You are on page 1of 104

Introduction to Control Engineering

Course code: ECEg4165

1
CHAPTER 4
FREQUENCY RESPONSE AND STABILITY ANALYSIS
4.1. Introduction

➢ Using time domain characteristic it is possible to obtain an idea about the


performance of a given system. If the order of the system increases, it is
very difficult to analyse the system in time domain approach.

➢ Frequency response is concerned with the steady-state response of a


system to a sinusoidal input of fixed amplitude and varying the frequency
over a certain range.

2
Cont. …
➢ If stable, linear, time-invariant system subjected sinusoidal input signal, the
steady-state output will also be a sinusoidal signal of the same frequency,
but with possibly different magnitude and phase angle of the input signal.

p(s) p(s)
➢ Let G s = = , and
q(s) s+s1 s+s2 …(s+sn )

3
Cont. …
𝐱 𝐭 = 𝐀 𝐬𝐢𝐧 𝛚𝐭 −is sinusoidal input signal and y t − is sinusoidal output
signal.

Y(s) Aω
= G s ⇒ Y s = G s X(s), X s =
X(s) s2 +ω2


Y s =G s X s = G(s)
s2 +ω2

k ഥ
k
Y s = + + partial fraction of G s term
s+jω s−jω
due to transfer function
due to input ↑
↑ transient response
steady state response (yss ) (it will die out as time increases
so we can neglect it)

4
Cont. …
➢ For a stable system, s1 , s2 , … , sn have negative real parts. Therefore, as time
approaches infinity, the terms e−s1 t , e−s2 t , … , e−sn t approach zero.

k ഥ
k
Yss (s) = + where, kത − complex conjugate of k.
s+jω s−jω

AωG(s) AG −jω A G(jω)


k = s + jω |s=−jω = − = − e−jϕ
s+jω s−jω 2j 2j

Since, G −jω = G −jω e∠G(−jω) = G jω e−∠G(jω) = G jω e−jϕ and

Imaginary part of G(jω)


ϕ = ∠𝐺 𝑗𝜔 = tan−1
Real part of G(jω)

AωG(s) AG jω A G(jω) A G(jω)


kത = s − jω |s=jω = = e∠G(jω) = ejϕ
s+jω s−jω 2j 2j 2j
5
Cont. …
A A
k ഥ
k − G jω e−jϕ G jω ejϕ
2j 2j
Yss s = + = +
s+jω s−jω s+jω s−jω

A G(jω) A G(jω)
Yss t = −e−jϕ 𝑒 −𝑗𝜔𝑡 + ejϕ 𝑒 𝑗𝜔𝑡 = −𝑒 −𝑗(𝜔𝑡+ϕ) + 𝑒 𝑗(𝜔𝑡+ϕ)
2j 2j

𝑒 𝑗(𝜔𝑡+ϕ) −𝑒 −𝑗(𝜔𝑡+ϕ)
Yss t = A G(jω) = A G(jω) sin 𝜔𝑡 + ϕ ,
2𝑗

Yss t = B sin 𝜔𝑡 + ϕ , where, B = A G(jω)

𝐞𝐣𝛉 − 𝐞−𝐣𝛉
= 𝐣𝐬𝐢𝐧𝛉
𝟐

6
Cont. …

Fig.4.1: Input and output sinusoidal signals.

Where, ϕ − phase shift angle between input and output, ω −Oscillation


frequency and A and B − are magnitudes of input and output respectively.

➢ A positive phase angle is called phase lead while negative phase angle is
called phase lag. 7
Cont. …
➢ Consider the network given by
1
s+
T1
G s =
1
s+
T2
For the sinusoidal input, x t = k sin ωt
a) Find the steady state output.
b) Determine whether this network is a lead network or lag network.
1
jω+T T2 (1+jωT1 ) T2 1+T2 1 ω2
Solution: G jω = 1
1
= , G jω = ,
jω+T T1 (1+jωT2 ) T1 1+T2 1 ω2
2

ϕ = ∠𝐺 𝑗𝜔 = tan−1 ωT1 − tan−1 ωT2


T2 1+T2 1 ω2
a) yss t = k sin ωt + tan−1 ωT1 − tan−1 ωT2
T1 1+T2 1 ω2

b) IF T1 > T2 , then tan−1 ωT1 > tan−1 ωT2 , so the network is lead network. Lag
network if T1 < T2 . 8
Cont. …
❖ Advantages of frequency response analysis

➢ Can infer performance and stability from same plot. It is very easy to
visualize the effects of noise disturbance and parameter variations.

➢ Design process is independent of system order (number of poles).

➢ When the transfer function for a component is unknown, the frequency


response can be determined experimentally with good accuracy and an
approximate expression for the transfer function can be obtained from the
graph of the experimental data.

➢ Graphical techniques (analysis/synthesis) are quite simple.


9
Cont. …
➢ Nyquist stability criterion enables one to investigate both the absolute and
relative stabilities of linear closed-loop systems from a knowledge of their
open-loop frequency-response characteristics.

❖ Dis-advantages of frequency response analysis

➢ These methods are basically applicable only to linear systems.

➢ It is possible to obtain frequency response for an existing system if the time


constants are up to few minutes.

➢ These are time-consuming process.

➢ These methods are back dated compared to the extensive methods


developed for digital computer simulation and modeling. 10
4.2. Frequency Domain Analysis
➢ The magnitude function and phase function are the two plots in frequency
response characteristics. The angular frequency (𝜔) is the common variable
for both these plots.

➢ There are three commonly used curves for representations of sinusoidal


transfer functions in frequency response analysis.

1) Bode diagram or logarithmic plot

2) Nyquist plot or polar plot

3) Log-magnitude-versus-phase plot (Nichols plots)

11
4.3. Frequency domain specifications
➢ Consider the negative feedback system shown in Figure 4.2.

Fig.4.2.

➢ The closed-loop transfer function is given by

C(s) G(s)
Gcl s = =
R(s) 1 + H s G(s)

12
Cont. …

G(jω)
Gcl jω =
1 + H jω G(jω)

❖ Resonant peak (𝐌𝐫 ): The maximum value of Gcl (jω) as ω is varied is termed
as resonance peak (Mr ). It is the peak (maximum) value of Gcl jω .

➢ The relative stability of the closed-loop system can be indicated by Mr . If Mr


is large, there will be large maximum overshoot in a step response.

Mr = Gcl jω |ω=ωr

13
Cont. …
❖ Resonant frequency (𝝎𝒓 ): It is the frequency at which the magnitude of the
frequency response has peak value for the first time. It can obtain from
𝑑
Gcl jω 𝜔=𝜔𝑟 =0
𝑑𝜔

❖ Bandwidth (𝝎𝒃 ): It is the range of frequencies over which, the magnitude of


1
Gcl jω drops to 70.7% (−3 𝑑𝐵 𝑜𝑟 ). It can obtain from
2

1
Gcl jω 𝜔=𝜔𝑏 = = 0.707
2

➢ Small bandwidth means lower frequencies are passed and hence response
is sluggish.
14
Cont. …
➢ If bandwidth is large, higher frequencies will pass through the system. This
indicates a faster rise in time. Large bandwidth indicates susceptibility to
noise.

➢ Therefore, bandwidth gives an idea about transient response. Bandwidth


also implies the sensitivity to parameter variations.

❖ Cut-off rate: It is the slop of the log-magnitude curve near the cut-off
frequency.

➢ The ability of the system to distinguish a signal from noise can be


determined by cut-off rate.
15
Cont. …
❖ Gain margin (𝐆𝐦 ): The value of gain to be added to system in order to bring the
system to the verge of instability.

➢ The factor by which the gain H(jω)G(jω) dB increases without making the system
unstable (the gain required to lift magnitude curve to 0dB).

➢ It can be obtained from the bode plot by

i. Locate phase crossover frequency (ωpc ) on bode plot.

1
ii. Determine, Gm = ; In decibels, Gm │dB = 20 log Gm = −20 log H(jω)G(jω) │ω=ωpc
H(jω)G(jω) │ω=ωpc

➢ Large gain margin is better stability.


➢ Phase crossover frequency or critical frequency (𝝎𝒑𝒄 ) is the frequency at which
phase curve crosses −180° axis.
16
Cont. …
❖ Phase margin (𝐏𝐦 ): Additional phase lag to be added at the gain crossover
frequency in order to bring the system to the verge of instability.

