You are on page 1of 129

Frequency Response Analysis

Assoc. Prof. Enver Tatlicioglu


Department of Electrical & Electronics Engineering
Izmir Institute of Technology

Chapter 8

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

1 / 129

Frequency response analysis


Advantages and disadvantages of root locus design approach

Advantages of root locus design approach:


Good indicator of transient response.
Explicitly shows location of closedloop poles.
Tradeoffs are clear.
Disadvantages of root locus design approach:
Requires the transfer function of the plant to be known.
Difficult to infer all performance values.
Hard to extract steady state response (for sinusoidal inputs).
Frequency response methods can be used to supplement root locus:
Can infer performance and stability from same plot.
Can use measured data when no model is available.
Design process is independent of system order (# poles).
Time delays are handled correctly (exp (s )).
Graphical techniques (analysis/synthesis) are quite simple.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

2 / 129

Frequency response
We want to know how a linear system responds to a sinusoidal input
in steady state.
Consider the system
Y (s)
= G (s) .
U (s)
Consider the input
u (t) = u0 sin (t) U (s) =

u0
.
+ 2

s2

For zero initial conditions, we obtain


Y (s) = G (s)

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

u0
.
+ 2

s2

EE362 Feedback Control Systems

Chapter 8

3 / 129

Frequency response
After doing a partial fraction expansion (assume distinct roots)
1
n
0
0
+ +
+
+
s a1
s an
s + j s j
y (t) = 1 exp (a1 t) + + n exp (an t) +2 |0 | sin (t + )
{z
}
|

Y (s) =

if stable, these decay to zero

from which, we have

yss

= 2 |0 | sin (t + )
= Au0 sin (t + )

where A is the amplitude gain and is the phase shift.


Important LTI system fact:
If the input to a stable LTI system is a sinusoid, then the steady state
output is a sinusoid of the same frequency but with different
amplitude and phase.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

4 / 129

Frequency response
Notice that, 0 is found using partial fraction expansion method, and
A and may be found from 0 as
A = |G (s)| |s=j

and

= |G (j)|
q
=
[R (G (j))]2 + [I (G (j))]2
= G (s) |s=j

= G (j)


I (G (j))
.
= tan1
R (G (j))

Transfer function along jaxis tells us response to a sinusoid.


But, it also tells us about the stability since jaxis is the stability
boundary!
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

5 / 129

Plotting frequency response


There are two common ways to plot a frequency response (i.e.,
plotting the magnitude and phase for all frequencies).
Consider the following simple circuit.

The transfer function is found as


G (s) =

1
.
1 + RCs

After letting RC = 1 and setting s = j, we obtain


G (j) =
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

1
.
1 + j

EE362 Feedback Control Systems

Chapter 8

6 / 129

Plotting frequency response


Plot method #1: Polar plot in splane

Evaluate G (j) at each frequency for 0 +.

Result will be a complex number in the form a + jb or A exp (j) for


each frequency.
Plot each point a + jb or A exp (j) on the complex plane obtained
for each frequency.
Result is a polar plot called a Nyquist plot.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

7 / 129

Plotting frequency response


Plot method #1: Polar plot in splane

For the RC circuit example, we obtain:

0.0
0.5
1.0
1.5
2.0
3.0
5.0
10.0
+

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

G (j)
1.0000.0
0.894 26.6
0.707 45.0
0.555 56.3
0.447 63.4
0.316 71.6
0.196 78.7
0.100 84.3
0.000 90.0

EE362 Feedback Control Systems

Chapter 8

8 / 129

Plotting frequency response


Plot method #1: Polar plot in splane

The polar plot is parametric in , so it is hard to read the frequency


response for a specific frequency from the plot.

Figure: Polar plot in splane

We will later see that the polar plot will help us determine stability
properties of the plant and the closedloop system.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

9 / 129

Plotting frequency response


Plot method #2: Magnitude and phase plots

We separate magnitude and phase information from rational


polynomials that are functions of j
magnitude of numerator
Magnitude =
magnitude of denominator
q
[R (num)]2 + [I (num)]2
= q
[R (den)]2 + [I (den)]2
and

Phase = phase of numerator phase of denominator






I (num)
I (den)
1
1
= tan
tan
.
R (num)
R (den)

For the RC circuit example, we obtain


1
tan1 () .
G (j) =
2
1+
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

10 / 129

Plotting frequency response


Plot method #2: Magnitude and phase plots

We can plot the data by separating magnitude and phase plots.

Figure: Magnitude versus frequency plot for the RC circuit example

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

11 / 129

Plotting frequency response


Plot method #2: Magnitude and phase plots

Figure: Phase versus frequency plot for the RC circuit example

These plots are in natural scale, but usually a loglog plot is preferred.
These are called Bode plots or Bode diagrams.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

12 / 129

Bode plots

Simplest way to display the frequency response of a rational


polynomial transfer function is to use Bode plots.
Bode magnitude plot is the plot of logarithmic |G (j)| versus
logarithmic .
Bode phase plot is the plot of G (j) versus logarithmic .
Since
log10

ab
cd

= log10 (a) + log10 (b) log10 (c) log10 (d)

the factors of the transfer function can be split up and their


contributions can be evaluated separately.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

13 / 129

Bode plots
Example

For
G (s) =

s +1
j + 1
G (s) |s=j = G (j) =
s/10 + 1
j/10 + 1

we obtain
|G (j)| =
and thus

|j + 1|
|j/10 + 1|

p
log10 |G (j)| = log10 1 + 2 log10

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

1+

 2
10

Chapter 8

14 / 129

Bode plots
Consider the following expression
s
log10

For n ,
log10

1+

1+

2

2

log10 1 = 0.

