Frequency Response

Techniques
Chapter 10

(c) 2019 Farrokh Sharifi 1

 Frequency Response Techniques
• So far we discussed Time Response techniques.

• Frequency response methods for analysis and design of feedback

control systems:
• Nyquist method - polar plots of the frequency response of the system.
• Bode method - separates plots of magnitude and phase of the frequency
response.

(c) 2019 Farrokh Sharifi 2

 Frequency Response Techniques
• Control systems are affected by signals- often unpredictable:
noise, disturbances, reference inputs.
• Frequency content can be a better way to characterize these
signals, as opposed to time description.
• The frequency content of any signal can be obtained analytically
or experimentally using Fourier analysis.
• Nyquist, Bode and Nichols introduced frequency response
methods in dynamics and control in the 1930’s and 40’s.
• The first step is to study how linear systems (transfer functions)
are affected by signals of various frequencies.

(c) 2019 Farrokh Sharifi 3

 Frequency Response Techniques
• If G(s) open loop transfer function & ω is the frequency vector, G(jω) is a
complex number. Then magnitude and phase vs ω can be plotted (Bode plot)
or position in complex plan (Nyquist plot).

(c) 2019 Farrokh Sharifi 4

 Frequency Response Techniques
• Dynamic system response is selective to frequency: amplitude
changes and phase shifts.
• Think of the TF as a frequency-dependent device which alters the
amplitude and phase of the incoming sine wave.
• The frequency response of a system is a description of the
amplitude and phase variations as a function of frequency
• We’ll learn how to construct frequency response diagrams.
• It can be shown that for t → ∞ (see Ogata’s Book):
Amplification Ratio= |G(jω)| Phase Shift= ∠G(jω)

(c) 2019 Farrokh Sharifi 5

 Frequency Response Techniques
• Time domain analysis- We applied a certain function of time
(e.g., step) and analyzed time response of the system. We found
ways to improve transient and steady-state performance.

• Frequency domain analysis - ?

Outputs might not catch the input as the frequency increases.
– Amplitude might drop (magnitude)
– The response might lag behind the input (phase)
Perfect systems output does not lag behind inputs and the output
magnitude does not drop for any frequency.

(c) 2019 Farrokh Sharifi 6

 Frequency Response Techniques

• Example: CD players, tape decks, etc.

• Measure how uniformly it reproduces sounds from the lowest
tones to the highest.
• You expect them to preserve the loudness relationship between
various instruments and voices and should not over or under-
emphasize any frequency or tone (concept of flat frequency
response).
• Example: Bracket structure design
http://www.youtube.com/watch?v=Ddc3Kr6GDgI
Frequency response for thermal imaging
http://www.youtube.com/watch?v=hHaLneXkec4 7
Frequency Response Techniques

(c) 2019 Farrokh Sharifi 8

Frequency Response Techniques

(c) 2019 Farrokh Sharifi 9

Frequency Response Techniques

(c) 2019 Farrokh Sharifi 10

 Frequency Response Techniques
• Definition:
“A measure of the effectiveness with which a device, or system
transmits the different frequencies applied to it”

“It is a phasor whose magnitude is the ratio of the magnitude of

the output signal to that of a sine-wave input, and whose phase is
that of the output with respect to the input.”

(c) 2017 Farrokh Sharifi 11

 Frequency Response Techniques
• Main advantages of the frequency response techniques:
 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 handled correctly (e sT ).
 Graphical techniques (analysis/synthesis) are “quite simple.”

(c) 2019 Farrokh Sharifi 12

 Frequency Response Techniques
• Main advantages of the frequency response techniques:
 Test involves measurements under steady state conditions
(simpler to analyze compared to measurement of transient
responses).
 Tests on open loop systems which are not subject to instability
issues.
 Results give convenient access to control system order, gain,
error constants, resonant frequencies.
 Modeling of transfer function from measurement data.
 Design of lead compensators to meet steady-state errors and
transient response requirements.
 Finding stability of nonlinear systems.
 Allow quick hand sketching of frequency response diagrams
13
for analysis and design.
 Frequency Response Techniques
• Main disadvantages of the frequency response techniques:
 Not always easy to capture transient response characteristics
from frequency response.
 It can be difficult to generate low frequency signals.