➢ The factor by which the phase in degree can be increased without making the
system unstable. It is a measure of relative stability.

➢ It can be obtained from bode plot by

i. Locate gain crossover frequency (𝜔𝑔𝑐 ) on bode plot.

ii. Calculate, ϕ jω |ω=ωgc

Pm = 180° + ϕ jω |ω=ωgc

➢ Gain crossover frequency (𝝎𝒈𝒄 ) is the frequency at which H(jω)G(jω) 𝑑𝐵 curve


crosses 𝜔 −axis (0 𝑑𝐵 axis or H(jω)G(jω) = 1). 17
4.4. Correlation between time and frequency domain specifications
for second order system
➢ Consider the closed-loop transfer function of the second order control system
Y(s) ω2 n 1
G s = = = s2 2ξ𝑠
R(s) s2 +2ξωn s+ω2 n + +1
ω2 n ωn
1
G jω = 𝐺(𝑠)│𝑠=𝑗𝜔 = ω2 2ξω
1− 2 +j
ω n ωn
2ξω
1 ωn
𝐺(𝑗𝜔) = and ∠𝐺 𝑗𝜔 = − tan−1 ω2
2 1− 2
ω2 2ξω 2 ω n
1− 2 +
ω n ωn

➢ Drive the resonant frequency (𝜔𝑟 ), resonant peak (𝑀𝑟 ) and bandwidth (𝜔𝑏 ).

2 −1Τ2
𝑑 𝑑 ω2 2ξω 2
G(jω) 𝜔=𝜔𝑟 =0 ⇒ 1− + │𝜔=𝜔𝑟 = 0
𝑑𝜔 𝑑𝜔 ω2 n ωn
18
Cont. …
ω
For simplicity let, u =
ωn

d 2 −1Τ2 │ ωr
1 − u2 2 + 2ξu u=ur = 0, where ur =
du ωn

1 2 −3Τ2
− 1 − u2 2
+ 2ξu 2 1 − u2 −2u + 2 2ξu 2ξ │u=ur = 0
2

1 2 −3Τ2
− 1 − u2 2 + 2ξu 4u u2 − 1 + 2ξ2 │u=ur = 0
2

−4u u2 −1+2ξ2
2 3Τ2 |u=ur = 0 ⇒ 4u u2 − 1 + 2ξ2 |u=ur = 0
2 1−u2 2 + 2ξu

4ur u2 r − 1 + 2ξ2 = 0 ⇒ ur = 1 − 2ξ2

ωr
ur = → ωr = ωn ur = ωn 1 − 2ξ2 19
ωn
Cont. …
1 1
Mr = G(jω) |ω=ωr = |ω=ωr = ,
2 2
ω2 2ξω 2 ω2 r 2ξωr 2
1− 2 + ω 1− 2 + ω
ω n n ω n n

Substituting for ωr = ωn 1 − 2ξ2 gives,


𝟏
𝐌𝐫 = for 0 ≤ ξ ≤ 0.707 and 𝐌𝐫 = 1 for, ξ > 0.707 & ωr = 0
𝟐𝛏 𝟏−𝛏𝟐

➢ Resonant frequency is a function of 𝜔𝑛 and ξ, the following conclusions can be


drawn

i. Mr approaches infinity and ωr approaches ωn when ξ approaches zero.

ii. Mr is greater than unity and ωr is less than ωn for 0 < ξ < 0.707
iii. There is no resonant peak and the maximum value of Mr is unity and ωr = 0 and
ξ > 0.707
20
Cont. …
1 1 1
G(jω) |ω=ωb = = 0.707 ⇒ |ω=ωb =
2 2 2
ω2 2ξω 2
1− 2 + ω
ω n n

2
1 1 𝜔2 𝑏 2ξωb 2 ωb
= ⇒ 1− + = 2, let 𝑢𝑏 =
2 2 ω2 n ωn ωn
𝜔2 2ξωb 2
1− 2 𝑏 + ωn
ω n

1 − u2 b 2 + 2ξub 2 = 2 ⇒ u4 b + 4ξ2 − 2 u2 b − 1 = 0, let u2 b = x

x 2 + 4ξ2 − 2 x − 1 = 0 ⇒ x = 1 − 2ξ2 ± 4ξ4 − 4ξ2 + 2

x = 1 − 2ξ2 ± 4ξ4 − 4ξ2 + 2

u2 b = x ⇒ ub = 1 − 2ξ2 ± 4ξ4 − 4ξ2 + 2

21
Cont. …

ωb
ub = ⇒ 𝜔𝑏 = 𝜔𝑛 𝑢𝑏 = ωn 1 − 2ξ2 ± 4ξ4 − 4ξ2 + 2
ωn

➢ Positive sign is Considering because bandwidth is positive quantity.

∴ ωb = ωn 1 − 2ξ2 + 4ξ4 − 4ξ2 + 2

➢ Bandwidth is the function of 𝜔𝑛 and ξ. The following conclusion can be drawn.

i. ωb ∝ ωn

ii. Bandwidth decreases as ξ increases.

iii. For 0 < ξ < 0.707, bandwidth and Mr are directly proportional.

iv. Bandwidth decreases due to increase in rise time. 22


Cont. …
Exercise 4.1:

1) A unity feedback system has the forward transfer function given by G s =


400
. Determine ωn , ξ, ωr , Mr & ωb .
s(s+6)

2) Determine the frequency domain specifications for second-order system


225
with unity feedback G s = .
s s+6

23
4.5. Bode plot and Bode stability criterion
Bode plot
➢ Bode diagram consists of two graphs:

➢ Plot of the logarithm of the magnitude of a sinusoidal transfer function and


plot of the phase angle. Both are plotted against the frequency on a
logarithmic scale.

➢ In the logarithmic representation, the curves are drawn on semi-log paper,


using the log scale for frequency and the linear scale for either magnitude
(in decibels dB) or phase angle (in degrees).

➢ The main advantage of using Bode diagram are: 24


Cont. …

a) Multiplication of magnitudes can be converted into addition.

b) A simple method for sketching an approximate log-magnitude curve is

available.

c) It is based on asymptotic approximations.

d) The experimental determination of a transfer function can be made simple if


frequency-response data are presented in the form of a Bode diagram.

➢ Consider the sinusoidal transfer function is given by

𝐺 𝑗𝜔 = 𝐺 𝑠 |𝑠=𝑗𝜔
25
Cont. …

➢ The basic factors that frequently occur in an arbitrary open-loop transfer

function H jω G(jω) are:

i. Constant gain k

ii. Integral or derivative factors (i.e., pole or zero at origin) jω ±

iii. First order factor (i.e., simple pole or zero) 1 + jωT ±

±
ω2 j2ζω
iv. Quadratic factor (i.e., quadratic pole or zero) 1 − 2 +
ωn ωn

26
Cont. …
➢ Once we become familiar with the logarithmic plots of these basic factors, it
is possible to utilize them to construct a composite logarithmic plot for any
general form of H jω G(jω) by sketching the curves for each factor and
adding individual curves graphically, because adding the logarithms of the
gains corresponds to multiplying them together.

Example: Consider the open-loop transfer function with unity feedback


given by

k T1 s + 1 T2 s + 1 … Tn s + 1
G s =
p1 s + 1 p2 s + 1 … pn s + 1

27
Cont. …
➢ The sinusoidal transfer function is given by

k jT1 ω+1 jT2 ω+1 … jTn ω+1


G jω = G s |s=jω =
jp1 ω+1 jp2 ω+1 … jpn ω+1

k jT1 ω+1 jT2 ω+1 … jTn ω+1


G jω =
jp1 ω+1 jp2 ω+1 … jpn ω+1

G jω |dB = 20 log k + 20 log jT1 ω + 1 + 20 log jT2 ω + 1 + … + 20 log|jTn ω +

1| − 20 log jp1 ω + 1 − 20 log jp2 ω + 1 − ⋯ − 20 log jpn ω + 1

ϕ jω = tan−1 k + tan−1 T1 ω + tan−1 T2 ω + ⋯ + tan−1 Tn ω − tan−1 p1 ω −

tan−1 p2 ω − ⋯ − tan pn ω
28
Cont. …
❖ Bode plot of constant gain 𝐤: Assume unity feedback system

➢ The open-loop transfer function H s G s = G s = k ⇒ G jω = G(s)|s=jω = k

G jω |dB = 20 log k = 20 log k and ϕ jω = 0°, as ω varies from 0 to ∞.

