EE3202B Control Systems

Lecture 6: Nyquist Analysis and Relative Stability

Sei Zhen Khong

National Sun Yat-sen University

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 1 / 27

Part I
Nyquist Stability Criterion

Part II
Relative Stability

This lecture covers material in Chapters 6.3 and 6.4 of the textbook.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 2 / 27

Introduction

Nyquist stability criterion relates the open-loop frequency response to the

number of closed-loop poles of the system in the RHP.
The criterion was developed by Harry Nyquist of the Bell Lab in 1932. The
criterion is based on a result from complex variable theory known as
“Cauchy’s principle of the argument”.
The criterion requires building a curve of the open-loop system, known as
“the Nyquist curve”, which is closely related to the Bode diagram.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 3 / 27

Argument principle – example 1

Consider F (s) = s + 1. Then

Im Im

× ×

× ×
−1 1 Re −1 1 Re
C2 F (C2 )

C1 F (C1 )

clockwise oriented curve C2 does not encircle the zero of F (s)

⇔ F (C2 ) does not encircle the origin.
clockwise oriented curve C1 encircles one zero of F (s)
⇔ F (C1 ) encircles the origin clockwise once.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 4 / 27

Argument principle – example 2
Consider F (s) = (s + 1)2

4 8

3 6

2 4

1 2

0 0

−1 −2

−2 −4

−3 −6

−4 −8
−2 −1 0 1 2 −5 0 5 10

clockwise oriented blue curve does not encircle any zero of F (s)
⇔ the curve mapped by F (s) does not encircle the origin.
clockwise oriented red curve encircles two zeros of F (s)
⇔ the curve mapped by F (s) encircle the origin clockwise twice.
Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 5 / 27
Argument principle – example 3
1
Consider F (s) = s+1

4 1.5

3
1

0.5
1

0 0

−1
−0.5

−2

−1
−3

−4 −1.5
−2 −1 0 1 2 −1 0 1 2

clockwise oriented blue curve does not encircle any pole of F (s)
⇔ the curve mapped by F (s) does not encircle the origin.
clockwise oriented red curve encircles one pole of F (s)
⇔ the curve mapped by F (s) encircle the origin counterclockwise once.
Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 6 / 27
Argument principle – example 4
(s+1)
F (s) = (s+1.5)2

4 2.5

2
3

1.5
2
1

1
0.5

0 0

−0.5
−1

−1
−2
−1.5

−3
−2

−4 −2.5
−2 −1 0 1 2 −5 −4 −3 −2 −1 0

clockwise oriented blue curve does not encircle any pole or zero of F (s)
⇔ the curve mapped by F (s) does not encircle the origin.
clockwise oriented red curve encircles one zero and two poles of F (s)
⇔ the curve mapped by F (s) encircle the origin counterclockwise once.
Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 7 / 27
Cauchy’s principle of argument

Consider a complex function F (s) and a clockwise oriented closed contour C1 .

The map of C1 by F (s) (i.e., the contour {F (z) | z ∈ C1 }) will encircle the origin
clockwise Z − P times, where
Z is the number of zeros of F (s) inside the contour C1 ;
P is the number of poles of F (s) inside the contour C1 .
Im
Application to Control Design

N
Recall that for stability we test for the
presence of roots of 1 + G (s)C (s) in
|s| = ∞
the RHP.
Re
Cauchy’s principle of argument tells us
we can obtain this information from a
plot of 1 + G (s)C (s) as s moves along
an appropriate contour N .

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 8 / 27

Nyquist stability criterion

Note that

No. of encirclement of the origin by 1 + G (s)C (s)

≡ No. of encirclement of −1 by G (s)C (s)

Consider the Nyquist plot of G (s)C (s); that is, the map of the contour N on the
previous frame by G (s)C (s). Suppose that
the map encircles −1 clockwise N times;
G (s)C (s) has P poles in the open right-half complex plane, and has no poles
on the imaginary axis.
Then the closed-loop system has Z := N + P poles in the open right-half plane.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 9 / 27

Plotting the Nyquist plot

Suppose G (s)C (s) has no poles on the imaginary axis.