For n ,
log10

1+

2

log10

So, we obtain two straight lines on log scale intersecting at = n .


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

15 / 129

Bode plots

Typically, we plot 20 log10 |G (j)|; which is in dB.


A transfer function is made up of:
first order zeros and poles,
complex zeros and poles,
constant gains,
delays.
We will see how to make straight line approximations for magnitude
and phase plots, and combine them to form Bode plots.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

Figure:

EE362 Feedback Control Systems

Chapter 8

16 / 129

Bode magnitude plots


Constant gain

We have 20 log10 |K |dB which is not a function of frequency.


So, it is a horizontal straight line.
If |K | < 1, then 20 log10 |K | is negative.
If |K | > 1, then 20 log10 |K | is positive.

Figure: Bode magnitude plot for constant gain


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

17 / 129

Bode magnitude plots


Zero at origin

For a zero at the origin, G (s) = s.


So, we obtain
20 log10 |G (j)| = 20 log10 |j| = 20 log10 dB.

Figure: Bode magnitude plot for a zero at origin


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

18 / 129

Bode magnitude plots


Pole at origin

For a pole at the origin, G (s) = 1/s.


So, we obtain
20 log10 |G (j)| = 20 log10 |j| = 20 log10 dB.

Figure: Bode magnitude plot for a pole at origin


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

19 / 129

Bode magnitude plots


Zero or pole at origin

For a zero or pole at the origin,


Both are straight lines with a slope of 20dB per decade of frequency.
Line intersects axis at = 1 (since log10 1 = 0).

For an nth order pole or zero at the origin, we obtain


20 log10 |(j)n | = 20 log10 n = 20n log10
which are still straight lines,
which still intersect axis at = 1,
which have a slope of 20n per decade.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

20 / 129

Bode magnitude plots


Zero on real axis, but not at origin

For a zero on the real axis (LHP or RHP), the standard Bode form is
G (s) =

s
1
n

which ensures unity DC gain.


If we are given something like
G (s) = s n
then we factor the DC gain n as
G (s) = n


s
1
n

and draw the constant gain n separately from the term (s/n 1).
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

21 / 129

Bode magnitude plots


Zero on real axis, but not at origin

An LHP or RHP zero has the following standard Bode form


 

s
1 G (j) = j
1.
G (s) =
n
n
So, we obtain

20 log10 |G (j)| = 20 log10


For n ,
20 log10

20 log10

For n ,

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

1+

1+

1+

2

20 log10

2

20 log10

EE362 Feedback Control Systems

2

1 = 0.

.
Chapter 8

22 / 129

Bode magnitude plots


Zero on real axis, but not at origin

So, we obtain two straight lines on log scale intersecting at = n .

Figure: Bode magnitude plot for a zero on real axis, but not at origin

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

23 / 129

Bode magnitude plots


Pole on real axis, but not at origin

For a pole on the real axis (LHP or RHP), standard Bode form is
G (s) =

  
1
1

s
G (j) = j
1
.
1
n
n

So, we obtain
s

1+

2

1+

2

20 log10 |G (j)| = 20 log10


= 20 log10

which is same as zero on real axis, but not at origin except for a
minus sign.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

24 / 129

Bode magnitude plots


Pole on real axis, but not at origin

So, we obtain two straight lines on log scale intersecting at = n .

Figure: Bode magnitude plot for a pole on real axis, but not at origin

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

25 / 129

Bode magnitude plots


Complex zero pair

For a complex zero pair (LHP or RHP), standard Bode form is


 2
 
s
s
G (s) =
2
+1
n
n
which ensures unity DC gain.
If we are given something like
G (s) = s 2 2n s + n2

then we factor the DC gain n2 as


" 
#
 
s 2
s
2
G (s) = n
+1
2
n
n
and
the constant gain n2 separately from the term
 draw
 
2
s
s

2
n
n + 1 .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

26 / 129

Bode magnitude plots


Complex zero pair

Complex zero pair does not lend themselves very well to straight line
approximation.
We write complex zero pair (LHP or RHP) as
" 
#
 
s 2
s
G (s) =
2
+1 .
n
n
At = n , magnitude value = 20 log10 (2) dB.
For n , magnitude value 0 dB.

For n , magnitude value 40 dB.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

27 / 129

Bode magnitude plots


Complex zero pair

Figure: Bode magnitude plots for complex zero pairs

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

28 / 129

Bode magnitude plots


Complex pole pair

We write complex pole pair (LHP or RHP) as


G (s) =

"

s
n

2

s
n

#1

+1

At = n , magnitude value = 20 log10 (2) dB.


For n , magnitude value 0 dB.

