EMS507 Control System

Analysis and Design

Week 6 – Frequency analysis

Yunpeng Zhu
Frequency analysis
© Copyright 2023 Yunpeng Zhu. All Rights Reserved
Edition: v1.1
Table of Contents
1 Frequency Response Analysis .................................................................- 1 -
1.1 Frequency Response .............................................................................- 1 -
1.2 Frequency Response Function (FRF) .....................................................- 3 -
2 Bode plot .................................................................................................- 6 -
2.1 Concept .................................................................................................- 6 -
2.2 Sketch Bode plots of simple factors ......................................................- 7 -
2.3 Sketch Bode plots of multiple factors .................................................. - 12 -
2.4 Bode plots of second order factors ..................................................... - 18 -

i
1 Frequency Response Analysis
1.1 Frequency Response

The response of a system can be partitioned into both the transient response
and the steady state response. The steady state response of a system for an
input sinusoidal signal is known as the frequency response.
If a sinusoidal signal is applied as an input to a Linear Time-Invariant (LTI)
system, then it produces the steady state output, which is also a sinusoidal
signal. The input and output sinusoidal signals have the same frequency, but
different amplitudes and phase angles.

Let the input signal generated by the building shaker be

u (t )  A sin(0t )

Consider the system steady state response only, the open loop transfer
function will be
G ( s )  G ( j )

where the real part of the poles  dominating transient response is omitted.
We can represent G( j ) in terms of magnitude and phase as shown below.

G ( j )  G ( j ) G ( j )

Substitute   0 in the above equation, G ( j0 )  G ( j0 ) G ( j0 ) .

The output signal is

-1-
 y (t )  A G ( j0 ) sin(0t  G ( j0 ))

 The amplitude of the output sinusoidal signal is obtained by multiplying

the amplitude of the input sinusoidal signal and the magnitude of G( j ) at
  0 .
 The phase of the output sinusoidal signal is obtained by adding the phase
of the input sinusoidal signal and the phase of G( j ) at   0 .

Consider a transfer function of a second order system:

4
G ( j ) 
s  3s  2
2

u (t )  sin(t )  y(t )  A sin(t   )

4 4 4 4
G ( j0 )     (1  3j)
( j0 )  3( j0 )  2 0 1 1  3j  2 1  3j 10
2

 A  1.2

   tan 3
1

Response is frequency dependent. Change the frequency and both A and 

will change:
u (t )  sin(2t )  y (t )  0.62sin(2t  1.8)

As the frequency of the input changes:

• The amplitude A of the output changes.

• The phase shift  of the output changes.

-2-
Frequency response is a description of how the amplitude and phase shift
depend upon the frequency of the input, that is, how do the characteristics of
the response, depend upon the frequency?
Phase shift is in radians – this corresponds to a time shift t s from input to
output via the frequency of oscillation  :
sin(t   )  sin(t  ts )

In y(t )  A sin(t   ) , A  A( ) is defined as the system gain and    ( ) is

defined as the system phase. Frequency response is simply the information
stored in A  A( ) and    ( ) .

Quiz 1.1: Find the gain and phase for the following system for the frequency of
  3 rad/s , that is u (t )  sin(3t ) .

The amplitude of oscillation is marked with the double green arrow; Clearly
1.6  0.6
A  0.5
2
The time shift (dotted double arrow) is
ts  50.8  51.7  0.9 s

Hence, the phase shift is: ts  3  (0.9)  2.7 rad

1.2 Frequency Response Function (FRF)

The transfer function is

N ( s ) b0 s m  b1s m1   bm1s  bm

G(s)  
D( s ) a0 s n  a1s n1   an1s  an

-3-
Let s  j , we obtain the FRF:

b0 ( j ) m  b1 ( j ) m1   bm1 ( j )  bm
G ( j ) 
a0 ( j ) n  a1 ( j ) n1   an1 ( j )  an

An important frequency domain specification is the resonant frequency.

Consider the FRF of the second order system as

n 2 n 2
G ( s)  2  G ( j ) 
s  2n s  n 2 s  j ( j ) 2  2n ( j )  n 2

The frequency let the FRF (or frequency response) has the maximum amplitude
is the resonant frequency.
dG ( j )
The resonant frequency is evaluated by solving  0 , yields
d

r  n 1  2 2
We have learned
Natural frequency ( n ),

Damped natural frequency ( d  n 1   2 ),

Resonant frequency ( r  n 1  2 2 ).
For small damping ratio, they are almost the same. That means, in practice, if
you drive a system at its natural frequency, the system will resonant.
[Watch the video: Resonant Frequency, 5 min]
How to get frequency responses from the FRF?
We know that the frequency response is
u (t )  A sin(0t )  y (t )  A G ( j0 ) sin(0t  G ( j0 ))

-4-
The only thing we need to do is to find the amplitude G ( j0 ) and the phase
angle G ( j0 ) .