Find the DC value of G (s)C (s): lim G (jω)C (jω)

ω→0
Find the value of G (s)C (s) when s is very large: lim G (s)C (s).
|s|→∞
For a strictly proper transfer function, lim G (s)C (s) = 0.
|s|→∞
For a biproper transfer function, lim G (s)C (s) = k, a real number.
|s|→∞

For ω ∈ [0, ∞), the curve of G (jω)C (jω) can be obtained from the Bode
diagram, if it is available.
The curve of G (jω)C (jω) for ω ∈ (−∞, 0] is the conjugate of the curve for
ω ∈ [0, ∞).

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 10 / 27

Example (1)

1
G (s)C (s) =
(s + 1)2
Nyquist Diagram

0.6

0.4

0.2

Imaginary Axis
0

−0.2

−0.4

−0.6

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1

Real Axis

No encirclement of −1: N = 0
G (s)C (s) is stable: P = 0.
Z = N + P = 0; thus the closed-loop system is stable

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 11 / 27

Example (2)
s +4
G (s)C (s) =
(s + 1)(s − 2)(s + 3) Nyquist Diagram

0.15

0.1

0.05

Imaginary Axis
0

−0.05

−0.1

−0.15

−1 −0.9 −0.8 −0.7 −0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0

Real Axis

No encirclement of −1: N = 0
G (s)C (s) has one unstable pole: P = 1.
Z = N + P = 1; thus the closed-loop system is unstable
Stability requires N = −1: may be achieved by increasing the gain
Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 12 / 27
Nyquist plot: system with poles on the imaginary axis
When plotting the Nyquist plot on slide 10, we assumed that G (s)C (s) has no
poles on the imaginary axis. If G (s)C (s) has poles on the imaginary axis, the
Nyquist contour N must be indented around these points. For instance, for a pair
of poles at s = ±ωc j we use the path described below, with ε → 0.
Im

N
ε
ωc ×

|s| = ∞

Re

ε
− ωc ×

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 13 / 27

What happens along the arc around ωc j

As ε → 0, when s moves along the arc around ωc j:

|G (s)C (s)| → ∞. This is clear because ωc j is a pole of G (s)C (s); therefore,
when s is very close to ωc j, the denominator of G (s)C (s) is very small.
The phase of G (s)C (s) changes −nπ, where n is the number of poles at ωc j.
Therefore, as s moves along the infinitesimal arc around ωc j, the polar plot
of G (s)C (s) moves in an infinite radius arc in the clockwise direction for n/2
revolutions.
The same applies to the arc around −ωc j, or any pole on the imaginary axis for
this matter.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 14 / 27

Example: a pole at the origin
1 1 j
G (s)C (s) = ⇔ G (jω)C (jω) = − 2 −
s(s + 1) ω +1 ω(ω 2 + 1)
G (−jε)C (−jε)

Im

Im

|s| = ∞ G (εeθ )C (εeθ )

× × Re Re
−1 −1
Ce := εeθ G (±∞)C (±∞)

G (jε)C (jε)

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 15 / 27

Summary of Part I

The Nyquist criterion is another classical method of stability analysis.