Fig.4.3: Bode plot of constant gain k, (a) Magnitude plot and (b) phase plot.
29
Cont. …
❖ Bode plot of Derivative or Integral factors (zero or pole at origin) 𝒋𝝎 ±:

Assume unity feedback system


➢ Consider open-loop transfer function H s G s = G s = s ⇒ G jω = G(s)|s=jω = jω

𝐺 𝑗𝜔 |𝑑𝐵 = 20 log ω = 20 log ω and ϕ 𝑗𝜔 = 90°, as 𝜔 varies from 0 to ∞.

If 𝜔 = 0.1, 𝐺 𝑗𝜔 |𝑑𝐵 = −20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒

If 𝜔 = 1, 𝐺 𝑗𝜔 |𝑑𝐵 = 0
If 𝜔 = 10, 𝐺 𝑗𝜔 |𝑑𝐵 = 20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒
1 1
➢ Consider open-loop transfer function G s = ⇒ G jω =
𝑠 𝑗𝜔
1
𝐺 𝑗𝜔 |𝑑𝐵 = 20 log = −20 log ω and ϕ 𝑗𝜔 = −90°, as 𝜔 varies from 0 to ∞.
𝜔2

If 𝜔 = 0.1, 𝐺 𝑗𝜔 |𝑑𝐵 = 20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒

If 𝜔 = 1, 𝐺 𝑗𝜔 |𝑑𝐵 = 0
If 𝜔 = 10, 𝐺 𝑗𝜔 |𝑑𝐵 = −20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒
30
Cont. …

1
Fig.4.4: Bode diagram of (a) 𝐺 𝑗𝜔 = jω, (b) 𝐺 𝑗𝜔 =
𝑗𝜔 31
Cont. …
➢ In Bode diagrams, frequency ratios are expressed in terms of octaves or decades. An
octave is a frequency band from ω1 to 2ω1 and decade is a frequency band from ω1 to
10ω1 , where ω1 , is any frequency value.

1
➢ The slopes of the log-magnitude curves for the factors jω and are 20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒

and −20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒 respectively.
1 𝑛
➢ Note: If the transfer function contains the factor jω 𝑛 and , the log magnitude

becomes, respectively,
1
20 log jω 𝑛
= 20n log jω = 20𝑛 log 𝜔 𝑑𝐵 and 20 log = −20n log jω = −20𝑛 log 𝜔 𝑑𝐵
jω 𝑛

Also the phase angle becomes ϕ 𝑗𝜔 = 𝑛90° and ϕ 𝑗𝜔 = −𝑛90° respectively.


1 𝑛
➢ The slopes of the log-magnitude curves for the factors jω 𝑛 and are

20𝑛 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒 and −20𝑛 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒 respectively.
32
Cont. …
❖ Bode plot of First order factor (zero or pole on real axis) 𝟏 + 𝒋𝝎𝑻 ± : Assume unity
feedback system.

➢ Consider open-loop transfer function H s G s = G(s) = Ts + 1 ⇒ 𝐺 jω = 1 + jωT

G jω |dB = 20 log 1 + jωT = 20 log 1 + T 2 ω2

For ωT ≪ 1 (ω → 0), G jω |dB = 20 log 1 = 20 log 1 = 0 dB

For ωT = 1 (ω = 1ΤT), G jω |dB = 20 log 2 = 0 dB but exact value is 𝟑 𝐝𝐁

ω = 1ΤT − is corner frequency or break frequency. It is frequency at which the two


asymptotes meet. It is low frequency and high frequency expression.

For ωT = 10 (ω = 10ΤT), G jω |dB = 20 log 101 = 20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒

For ωT ≫ 1 (ω ≫ 1ΤT), G jω |dB = 20 log ωT dB 33


Cont. …
−1 ωT
ϕ 𝑗𝜔 = tan = tan−1 ωT
1

For ωT ≪ 1, ϕ 𝑗𝜔 = tan−1 0 = 0°

For ωT = 1, ϕ 𝑗𝜔 = tan−1 1 = 45°

For ωT ≫ 1, ϕ 𝑗𝜔 = tan−1 ∞ = 90°

1 1
➢ Consider open-loop transfer function G s = ⇒ 𝐺 jω =
Ts++1 1+jωT

1 1
G jω |dB = 20 log = 20 log = −20 log 1 + T 2 ω2 𝑑𝐵
1+jωT 1+T2 ω2

For ωT ≪ 1 (ω → 0), G jω |dB = −20 log 1 = −20 log 1 = 0 dB

For ωT = 1 (ω = 1ΤT), G jω |dB = −20 log 2 = 0 dB but exact value is −𝟑 𝐝𝐁


34
Cont. …
ω = 1ΤT − is corner frequency or break frequency. It is frequency at which the two

asymptotes meet. The low frequency and high frequency expression.

➢ Cut off frequency (corner frequency or break frequency) is the frequency at which the

magnitude of closed loop system is 3 𝑑𝐵 less than its maximum.

For ωT = 10 (ω = 10ΤT), G jω |dB = −20 log 101 = −20 𝑑𝐵 Τ𝑑𝑒𝑐𝑎𝑑𝑒

For ωT ≫ 1 (ω ≫ 1ΤT), G jω |dB = −20 log ωT dB


−1 ωT
ϕ 𝑗𝜔 = − tan = − tan−1 ωT
1

For ωT ≪ 1, ϕ 𝑗𝜔 = − tan−1 0 = 0°

For ωT = 1, , ϕ 𝑗𝜔 = − tan−1 1 = −45°

For ωT ≫ 1, ϕ 𝑗𝜔 = − tan−1 ∞ = −90° 35


Cont. …

Fig.4.5:Log-magnitude curve, together with the asymptotes, and phase-angle curve


1
of, (a) 𝐺 jω = 1 + jωT , (b) 𝐺 jω = . 36
1+jωT
Cont. …
❖ Bode plot of quadratic factors (complex zero pair or complex pole pair)
±
𝝎𝟐 𝒋𝟐𝜻𝝎
𝟏 − 𝟐 + : Assume unity feedback system.
𝝎𝒏 𝝎𝒏

𝑗𝜔 𝑗𝜔 2 𝜔2 𝑗2𝜁𝜔
➢ Consider second order zero 𝐺 𝑗𝜔 = 1 + 2𝜁 + =1− +
𝜔𝑛 𝜔𝑛 𝜔𝑛 2 𝜔𝑛

• If 𝜁 > 1, this quadratic factor can be expressed as a product of two first-order


factors with real roots. If 𝟎 < 𝜻 < 𝟏 , this quadratic factor is the product of two
complex conjugate factors.

• The magnitude and phase of the quadratic factor depend on both the corner
frequency and the damping ratio.

• Damping ratio determine magnitude resonant peak. It is large for small value of ζ.
37
Cont. …
2
𝜔2 𝑗2𝜁𝜔 𝜔 2 2𝜁𝜔 2
G jω |dB = 20 log 1 − + = 20 log 1− + 𝑑𝐵
𝜔𝑛 2 𝜔𝑛 𝜔𝑛 𝜔𝑛

𝜔 2
For 𝜔 → 0 𝜔 ≪ 𝜔𝑛 ⇒ ≪ 1 , G jω |dB = 20 log 1 = 20 log 1 = 0 𝑑𝐵
𝜔𝑛

𝜔 2 𝜔 4
For high frequency 𝜔 ≫ 𝜔𝑛 , 4𝜁 2 ≪ because 𝜁 is very low.
𝜔𝑛 𝜔𝑛

𝜔 4 𝜔
G jω |dB = 20 log = 40 log dB (this the straight line of slope 40 𝑑𝐵Τ𝑑𝑒𝑐𝑎𝑑𝑒.
𝜔𝑛 𝜔𝑛
corner frequency is 𝜔 = 𝜔𝑛 , G jω |dB = 0 dB).
2𝜁𝜔
−1 𝜔𝑛
ϕ 𝑗𝜔 = tan
𝜔 2
1−
𝜔𝑛
0
For 𝜔 → 0, ϕ 𝑗𝜔 = tan−1 = 0°
1
38
Cont. …
𝜔 2𝜁 2𝜁
For 𝜔 = 𝜔𝑛 ( = 1), ϕ 𝑗𝜔 = tan−1 = tan−1 = tan−1 ∞ = 90°
𝜔𝑛 1−1 0

𝜔
For → ∞, ϕ 𝑗𝜔 = 180°
𝜔𝑛

𝜔
For > 1, ϕ 𝑗𝜔 − is becomes negative. Thus after this, the angle contribution
𝜔𝑛

obtained by adding 180° on the resulting negative angle.