For n , magnitude value 40 dB.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

29 / 129

Bode magnitude plots


Complex pole pair

Figure: Bode magnitude plots for complex pole pairs

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

30 / 129

Bode magnitude plots


Time delay

The transfer function for time delay is given by


G (s) = exp (s ) |G (j)| = |exp (j )| = 1.
So, we obtain
20 log10 |G (j)| = 20 log10 1 = 0 dB
which obviously does not change the magnitude response.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

31 / 129

Bode magnitude plots


Example

Draw the Bode magnitude response for


G (s) =

2000 (s + 0.5)
.
s (s + 10) (s + 50)

First, rewrite in standard Bode form


G (s) =
=

2000 0.5
10 50 s
2

s
10

G (j) = 2
j
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

+1


1
.
s
50 + 1

s
0.5 +

+1

Next, make the change s = j

s
10

s
0.5 +

s
50


+1


+1

.
+ 1 j
+
1
50

j
10


1

j
0.5

EE362 Feedback Control Systems

Chapter 8

32 / 129

Bode magnitude plots


Example

The components of the Bode magnitude plot may be obtained


separately.
For the DC gain of 2:
20 log10 2 = 6 dB.
For the real zero not at origin (n = 0.5):
For 0.5, a straight line with a slope of 0 dB.

For 0.5, a straight line with a slope of 20 dB.

So, we obtain two straight lines on log scale intersecting at = 0.5.


For the pole at origin:
A straight line with a slope of 20 dB per decade of frequency and
intersecting the axis at = 1.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

33 / 129

Bode magnitude plots


Example

Two real poles not at origin (n = 10 and n = 50):


For n = 10:
For 10, a straight line with a slope of 0 dB.

For 10, a straight line with a slope of 20 dB.

So, we obtain two straight lines on log scale intersecting at = 10.


For n = 50:
For 50, a straight line with a slope of 0 dB.

For 50, a straight line with a slope of 20 dB.

So, we obtain two straight lines on log scale intersecting at = 50.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

34 / 129

Bode magnitude plots


Example

Figure: Bode magnitude plot for G (s) =


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

2000(s+0.5)
s(s+10)(s+50)
Chapter 8

35 / 129

Bode magnitude plots


Steady state errors

Recall our discussion of steady state errors to step/ramp/parabolic


inputs versus system type.
Consider a unity feedback system.
If the openloop plant transfer function has N poles at s = 0, then
the system is type N.
Also recall that,
Kp is the error constant for type 0 systems,
Kv is the error constant for type 1 systems,
Ka is the error constant for type 2 systems.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

36 / 129

Bode magnitude plots


Steady state errors

For a unity feedback system, Kp = lims0 G (s) is nonzero for a type


0 system.
At low frequencies, a type 0 system will have G (s) = Kp .
We can read this off from the Bode magnitude plot directly.
Kp is equal to the horizontal y intercept at low frequencies.
So, we will have ess = 1/ (1 + Kp ) for a step input.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

37 / 129

Bode magnitude plots


Example

Consider the below Bode magnitude diagram:

Horizontal as 0, so we know this system is type 0.

Intercept the y axis at 6 dB, so Kp = 6 dB or Kp = 2 in linear units.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

38 / 129

Bode magnitude plots


Steady state errors

For a unity feedback system, Kv = lims0 sG (s) is nonzero for a


type 1 system.
At low frequencies, a type 1 system will have G (s) Kv /s.

At low frequencies, |G (j)| Kv / results in a slope of 20


dB/decade.
Use the above approximation to extend the low frequency asymptote
to = 1.
The asymptote (not the original |G (j)|) evaluated at = 1 is Kv .
So, we will have ess = 1/Kv for a ramp input.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

39 / 129

Bode magnitude plots


Example

Consider the below Bode magnitude diagram:

Slope is 20 dB/decade as 0, so we know the system is type 1.

Extend the slope at low frequency to = 1, and read the magnitude


as 20 dB, so Kv = 20 dB or Kv = 10 in linear units.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

40 / 129

Bode magnitude plots


Steady state errors

For a unity feedback system, Ka = lims0 s 2 G (s) is nonzero for a


type 2 system.
At low frequencies, a type 2 system will have G (s) Ka /s 2 .

At low frequencies, |G (j)| Ka / 2 results in a slope of 40


dB/decade.
Use the above approximation to extend the low frequency asymptote
to = 1.
The asymptote (not the original |G (j)|) evaluated at = 1 is Ka .
So, we will have ess = 1/Ka for a parabolic input.
Similar for higher order systems.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

41 / 129

Bode phase plots

Bode plots we have seen so far consist of only the magnitude plots.
There are also phase plots.
They differ depending on whether the dynamics (the poles and the
zeros) are at the RHP or the LHP.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

42 / 129

Bode phase plots


Finding the phase of a complex number

Plot the location of the number as a vector in the complex plane.


Use trigonometry to find the phase.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

43 / 129

Bode phase plots


Finding the phase of a complex number

Finding the phase of a complex function of is the same as finding


the phase of a complex number.
For complex numbers with positive real part,


I (Complex number)
1
(Complex number) = tan
.
R (Complex number)
For complex numbers with negative real part,


I (Complex number)
1
.
(Complex number) = 180 + tan
R (Complex number)
Note that

ab
cd

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

= (a) + (b) (c) (d) .