Rules:

Consider a simple case for a single factor of the form (s+a) or 1/(s+b)

Amplitude: j  a   2  a 2
s  a  j  a

Phase: ( j  a)  tan 1
a
1 1
Amplitude: 
1 1 j  b  2  b2

sb j  b 1 
Phase:    tan 1
j  b b

For example, frequency response with three factors uses the simple properties
of complex numbers:

j  2 2  4
Amplitude: 
s2 ( j  3)( j  4)  2  9  2  16
G(s) 
( s  3)( s  4) j  2   
Phase:   tan 1  tan 1  tan 1
( j  3)( j  4) 2 3 4

-5-
Quiz 1.2: Find the amplitude and the phase of the FRF from
4
G (s) 
s( s  4)

4 4 4
Amplitude:  
( j )( j  4)  2  2  16   2  16
4
G (s) 
s( s  4) Phase:
4 0    
  tan 1  tan 1  tan 1    tan 1
( j )( j  4) 2 0 4 2 4

2 Bode plot
2.1 Concept

The Bode plot or the Bode diagram is one of the most

important methods in classical control theory. It was
originally conceived by Hendrik Wade Bode in the 1930s. The
Bode plot consists of two plots:
 Magnitude plot
 Phase plot

In both the plots,

 Horizontal axis represents angular frequency (logarithmic scale), obtained
as log10  or  on a log axis.
 Vertical axis represents either

-6-
  the magnitude of open loop FRF in the magnitude plot: 20log10 G ( j )
(linear scale in decibels (dB))
 the phase angle of the open loop FRF in the phase plot: Arg(G ( j ))
(linear scale in degrees)
We can generate a Bode Plot using MATLAB.
MATLAB command:
num = [1, a1, a2, …];
den = [1, b1, b2, …];
G = tf(num, den);
bode (G);
For example, consider
1
G( s) 
s  2s  5
2

Bode plot:

2.2 Sketch Bode plots of simple factors

In many cases, we want to understand the system’s properties without using a

computer. Sketching a Bode plot is useful and important for researchers to
gain insights and design a control system.
-7-
We often use the approximated bode plot of some typical factors to sketch the
whole Bode plot.
The tail-1 standard form of a transfer function is (Lecture 2)
m1 m2

 ( k s  1) ( l 2 s 2  2 l s  1)
G (s)  K k 1
n1
l 1
n2

 (T s  1) (T
i 1
i
j 1
j s  2 T j s  1)
2 2

where  k , l , Ti , T j and  are real numbers, K is the gain for transfer function.

This standard form indicates that a transfer function is composed by some

simple factors:
1 1 1
K , s, , ( s  a ), , 2 , s 2  2n s  n 2
s s  a s  2n s  n 2

To make things simple, we only focus on first order factors with a  0 . Then,
the constant and the second order factors will be introduced briefly.
(1) Factor G( s)  s

Substitute s  j in G ( s ) as we want to study the frequency responses. Use

the definition of the Bode plot, we have:
Magnitude plot: 20log10 j against log10 ;

20log10 j  20log10  vs log10 . It is clear that the gradient (slope) is 20 .

Math: log a x  y  a y  x

At   0 , 20log10   

At   0.1 , 20log10   20

At   1, 20log10   0

Phase plot: j against log10 ;

-8-
j  90


0  j  tan    
0

   90
2
1
(2) Factor G ( s) 
s
Substitute s  j in G ( s ) , we have:

1
Magnitude plot: 20log10 against log10 ;
j

1
20log10  20log10  1 vs log10 . It is clear that the gradient (slope) is 20 .
j

Math: log a x n  n log a x

At   0 , 20log10  1  

At   0.1 , 20log10  1  20

At   1, 20log10  1  0

Phase plot: j against log10 ;

1
  90
j
1
0  0  j 1
j

 tan    
0

    90
2

(3) Factor G( s)  s  a
-9-
Substitute s  j in G ( s ) , we have:

Magnitude plot: 20log10 a  j against log10 ;

If  a,
20log10 a  j  20log10 a
which is a constant
If  a,
20log10 a  j  20log10 j
which is a straight line with
slope 20
If   a ,

20log10 a  ja  20log10 2a

Phase plot: (a  j ) against log10 ;

If  a,
( a  j  )   a
0
 tan    0    0
a
If  a,
( a  j  )   j 

 tan        90
0
If   a ,
(a  j )  (a  ja )
a
 tan    1    45
a

1
(4) Factor G ( s) 
sa
Substitute s  j in G ( s ) , we have:
- 10 -
 1
Magnitude plot: 20log10 against log10 ;
a  j