• This is applicable if roots of quadratic is complex conjugate (0 < 𝜁 < 1) otherwise


factorize roots and treat separately as simple factor.

1 1
❖ Consider second order zero 𝐺 𝑗𝜔 = 𝑗𝜔 𝑗𝜔 2
= 𝜔2 𝑗2𝜁𝜔
1+2𝜁 + 1− 2 +
𝜔𝑛 𝜔𝑛 𝜔𝑛 𝜔𝑛

39
Cont. …
2
1 𝜔 2 2𝜁𝜔 2
G jω |dB = 20 log 𝜔2 𝑗2𝜁𝜔
= −20 log 1− + 𝑑𝐵
1− 2 + 𝜔 𝜔𝑛 𝜔𝑛
𝜔𝑛 𝑛
𝜔
For ≪ 1, G jω |dB = −20 log 1 = −20 log 1 = 0 𝑑𝐵
𝜔𝑛

𝜔 𝜔 2 𝜔 4
For high frequency 𝜔 ≫ 𝜔𝑛 ( ≫ 1), 4𝜁 2 ≪ because 𝜁 is very low.
𝜔𝑛 𝜔𝑛 𝜔𝑛

𝜔 4 𝜔
G jω |dB = −20 log = − 40 log (this the straight line of slope
𝜔𝑛 𝜔𝑛
− 40 𝑑𝐵Τ𝑑𝑒𝑐𝑎𝑑𝑒. Corner frequency is 𝜔 = 𝜔𝑛 , G jω |dB = 0 dB.
2𝜁𝜔
𝜔𝑛
ϕ 𝑗𝜔 = − tan−1 𝜔 2
1−
𝜔𝑛

0
For 𝜔 → 0, ϕ 𝑗𝜔 = tan −1 = 0°
1

40
Cont. …
𝜔 2𝜁 2𝜁
For 𝜔 = 𝜔𝑛 ( = 1), ϕ 𝑗𝜔 = − tan−1 = − tan−1 = − tan−1 ∞ = −90°
𝜔𝑛 1−1 0

𝜔
For → ∞, ϕ 𝑗𝜔 = −180°
𝜔𝑛

𝜔
For > 1, ϕ 𝑗𝜔 − is becomes positive. Thus in such case the angle contribution
𝜔𝑛

obtained by subtracting 180° from this positive ϕ 𝑗𝜔 angle.

➢ This is applicable only if the roots of quadratic are complex conjugate.

𝜔
Example: = 2, 𝜁 = 0.3, ϕ 𝑗𝜔 = − tan−1 −0.4 = 21.8°. Thus the actual angle is
𝜔𝑛

21.8° − 180° = −158.2°

41
Cont. …

Fig.4.6: Log-magnitude curve, together with the asymptotes, and phase-angle


𝜔2 𝑗2𝜁𝜔 1
curve of, (a) 𝐺 jω = 1 − 𝜔 2 + 𝜔 , (b) 𝐺 jω = 𝜔2 𝑗2𝜁𝜔
𝑛 𝑛 1− 2 +
𝜔𝑛 𝜔𝑛 42
Cont. …
❖ General procedures for plotting Bode diagrams
➢ First rewrite the sinusoidal transfer function 𝐻 𝑗𝜔 𝐺(𝑗𝜔) as a product of
basic factors discussed above.

➢ Then identify the corner frequencies associated with these basic factors.

➢ Finally, draw the asymptotic log-magnitude curves with proper slopes


between the corner frequencies.

➢ The exact curve, which lies close to the asymptotic curve, can be obtained
by adding proper corrections.

➢ Similarly, the phase-angle curve of 𝐻 𝑗𝜔 𝐺(𝑗𝜔) can be drawn by adding the


phase-angle curves of individual factors.
43
Cont. …
Example:
1) Draw the Bode diagram for the unity feedback system with the forward transfer
function given by :
10(𝑠 + 4)
𝐺 𝑠 =
𝑠 + 2 𝑠 + 10
Solution:
𝑗𝜔
10 𝑗𝜔+4 10×4 4 +1 2 𝑗0.25𝜔+1
H j𝜔 𝐺 𝑗𝜔 = 𝐺(𝑗𝜔) = 𝐺 𝑠 |𝑠=𝑗𝜔 = = 𝑗𝜔 𝑗𝜔 =
𝑗𝜔+2 𝑗𝜔+10 2×10 +1 +1 𝑗0.5𝜔+1 𝑗0.1𝜔+1
2 10

➢ Constant: 𝐺 𝑗𝜔 = k = 2 , 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 2 = 6.02 dB

➢ Zero: 𝐺 𝑗𝜔 = 𝑗0.25𝜔 + 1, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 1 + 0.25𝜔 2

For 𝜔 → 0, 𝐺 𝑗𝜔 𝑑𝐵 = 0 dB
For 𝜔 = 4, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 1 + 0.25 × 4 2 = 20 log 2 = 3 dB
𝜔 = 4 − is corner frequency
44
Cont. …

1 1
➢ Pole: G jω = and G jω = , G jω dB = −20 log 1 + 0.5ω 2 and
j0.5ω+1 j0.1ω+1

G jω = −20 log 1 + 0.1ω 2


dB

For ω → 0, both of them are G jω dB = 0 dB

ω = 2 and ω = 10 are their corner frequency respectively.

➢ The overall magnitude is given by

G jω = 20 log 2 + 20 log 1 + 0.25ω 2 − 20 log 1 + 0.5ω 2 − 20 log 1 + 0.1ω 2


dB

45
Cont. …

46
Cont. …
ϕ 𝑗𝜔 = tan−1 0 + tan−1 0.25𝜔 − tan−1 0.5𝜔 − tan−1 0.1𝜔

47
Cont. …
2) Draw the Bode diagram for the unity feedback system with forward transfer
100(𝑠+1)
function given by: 𝐺 𝑠 =
𝑠 2 +110𝑠+1000

100(𝑠+1) 100(𝑠+1)
Solution:H s 𝐺 𝑠 = 𝐺(𝑠) = =
𝑠 2 +110𝑠+1000 (𝑠+10)(𝑠+100)

100(𝑗𝜔+1) 100(𝑗𝜔+1) (𝑗𝜔+1)


𝐺 𝑗𝜔 = 𝐺 𝑠 |𝑠=𝑗𝜔 = = 𝑗𝜔 𝑗𝜔 = 0.1
(𝑗𝜔+10)(𝑗𝜔+100) 10×100( +1)( +1) (𝑗0.1𝜔+1)(𝑗0.01𝜔+1)
10 100

➢ Constant: 𝐺 𝑗𝜔 = k = 0.1 , 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 0.1 = −20 dB

➢ Zero: 𝐺 𝑗𝜔 = 𝑗𝜔 + 1, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 1 + 𝜔 2

For 𝜔 → 0, 𝐺 𝑗𝜔 𝑑𝐵 = 0 dB
For 𝜔 = 1, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 1 + 1 2 = 20 log 2 = 3 dB
𝜔 = 1 − is corner frequency 48
Cont. …
1 1
➢ Pole: 𝐺 𝑗𝜔 = and 𝐺 𝑗𝜔 = , 𝐺 𝑗𝜔 𝑑𝐵 = −20 log 1 + 0.1𝜔 2
(𝑗0.1𝜔+1) (𝑗0.01𝜔+1)

and 𝐺 𝑗𝜔 𝑑𝐵 = −20 log 1 + 0.01𝜔 2

For 𝜔 → 0, both of them are 𝐺 𝑗𝜔 𝑑𝐵 = 0 dB


𝜔 = 10 and 𝜔 = 100 are their corner frequency respectively.