EE362 Feedback Control Systems

Chapter 8

44 / 129

Bode phase plots


Constant gain

For G (s) = K ,
(K ) =

0
for K 0
180 for K < 0

which has a constant phase of 0 or 180 .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

45 / 129

Bode phase plots


Zero or pole at origin

For a zero at origin, we have G (s) = s, so


G (j) = j = 90 .
For a pole at origin, we have G (s) = 1/s, so
G (j) =

j
1
1
= = 90 .
j

So both of them have constant phase of 90 .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

46 / 129

Bode phase plots


Real LHP zero



For a real LHP zero, we have G (s) = si + 1 .
So,


 
j

1
+ 1 = tan
G (j) =
.
i
i

Figure: Bode phase plot of a real LHP zero


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

47 / 129

Bode phase plots


Real LHP pole

For a real LHP pole, we have G (s) =


So,
G (j) = (1)

s
i

1 
.
+1


 
j

1
+ 1 = tan
.
i
i

Figure: Bode phase plot of a real LHP pole


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

48 / 129

Bode phase plots


Real RHP zero



For a real RHP zero, we have G (s) = si 1 .
So,




 
j

1
G (j) =
1 = 180 +tan

= 180 tan
.
i
i
i

Figure: Bode phase plot of a real RHP zero


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

49 / 129

Bode phase plots


Real RHP pole

For a real RHP pole, we have G (s) =


So,
G (j) = (1)

s
i

1 
.
1





j

1
1 = 180 + tan

.
i
i

Figure: Bode phase plot of a real RHP pole


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

50 / 129

Bode phase plots


Complex LHP zero pair

Complex LHP zero pair causes phase to go from 0 to 180 .

Figure: Bode phase plots for complex LHP zero pairs


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

51 / 129

Bode phase plots


Complex LHP pole pair

Complex LHP pole pair causes phase to go from 0 to 180 .

Figure: Bode phase plots for complex LHP pole pairs


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

52 / 129

Bode phase plots


Complex RHP zero pair

Complex RHP zero pair causes phase to go from 360 to 180 .

Figure: Bode phase plots for complex RHP zero pairs


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

53 / 129

Bode phase plots


Complex RHP pole pair

Complex RHP pole pair causes phase to go from 360 to 180 .

Figure: Bode phase plots for complex RHP pole pairs


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

54 / 129

Bode phase plots


Time delay

The transfer function for time delay is given by


so

G (s) = exp (s ) G (j) = exp (j ) = 1


G (j) = in radians = 56.3 in degrees.

Figure: Bode phase plot for time delay

Notice that, a line becomes a curve in log scale.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

55 / 129

Bode phase plots


Example

Draw the Bode phase response for


G (s) =

2000 (s + 0.5)
.
s (s + 10) (s + 50)

First, rewrite in standard Bode form


G (s) =
=

2000 0.5
10 50 s
2

s
10

G (j) = 2
j
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

+1


1
.
s
50 + 1

s
0.5 +

+1

Next, make the change s = j

s
10

s
0.5 +

s
50


+1


+1

.
+ 1 j
+
1
50

j
10


1

j
0.5

EE362 Feedback Control Systems

Chapter 8

56 / 129

Bode phase plots


Example

The components of the phase plot can be obtained separately.


For the DC gain of 2:
Zero phase contribution.
One real LHP zero (i = 0.5):
Phase changes from 0 at 0.1i = 0.05 to 90 at 10i = 5.
Pole at origin:
Constant phase contribution of 90 .

Two real LHP poles (i = 10 and i = 50):


For i = 10:

Phase changes from 0 at 0.1i = 1 to 90 at 10i = 100.


For i = 50:

Phase changes from 0 at 0.1i = 5 to 90 at 10i = 500.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

57 / 129

Bode phase plots


Example

Figure: Bode phase plot for G (s) =


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

2000(s+0.5)
s(s+10)(s+50)
Chapter 8

58 / 129

Nonminimum phase systems


A system is called nonminimum phase if it has pole(s) or zero(s) in
the RHP.
Consider
s+1
G1 (s) = 10 s+10
s1
G2 (s) = 10 s+10

o
o

zero at
pole at
zero at
pole at


1
10 
+1
10

minimum
phase
nonminimum
phase

The magnitude responses of these two systems are found as

|j+1|
+1
|G1 (j)| = 10 |j+10|
= 10 2 +100
|j1|
=
|G2 (j)| = 10 |j+10|

2 +1
10 2 +100

which are the same.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

59 / 129

Nonminimum phase systems

s+1
s1
Figure: Bode magnitude plot for G1 (s) = 10 s+10
and G2 (s) = 10 s+10

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

60 / 129

Nonminimum phase systems


However, the phase responses are very different:

s+1
s1
Figure: Bode phase plots for G1 (s) = 10 s+10
and G2 (s) = 10 s+10

Note that, the change in phase of G1 is much smaller than the change
of phase in G2 .
Hence G1 is minimum phase and G2 is nonminimum phase.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

61 / 129

Nonminimum phase systems

Nonminimum phase is usually associated with delay.


After writing G2 (s) in terms of G1 (s), we obtain the following
expression
s 1
G2 (s) = G1 (s)
+ 1}
|s {z
Delay

s1
where we notice that s+1
is very similar to a first order Pade
approximation of a delay.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

62 / 129

Bode plots
Stability revisited

If we know the closedloop transfer function of a system in rational


polynomial form, we can use Routh test to find stable ranges for K.
What if we only have openloop frequency response?
Consider the following system (for now, we assume that we know the
transfer function of the system, so that we can plot the root locus)

where
D (s) = K , G (s) =

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

1
s (s + 1)2

EE362 Feedback Control Systems

Chapter 8

63 / 129

Bode plots
Stability revisited

Root Locus
3

Imaginary Axis

3
5

Real Axis

Figure: Root locus plot for G (s) =


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

1
s(s+1)2

EE362 Feedback Control Systems

with D (s) = K
Chapter 8

64 / 129

Bode plots
Stability revisited

We can find that there is neutral stability at K = 2 (i.e., the system is


stable for K < 2 and unstable for K > 2).
Recall that a point is on the root locus if |KG (s)| = 1 and
G (s) = 180 .

