to indicate separation between controller and plant
6. Signal Flow Graph โ Example 2
๏จ Revisit the block diagram from earlier
๐ Convert to a signal flow graph
๏จ Label all signals, then place a node for each
8. Signal Flow Graph โ Example 2
๏จ Simplify โ eliminate X5 ๐ , X6 ๐ , X7 ๐ , and X8 ๐
9. 40
Masonโs Rule
๏จ Weโve seen how to reduce a complicated block
diagram to a single input-to-output transfer
function
๐ Many successive simplifications
๏จ Masonโs rule provides a formula to calculate the
same overall transfer function
๐ Single application of the formula
๐ Can get complicated
๏จ Before presenting the Masonโs rule formula, we
need to define some terminology
10. 41
Loop Gain
๏จ Loop gain โ total gain (product of individual gains) around
any path in the signal flow graph
๐ Beginning and ending at the same node
๐ Not passing through any node more than once
๏จ Here, there are three loops with the following gains:
1. โ๐บ1๐ป3
2. ๐บ2๐ป1
3. โ๐บ2๐บ3๐ป2
11. 42
Forward Path Gain
๏จ Forward path gain โ gain along any path from the input
to the output
๐ Not passing through any node more than once
๏จ Here, there are two forward paths with the following
gains:
1. ๐บ1๐บ2๐บ3๐บ4
2. ๐บ1๐บ2๐บ5
12. 43
Non-Touching Loops
๏จ Non-touching loops โ loops that do not have any
nodes in common
๏จ Here,
1. โ๐บ1๐ป3 does not touch ๐บ2๐ป1
2. โ๐บ1๐ป3 does not touch โ๐บ2๐บ3๐ป2
13. 44
Non-Touching Loop Gains
๏จ Non-touching loop gains โ the product of loop gains
from non-touching loops, taken two, three, four, or
more at a time
๏จ Here, there are only two pairs of non-touching loops
1.
2.
โ๐บ1๐ป3 โ ๐บ2๐ป1
โ๐บ1๐ป3 โ โ๐บ2๐บ3๐ป2
14. 45
Masonโs Rule
๐ ๐
๐
๐ ๐ 1
= = ๏ฟฝ ๐๐ฮ๐
๐ ๐ ฮ
๐=1
where
๐ = # of forward paths
๐๐ = gain of the ๐๐กโ forward path
ฮ = 1 โ ฮฃ(loop gains)
+ฮฃ(non-touching loop gains taken two-at-a-time)
โฮฃ(non-touching loop gains taken three-at-a-time)
+ฮฃ(non-touching loop gains taken four-at-a-time)
โฮฃ โฆ
ฮ๐ = ฮ โ ฮฃ(loop gain terms in ฮ that touch the ๐๐กโ forward path)
16. 47
Masonโs Rule โ Example - ฮ๐
๏จ Simplest way to find ฮ๐ terms is to calculate ฮ with the ๐๐กโ
path removed โ must remove nodes as well
๏จ ๐ = 1:
๏จ With forward path 1 removed, there are no loops, so
ฮ1 = 1 โ 0
ฮ1 = 1
17. 48
Masonโs Rule โ Example - ฮ๐
๏จ ๐ = 2:
๏จ Similarly, removing forward path 2 leaves no loops, so
ฮ2 = 1 โ 0
ฮ2 = 1
20. 51
Controller Design โ Preview
๏จ We now have the tools necessary to determine the
transfer function of closed-loop feedback systems
๏จ Letโs take a closer look at how feedback can help us
achieve a desired response
๐ Just a preview โ this is the objective of the whole course
๏จ Consider a simple first-order system
๏จ A single real pole at ๐ = โ2
๐๐
๐
๐ ๐
๐
21. 52
Open-Loop Step Response
๏จ This system
exhibits the
expected first-
order step
response
๐ No overshoot or
ringing
22. 53
Add Feedback
๏จ Now letโs enclose the system in a feedback loop
๏จ Add controller block with transfer function ๐ท ๐
๏จ Closed-loop transfer function becomes:
๐ ๐
๐ท ๐
1
1 + ๐ท ๐
1
๐ + 2
= ๐ + 2 =
๐ท ๐
๐ + 2 + ๐ท ๐
๏จ Clearly the addition of feedback and the controller
changes the transfer function โ but how?