49
Cont. …
ϕ 𝑗𝜔 = tan−1 0 + tan−1 𝜔 − tan−1 0.1𝜔 − tan−1 0.01𝜔 (deg)

50
Cont. …
3) Draw the Bode Diagram for the unity feedback system with forward transfer
120𝑆
function given by: 𝐺 𝑠 =
(𝑆+3)(𝑆+6)

120𝑆
Solution: H s 𝐺 𝑠 = 𝐺(𝑠) =
(𝑆+3)(𝑆+6)

120 𝑗𝜔 120(𝑗𝜔) 𝑗𝜔
𝐺 𝑗𝜔 = 𝐺 𝑠 |𝑠=𝑗𝜔 = = 𝑗𝜔 𝑗𝜔 = 20Τ3 𝑗𝜔 𝑗𝜔
(𝑗𝜔+3)(𝑗𝜔+6) 3×6( +1)( +1) ( +1)( +1)
3 6 3 6

➢ Constant: 𝐺 𝑗𝜔 = k = 20Τ3 , 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 20Τ3 = 16.5 dB

➢ Zero: 𝐺 𝑗𝜔 = 𝑗𝜔, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 𝜔


For 𝜔 → 0.1, 𝐺 𝑗𝜔 𝑑𝐵 = −20 dB
For 𝜔 = 1, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 1 = 0 dB
𝜔 = 1 − is corner frequency
51
Cont. …

1 1 𝜔2
➢ Pole: 𝐺 𝑗𝜔 = 𝑗𝜔 and 𝐺 𝑗𝜔 = 𝑗𝜔 , 𝐺 𝑗𝜔 𝑑𝐵 = −20 log 1 + and
( 3 +1) ( 6 +1) 9

𝜔2
𝐺 𝑗𝜔 𝑑𝐵 = −20 log 1 +
36

For 𝜔 → 0, both of them are 𝐺 𝑗𝜔 𝑑𝐵 = 0 dB

𝜔 = 3 and 𝜔 = 6 are their corner frequency respectively.

−1 𝜔 −1 𝜔
• ϕ 𝑗𝜔 = 90° − tan − tan
3 6
52
Cont. …

53
Cont. …
4) Draw the Bode Diagram for the unity feedback system with the forward transfer
120𝑠 2
function given by: 𝐺 𝑠 =
(𝑆+3)(𝑆+6)

100𝑠 2
Solution: H(s)𝐺 𝑠 =
(𝑆+3)(𝑆+6)

100 𝑗𝜔 2 100 𝑗𝜔 2 −𝜔2


𝐺 𝑗𝜔 = 𝐺 𝑠 |𝑠=𝑗𝜔 = = 𝑗𝜔 𝑗𝜔 = 50Τ9 𝑗𝜔 𝑗𝜔
(𝑗𝜔+3)(𝑗𝜔+6) 3×6( 3 +1)( 6 +1) ( 3 +1)( 6 +1)

➢ Constant: 𝐺 𝑗𝜔 = k = 50Τ9 , 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 50Τ9 = 14.9 dB

➢ Zero: 𝐺 𝑗𝜔 = −𝜔2 , 𝐺 𝑗𝜔 𝑑𝐵 = 20 log −𝜔2 = 20 log 𝜔2 = 40 log 𝜔


For 𝜔 → 0.1, 𝐺 𝑗𝜔 𝑑𝐵 = −40 dB
For 𝜔 = 1, 𝐺 𝑗𝜔 𝑑𝐵 = 20 log 1 = 0 dB
𝜔 = 1 − is corner frequency 54
Cont. …

1 1 𝜔2
➢ Pole: 𝐺 𝑗𝜔 = 𝑗𝜔 and 𝐺 𝑗𝜔 = 𝑗𝜔 , 𝐺 𝑗𝜔 𝑑𝐵 = −20 log 1 + and
( 3 +1) ( 6 +1) 9

𝜔2
𝐺 𝑗𝜔 𝑑𝐵 = −20 log 1 +
36

For 𝜔 → 0, both of them are 𝐺 𝑗𝜔 𝑑𝐵 = 0 dB

𝜔 = 3 and 𝜔 = 6 are their corner frequency respectively.

−1 𝜔 −1 𝜔
• ϕ 𝑗𝜔 = 180° − tan − tan
3 6
55
Cont. …

➢ The bode plot is given as follow:

56
Cont. …

Exercise 4.2: Draw the bode diagram of the following unity feedback system with

the forward transfer functions given by,

10(s+3) 20
a) G s = b) 𝐺 𝑠 =
s s+2 s2 +s+2 𝑠2 𝑠+3 𝑠+4

10(s+4) 100𝑠2 𝑠+1


c) G s = d) 𝐺 𝑠 =
s+2 s+10 𝑠+4 𝑠2 +6𝑠+5

s(s+2)
e) G s =
s+2 s2 +4s+3

Exercise 4.3: Give your own examples and show clearly how the transfer function
is obtained from bode plot of the system.
57
Cont. …

Bode stability criterion


➢ The system is stable if the magnitude of H(jω)G(jω) at phase crossover
frequency (ωpc ) is negative ( i.e. additional gain is required to bring
𝐻(𝑗𝜔)G(jω) to 0 dB). This means if gain margin is positive the system is
stable and if the gain margin is negative (i.e. the magnitude of H(jω)G(jω) is
positive) the system is unstable.

➢ If the phase margin is positive then the system is stable otherwise unstable.

➢ For the system to be stable both gain margin and phase margin needs to be
greater than zero.
58
Cont. …
Bode stability requirement
❖ For a given open-loop transfer function (𝐻 𝑠 𝐺(𝑠)), the closed-loop system is stable
if the following conditions are met.

➢ Open-loop transfer function is strictly proper (more poles than zeros).

➢ Minimum Phase System: has no poles located on or to the right of the imaginary
axis, except of a single pole at the origin.

➢ Assume that the open-loop frequency response has only a single phase crossover
frequency and a single gain crossover frequency.

❖ Consider the bode diagram of a unity feedback system shown in Figure 4.7 below

59
Cont. …

Fig.4.7: a) Stable system b) Unstable system 60


Cont. …

Exercise 4.4: For a unity feedback system with forward transfer function

50
𝐺 𝑠 =
𝑠(0.4𝑠+1)

a) Find the gain crossover frequency (𝜔𝑔𝑐 ) and phase margin (𝑃𝑚 ),

b) Find ωn , ξ, ωr , Mr & ωb

61
4.6. Log-magnitude-versus-phase plots (Nichols Plots) and Stability
❖ Log-magnitude-versus-phase plots

➢ It is a plot of the logarithmic magnitude in decibels versus the phase angle or phase
margin for a frequency range of interest.

➢ The phase margin is the difference between the actual phase angle ϕ jω and −180°, that
is, ϕ jω − −180° = 180° + ϕ jω .

➢ Such log-magnitude-versus-phase plots are commonly called Nichols plots.

➢ A change in the gain constant of G jω , merely shifts log-magnitude-versus-phase plot


curve up (for increasing gain) or down (for decreasing gain), but the shape of the curve
remains the same.

➢ Advantages of the log-magnitude vs phase plot are the relative stability of the closed-loop
system can be determined quickly and compensation can be worked out easily. 62
Cont. …
Example: For the following sinusoidal transfer function sketch the log-magnitude
in decibels versus phase plot in degree.
1 1
a) G jω = b) G jω = c) G jω = 1 + jωT
jω 1+jωT
1 1
d) G jω = e−jωL e) G jω = f) G jω = ω2 2ξω
for 𝛏 > 0
jω(1+jωT) 1− 2 +j
ω n ωn

Solution:
1 1
a) G jω dB = 20 log = 20 log = −20 log ω and ϕ jω = −90°
ω2 ω

If ω → 0, G jω dB → +ve

If ω = 1, G jω dB =0

If ω → ∞, G jω dB → −ve
63
Cont. …
1
b) G jω dB = 20 log = −20 log 1 + (ωT)2 and ϕ jω = − tan−1 ωT
1+(ωT)2

If ω = 0, G jω dB = 0 dB and ϕ jω = 0°

If ωT = 1, G jω dB = −3 dB and ϕ jω = −45°

If ω → ∞, G jω dB → −∞ and ϕ jω = −90°

64
Cont. …
d) G jω dB = 20 log 1 = 0 dB and ϕ jω = −ω𝐿

If ω = 0, G jω dB = 0 dB and ϕ jω = 0°

If ω → ∞, G jω dB = 0 dB and ϕ jω = −∞

65
Cont. …
2ξω
1 ω2 2 2ξω 2 ωn
f) G jω dB = 20 log = −20 log 1− +( ) and ϕ jω = − tan−1 ω2
2 ω2 n ωn 1− 2
ω2 2ξω 2 ω n
1− 2 +( )
ω n ωn

If ω = 0, G jω dB = 0 dB and ϕ jω = 0°
If ω = ωn , G jω dB = −20 log 2𝜉 𝑑𝐵 and ϕ jω = −90°
𝜔
If → ∞, G jω dB = −∞ and ϕ jω = −180°
𝜔𝑛

If ω = ωr , G jω dB = Mr

66
Cont. …

❖ Nichols Plot Stability Criterion


➢ From Nichols plot based on gain margin and phase margin, we can determine the
stability of the system.

