Frequency Response

Frequency Response Analysis

• Is the response of a process to a sinusoidal
input
• Considers the effect of the time scale of the
input.
• Important for understanding the propagation
of variability through a process.
• Important for terminology of the process
control field.
• But it is NOT normally used for tuning or
design of industrial controllers.
Process Exposed to a Sinusoidal
Input
Bode Plot: A Convenient Means of Presenting Ar
and f versus w
10

1
Ar

0.1

0.01
0
0.01 0.1 1 10 100
w
-90

-180

-270

-360
0.01 0.1 1 10 100
w
 Ways to Generate Bode Plot
• Direct excitation of process.
• Combine transfer function of the process
with sinusoidal input.
• Substitute s=i w into Gp(s) and convert into
real and imaginary components which
yield Ar(w) and f(w).
• Apply a pulse test.
Developing a Bode Plot from the
Transfer Function

G p (i w )  R (w )  i I w 

Ar (w )  R 2 (w )  I 2 (w )

 I (w ) 
f (w )  tan 
1

 R (w ) 
Derivation of the Bode Plot for a
First Order Process
Kp Kp
G p (s)  G p (iw ) 
 p s 1 iw p  1
After rationalization
Kp K pw  p
G p (iw )  2 2 i 2 2
w p 1 w p 1
K K w 
2
p
2
p
2 2
p Kp
Ar (w )  
w  1 2 2
p w 2 2p  1
f (w )  tan (w p )
1
 Properties of Bode Plots

Ga ( s ) Gb ( s )
Consider : G p ( s) 
Gc ( s ) Gd ( s )
Ga ( s ) Gb ( s )
Ar  or
Gc ( s ) Gd ( s )
ln[ Ar (w )]  ln Ga (iw )  ln Gb (iw )  ln Gc (iw )  ln Gd (iw )

f (w )  Ga (iw )  Gb (iw )  Gc (iw )  Gd (iw )

 Bode Stability Criterion
• A system is stable if Ar is less than 1.0 at
the critical frequency (i.e., w that
corresponds to f=-180º)
• Closed loop stability of a system can be
analyzed by applying the Bode Stability
Criterion to the product of the transfer
functions of the controller and the process,
i.e., Gc(s)Gp(s).
 Gain Margin
10

1
M
Ar

0.1

0.01
00.01 0.1 1 10 100
w
-90

-180

-270

-360
wc
0.01 0.1 1 10 100
w
 Gain Margin
• Gain Margin = 1/Ar*
– Where Ar* is the amplitude ratio at the critical
frequency.
– Controllers are typically designed with gain
margins in the range of 1.4 to 1.8 which
implies that Ar at the critical frequency varies
between 0.7 and 0.55, respectively.
 Phase Margin
10
1

0.1
Ar

0.01

0.001

0.0001
w co
0.01 0.1 1 10 100
0 w

-45

-90

-135
PM
-180
0.01 0.1 1 10 100
w
 Phase Margin
• PM = f*  180
– Where f* is f at the crossover frequency.
– Controllers are typically designed with a PM
between 30º to 45º.
Graphic expression of the frequency response
The shortage of the polar plot and the rectangular
coordinates plot: to synchronously investigate the cases
of the lower and higher frequency band is difficult.
Idea: How to enlarge the lower frequency band and shrink
(shorten) the higher frequency band？
Bode diagram(logarithmic plots)
Plot the frequency characteristic in a semilog coordinate:
Magnitude response — Y-coordinate in decibels: 20 log G( jw )
X-coordinate in logarithm of ω: logω
Phase response — Y-coordinate in radian: G( jw )
X-coordinate in logarithm of ω: logω
First we discuss the Bode diagram in detail with the frequency
response of the typical elements.
Frequency Response of The Typical
Elements
The typical elements of the linear control systems —
refer to Chapter 2.
1. Proportional element
Transfer function: C ( s)
G( s)  K
R( s )
Frequency response:
 G ( jw )  K  L(w )  20 log G ( jw )  20 log K
G ( jw )  K  
  (w )  G ( jw )  0 o