If system is neutrally stable, jaxis will have a point (or points)


where |KG (j)| = 1 and G (j) = 180 .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

65 / 129

Bode plots
Stability revisited

Consider the Bode magnitude plots of KG (s) for K = 0.1, K = 2 and


K = 10.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

66 / 129

Bode plots
Stability revisited

Consider the Bode phase plot of KG (s).

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

67 / 129

Bode plots
Stability revisited

A neutral stability condition from Bode plots is obtained when


|KG (j0 )| = 1 and G (j0 ) = 180 at the same frequency 0 .

In this case, increasing K results in instability (which also means that


stability is obtained when |KG (j)| < 1 at G (j0 ) = 180 ).
In some cases, decreasing K results in instability (which also means
that stability is obtained when |KG (j)| > 1 at G (j0 ) = 180 ).
We can find neutral stability point on Bode plot, but we dont (yet)
have a way of determining if the system is stable or not.
Nyquist found a frequency domain method to do so.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

68 / 129

Nyquist stability
If the poles of a closedloop transfer function are in RHP, then the
system is unstable.
Nyquist found a way to count the closedloop poles in RHP.
If the count is greater than zero, then the system is unstable.
Idea:
First, find a way to count closedloop poles inside a contour.
Second, make the contour equal to the RHP.
Nyquists technique is a graphical method to determine:
system stability,
regions of stability,
margins of stability.
It involves graphing complex functions of s as a polar plot.
Counting is related to complex functional mapping.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

69 / 129

Nyquist stability
Complex functional mapping

Plotting a real function f (x) of a real variable x.

This can be done.


Plotting a complex function F (s) of a complex variable s.

No, this is wrong!


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

70 / 129

Nyquist stability
Complex functional mapping

We must draw mapping of points or lines from splane to F (s)plane.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

71 / 129

Nyquist stability
Complex functional mapping

Map points A, B, C, D by F (s) = 2s + 1.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

72 / 129

Nyquist stability
Complex functional mapping

Map a square contour (closed path) by F (s) = s/ (s + 2).

Idea:
By drawing maps of a specific contour, using a mapping function
related to the plant openloop frequency response, we will be able to
determine closedloop stability of systems.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

73 / 129

Nyquist stability
Mapping function: Poles of the function

When we map a contour containing (encircling) poles and zeros of


the mapping function, this map will give us information about how
many poles and zeros are encircled by the contour.
We practice drawing maps when we know poles and zeros.
Notice that, when we evaluate G (s) at s0
G (s) |s=s0

= G (s0 )

where
=

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

= |G (s0 )| exp (j)

zeros

poles.

EE362 Feedback Control Systems

Chapter 8

74 / 129

Nyquist stability
Mapping function: Poles of the function

Consider the below plots

where there are no zeros or poles inside the contour c1 .


The phase increases and decreases, but never undergoes a net
change of 360 (i.e., F (c1 ) does not encircle the origin).

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

75 / 129

Nyquist stability
Mapping function: Poles of the function

Consider the below plots

where there is one pole inside the contour c2 .


Resulting map undergoes 360 net phase change (i.e., F (c2 ) encircles
the origin).

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

76 / 129

Nyquist stability
Mapping function: Poles of the function

Consider the below plots

where there are two poles inside the contour c3 .


The map F (c3 ) encircles the origin twice.
These examples give heuristic evidence of the general rule known as
Cauchys theorem.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

77 / 129

Nyquist stability
Cauchys Theorem

Theorem
Let F (s) be the ratio of two polynomials in s.
Let the closed curve C in the splane be mapped into the complex plane
through the mapping F (s).
If the curve C does not pass through any zeros or poles of F (s) as it is
traversed in the CW direction, the corresponding map in the F (s)plane
encircles the origin N = Z P times in the CW direction, where
Z

= # of zeros of F (s) in C ,

P = # of poles of F (s) in C .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

78 / 129

Nyquist stability
Cauchys Theorem

Consider the system

where the transfer function is


T (s) =

D (s) G (s)
.
1 + D (s) G (s) H (s)

For closedloop stability, we want no poles of T (s) to be in RHP.


This is equivalent to saying, we want no zeros of 1 + D (s) G (s) H (s)
to be in RHP.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

79 / 129

Nyquist stability
Cauchys Theorem

Let F (s) = 1 + D (s) G (s) H (s), and count the number of zeros of
F (s) in RHP using Cauchys theorem where the contour is the entire
RHP.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

80 / 129

Nyquist stability
Cauchys Theorem

Note that, for F (s) = 1 + D (s) G (s) H (s), from Cauchys Theorem,
N = # of encirclements of origin.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

81 / 129

Nyquist stability
Nyquist criterion

After a simple modification, Nyquist proposed


F (s) = D (s) G (s) H (s) and N = # of encirclements of 1.