➢ If both gain margin and phase margin are positive the system is stable otherwise
unstable.

67
Cont. …
Exercise 4.5: For the following open-loop transfer functions sketch log-magnitude
versus phase plot.

2 50
a) G s = b) G s =
s(s+1)(s+2) s(s+5)

68
4.6. Polar Plot and stability
❖ Polar plot

➢ Polar plot of a sinusoidal transfer function G(jω) is a plot of the magnitude of

G(jω) versus the phase angle of G(jω) on polar coordinates as ω is varied

from zero to infinity.

➢ Thus, the polar plot is the locus of vectors G(jω) ∠G(jω) as ω is varied from

zero to infinity.

➢ Note that in polar plots a positive (negative) phase angle is measured

counterclockwise (clockwise) from the positive real axis. 69


Cont. …
➢ The polar plot is often called the Nyquist plot.

➢ One disadvantage is that the plot does not clearly indicate the contributions
of each individual factor of the open-loop transfer function.

❖ The steps to draw the polar plot:

➢ Given 𝐺(𝑠)

i. Obtain 𝐺 𝑗𝜔 = 𝐺(𝑠)│𝑠=𝑗𝜔

ii. Write 𝐺 𝑗𝜔 both in rectangular(by rationalizing) and polar form as


𝐺 𝑗𝜔 = 𝑅𝑒 𝐺 𝑗𝜔 + 𝑗𝐼𝑚 𝐺 𝑗𝜔 = 𝐺 𝑗𝜔 ∠𝐺 𝑗𝜔

70
Cont. …
iii. Determine 𝐺 𝑗𝜔 as well as ∠𝐺 𝑗𝜔 at frequency 𝜔 = 0 and 𝜔 = ∞ using limit.

𝐺 𝑗𝜔 𝜔→0 𝐺 𝑗𝜔 𝜔→∞
ቊ and ቊ
∠𝐺 𝑗𝜔 𝜔→0 ∠𝐺 𝑗𝜔 𝜔→∞

iv. Obtain R e −axis and jIm − axis intercepts.


❖ To obtain 𝐑 𝐞 −axis intercept:
jIm G jω = 0, so that find the corresponding frequency let say ωx
➢ Then find, R e G jω ω=ωx

❖ To obtain 𝐣𝐈𝐦 −axis intercept:


R e G jω = 0, so that find the corresponding frequency let say ωy
➢ Then find, jIm G jω ω=ωy

v. Draw the polar plot from step (iii) & (iv). 71


Cont. …
Example: Draw the polar plot of the following sinusoidal transfer functions.

1
a) G jω = b) G jω = jω

1
c) G jω = d) G jω = 1 + jωT
1+jωT
e−jωL 1
e) G jω = f) G jω = jω 2
, for ξ > 0
1+jωT jω
1+2ξ +
ωn ωn

Solution:
1 1 −jω 1
a) G jω = = × = −j
jω jω −jω ω
1 1
G jω = −j = ∠ − 90°
ω ω
lim G jω = −𝑗∞ = ∞∠ − 90°
𝜔→0
lim G jω = −𝑗0 = 0∠ − 90°
𝜔→∞
72
Cont. …
b) G jω = jω = ω∠90°

lim G jω = 𝑗0 = 0∠90°
𝜔→0

lim G jω = 𝑗∞ = ∞∠90°
𝜔→∞
1
c) G jω =
1+jωT
1 1−jωT
G jω = ×
1+jωT 1−jωT
1 ωT 1 −1 ωT
G jω = − j = ∠ − tan
1+ ωT 2 1+ ωT 2 1+ ωT 2

lim G jω = 1 − j0 = 1∠0°
ω→0
lim G jω = 0 − j0 = 0∠ − 90°
ω→∞
1
lim G jω = 1Τ2 − j 1Τ2 = ∠ − 45° 73
ωT=1 2
Cont. …
d) G jω = 1 + jωT
G jω = 1 + jωT = 1 + ωT 2 ∠ tan−1 ωT
lim G jω = 1 + j0 = 1∠0°
ω→0
lim G jω = 1 + j∞ = ∞∠90°
ω→∞

e−jωL
e) G jω =
1+jωT

1 1−jωT
G jω = e−jωL ×
1+jωT 1−jωT

1 ωT 1
G jω = −j e−jωL = ∠ − ωL − tan −1 ωT
1+ ωT 2 1+ ωT 2 1+ ωT 2

lim G jω = 1 − j0 = 1∠0°
ω→0

lim G jω = 0 − j0 = 0∠ − ∞°
ω→∞
74
Cont. …
1
f) G jω = jω jω 2
, for ξ > 0
1+2ξ +
ωn ωn

ω 2 ω
1 1 1− −j2ξ
ωn ωn
G jω = ω 2 ω
= ω 2 ω
× ω 2 ω
1− ωn
+j2ξωn
1− ωn
+j2ξωn
1− ωn
−j2ξ ωn

ω 2 ω
1− ω 2ξω
n n
G jω = 2 −j 2
ω 2 ω 2 ω 2 ω 2
1− + 2ξω 1− + 2ξω
ωn n ωn n

ω
1 2ξω
G jω = ∠ − tan−1 n
ω 2
2 1−
ω 2 ω 2 ωn
1− ωn
+ 2ξω
n

lim G jω = 1 − j0 = 1∠0° lim G jω = 0 − j 1Τ2ξ = 1Τ2ξ ∠ − 90°


ω→0 ω=ωn
ω
2ξω r
lim G jω = −0 − j0 = 0∠ − 180° lim G jω = Mr ∠ − tan−1 n
ωr 2
ω→∞ ω=ωr 1− ωn 75
Cont. …
Exercise 4.6: Draw the polar plot of the following transfer functions.

1 1
a) G s = b) G s =
s(Ts+1) T1 s+1 T2 s+1

1 k
c) G s = d) G s =
T1 s+1 T2 s+1 (T3 s+1) s(Ts+1)

1 1
e) G s = f) G s =
s2 T1 s+1 T2 s+1 s+4 s+2

jω jω 2 20
g) G jω = 1 + 2ξ + , for ξ > 0 h) 𝐺 𝑠 =
ωn ωn 𝑠(𝑠+1)(𝑠+2)

20
i) G s =
s(s+1)

76
Cont. …
❖ Polar plot stability criterion
➢ Gain margin and phase margin can be determined from, the polar plot by drawing
unit circle.

➢ In the polar plot, a line may be drawn from the origin to the point at which the unit
circle crosses the 𝐺(𝑗𝜔)1ocus. The angle from the negative real axis to this line is
the phase margin. If the phase margin is positive the system is stable and if phase
margin is negative the system is unstable.

Pm = 180° + ϕ jω │ω=ωgc

➢ Gain margin is the reciprocal of the magnitude 𝐺(𝑗𝜔) at phase crossover


frequency at which the phase angle of the open-loop transfer function equals
− 180°. 77
Cont. …
1
Gm =
G(jω) │𝜔=𝜔𝑝𝑐

➢ In terms of decibels the gain margin is given by

𝐺𝑚 │𝑑𝐵 = 20 log 𝐺𝑚 = −20 log G(jω) │𝜔=𝜔𝑝𝑐

➢ The gain margin expressed in decibels is positive if 𝐺𝑚 > 1 or G(jω) │𝜔=𝜔𝑝𝑐 < 1

and negative if 𝐺𝑚 < 1 or G(jω) │𝜔=𝜔𝑝𝑐 > 1.