If  a,

1 1
20log10  20log10
a  j a
which is a constant
If  a,

1 1
20log10  20log10
a  j j
which is a straight line with
slope -20
If   a ,

1 1
20log10  20log10
a  ja 2a

1
Phase plot: ( ) against log10 ;
a  j

If  a,
1 1
 
a  j a
0
 tan   1  0    0
a
If  a,
1 1
 
a  j j

 tan        90
0
If   a ,

- 11 -
 1 1
   (a 1  ja 1 )
a  j a  ja
a 1
 tan    1  1    45
a
Quiz 2.1: Sketch the Bode plot of
1
G(s) 
s  0.1
For  0.1 , the magnitude plot is a constant: 20log10 0.1  20 ;

the phase is   0
For  0.1 , the magnitude plot is line with the scope 20 ;
the phase is   90
1
For   0.1 , the magnitude plot 20log10  17 ;
0.1 2
the phase is   45

2.3 Sketch Bode plots of multiple factors

It’s like building block or LEGO.

- 12 -
When you buy a LEGO, you will find a book inside - instruction. That’s the rules
you need to follow to build up your product.

There are also rules to construct the Bode plot of a complex FRF.

For example,
1) Consider the transfer function

- 13 -
 1
H ( s) 
( s  1)( s  2)
This transfer function is composed by two simple factors:
1 1
G ,M 
s 1 s2
For the magnitude plot, we know:
There are two corner frequencies   1 and   2
If  1,

1 20log10 H ( j )  20log10 G  20log10 M

20log10  20log10 1  0
1  j  6

1 1 The constant is 0+(-6)=-6

20log10  20log10  6
2  j 2

If 1    2 ,

1 20log10 H ( j )  20log10 G  20log10 M

20log10  20log10 
1  j  20log10   6

1 1 The slope is -20

20log10  20log10  6
2  j 2

If  2,

1 20log10 H ( j )  20log10 G  20log10 M

20log10  20log10 
1  j  40log10 

1 The slope is -20-20=-40

20log10  20log10 
2  j

- 14 -
For the phase plot, we know
If  1,
1 H ( j )  G  M  0
  1  0
1  j
The phase is 0+0=0
1 1
    0
2  j 2

If 1    2 ,
1 1 H ( j )  G  M  90
    j  90
1  j 
The phase is -90-0=-90
1 1
    0
2  j 2

If  2,
1 1 H ( j )  G  M  180
    j  90
1  j 
The phase is -90-90=-180
1 1
    j  90
2  j 

- 15 -
2) Consider the transfer function
s 1
H ( s) 
( s  4)( s  2)
This transfer function is composed by three simple factors:
1 1
G1  s  1, G2  , G3 
s4 s2
For the magnitude plot, we have 3 corner frequencies   1 ,   2 and   4
1 1
If  1 20log10 H ( j )  20log10 1  20log10  20log10  18
4 2
1 1
If 1    2 20log10 H ( j )  20log10   20log10  20log10
4 2
1
20log10 H ( j )  20log10   20log10   20log10
2
If 2    4
1
 20log10
2
20log10 H ( j )  20log10   20log10   20log10 
If  4
 (20  20  20)log10   20log10 

For the phase plot, we have

If  1 H ( j )  G1  G2  G3  0  0  0  0

If 1    2 H ( j )  G1  G2  G3  90  0  0  90

If 2    4 H ( j )  G1  G2  G3  90  90  0  0

If  4 H ( j )  G1  G2  G3  90  90  90  90

- 16 -
Quiz 2.2: Sketch the Bode plot of
1
G(s) 
s( s  1)
20log10 G ( j )  20log10   20log10 1  20log10 

If  1   0.01  20log10   40;   1  20log10   0

G( j )  90  0  0

20log10 H ( j )  20log10   20log10   40log10 

If  1
G( j )  90  90  180

- 17 -
2.4 Bode plots of second order factors

This is a simple introduction of the constant and the second order factors’
Bode plot.
(1) Factor G ( s)  K

(2) Factor s 2  2n s  n 2

1
(3) Factor
s  2n s  n 2
2

- 18 -
Quiz 2.3: Plot the FRF of a standard second order system:

n 2
G ( s)  2
s  2n s  n 2

3 Nyquist plot
A Nyquist diagram is a plot of G( j ) , in the complex plane, for
all real values of  . It was discovered by the Swedish-American
electrical engineer Harry Nyquist at Bell Telephone Laboratories
in 1932 is a graphical technique for determining the stability of
a dynamical system.
The definition of a Nyquist diagram is very simple, just compute G( j ) for
several frequencies, tabulate the real and imaginary parts and then plot in an
Argand diagram. Clearly however, this is somewhat tedious in practice!

- 19 -
MATLAB command:
num = [1, a1, a2, …];
den = [1, b1, b2, …];
G = tf(num, den);
nyquist (G);
Nyquist can also be used to assess the system stability. We won’t discuss about
these in this module. Please watch the videos if you are interested:
https://www.youtube.com/watch?v=sof3meN96MA
https://www.youtube.com/watch?v=tsgOstfoNhk

- 20 -