Im L(w ),  (w )
 (w )  0o
K Re
L(w )  20 log K dB
0dB, 0o w (log w )
0.1 1 10 100
Polar plot Bode diagram
Frequency response of the typical
elements
2. Integrating
element C ( s) 1
Transfer function: G ( s )  
R( s ) s
Frequency response:
 1
1  G ( jw )   L(w )  20 log G ( jw )  20 log w
G ( jw )   w
jw  o
  (w )   G ( j w )   90
Im L(w ),  (w )
w 

Re L(w ) :  20dB / dec

0dB, 0o w (log w )
w 0 0.1 1 10 100  (w )  90o

Polar plot
Bode diagram
Frequency response of the typical elements
3. Inertial element
C ( s) 1 1
Transfer function: G ( s )   G ( j w ) 
R( s ) Ts  1 jw T  1
0 w  1 T
1 
G ( jw )   L(w )  20 log 1  (wT ) 2    3dB w 1 T
1  (wT ) 2  20 log(wT ) w  1 T

1
 (w )   tg (wT )
K 1
1/T: break frequency G(s )  :
T2s 1 T2
L(w ),  (w )
1
Im w 0 20 log K T
w  0dB, 0o w (logw )
1 Re 0.1 1 10 100
 45 o  20 dB / dec

 90 o

Polar plot Bode diagram

Frequency response of the typical elements
4. Oscillating element
C ( s) 1
Transfer function: G ( s )   0 1
R( s ) T 2 s 2  2Ts  1
1
G ( jw ) 
(1  w 2T 2 )  j 2Tw
1 2Tw
G ( jw )   (w )  tg 1 ( )
1  (w T )  (2Tw )
2 2 2 2 1  w 2T 2
0 w  w n (  1 T )

L(w )  20 log (1  w 2T 2 ) 2  ( 2Tw ) 2    20 log( 2 ) w  wn
 40 log(wT ) w  w
 n
maximum value of G( jw )
d 2 2
Make: ( G ( jw ) )  0  w  w r  w n 1  2 (0    )
dw 2
w r  resonant f requency  M  G ( jw )  1
r r
M r  resonant p eak 2 1   2
Bode diagram of the open loop systems
Plotting methods of the Bode diagram of the open loop
systems
Assume: G ( s )  G1 ( s )  G 2 ( s )  G 3 ( s )...
here : G i ( s )  the transf er functio n of the t ypical elements
We have:
 (w )  G ( jw )  G1 ( jw )  G2 ( jw )  G3 ( jw )  ...

L(w )  20 log G  2o log G1  20 log G2  20 log G3  ...

That is, Bode diagram of a open loop system is the superposition
of the Bode diagrams of the typical elements.

10( s  1)
Example 1 G( s ) H ( s ) 
s 2 (0.01s  1)
Bode diagram of the open loop systems
G(s)H(s) could be regarded as:

10( s  1) 11
G ( s) H ( s)   10  (s  1)  
2
s ( 0.01s  1) s2 0.01s  1
Then we have:
L(w ),  (w ) 20dB/dec
②
－40dB/dec
40dB, 90o
①
20dB, 45o
－20dB/dec
0dB, 0o w (log w )
0.1 1 10 100 －20dB/dec
-20dB, -45o ④
-40dB/dec
-40dB, -90o
－40dB/dec
-60dB.-135o
③
-80dB,-180o
Facility method to plot the magnitude response
of the Bode diagram
Summarizing example 1, we have the facility method to plot the
magnitude response of the Bode diagram:
1) Mark all break frequencies in theω-axis of the Bode diagram.
2) Determine the slope of the L(ω) of the lowest frequency band
(before the first break frequency) according to the number of the
integrating elements:
－20dB/dec for 1 integrating element
－40dB/dec for 2 integrating elements …
3) Continue the L(ω) of the lowest frequency band until to the
first break frequency, afterwards change the the slope of the L(ω)
which should be increased 20dB/dec for the break frequency of
the 1th-order differentiating element .
The slope of the L(ω) should be decreased 20dB/dec for the
break frequency of the Inertial element …
Facility method to plot the magnitude response
of Bode diagram
Plot the L(ω) of the rest break frequencies by analogy .
Example 2 10( s  1)
G ( s) 
s( 0.1s  1)( 0.012 s 2  0.01s  1)
 20 log10  20 logω (w  1)  (w )  90o  tg 1w  tg 1 ( 0.1w )
 20log10  20logw
  0.01w
 (1  w  10)  tg 1
  20logw  2
 1  ( 0 . 01w )