➢ Thus, a positive gain margin (in decibels) means that the system is stable, and a
negative gain margin (in decibels) means that the system is unstable.

Example: Consider the polar plot shown below with unit circle.
78
Cont. …

➢ The point ‘A’ corresponds to phase crossover frequency (i.e., ϕ jω = −180°), while
point ‘B’ corresponds to gain crossover frequency (i.e., G(jω) = 1).

➢ Depending on the location of A and B there are four possible situations:

i. The plot may not cross real axis (point ‘A’ doesn’t exists).

➢ Gm = ∞, the system is more stable (larger gain margin is more stable).


79
Cont. …
➢ The gain margin of a first- or second-order system is infinite since the polar plots
for such systems do not cross the negative real axis. Thus, theoretically, first or
second order systems cannot be unstable. However, first or second-order systems
are only approximations in the sense that small time lags are neglected in deriving
the system equations and not truly first or second-order systems.

➢ If these small lags are accounted for, the so-called first or second-order systems
may become unstable.

ii. The plot may cross 𝑅𝑒 −axis between −1 + 𝑗0 and 0 + 𝑗0.

Example: The plot cross 𝑅𝑒 −axis at − 1Τ2 + 𝑗0 , then G(jω) = 1Τ2

𝐺𝑚 │𝑑𝐵 = −20 log G jω = −20 log 1Τ2 = 20 log 2, Since 𝐺𝑚 > 0, the system is stable.
80
Cont. …
iii. The plot may cross 𝑅𝑒 −axis at −1 + 𝑗0. Then G(jω) = 1
𝐺𝑚 │𝑑𝐵 = −20 log G jω = −20 log 1 = 0 dB , the system is marginally stable.
iv. The plot may cross 𝑅𝑒 −axis beyond −1 + 𝑗0.
Example: The plot cross 𝑅𝑒 −axis at −2 + 𝑗0 , then G(jω) = 2

𝐺𝑚 │𝑑𝐵 = −20 log G jω = −20 log 2 = −6 dB, Since 𝐺𝑚 < 0, the system is unstable.

Exercise 4.7: Identify which of the following system is stable and unstable.

81
Cont. …
Exercise 4.8: For the system shown below by block diagram, sketch the
polar plot for k = 30, 60 & 120 and from polar plot determine whether the
closed-loop system is stable for each value of this gain.

82
4.8. Nyquist plot and Nyquist stability criterion
❖ Nyquist plot
➢ Nyquist plot is the mirror image of polar plot. It is based on polar plot.
k
Example: Draw the Nyquist plot of G s =
s(Ts+1)
k
G jω =
jω jωT+1

k −jω 1−jωT
G jω = ×
jω 1+jωT −jω 1−jωT

kT k k −1 ωT
G jω = − −j = ∠ − 90° − tan
1+ ωT 2 ω 1+ ωT 2 ω 1+ ωT 2

lim 𝐺(𝑗𝜔) = −kT − j∞ = ∞∠ − 90°


𝜔→0
lim 𝐺(𝑗𝜔) = −0 − j0 = 0∠ − 180°
𝜔→∞

➢ Then drawing the mirror image of polar plot.


83
Cont. …
❖ Nyquist stability criterion
➢ Using Nyquist stability criterion, it is possible to determine the stability of
closed-loop system from open-loop poles without knowing the roots of
characteristic equation.

➢ Consider the closed-loop system shown in Figure 4.8.

Fig.4.8: Closed-loop system. 84


Cont. …
➢ The closed-loop transfer function is

Y(s) G(s)
=
R(s) 1 + H s G(s)

➢ The characteristic equation is given by


Q s =1+H s G s =0

➢ For stability of the closed-loop system all the roots of the characteristic
equation must lie to the left half of s-plane.

k s+z1 s+z2 … s+zm


Let, H s G s = , m≤n (4.1)
s+p1 s+p2 … s+pn

85
Cont. …
➢ The characteristic equation is

k s + z1 s + z2 … s + zm
𝑄 𝑠 =1+H s G s =1+ =0
s + p1 s + p2 … s + pn

s + p1 s + p2 … s + pn + 𝑘 s + z1 s + z2 … s + zm
𝑄 𝑠 = =0
s + p1 s + p2 … s + pn

𝐬+𝐳 ′ 𝟏 𝐬+𝐳 ′ 𝟐 … 𝐬+𝐳 ′ 𝐦


𝐐 𝐬 = =𝟎 (4.2)
𝐬+𝐩𝟏 𝐬+𝐩𝟐 … 𝐬+𝐩𝐧

➢ From equation (4.1) and (4.2), we find that the poles of Q(s) are the same as the
poles of open-loop transfer function and zeros of Q(s) are the root of characteristic
equation. 86
Cont. …

➢ The system is stable if all the roots of characteristic equation are lie in the

left half of s-plane otherwise unstable.

➢ Therefore for the system to be stable the zeros of 𝑄(𝑠) lie in the left half of s-
plane.

➢ Nyquist combined this fact and Cauchy’s principle of argument to construct


a closed contour 𝐶 such that all the right half of s-plane is encircled to detect
the presence of poles and zeros of 𝑄(𝑠) with positive real parts.

➢ The contour 𝐶 is referred to as Nyquist contour (Nyquist path) which is


directed in clockwise direction which composed of three segments as
shown in Figure 4.9 below. 87
Cont. …

Fig.4.9: Nyquist contour (path).

➢ Segment 𝐶2 and 𝐶3 cover the whole imaginary axis from −𝑗∞ to 𝑗∞.

➢ Segment 𝐶1 is semicircle of radius infinity that encloses the entire right half of s-
plane.
88
Cont. …
➢ Mapping of C into Q(s) is obtained for segment C2 and C3 by substituting s = jω in
Q(s) and varying ω from 0 to j∞ and −j∞ to 0 respectively.
➢ Segment C1 can be mapped into Q(s) plane by substituting s = Rejθ and varying θ
from + πΤ2 to − πΤ2 and radius approach infinity.
❖ With mapping Q(s) of closed path C, we can find the following on the basis of
principle of argument.

1) If the total number of zeros of Q(s) in the right half of s-plane is equal to z, then the
number of clockwise encirclements of the origin of Q s −plane is z, hence the
system is unstable (i.e. there is no open-loop poles (𝑝 = 0) in the right half -plane).

2) If the total number of poles of Q(s) in the right half of s-plane is equal to p, then the
number of counterclockwise encirclements of the origin of Q s −plane is p, hence
the system is stable (i.e. there is no zeros of Q(s) (𝑧 = 0) in the RH s-plane ).
89
Cont. …
3) If the difference between number of poles and zeros of Q(s) in the right half of s-
plane is N ( i.e., N = z − p ), then the number of encirclements of the origin of
Q s − plane is N.

➢ If N is positive clockwise encirclement and if N is negative counterclockwise


encirclement directions.

➢ For closed-loop system to be stable the number of zeros of Q(s) that lie to the right
of s-plane is zero. That is z = 0, which gives N = −p. Therefore the system is stable,
the number of counterclockwise encirclements of origin Q s −plane must be
equal to p.

➢ If N = 0, then z = p, the system may be stable or unstable depending on p. If p = 0,


the system is stable.
90
Cont. …
❖ We know that,
Q s =1+H s G s =0
⇒1+H s G s =0

⇒ H s G s = −1 (4.3)

➢ From the above Eq.(4.3) the contour Cq in Q s − plane or 1 +


H s G s −plane is identical to the contour CHG in the H s G s −plane, if we
shift the origin from 0 + j0 to −1 + j0.

➢ Hence Nyquist plot is mapping of Nyquist contour in the H(s)G 𝑠 −plane


and shifting the origin from 0 + j0 to −1 + j0.
91
Cont. …
➢ Nyquist contour C, Nyquist contour into Q s −plane and Nyquist contour into
H s G s −plane as shown in Figure 4.10 below respectively.