The Nyquist criterion simplifies Cauchys criterion for feedback


D(s)G (s)
systems of T (s) = 1+D(s)G
(s)H(s) form.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

82 / 129

Nyquist stability test


We think of Nyquist path as four parts

I. The origin (which is sometimes a special case as we will see later),


II. +jaxis which is the frequency response of the openloop system
(we will plot it in polar form),
III. For physical systems = 0,
IV. jaxis (i.e., complex conjugate of the region in part II).
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

83 / 129

Nyquist stability test


So, for most physical systems, the Nyquist plot, used to determine
closedloop stability, is basicly a polar plot of loop frequency response
D (j) G (j) H (j).
We dont even need a mathematical model of the system. Measured
data of G (j) combined with known D (j) and H (j) are enough
to determine the closedloop stability.
N = # of encirclements of 1 for F (s) = D (s) G (s) H (s).

P = # of poles of 1 + F (s) in RHP = # of openloop unstable poles


(assuming that H (s) is stable which is a reasonable assumption).
Z = # of zeros of 1 + F (s) in RHP = # of closedloop unstable
poles
Z = N + P.
The system is stable iff Z = 0.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

84 / 129

Nyquist stability test


You need to be careful when counting encirclements.
A safe way is to draw a line from 1 in any direction, and to count #
of crossings of line and diagram, and
N = # of clockwise crossings # of counterclockwise crossings.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

85 / 129

Enhanced Nyquist stability test

Note that, changing the gain K of F (s) magnifies the entire plot.
This magnification property allows an enhanced test to obtain ranges
of K for stability.
For the loop transfer function KD (s) G (s) H (s), N = # of
encirclements of 1/K point when F (s) = D (s) G (s) H (s) where
the rest of test is the same.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

86 / 129

Nyquist stability
Example

Consider the system with


G (s) =

5
(s + 1)2

, D (s) = H (s) = 1.

We have:
I. For s = 0, G (s) = 5.
II. For s = j, G (j) = 5/ (j + 1)2 .
III. For |s| = , G (s) = 0.

IV. For s = j, G (j) = 5/ (j + 1)2 = G (j).

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

87 / 129

Nyquist stability
Example

R (G (j)) I (G (j))
0.0000
5.0000
0.0000
0.0019
4.9999
0.0186
0.0040
4.9998
0.0404
0.0088
4.9988
0.0879
0.0191
4.9945
0.1908
0.0415
4.9742
0.4135
0.0902
4.8797
0.8872
0.1959
4.4590
1.8172
0.4258
2.9333
3.0513

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

R (G (j)) I (G (j))
0.9253
0.2086
2.6856
2.0108
0.5983
0.7906
4.3697
0.2241
0.1082
9.4957
0.0536
0.0114
20.6351
0.0117
0.0011
44.8420
0.0025
0.0001
97.4460
0.0005
0.0000
500.0000 0.0000
0.0000

EE362 Feedback Control Systems

Chapter 8

88 / 129

Nyquist stability
Example

The Nyquist plot is obtained as:

No encirclements of 1: N = 0.
No openloop unstable poles: P = 0.
So, Z = N + P = 0, thus, closedloop system is stable.
No encirclements of 1/K for any K > 0.
So, the system is stable for any K > 0.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

89 / 129

Nyquist stability
Example

Lets confirm by checking Routh array.


Check the stability of a (s) = 1 + KG (s) = s 2 + 2s + 1 + 5K .
The routh array is obtained as
1
s2
1
s
2
s 0 1 + 5K

1 + 5K

so the system is stable for any K > 0.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

90 / 129

Nyquist stability
Example

Consider the system with


G (s) =

50
2

(s + 1) (s + 10)

, D (s) = H (s) = 1.

We have:
I. For s = 0, G (s) = 50/10 = 5.
II. For s = j, G (j) =

50
.
(j+1)2 (j+10)

III. For |s| = , G (s) = 0.

IV. For s = j, G (j) =

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

50
(j+1)2 (j+10)

EE362 Feedback Control Systems

= G (j).

Chapter 8

91 / 129

Nyquist stability
Example

R (G (j)) I (G (j))
0.0
5.0000
0.0000
0.1
4.9053
0.8008
0.2
4.4492
1.8624
0.5
2.4428
3.2725
1.2
0.5621
2.0241
2.9
0.4764
0.1933
7.1
0.0737
0.0262
17.7 0.0046
0.0064
43.7 0.0002
0.0006
100
0.0000
0.0000

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

92 / 129

Nyquist stability
Example

The Nyquist plot is obtained as:

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

93 / 129

Nyquist stability
Example

Note the loop to the left of the origin

so the system is not stable for all K > 0.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

94 / 129

Nyquist stability
Pole at the origin

Consider the system with


G (s) =

1
, D (s) = H (s) = 1.
s ( s + 1)

Note that there is a pole at the origin.


Warning: We cannot blindly follow procedure.
Nyquist path goes through a pole at the origin.
Remember from Cauchys theorem that the path cannot pass directly
through a pole or a zero.
Remember that we want to count the closedloop poles inside a
contour that encompasses the RHP.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

95 / 129

Nyquist stability
Pole at the origin

So, we use a slightly modified Nyquist path

where the bump at the origin makes a detour around the offending
pole.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

96 / 129

Nyquist stability
Pole at the origin

Bump is defined by the curve


s = lim exp (j)
0

where 0 90 .
From above,

G (s) |s= exp(j) =

1
exp (j) ( exp (j) + 1)

where 0 90 .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

97 / 129

Nyquist stability
Pole at the origin

Magnitude when 0 is


lim G (s) |s= exp(j) =

1
1
.
| exp (j) + 1|

Phase when 0 is

lim G (s) |s= exp(j) = ( exp (j) + 1) .