L(w )   20log10  20logw     51.3o
 (10  w  100) w 1
20logw  20log(0.1w ) 
    56.5o w  10
 20log10  20logw    (w )   o

 20logw  20log(0.1w ) (100  w )   174 .9 w  100
   179.6o w  104
 40log( 0.01w )  

The Bode diagram is shown in following figure:

 Facility method to plot the magnitude response
of the Bode diagram
L(w ),  (w )
G ( s) 
－20dB/dec
10( s  1) 40dB, 90o －20dB/dec
2 2 1.25dB
s( 0.1s  1)( 0.01 s  0.01s  1) 20dB, 45o
0dB, 0o w (log w )
0.1 1 10 100
-20dB, -45o
－60dB/dec
-40dB, -90o wr
-60dB.-135o
-80dB,-180o
-100dB,-225o
-120dB,-270o

There is a resonant peak Mr at: w  w r  w n 1  2 2

1
Mr   1.154  1.25 dB  100 1  2  0.5 2  70.7
2 1   2
Graphic expression of the frequency response
Polar plot
The polar plot is easily useful for investigating system stability.

Example 5.1.3 K K
G( s)   G ( jw )  G ( s ) 
s (Ts  1) s  jw j w ( j w T  1)
The magnitude and phase response:
K
A(w )  G( jw )  ;  (w )  G( jw )  [90o  tg 1 (wT)]
w 1  (wT )2
Im
Calculate A(ω) and  (w ) for different ω: KT w 
2
1 1 Re
w 0 
2T T w  1
A(w )   4KT KT 0 4KT T
5 2 5 -135o
w  1 2T
 (w )   90o  117o  135o  180o
w 0 -117o
Polar plot for determination of GM and PM

Im Im
G0 ( jw g )

kg wc
•
wg
-1 w g Re -1  Re
• •
wc •
kg
G 0 ( jw g )
Nyquist criterion and the relative stability
(Relative stability of the control systems)
In frequency domain, the relative stability could be
described by the “gain margin” and the “phase margin”.
1. Gain margin Kg
1
Kg  K g ( dB )   20 log G ( jw ) H ( jw )
G ( jw ) H ( jw ) w w g
w w g

wg : G ( jw ) H ( jw ) w w  1800  Phase-cros sover freq uency

g
2. Phase margin γc
 c  G ( jw ) H ( jw ) w w  ( 1800 )  G ( jw c ) H ( jw c )  1800
c
ωc : G ( jw ) H ( jw )  1  Gain-crossover frequ ency
w w c
3. Geometrical and physical meanings of the Kg and γc
Nyquist criterion and the relative stability
The geometrical meanings is shown in Fig. 5.4.8.
Im
The physical signification : 1/Kg

Kg— amount of the open-loop gain in

decibels that can be allowed to increase
－1 Re
before the closed-loop system reaches to
be unstable. γc
For the minimum phase system:
Kg>1
the closed loop system is stable . stable
unstable
γc —amount of the phase shift CriticalFig. 5.4.8
of G(jω)H(jω) to be allowed before stability

the closed-loop system reaches to be unstable.