Fig.4.10: a) Nyquist contour, b) Contour Cq in Q s −plane c) ) Contour CHG in


H s G(s) −plane.
92
Cont. …
❖ Nyquist stability criterion stated as follow:

➢ If the contour CHG of the open-loop transfer function H s G(s) in the


HG s −plane corresponding to Nyquist contour in the s-plane encircles the
point −1 + j0 in the counterclockwise direction as many times as the right
half of s-plane poles of H s G(s), then the closed-loop system is stable.

Example: when a system has one open-loop pole (p = 1) in the right half of
s-plane and has one clockwise encirclement of −1 + j0, i.e., N = 1, then
number of closed-loop pole is two (z = 2) in the right half of s-plane. N = z −
p ⇒ z = N + P = 1 + 1 = 2, hence the system is unstable.

93
Cont. …
➢ If open-loop pole is one ( 𝑝 = 1 ) in the right s-plane and has one
counterclockwise encirclement of −1 + 𝑗0, i.e., 𝑁 = −1, 𝑁 = 𝑧 − 𝑝 ⇒ 𝑧 =
𝑁 + 𝑝 = −1 + 1 = 0. Therefore there is no closed-loop poles in the right half
of s-plane , hence the system is stable but open-loop is unstable.

❖ Special case when 𝐇 𝐬 𝐆(𝐬) involves poles and/or zeros on the 𝐣𝛚 −


axis
➢ Nyquist path not pass through poles or zeros. Hence we have to modifies the
closed-contour/path.

➢ In modified closed contour/path only a very small portion of the plane is not
encircled. 94
Cont. …
➢ This small portion approaches zero as the radius of semicircle 𝜖 → 0, thus
ensuring induction of all poles and zeros in the right half of s-plane.

Fig.4.11: Nyquist contour, when poles and/or zeros on jω −axis. 95


Cont. …
❖ The mapping of Nyquist contour into contour CHG can be done as follow:

1) First substitute s = jω in open-loop transfer function H s G s . This converts


transfer function into frequency function H jω G jω .

2) Varies ω from −∞ to ∞ and map all points on HG −plane. This can be done by
plotting polar plot for a given transfer function by varying ω from 0 to ∞. Then plot
for ω = −∞ to 0 is the mirror image of the polar plot with respect to real axis.

3) The mapping of infinite semicircle arc can be done by substituting s = Rejθ and
varying θ from πΤ2 to −πΤ2 and tending R to infinity.

4) If the open-loop transfer function has any poles on jω −axis, then map the segment
by letting s = ϵejθ , (i. e. ϵ → 0) and varying θ from −πΤ2 to πΤ2.
96
Cont. …
Example: For the system shown in figure 4.12(a) below, sketch the
Nyquist plot for k = 2 and use Nyquist criterion to determine whether
the closed-loop system is stable for this gain. Find the range of k for
the system to be stable.

Fig.4.12: a) unity feedback system b) Nyquist contour/path.


97
Cont. …
Solution: Draw the Nyquist contour/path as shown in figure 4.12(b).

k 1−s 2 1−s
H s G s = = , for k = 2
s+2 s+3 s+2 s+3

Now let as map each section into HG −plane

➢ Mapping of section 𝐶1 : In this section, 𝜔 − varies from 0 to ∞. The mapping of this


section into 𝐻𝐺 −plane is equal to polar plot of 𝐻(𝑠)𝐺(𝑠), let 𝑠 = 𝑗𝜔

𝟐 𝟏−𝐣𝛚 𝟐 𝟏−𝐣𝛚 𝟐−𝐣𝛚 𝟑−𝐣𝛚


𝐇 𝐣𝛚 𝐆 𝐣𝛚 = = × , 𝐇 𝐣𝛚 𝐆(𝐣𝛚) = 𝐆(𝐣𝛚), b/c 𝐇 𝐬 = 𝟏
𝐣𝛚 +𝟐 𝐣𝛚 +𝟑 𝐣𝛚 +𝟐 𝐣𝛚 +𝟑 𝟐−𝐣𝛚 𝟑−𝐣𝛚

𝟏𝟐 𝟏−𝝎𝟐 𝟐𝝎 𝝎𝟐 −𝟏𝟏 𝟐 𝟏+𝝎𝟐 −𝟏 −𝝎 − tan−𝟏 𝝎Τ𝟐 − tan−𝟏 𝝎Τ𝟑


𝐇 𝐣𝛚 𝐆 𝐣𝛚 = +𝒋 = ∠ tan
𝝎𝟐 +𝟒 𝝎𝟐 +𝟗 𝝎𝟐 +𝟒 𝝎𝟐 +𝟗 𝟒+𝝎𝟐 𝟗+𝝎 𝟐

lim G jω = 12Τ36 + j0 = 0.33∠0°


ω→0
98
Cont. …
lim G jω = −0 + j0 = 0∠ − 270°
ω→∞
𝟏𝟐 𝟏−𝛚𝟐
Im −intercept, R e G jω =0 ⇒ = 0 ⇒ ω = ωy = 1
𝛚𝟐 +𝟒 𝛚𝟐 +𝟗

lim G jω = 0 − j0.4 = o. 4∠ − 90°


ωy =1

𝟐𝛚 𝛚𝟐 −𝟏𝟏
R e −intercept, Im G jω =0 ⇒ = 0 ⇒ ω = ωx = 11
𝛚𝟐 +𝟒 𝛚𝟐 +𝟗

lim G jω = −0.4 + j0 = o. 4∠ − 180°


ωx = 11

➢ Mapping of section 𝐶3 : It is the mirror image of the above polar plot.

99
Cont. …
➢ Mapping of section C2 : This section in s-plane is mapped to HG −plane by

substituting s = lim Rejθ in H(s)G(s) and varying θ from+ πΤ2 to − πΤ2


R→∞

jθ 2 1−Rejθ
G Re = G s │s=Rejθ =
Rejθ +2 Rejθ +3

jθ 2 Rejθ 1ΤRejθ −1
lim G Re = lim = lim ϵe−jθ
R→∞ R→∞ R2 ej2θ 1+2ΤRejθ 1+3ΤRejθ ∈→0

when θ = πΤ2, G Rejθ = lim ϵe−jπΤ2 and when θ = − πΤ2, G Rejθ = lim ϵejπΤ2
∈→0 ∈→0

➢ From the above value, section C2 in s-plane is mapped into a circular arc of radius
zero (0) at origin in the HG −plane.

100
Cont. …
➢ The complete Nyquist plot is

➢ The intersection of Nyquist plot with real axis is obtained by equating imaginary
part of H(jω)G(jω) to zero as solved above.
𝟐
𝟐𝛚 𝛚𝟐 −𝟏𝟏 𝟐 𝟏+ 𝟏𝟏
= 0 ⇒ ω = ωx = 11 radΤs, ∴ H(jω)G(jω) │ω=ωx = 11 = = 0.4
𝛚𝟐 +𝟒 𝛚𝟐 +𝟗 𝟐 𝟐
𝟒+ 𝟏𝟏 𝟗+ 𝟏𝟏
101
Cont. …
➢ Therefore the Nyquist plot cuts the real axis at 𝑅𝑒 𝐻𝐺 = −0.4 and the number of
encirclement of −1 + 𝑗0 point is zero. That is 𝑁 = 0, the number of open-loop poles in
the right of s-plane is zero. Z= 𝑁 + 𝑃 = 0 + 0 = 0

➢ Thus for 𝑘 = 2 the number of zeros of 1 + 𝐻 𝑠 𝐺 𝑠 = 0 (closed-loop poles) in the right


of s-plane is zero and hence the system is stable.
❖ To find the range of k: The magnitude of H(jω)G(jω) at ω = 11 for arbitrary value of
open-loop gain 𝑘 is

𝟐
𝒌 𝟏+ 𝟏𝟏
H(jω)G(jω) │ω= 11 = = k(0.2)
𝟐 𝟐
𝟒+ 𝟏𝟏 𝟗+ 𝟏𝟏

➢ For the system to be stable, the number of encirclements of Nyquist plot with −1 + 𝑗𝑜
point must be zero.
∴ k 0.2 < 1 ⇒ 𝑘 < 5, so that the range of k is, 0 < 𝑘 < 5. 102
Cont. …
Exercise 4.9:

1) The open-loop transfer function of a unity feedback system is given by

5
H s G s =
s(s + 1)(s + 2)

Draw the Nyquist plot and determine whether the system is stable or
unstable.

103
End

104

You might also like