(c) 2019 Farrokh Sharifi 14

 Introduction: Phase

• Phase of a complex number # : plot the location of the

number
• Numbers with positive real part

• Numbers with negative real part

• Also
(c) 2017 Farrokh Sharifi 15
 Introduction:
The Basic Concept of Frequency Response (section 10.1)
• The steady state response of a linear system to a sinusoidal input
is a sinusoid with the same frequency.
• Sinusoid: 𝑀1 cos 𝜔𝑡 + 𝜙1
can be represented as complex number: phasor.
 Amplitude of the sinusoid M1 is the magnitude of the
complex number;
 Phase of the sinusoid 1 is the angle of the complex number;

Or, 𝑀1 cos 𝜔𝑡 + 𝜙1 is represented as 𝑀1 ∠𝜙1 .

The frequency  of the response remain unchanged.

(c) 2019 Farrokh Sharifi 16

 Introduction:
The Basic Concept of Frequency Response
• Consider the following system with an input force f(t), and the
output response x(t).

• The input and output can be represented as complex numbers, or

phasors, as and .
• If we assume that the mechanical system is represented by the
complex number , then we can depict the system
with the input and output as:

17
 Introduction:
The Basic Concept of Frequency Response
• Consider a simple scenario:

I  M i e(t i )
G  Me
O  IG  M i Me(t i  )  M o e(t o )
O M o e(t o ) M o (o i )
G  ( t i )
 e
I Mie Mi
 Introduction:
The Basic Concept of Frequency Response
• We know that if the input signal {the force f (t)} is sinusoidal,
then the steady-state output response is sinusoid with the same
frequency (different magnitude and phase).

• Therefore, the output sinusoid: product of the complex number

representation of the input sinusoid and the system

(c) 2019 Farrokh Sharifi 19

 Introduction:
The Basic Concept of Frequency Response
• The magnitude frequency response of the system is

• and phase frequency response of the system is

• The combination of the magnitude and phase is called

frequency response of the system, defined as M ()  () .

Note that both the magnitude M and the phase response  are
functions of the frequency .
We study how the input will be varied to get an output at different
frequencies.
(c) 2019 Farrokh Sharifi 20
 Introduction:
Analytical Expressions for Frequency Response
• Now consider a system with a general sinusoidal input in the
form

• The input can be represented in 3 ways:

a) polar form M i ( ) i ( )  A2  B 2   tan 1  B 
 A

b) rectangular form M i ( ) i ( )  A  jB

j i
c) by Euler’s formula M i ( ) i ( )  M i e
21
(c) 2019 Farrokh Sharifi
 Introduction:
Analytical Expressions for Frequency Response
• The output C (s) of the system is

• By performing partial fraction expansion we obtain

where
• K1 is the complex conjugate of K1

22
 Introduction:
Analytical Expressions for Frequency Response
• The steady state response comes from the first two terms
(associated to the input sinusoid)

and substituting the values for K1 and K1 yields

• The inverse Laplace transform of this equation gives

• In the phasor form, the response can be represented as

M o ( ) o ( )   M i ( ) i ( )  M G ( ) G ( )  23
 Introduction:
Analytical Expressions for Frequency Response
• Since

<To be used for steady-state error, transient response and stability

analysis>.
Frequency response can be plotted as:
• Polar plots: the phasor’s length is the magnitude, and phasor’s
angle is the phase.
• Bode plots: separate plots for the magnitude and phase as a
function of the frequency .
The frequency range could be quite large. The separate plots are
usually plotted on a logarithmic scale for the frequency: log

Magnitude could be large and is usually expressed in decibels (dB):

dB  20log M 24
Introduction: Typical Bode Plot

(c) 2019 Farrokh Sharifi 25

 Introduction:
Plotting Frequency Response
• Example: Find the analytical expression for the frequency
response of G ( s )  1  s  2  . Plot the polar and the separate
magnitude and phase plots of the frequency response.
1 1 2  j 2  j
s  j   G ( j )   .  2
j  2 2  j 2  j   4
| 2  j | 2  4 1
| G ( j ) | 2  2  
|  4|   4  4
2


in dB: 20 log | G ( j ) | 20 log 
1 
  20 log
  4 
2  2  4 
 
G ( j )  (2  j )  ( 2  4)  tan 1 ( )  0   tan 1 ( )
2 2
(c) 2019 Farrokh Sharifi 26
 Introduction:
Plotting Frequency Response
Example cont.:
• The magnitude diagram is is , and the
phase diagram is , both plotted over log.
As the frequency increases
The output amplitude will drop

As the frequency increases

The output will lag more

27
 Introduction:
Plotting Frequency Response
Example cont.:
• Polar plot:

1
  2  M (2)   0.36,    tan 1 (1)  45
8
1
  1  M (1)   0.447,    tan 1 (0.5)  26
5

28
 Introduction:
Plotting Frequency Response
Example cont.:
• For  = 1 rad/s, magnitude is 20log M  7 dB , with phase
of –26 degrees. It follows: M = 107/20  0.447 .