The Nyquist path is mapped from the s-plane by the open loop transfer
function.
The resulting contour provides information about the number of unstable
closed-loop poles.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 16 / 27

Part I
Nyquist Stability Criterion

Part II
Relative Stability

This lecture covers material in Chapters 6.3 and 6.4 of the textbook.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 17 / 27

Relative stability: stability margins and sensitivity peak

In control system design, one often needs to go beyond the issue of closed
loop stability.
It is usually desirable to assess the relative stability; i.e., to obtain some
quantitative measure of how far from instability the closed-loop system is.
These measures give quantitative indication of how uncertainty affects the
system, in particular, the stability property.
Such measures describe the distance from the nominal open loop frequency
response to the critical stability point in the complex plane.
Examples include the gain and phase margins and the peak sensitivity.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 18 / 27

Gain and phase margins
Im
m

−1
Re
φ

G (jω)C (jω)

Assume G (s)C (s) has no unstable poles and the Nyquist curve does not
encircle −1
1
Gain margin Mg := or −20 log m dB
m
Phase margin Mf := φ

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 19 / 27

Identify gain and phase margins from the Bode diagram

Bode Diagram
10

0
Mg
−10
Magnitude (dB)

−20

−30

−40

−50

−60
0

−45

−90
Phase (deg)

−135
Mf
−180

−225

−270
−1 0 1
10 10 10
Frequency (rad/sec)

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 20 / 27

Physical interpretation of gain and phase margins

Stable nominal closed-loop, C (s)G (s) has no unstable poles:

r C (s) G (s) y
−

The gain margin indicates the additional open-loop gain that would take
the closed-loop system to the critical stability condition.

The phase margin indicates the pure phase delay that can be added before
the closed-loop system reaches the critical stability condition.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 21 / 27

Peak sensitivity

Im

−1
Re
η

G (jω)C (jω)

1 1
Peak sensitivity: max = ;
ω |1 + G (jω)C (jω)| η
Furthermore, it can be shown that the phase margin Mf ≥ 2 sin−1 ( η2 ).

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 22 / 27

Sensitivity function

Recall that the sensitivity function S(s)

is the transfer function do (t) 7→ y (t) and r (t) 7→ e(t):
do

e u +
r C (s) G (s) y
−

relates the variation of the open loop and the variation of the closed-loop:

δ[T (s)] δ[G (s)C (s)]

≈ S(s)
T (s) G (s)C (s)

where
T (s): nominal complementary sensitivity function R(s)/Y (s)
δ[G (s)C (s)]: variation of open-loop G (s)C (s)
δ[T (s)]: variation of T (s) due to δ[G (s)C (s)].

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 23 / 27

Example: stability margins

Compute the stability margins and peak sensitivity for a system with open loop
transfer function
s + 10
G (s)C (s) = 3
s + 6s 2 + 11s + 6
The sensitivity function is

1 s 3 + 6s 2 + 11s + 6
S(s) = = 3
1 + G (s)C (s) s + 6s 2 + 12s + 16

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 24 / 27

Gain and phase margins
Use MATLAB:
>> s = tf([1,0],1);
>> GC = (s+10)/(s*s*s+6*s*s+11*s+6);
>> margin(G)
Bode Diagram
Gm = 23.522 dB (at 5.099 rad/sec), Pm = 95.329 deg (at 1.0061 rad/sec)
20

0
Magnitude (dB)

−20

−40

−60

−80
0

−45
Phase (deg)

−90

−135

−180

−225
−1 0 1 2
10 10 10 10
Frequency (rad/sec)

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 25 / 27

Peak sensitivity
To find the peak sensitivity, plot the magnitude of S(jω) against ω.
>> Sen = 1/(1+GC);
>> [mag,phase,omega] = bode(Sen);
>> semilogx(omega,20*log10(squeeze(mag)));
4

2.343 dB
2

−2

−4

−6

−8

−10
−1 0 1 2
10 10 2.37 10 10

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 26 / 27

Summary of Part II

For practical systems, where modelling errors exist, the notion of nominal
stability is not sufficient.
Relative stability allows us to quantify how far from instability a system is.
Three indicators were discussed in this lecture: gain margin, phase margin,
and peak sensitivity.
These indicators give us insights into how perturbations may affect the
stability property of closed-loop systems.

Sei Zhen Khong Lecture 6: Nyquist Analysis and Relative Stability 27 / 27