For the minimum phase system: γc>0 the closed loop system is
stable .
 System analysis based on the frequency
response
Generally Kg and γc could be concerned with the resonance
peak Mr : Kg and γc ↑ —— Mr ↓.
ωc could be concerned with the resonance frequency ωr and
bandwidth ωb : ωc↑ —— ωr and ωb↓.
5.5.2 Relationship of the performance specifications between the
frequency and time domain
The relationship between the frequency response and the time
response of a system can be expressed by following formula:
1 
C ( t )  L C ( jw ) 
1 jwt
 C ( jw )e dw
2  
G ( jw )
here : C ( jw )  f ( jw ) R( jw )  R( jw )
1  G ( jw ) H ( jw )
But it is difficult to apply the formula .
Relationship of the performance specifications between
the frequency and time domain
(1) Bandwidth ωb(or Resonance frequency ωr)  Rise time
tr
Generally ωb(or ωr )↑—— tr ↓ because of the “time scale”
theorem: 1 
In terms of c ( t )  L C ( jw ) 
1
 C ( jw ) e jwt
dw
2  
If : w   w, C ( jw )  C ( j w)
1  jw t /   1 1
Then : c ( t )   C ( j w)e d  w  c ( t /  )
2    
That is : βω  t So ωb(or ωr )↑—— tr ↓
β
alike : ωc↑—— tr ↓ because of ωc≈ ωb .
For the large ωb , there are more high-frequency portions in c(t),
which make the time response to be faster.
Relationship of the performance specifications between the
frequency and time domain
(2) Resonance peak Mr  overshoot σp%
Normally Mr ↑ —σp% ↑ because of the large unbalance of the
frequency signals passing to c(t) .
Kg and γc ↓ —σp% ↑is alike because of Kg and γc ↓—Mr ↑.

Some experiential formulas:

Overshoot  p %  0.16  0.4( M r  1)  (1.1  M r  1.8)
1
and Mr 
sin  c
For mostdesignproblem, an optimum value of M r : 1.1  M r  1.5

Settling time ts 
k
wc
 1
, k  2  1.5
  1
 1  2.5


 1  350   c  900 
 sin  c   sin  c 
Relationship of the performance specifications between
the frequency and time domain
(3) A(0) → Steady state error ess
G ( jw )
A( 0 )  A(w ) w  0 
1  G ( jw ) H ( jw ) w  0
1
assume : G ( s )  KG   G0 ( s ) H ( s )  K H  H 0 ( s )
v
s
 1 For the unity feedback system, H(s)  1 :
 K v 1
then : A( 0)   H  1 v 1
1 
 v0 A( 0)   KG
 KG K H v0
1  KG

So for the unity feedback systems: A( 0)  1, A(0)   e ss 

Relationship of the performance specifications between
the frequency and time domain
(4) Reproductive bandwidth ωM → accuracy of
Reproducing r(t) A(ω)
Reproductive bandwidth ωM : Mr

G ( jw ) △
A(w )   A( 0)   A(0)
1  G ( jw ) H ( jw ) w w 0.707A(0)
M
ω
 : allowed reproducin g error 0
ωM ωr ωb
Fig. 5.5.2

for a given ωM , △↓—higher accuracy of reproducing r(t) .

for a given △, ωM ↑ —higher accuracy of reproducing r(t) .
Demonstration
assume : E ( jw )  R( jw )  C ( jw )  f e ( jw ) R( jw )
and : f e ( jw )  
Relationship of the performance specifications between the
frequency and time domain
For the frequency spectrum of r(t) shown in Fig.5.5.3 .
1 w M jwt
e(t )   f e ( jw ) R ( j w ) e dw R( jw )
2 w
M
 w M jwt
  R( jw )e dw    r ( t ) ω
2 w 0
ωM
M
Fig. 5.5.3
That is : e ( t )    r ( t )    e ( t ) 
Relationship of the performance specifications between the
frequency and the time domain: for the typical 2th-order system
For the typical 2th-order system:
w n2 w n2
G( s)   f ( s) 
s( s  2w n ) s 2  2w n s  w n 2
 Relationship of the performance specifications between
the frequency and the time domain: for the typical 2th-
order system
We have:

w b  w n (1  2 2 )  2  4 2  4 4

2 2 
w r  w n 1  2 (0    ) 
2    , w n   p % , t s , t r ...
1 
Mr  
2 1   2

wc  wn 1  4 4  2 2

2 
 c  tg 1 
   , w n   p % , t s , t r ...
1  4 4  2 2 
Kg   