-26°

0.447

(c) 2019 Farrokh Sharifi 29

Asymptotic Approximations: Bode Plots
(section 10.2 in the book)

• Bode Plots: The log-magnitude and phase frequency response

plots versus log.
• By approximating the Bode plots with straight lines, they can
be sketched manually.
• Approximations: let’s analyze a general transfer function

• Phase angles and log magnitudes of individual factors are

added together.

(c) 2019 Farrokh Sharifi 30

Asymptotic Approximations: Bode Plots

• The magnitude frequency response is

and if expressed in dB, the magnitude response is obtained as

sums of the contributions by the individual poles and zeros

(c) 2019 Farrokh Sharifi 31

Asymptotic Approximations: Bode Plots
 1. Obtain the characteristic equation of the closed loop system: 1+G(s)H(s) =
0 (loop transfer function L = GH).
K  s  z1  s  z2   s  zm   s 2  2 1n1s  n21   s 2  2 mnm s  nm2 
L( s )  G ( s ) H ( s )   N
s  s  p1  s  p2   s  pn   s 2  2 1n1s  n12   s 2  2 nnn 2 
 s  nn

 2. Write the transfer function L in Bode scaled-normalized form by dividing

the terms to zeros zi , poles pi , and squares of natural frequency n (for the
second-order expressions) and calculate K 0 :

s  s   s  s 2 s   s2 s 
K 0   1  1   1 2  2 1  1  2  2 m  1
L( s )   z1  z2   zm  n1 n1   nm nm 
N  s  s   s  s 2 s   s2 s 
s   1  1   1 2  2 1  1  2  2 n  1
 p1  p2   pn  n1 n1   nn  
nn 

(c) 2019 Farrokh Sharifi 32

Asymptotic Approximations: Bode Plots
z1 z2 n21n22
K0  K  lim s N L( s)
p1 p2  '2n1  '2n 2 s 0

 3. Substitute s = j.
 4. Expand the 20log10 | L j)| of the magnitude and add their contributions
one by one.

j j
20 log L( j )  20 log K 0  20 log 1   20 log 1
z1 zm
2  2 
 20 log  2  j 2 1 1   20 log  2  j 2 m 1
n1 n1 nm nm
j j
20 log  j   20 log 1   20 log 1
N

p1 pn
2  2 
 20 log  2  j 2 1 '  1   20 log  2  j 2 n '  1
n1 n1 
nn nm

(c) 2019 Farrokh Sharifi 33

 Asymptotic Approximations: Bode Plots
• If we approximate each term in the previous equation with a
straight line, then we can obtain the plot of the total response as
a graphical addition of points.
• Similar reasoning holds for the phase diagrams.
• Case 1: Gain
20log M  20log K
Magnitude: not a function of frequency
a constant value
Phase: not a function of frequency
 0 K 0
K  
. 180 K 0 34
Asymptotic Approximations: Bode Plots
• Case 2: zero at origin G (s) = s
G ( j )  j
decade decade decade
20 log M  20 log 
• Magnitude: 20 log ω
constant 20 dB/decade slope,
at  = 1, 20 log M = 0.
straight line intersects 0dB
at 1rad/s

• Phase: constant +90 °.

35
Asymptotic Approximations: Bode Plots
• Case 2(b): multiple zeros at origin
G(s)  s n
G ( j )  ( j ) n
20 log M  20n log 
• Magnitude: 20n log ω
constant 20n dB/decade slope,
at  = 1, 20 log M = 0.
straight line intersects 0dB at 1rad/s

• Phase: constant +90 °×n.

(c) 2019 Farrokh Sharifi 36

Asymptotic Approximations: Bode Plots
• Case 3: pole at origin G (s) = 1/s
G ( j )  1/ j
20 log M  20 log 

• Magnitude: -20 log ω

constant –20 dB/decade slope,
at  = 1, 20 log M = 0.

• Phase: constant –90 °.

(c) 2019 Farrokh Sharifi 37

Asymptotic Approximations: Bode Plots
• Case 3(b): multiple poles at origin
G ( s)  1/ s n
G ( j )  1/ ( j ) n
20 log M  20n log 
• Magnitude: - 20n log ω
constant - 20n dB/decade slope,
at  = 1, 20 log M = 0.
straight line intersects 0dB at 1rad/s

• Phase: constant - 90 °×n.