So,
lim G ( exp (j)) lim


1
+

This is an arc of infinite radius, sweeping from 0 to 90+ (a little


more than 90 because of contribution from 1/ ( s + 1) term).
We cannot draw this to scale!
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

98 / 129

Nyquist stability
Pole at the origin

The Nyquist plot is obtained as:

N = # of encirclements of 1: N = 0.

P = # of loop transfer function poles inside the modified contour:


P = 0.
So, Z = N + P = 0, thus, the closedloop system is stable.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

99 / 129

Nyquist stability
Example

Consider the system with


G (s) =

1
, D (s) = H (s) = 1.
s 2 (s + 1)

We will use the modified Nyquist path again.


Near the origin, we will have a bump defined by curve
s = lim exp (j)
0

where 0 90 .
From above,
G (s) |s= exp(j) =

1
2 exp (j2) ( exp (j)

+ 1)

where 0 90 .
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

100 / 129

Nyquist stability
Example

Magnitude when 0 is
lim |G ( exp (j))| =

1
1
2.
2 | exp (j) + 1|

Phase when 0 is
lim G (s) |s= exp(j) = 2 ( exp (j) + 1) 2+ .

So,
lim G ( exp (j)) lim

where

90 .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

1
2+
2

EE362 Feedback Control Systems

Chapter 8

101 / 129

Nyquist stability
Example

This is an arc of infinite radius, sweeping from 0 to 180+ (a little


more than 180 because of contribution from 1/ (s + 1) term).

N = # of encirclements of 1: N = 2.
P = # of loop transfer function poles inside the modified contour:
P = 0.
So, Z = N + P = 2, thus, closedloop system is unstable for K = 1.
In fact, the system is unstable for any K > 0.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

102 / 129

Stability margins
A large fraction of systems to be controlled are stable for small gain
but become unstable if gain is increased beyond a certain point.
The distance between the current (stable) system and an unstable
system is called a stability margin.
We can have a gain margin and a phase margin.
Gain Margin: Factor by which the gain is less than the neutral
stability value.
Gain margin measures How much can we increase the gain of the
loop transfer function D (s) G (s) H (s) and still have a stable
system?
Phase Margin: Phase factor by which phase is greater than neutral
stability value.
Phase margin measures How much delay can we add to the loop
transfer function and still have a stable system?
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

103 / 129

Stability margins
Many Nyquist plots are like this one:

Increasing loop gain magnifies the plot.


GM = 1/(distance between origin and place where Nyquist map
crosses real axis).
If we increase gain, Nyquist map stretches and we may encircle 1.
For a stable system, GM > 1 (linear units) or GM > 0 dB.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

104 / 129

Stability margins
Many Nyquist plots are like this one:

PM = Angle to rotate Nyquist plot to achieve neutral stability =


intersection of Nyquist with circle of radius 1.
If we increase openloop delay, Nyquist map rotates and we may
encircle 1.
For a stable system, PM > 0 .
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

105 / 129

Stability margins
This is usually easier to check on Bode plot, even though derived on
Nyquist plot.
Define gain crossover as frequency where Bode magnitude is 0 dB.
Define phase crossover as frequency where Bode phase is 180 .
PM = Bode phase at gain crossover (180 ).

GM = 1/(Bode gain at phase crossover frequency) if Bode gain is


measured in linear units = (Bode gain at phase crossover frequency)
[dB] if Bode gain measured in dB.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

106 / 129

Stability margins

We can also determine stability as K changes.


Instead of defining gain crossover where |G (j)| = 1, use the
frequency |KG (j)| = 1.
You need to be careful using this test.

It works only if the system is minimum phase.


Nyquist plot is the safest way for this test.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

107 / 129

Bode compensator design

The frequency response methods we have seen so far largely tell us


about stability and stability margins of a closedloop system based on
openloop response.
Now, we look at frequency response based design methods which
primarily aim at improving stability margins.
Also, given the relationship between PM and performance, we have
some idea of transient response as well.
Start thinking of Bode magnitude and Bode phase plots as lego to
make the frequency response we want.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

108 / 129

Bode compensator design


P control

Consider the system

where
D (s) = K .
Similar to the root locus method, we use the characteristic equation
in the form
cl (s) = 1 + KG (s)
to examine the stability and the performance of Bode plot parameters.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

109 / 129

Bode compensator design


P control

Consider the Bode plots

Gain crossover frequency is denoted by p and is equal to the


frequency where |KG (j)| = 0 dB.
Phase crossover frequency is denoted by and is equal to the
frequency where KG (j) = 180 .
Recall that PM = 180 + G (jp ).
If PM < 0 , the closedloop system is unstable.
Recall that GM = 1/ |KG (j )| (linear units).
If GM < 1 (linear units), the closedloop system is unstable.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

110 / 129

Bode compensator design


P control

A bonus of computing PM from the openloop frequency response


graph is that it can help us predict the closedloop performance of
the system.
The Bode design parameters are related to the root locus design
parameters according to
qp
p = n
1 + 4 4 2 2

PM = tan1 qp
1 + 4 4 2 2

where n is the natural frequency and is the damping ratio.