๐ Letโs consider a couple of example cases for ๐ท ๐
23. 54
Add Feedback
๏จ First, consider a simple gain block for the controller
๐ ๐
๏จ Error signal, ๐ธ ๐ , amplified by a constant gain, ๐พ๐ถ
๐ A proportional controller, with gain ๐พ๐ถ
๏จ Now, the closed-loop transfer function is:
๐พ๐ถ
1 +
๐พ๐ถ
๐ + 2
= ๐ + 2 =
๐พ๐ถ
๐ + 2 + ๐พ๐ถ
๏จ A single real pole at ๐ = โ 2 + ๐พ๐ถ
๐ Pole moved to a higher frequency
๐ A faster response
24. 55
Open-Loop Step Response
๏จ As feedback gain
increases:
๐ Pole moves to a
higher frequency
๐ Response gets
faster
25. 56
First-Order Controller
๏จ Next, allow the controller to have some dynamics of its own
๏จ Now the controller is a first-order block with gain ๐พ๐ถ and a pole at
๐ = โ๐
๏จ This yields the following closed-loop transfer function:
๐ ๐
๐พ๐ถ 1
1 +
๐พ๐ถ 1
๐ + ๐ ๐ + 2
=
๐ + ๐ ๐ + 2
=
๐พ๐ถ
๐ 2 + 2 + ๐ ๐ + 2๐ + ๐พ๐ถ
๏จ The closed-loop system is now second-order
๐ One pole from the plant
๐ One pole from the controller
26. 57
First-Order Controller
๏จ Two closed-loop poles:
๐ 1,2 = โ
๐ + 2
ยฑ
2 2
๐2 โ 4๐ + 4 โ 4๐พ๐ถ
๏จ Pole locations determined by ๐ and ๐พ๐ถ
๐ Controller parameters โ we have control over these
๐ Design the controller to place the poles where we want them
๏จ So, where do we want them?
๐ Design to performance specifications
๐ Risetime, overshoot, settling time, etc.
๐ ๐ =
๐พ๐ถ
๐ 2 + 2 + ๐ ๐ + 2๐ + ๐พ๐ถ
27. 58
Design to Specifications
๏จ The second-order closed-loop transfer function
๐ ๐ =
can be expressed as
๐พ๐ถ
๐ 2 + 2 + ๐ ๐ + 2๐ + ๐พ๐ถ
๐ ๐ =
๐พ๐ถ
๐ 2 + 2๐๐๐๐๐ + ๐2
=
๐ ๐
๐พ๐ถ
๐ 2 + 2๐๐ + ๐2
๏จ Letโs say we want a closed-loop response that satisfies the
following specifications:
๐ %๐๐ โค 5%
๐ ๐ก๐ โค 600 ๐๐ ๐๐
๏จ Use %๐๐ and ๐ก๐ specs to determine values of ๐
๐and ๐
๐ Then use ๐
๐
and ๐ to determine ๐พ๐ถ and ๐
28. 59
Determine ๐
๐
from Specifications
๏จ Overshoot and damping ratio, ๐
๐
, are related as
follows:
๐
๐=
โ ln ๐๐
๐2 + ln2 ๐๐
๏จ The requirement is %๐๐ โค 5%, so
โ ln 0.05
๐2 + ln2 0.05
๐
๐โฅ = 0.69
๏จ Allowing some margin, set ๐
๐= 0.75
29. 60
Determine ๐ from Specifications
๐ก๐ โ
๏จ Settling time (ยฑ1%) can be approximated from ๐ as
4.6
๐
๏จ The requirement is ๐ก๐ โค 600 ๐๐ ๐๐
๏จ Allowing for some margin, design for ๐ก๐ = 500 ๐๐ ๐๐
๐ก๐ โ
4.6
๐
= 500 ๐๐ ๐๐ โ
4.6
๐ =
500 ๐๐ ๐๐
which gives
๐ = 9.2
๐๐๐
๐ ๐๐
๏จ We can then calculate the natural frequency from ๐
๐
and ๐
๐ 9.2 ๐๐๐
๐๐ =
๐
๐
=
0.75
= 12.27
๐ ๐๐
30. 61
Determine Controller Parameters from ๐ and ๐๐
๏จ The characteristic polynomial is
๐
๐ 2 + 2 + ๐ ๐ + 2๐ + ๐พ๐ถ = ๐ 2 + 2๐๐ + ๐2
๏จ Equating coefficients to solve for ๐:
2 + ๐ = 2๐ = 18.4
๐ = 16.4
and ๐พ๐:
๐
2๐ + ๐พ๐ถ = ๐2 = 12.27 2 = 150.5
โ 118
๐ท ๐ =
๐พ๐ถ = 150.5 โ 2 โ 16.4 = 117.7
๐พ๐ = 118
๏จ The controller transfer function is
118
๐ + 16.4
31. 62
Closed-Loop Poles
๏จ Closed-loop system
is now second order
๏จ Controller designed
to place the two
closed-loop poles at
desirable locations:
๐ ๐ 1,2 = โ9.2 ยฑ ๐๐๐.13
๐ ๐
๐= 0.75
๐ ๐๐ = 12.3
Controller
pole
Plant
pole
32. 63
Closed-Loop Step Response
๏จ Closed-loop step
response satisfies
the specifications
๏จ Approximations
were used
๐ If requirements not
met - iterate