(c) 2019 Farrokh Sharifi 38

 Asymptotic Approximations: Bode Plots
Observation for poles (zeros) at origin: G ( s )  1/ s n
(G ( s )  s n
)
• Magnitude (dB) of a simple pole (zero) is zero at   1 rad/sand
has a constant slope of -20 dB/decade (+20 dB/decade).
• Magnitude (dB) of n simple poles (zeros) is zero at   1 rad/s and
has a constant slope of -20n dB/decade (+20n dB/decade).
• Phase (deg) of simple pole (zero) is constant -90° (+ 90°).
• Phase (deg) of n simple poles (zeros) is constant -90°× n
(+ 90°× n).

(c) 2017 Farrokh Sharifi 39

 Asymptotic Approximations: Bode Plots
• Case 4: simple zero
s 
G ( s )    1
a 
j   
2

G ( j )    1  20 log M  20 log | G ( j ) | 20 log 1   
 a  a

 
2

 << a  20 log 1     20 log 1  0

a

     
2 2

 >> a  20 log 1     20 log    20 log  

a a a
   
  a  20 log    0,   10 a  20 log    20
a a
• Magnitude: 0 dB line before α & constant + 20 dB/decade slope
line after α
(c) 2019 Farrokh Sharifi 40
 Asymptotic Approximations: Bode Plots
Case 4 cont.:
• Note that for  = α, high frequency approximation is equal to
low frequency approximation.  = α is called break frequency.
• For the high frequencies the line has a slope of 20.
• The straight lines approximations are called asymptotes.
Each doubling of frequency (octave) causes 20log 
to increase 6 dB:  6dB/octave
20 log 2  20 log 2  20 log   6  20 log 
20 log 4  20 log 4  20 log   12  20 log 

And 20dB/decade:
20log10  20log10  20log   20  20log 41
Asymptotic Approximations: Bode Plots
• Now consider the phase frequency response.
 j   
G ( j )     1  tan 1  
 a  a
• For  = α , G( ja)  ja  a   tan 1 a  tan 1 1  45 .
a
• For  ≈ 0, the phase is 0, and for   a the phase is 90°.
• Start one decade below the break frequency, i.e.  = 0.1 α , and
draw 45° line to one decade above the break frequency,  = 10 α

42
 Asymptotic Approximations: Bode Plots
• The maximum difference of the approximation for the magnitude is at the
break frequency.
• For the phase, the maximum difference is at one decade below and above the
break frequency.

(c) 2019 Farrokh Sharifi 43

 Asymptotic Approximations: Bode Plots
• Case 5: simple pole
1
G (s) 
s 
  1
a 
 j   
2

G ( j )  1/   1  20 log M  20 log | G ( j ) | 20 log 1   

 a  a

 
2

 << a  20 log 1     20 log 1  0

a

     
2 2

 >> a  20 log 1     20 log    20 log  

a a a
   
  a  20 log    0,   10 a  20 log    20
a a
• Magnitude: 0 dB line before α & constant ˗ 20 dB/decade slope
line after α
(c) 2019 Farrokh Sharifi 44
 Asymptotic Approximations: Bode Plots
Case 5 cont.:
• Normalized Bode plots G ( s) 
1
s / a 1

• Phase frequency plot:

It starts at 0°, and

decreases by 45° from G ( s) 
1
s / a 1
one decade lower until
one decade above the
break frequency.

(c) 2019 Farrokh Sharifi 45

 Asymptotic Approximations: Bode Plots
Observation for simple pole (zero): G = 1/(s/ α + 1) (G = (s/ α + 1))
• Magnitude (dB) of simple pole (zero) starts at zero till break
frequency of  / a  1 and decrease (increases) with the slope of
˗ 20 dB/decade (+20 dB/decade).
• Phase (deg) of simple pole (zero): s = - α is 0 at low frequencies till
 / a  0.1 and decreases (increases) with the slope of -45 deg/decade
(+45 deg/decade) till  / a  10 and after that remains -90 (or +90
deg)