The phase margin is also approximately related to the damping ratio
as follows
PM/100 for PM 60 .
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

111 / 129

Bode compensator design


P control

Remember that is a measure of relative stability:

Hence, PM is also a measure of relative stability.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

112 / 129

Bode compensator design


Example

Consider the system

where
G (s) =

s
2

1
 , D (s) = K .
+1

Find K so that PM = 45 for the closedloop system.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

113 / 129

Bode compensator design


Example

1.
2.
3.
4.
5.
6.

7.

Outline of the solution:


Find the closedloop denominator in the form cl (s) = 1 + KG (s).
Plot |KG (j)| and G (j) by assuming K = 1.
From the given (desired) PM, calculate the desired G (jp ) (i.e.,
desired G (jp ) =PM180 ).
From the desired G (jp ) and the G (j) plot, find the desired p .
Raise or lower the magnitude plot so that it crosses the 0 dB level at
the desired value of p .
Find the distance between the desired magnitude plot and the original
magnitude plot at the desired value of p (i.e., desired + original =
shift in dB).
Calculate K by using the following formula derived from the
magnitude condition
K = 10(shift/20) .

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

114 / 129

Bode compensator design


Example

1. The closedloop denominator is in the root locus form


cl (s) = 1 + K

2. The Bode plot for KG (s) when K = 1:

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

s
2

1
.
+1

EE362 Feedback Control Systems

Chapter 8

115 / 129

Bode compensator design


Example

3. Desired G (jp ) = PM180 = 45 180 = 135 .


4. The desired p is found from the desired G (jp ) and the G (j)
plot.

Also note that, the actual PM = 180 118 = 62 for K = 1.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

116 / 129

Bode compensator design


Example

5. Raise or lower magnitude plot by changing K :

6. The difference between the desired magnitude plot and the original
magnitude plot is
shift = desired + original = 0 9 = 9 dB.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

117 / 129

Bode compensator design


Example

7. Calculate K by using the following formula derived from the


magnitude condition
K = 10(shift/20) = 10(9/20) = 2.8.
Note that, for this problem, the below specifications are equivalent:
PM = 45 ,

= 0.45,
Desired closedloop poles are at 1 j2.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

118 / 129

Bode compensator design


Example

Consider the system

where
G (s) =

(s + 1)

s
10

1

+1

s
100

 , D (s) = K .
+1

Find K so that PM = 45 for the closedloop system.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

119 / 129

Bode compensator design


Example

1. The closedloop denominator is in the root locus form


cl (s) = 1 + K

(s + 1)

s
10

2. The Bode plot for KG (s) when K = 1:

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

1

+1

EE362 Feedback Control Systems

s
100

.
+1

Chapter 8

120 / 129

Bode compensator design


Example

3. Desired G (jp ) = PM180 = 45 180 = 135 .


4. The desired p is found from the desired G (jp ) and the G (j)
plot.

Also note that, the actual PM = 180 0 = 180 for K = 1.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

121 / 129

Bode compensator design


Example

5. Raise or lower magnitude plot by changing K :

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

122 / 129

Bode compensator design


Example

6. The difference between the desired magnitude plot and the original
magnitude plot is
shift = desired + original = 0 20 = 20 dB.
7. Calculate K by using the following formula derived from the
magnitude condition
K = 10(shift/20) = 10(20/20) = 10.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

123 / 129

Bode compensator design


PD control

Controller has the form D (s) = K (1 + TD s).


We have seen from root locus that this has a stabilizing effect.
Magnitude and phase effects of a PD controller:

PD control increases phase for frequencies over 0.1/TD .


Locate 1/TD < crossover so that phase margin at crossover is better.
Problem: Magnitude response continues to increase as frequency
increases. This amplifies high frequency sensor noise.
Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

124 / 129

Bode compensator design


Lead compensator

Lead compensator has the form D (s) = K

Ts+1
Ts+1

where < 1.

This is an approximate of PD control for frequencies up to


= 1/ (T ).
Magnitude and phase effects of a lead compensator:

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

125 / 129

Bode compensator design


Lead compensator

The phase contributed at frequency is


= tan1 (T ) tan1 (T ) .
If we need more phase improvement, we can use a doublelead
compensator


Ts + 1 2
D (s) = K
.
Ts + 1
Need to compromise between good phase margin and good sensor
noise rejection at high frequency.

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

126 / 129

Bode compensator design


PI control

In many problems, it is important to keep bandwidth low, and also


reduce steady state error.


PI controller has the form D (s) = K 1 + T1I s .
Magnitude and phase effects of a PI controller:

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

127 / 129

Bode compensator design


PI control

Infinite gain at zero frequency.


Reduces steady state error to step, ramp, etc reference inputs.
But also has integrator antiwindup problems.
Adds phase below break point.
We want to keep break point frequency very low to keep from
destabilizing system
1
c .
TI

Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

128 / 129

Bode compensator design


Lag compensator

Approximates PI controller, but without integrator


overflow.

Ts+1
Lag compensator has the form D (s) = K Ts+1 where > 1.

Primary objective of lag compensator design is to add 20 log10 dB


gain to low frequencies without changing PM.
Magnitude and phase effects of a lag compensator:

Steady state response improves with little effect on transient response.


Assoc. Prof. Enver Tatlicioglu (EEE@IYTE)

EE362 Feedback Control Systems

Chapter 8

129 / 129

You might also like