(c) 2019 Farrokh Sharifi 46

Asymptotic Approximations: Bode Plots
Case 6: second order polynomials in numerator
 s2 s   2  
G ( s)   2  2  1  G ( j )  1  2  2 j
 n n   n n 
2
 s 
if   1    1
 n 
• Double real zero at n . Magnitude slope of +40 dB/decade after n . Phase
slope of +90 deg/decade between 0.1n and 10n
• For   1 there will be overshoot or undershoot at almost n (breakaway
frequency)
2 2  2  2
| G ( j ) | (1  2 )  4( ) G ( j )  tan (2 ) / (1  2 )
1

n n n n
 << n  20 log M  20 log 1  0 G ( j )  tan 1 (0)  0
  n  20 log M  20 log 2 G( j )  tan 1 ()  90
 >> n  G ( j )  tan 1 ( 0)  180 47
Asymptotic Approximations: Bode Plots
Case 6 cont.:

20log( M )

48
Asymptotic Approximations: Bode Plots
• The actual and approximated plots for the second order
systems depend on the value of ζ, and when these values are
low the difference between the plots are increasing.

• Dip frequency at
d  n 1  2 2 |G( jd ) |dB  20log(2 1   2 ) (dip only if 0    0.707)

(c) 2019 Farrokh Sharifi 49

Asymptotic Approximations: Bode Plots
• Therefore, it is required to apply corrections on the
approximated Bode plots for second-order systems with low
values of ζ.
• The tables with the correction factors are provided in the book
(see end of this chapter-slides).

50
(dip only if 0    0.707)
Asymptotic Approximations: Bode Plots
Case 7: second order polynomial in denominator
Similar to the previous example (opposite). G ( s)  1/  s  2 s  1
2


 n
2
n 
• Magnitude: – 40 dB/decade;
• Phase: from 0 to – 180°.

(peak only if 0    0.707) (c) 2019 Farrokh Sharifi 51

 Asymptotic Approximations: Bode Plots
Observation for second order system in the denominator
(numerator):  s2 s   s2 s 
G ( s )  1/  2  2  1 G ( s)   2  2  1
 n n   n n 
• Magnitude (dB) at low frequency
is 0 dB and decreases (increases) -40 dB/decade (+40 dB/decade)
after break frequency ( n ).

• Phase (deg) of system at low frequency

(0.1 n ) is 0° and decreases (increases) at -90° /decade (+ 90°
/decade) to -180° (+180°) till 10 n .

(c) 2019 Farrokh Sharifi 52

Asymptotic Approximations: Bode Plots
• Case 8: Time delay

G ( s)  e sT
G ( j )  e jT  20 log | e  jT | 20 log1  0 dB No effect on amplitude

G( j )  T

(c) 2019 Farrokh Sharifi 53

Asymptotic Approximations: Bode Plots
• Example (Bode plots of functions with first-order polynomials
of s):
( s  3)
G(s) 
s( s  1)( s  2)

• Solution: normalized function

3
( s / 3  1)
G(s)  2
s( s  1)( s / 2  1)
• Four terms: break frequencies at 1, 2, and 3 rad/s.
• Plot should include at least 1 decade below the lowest break
frequency (0.1 rad/s), and extend to at least one decade above
the highest break frequency (30 rad/s).
(c) 2019 Farrokh Sharifi 54
 Asymptotic Approximations: Bode Plots
Example cont.:
• Note the effect of K=3/2 on the plot:
- it does not affect the phase
- moves up the magnitude curve by 20 log K.
• For  = 0.1  | j |  20 log( 0.1 )  20 log15  23.52dB
3/ 2 3/ 2
dB

• Add the effect of each component.

• For simplicity, use a table with important points for magnitude.
(where slope changes for each component) and write the
slope contribution for any portion.
• Add the slopes for each point.
• Start from constant point calculated and add slopes as
calculated at each important point. 55
 Asymptotic Approximations: Bode Plots
Example cont.:

Zero

Zero

3
( s / 3  1)
G(s)  2
s( s  1)( s / 2  1)

(c) 2019 Farrokh Sharifi 56

 Asymptotic Approximations: Bode Plots
Example cont.:
23.52
Asymptotic Approximations: Bode Plots
Example cont.:
• Phase table for the slope contribution of each term (3/2K has no
effect and effect of pole at origin (1/s) is constant – 90 deg) :

1
Zero

Zero

3
( s / 3  1)
G(s)  2
(c) 2019 Farrokh Sharifi s( s  1)( s / 2  1) 58
Asymptotic Approximations: Bode Plots
Example cont.:
Asymptotic Approximations: Bode Plots
• Example:

• Solution: the normalized function is

s  s 
3   1   1
G(s)   3 

3 3 
s   s2 2  50  s   s 2 2 
(2)   1  25    s  1   1   s  1 
2   25 25   2   25 25 
• Magnitude table:
Asymptotic Approximations: Bode Plots
s 
  1
Example cont.: G(s) 
3 3 
50  s   s 2 2 
• Phase table:   1   s  1
 2   25 25 

• Note the correction of the second-order component in the plot

on the next slide

(c) 2019 Farrokh Sharifi 61

Asymptotic Approximations: Bode Plots
Example cont.:

3
20 log  24.4dB
50
Asymptotic Approximations: Bode Plots

Example cont.:
Asymptotic Approximations: Bode Plots
Example: Phase plot (identify)

• Constant K=2 no contribution to phase

• Pole at origin: -90deg contribution
• Two LHS poles of -10 and -50
(0 deg to -90 deg each)
• One LHS zero at -0.5 (0 to +90 deg)
Asymptotic Approximations: Bode Plots
• Algorithmic rules for drawing the Bode plots
 1. Obtain the characteristic equation of the closed loop system: 1+G(s)H(s) =
0 (loop transfer function L = GH).

K  s  z1  s  z2   s  zm   s 2  2 1n1s  n21   s  2  s    e
2 2
Td s
L( s )  G ( s ) H ( s )  N
m nm nm

s  s  p1  s  p2   s  pn   s 2  2 1n1s  n12   s  2  s   

2
n nn
2
nn

 2. Write the transfer function L in Bode scaled-normalized form by dividing

the terms to zeros zi , poles pi , and squares of natural frequency n (for the
second-order expressions) and calculate K 0 :
s  s   s  s 2 s   s2 s 
K 0   1  1   1 2  2 1  1  2  2 m  1
 z1  z2   zm  n1 n1   nm nm  Ts
L( s )  e
 s  s   s  s 2
s   s 2
s 
s N   1  1   1 2  2 1  1  2  2 n  1
 p1  p2   pn  n1 
n1   nn  
nn 

(c) 2019 Farrokh Sharifi 65

Asymptotic Approximations: Bode Plots
z1 z2 n21n22
K0  K  lim s N L( s)
p1 p2  '2n1  '2n 2 s 0

 3. Substitute s = j.
 4. Expand the 20log10 | L j)| of the magnitude and add their contributions
one by one.

j j
20 log L( j )  20 log K 0  20 log 1   20 log 1
z1 zm
2  2 
 20 log  2  j 2 1 1   20 log  2  j 2 m 1  0
n1 n1 nm nm
j j
20 log  j   20 log 1   20 log 1
N

p1 pn
2  2 
 20 log  2  j 2 1 '  1   20 log  2  j 2 n '  1
n1 n1 
nn nm
Asymptotic Approximations: Bode Plots
• a) Choose the frequency axis range (two decades below the smallest
frequency (× 0.01) and two decades above the largest frequency (×100).
• b) For a constant: 20log K0  constant
• c) For zero or pole at origin: 20log  j  N  20 N log 
(lines of slopes  N × 20 dB/decade passing through 1 rad/sec).
• d) For simple zero or pole 1  0   a
 j  
20 log   1   
 a   20 log   a
 a
(Locate corner frequency or break point  = α , and draw line of
20dB/decade. Draw a horizontal line of 0 dB. Two lines intersect at  = α )
• e) For quadratic zeros or poles
1  0   n
 2   
20 log   2  j 2  1   
 n n   40 log   n
 n

(c) 2019 Farrokh Sharifi 67

Asymptotic Approximations: Bode Plots
Similar to the previous case except that the break point is now  = n . Draw a
horizontal line at 0 dB. Draw line of 40dB/decade at break frequency. Two
lines intersect at the  = n. Correct for the effect of damping ratio according
to: magnitude ≈ ½ ζ at the break point or more accurately using the Table).
• For Time delay (eT s). Nothing is required (no effect on magnitude).
d

 5. Expand the L( j ) of the phase and add their contributions one by one.
 j   j 
L( j )  K 0     1     1
 z1   zm 
 2    2  
    2  j 2 1  1      2  j 2 m  1  Td 
 n1 n1   nm nm 
 j   j 
   j      1     1
N

 1
p   n
p 
 2    2  
    2  j 2 1  1      2  j 2 n  1
 n1 n1  
 nn 
nm 
Asymptotic Approximations: Bode Plots

 0 K0  0
• a) For a constant K 0   
180 K 0  0

• b) For zero or pole at origin   j  N  90 N

(horizontal lines of 90N magnitude.)
1
 j  
• c) For simple zero or pole:   1   tan 1
 a  a
(Locate corner frequency or break point = α . Draw line from 0° at (0.1 α ) (1
decade below break point) to 90° at (10 α ) (1 decade above break point).
Draw a horizontal line of 0. Two lines intersect at  = 0.1 α.)
Asymptotic Approximations: Bode Plots
• d) For quadratic zeros or poles:
1
 2    2  2 
   2  2 j  1   tan 1  1  2  
 n n   n  n  

Draw 0° up to  = 0.1n . Draw ±180° after  = 10n . Connect them.

Correct for the damping ratio ζ effect).
• e) For pure time delay ( eT s ):
d

Subtract – Td from the phase graph for all ’s.

e jTd  Td

(c) 2019 Farrokh Sharifi 70

Asymptotic Approximations: Bode Plots
• Example: Plot the Bode diagrams for the negative unity
feedback system given with the loop transfer function
6
G ( s) 
 s 2  2s  2   s  2 
• Solution:
 Step 1: since we have unity feedback, i.e., H(s) = 1, it follows
6
L( s )  G ( s ) H ( s ) 
 s 2  2s  2   s  2 
 Step 2: represent the transfer function in scaled-normalized
form: 1 6 1
L( s )   1.5
s2
  s  (2) (2)  s2  s 
  s  1    1   s  1    1
2  2  2  2 
71
(c) 2019 Farrokh Sharifi
Asymptotic Approximations: Bode Plots
• Example cont.:
 Step 3: L( j )  1.5 1
 1.5
1
 ( j ) 2   j   2   j 
  j  1   1    j  1   1
 2   2   2   2 
 2

 Step 4: 20 log L( j )  20 log 1.5  20 log j / 2  1  20 log   j  1 

2
2
3.5218  20 log j / 2  1  20 log   j  1
2

a) frequency axis ranges:

break frequencies at 2 and 2  1.41 rad/s, so it follows:
1 decade below the lowest break frequency (0.14 rad/s), and
1 decade above the highest break frequency (20 rad/s).

(c) 2019 Farrokh Sharifi 72

Asymptotic Approximations: Bode Plots
• Example cont.:
• b) for the constant: 20 log K 0  20 log1.5  3.5218
• c) there are no zeros or poles at the origin .
• d) there is one simple pole and break frequency 2:
 0   2

20 log  j / 2  1   
• e) quadratic poles:  20 log   2
2
1  0   1.41
 2  
20 log    j  1   
 2    40 log   1.41
 2
break point:   2  1.41 .
n

1
the damping ratio: 2  2  ,  1.41
n  0.707 is high.

(c) 2019 Farrokh Sharifi 73

Asymptotic Approximations: Bode Plots
• Example cont.:

Description Frequency (rad/s)

0.1 (starts: plot) 1.41 (ωn: 1.41 ) 2 (starts: pole at -2)
Pole at -2 0 0 -20
ωn =1.41 0 -40 -40
Total slope (dB/dec) 0 -40 -60

1
L( s)  1.5
 s2  s 
  s  1    1
2  2 

(c) 2019 Farrokh Sharifi 74

 Asymptotic Approximations: Bode Plots
Bode Diagram
• Example cont.: Magnitude plot:

1/(s2/2+s+1) 1/(s/2+1) 3.5218

20 dB
Magnitude (dB)

0 dB

-20 dB

-40 dB

3.5218 / (s/2+1)(s2/2+s+1)
-60 dB
2
1.41

0.1 1 10 100
Frequency (rad/s)
Asymptotic Approximations: Bode Plots
• Example cont.:
 Step 5: For the phase contributions we have:
 2 
L( j )  1.5    j / 2  1      j  1
 2 

• a) for the constant component: 1.5  0

• b) there are no zeros or poles at the origin.
• c) for the simple pole:   j / 2  1  tan 1  / 2 (0° to start 0.1 of
2 (0.2 rad/s) and then -90° after 20 rad/s)
  2
     2

• d) for the quadratic poles:  2

   j  1 

 tan   1 
1




 2
0° at 0.141 rad/s and -180° at 14.1 rad/s (-90 deg/decade).
76
Asymptotic Approximations: Bode Plots
• Example cont.:

Description Frequency (rad/s)

0.141 (start: 0.2 14.1 20
n  1.41 ) (start : (End: ωn= 1.41 (End: pole -2)
pole -2) )

Pole at -2 0 -45 -45 0

ωn= 1.41 -90 -90 0 0
Total slope -90 -135 -45 0
(deg/dec)

1
L( s)  1.5
 s2  s 
  s  1    1
2  2  (c) 2019 Farrokh Sharifi 77
 Asymptotic Approximations: Bode Plots
Bode Diagram
• Example cont.: Phase plot:

1/(s2/2+s+1) 1/(s/2+1)

0
Magnitude (dB)

-90

-180

-270

3.5218 / (s/2+1)(s2/2+s+1)
-360
0.141 14.1 20
0.2

0.1 1 10 100
Frequency (rad/s)
Asymptotic Approximations: Bode Plots
• Example cont.: For comparison the next figure shows the
Bode plots for the same example generated by MATLAB.
Bode Diagram
50

0
Magnitude (dB)

-50

-100

-150
0

-45
Phase (deg)

-90

-135

-180

-225

-270
-2 -1 0 1 2
10 10 10 10 10 79
Frequency (rad/sec)
 Stability, gain margin and phase
margin via Bode plots
(section 10.7 in the book)

• These concepts are based on the Nyquist criterions, which we

will develop in the next section.
• The main advantage of the Bode plots is that the
approximations of the plots can be quickly drawn manually.
• Simplified Concept: KG
– Consider closed loop transfer function 1  KGH
– Closed loop system becomes unstable when
1  KGH  0  KGH  1 | KGH | 1, ( KGH )  180
– For stable systems
| KGH | 1, or 20log | KGH | 0 , and ( KGH )  180
– As we increase gain K, magnitude increases. How much can we
increase K before |KGH| =1 or 20log|KGH|=0 or 20log K +
20log|GH|=0 ? (Gain Margin Concept)
80
(c) 2019 Farrokh Sharifi
 Stability, gain margin and phase
margin via Bode plots
• How much can we decrease phase before ( KGH )  180 ? (Phase
Margin Concept)
• Definitions of Gain Margin (GM) and Phase Margin (PM):

GM: difference between the

magnitude and 0 dB, at the
frequency of –180° phase angle.

PM: difference between the

phase angle and –180°, at the
frequency for which the
magnitude is 0 dB.
 Stability, gain margin and phase
margin via Bode plots
(section 10.7 in the book)

• Example: Find the conditions for stability of the system given

with the open-loop transfer function
K
G (s) 
 s  2  s  4  s  5
• Since all the poles are in the left-half plane, the open-loop is
stable. Therefore, we know that the closed loop will be stable
if the gain is less than unity when the phase angle is –180°.
K K K 1
G (s)   
 s  2  s  4  s  5   s  s  s  40  s  s  s 
(2)(4)(5)   1  1  1   1  1  1
 2  4  5   2  4  5 

K 1
20 log M  20 log( )  20 log
40  s  s  s 
  1  1  1
 2  4  5  82
 Stability, gain margin and phase
margin via Bode plots
Example cont.:
• For the magnitude plot: begin by setting K = 40 or in other
words plot 20logM +20log(K/40) , so that the magnitude at
the low frequencies is zero.
• Break points: 2, 4 and 5. GM=- 20 dB

Each factor contributes:

–20 dB for the magnitude

–80° for the phase

PM=- 80 deg

7 rad/s
 Stability, gain margin and phase
margin via Bode plots
Example cont.:
• From the previous figure we see that the phase angle of -180 is
at 7 rad/s.
• The magnitude at 7 rad/s is –20 dB. Therefore, to maintain less
than unity magnitude (equal to 0 dB) for stability, we can
increase the gain by up to 20 dB.
• What is the gain for the magnitude of 20 dB?
20 = 20 log x → 1 = log x → x = 10,
so the gain for 20 dB is 10.
• Remember that in the beginning we set K = 40, therefore the
range of the gain for stability of the system is:
0< K/40 < 10 or 0 < K < 40×10 = 400.
• The results from the actual plots are:  = 6.16, K = 378.
→ good approximation. 84
 Stability, gain margin and phase
margin via Bode plots
• Example: find the gain and phase margins for the previous
example with K = 200.
GM=-6.02
• The previous plot was for K=40
K=200 is 5 times more thus the
magnitude in 20 log will be
shifted 20log 5 =13.98 dB up.
The gain –20 dB +13.98 = – 6. 02 dB.
Gain margin: 6.02 dB

• Phase margin is at 5.5 rad/s,

(angle –165°) is 15 °.
PM=15 deg

(c) 2019 Farrokh Sharifi 85