MECN5015A – Chapter 5: Stability of

Feedback Control Systems

School of Mechanical, Industrial and Aeronautical

Engineering,
University of the Witwatersrand
Naadira Hassim
Naadira.Hassim@gmail.com
Naadira.hassim@wits.ac.za

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 1 of 32

PRESENTATION OUTLINE

1. Introduction
1. Introduction to Stability

2. General stability theory

1. Stability Specifications

3. Stability Criteria
1. Routh-Hurwitz Stability Criterion
2. Nyquist Stability Criterion

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 2 of 32

2. GENERAL STABILITY THEORY

General Stability Theory

 A complete and general stability theory is based on the locations of the
roots of the closed-loop characteristic equation in the complex s-plane.
 Stable systems have all of their roots in the left half plane (LHP).
 The analytical study of stability becomes the solutions of the closed-
loop system differential equations.
 There are two methods use for stability analysis:
1. Routh-Hurwitz.
2. Nyquist.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 3 of 32

1. INTRODUCTION

Absolutely Stable
If a system in equilibrium is momentarily excited by command and/or
disturbance inputs and those inputs are then removed, the system
must return to equilibrium if it is absolutely stable.

A stable system is a dynamic system with a bounded response to

a bounded input.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 4 of 32

2. GENERAL STABILITY THEORY

Stability Specifications
 If as t → +∞, the trajectory of the describing point tends to the steady-
state/ equilibrium / origin of the coordinate system, then the system is
said to be asymptotically stable.

Examples of Stable Responses

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 5 of 32

2. GENERAL STABILITY THEORY

Stability Specifications
 If as t → +∞, the trajectory of the describing point moves infinitely away
from the origin of the coordinate system, then the system is said to be
unstable.

Examples of Unstable Responses

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 6 of 32

1. INTRODUCTION

Absolutely Unstable
 If the action persists indefinitely after the excitation is removed, the
system is judged to be absolutely unstable.

 If a system is stable, how close is it to becoming unstable? That is,

what is the degree of stability? A fighter aircraft such as F-16, is less
stable than a commercial transport, hence it can maneuver more
quickly.
 Gain and Phase Margins are Relative Stability Indicators.
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 7 of 32
2. GENERAL STABILITY THEORY

Types of Stability
There are two types of stabilities of interest to us in mechatronic systems:
1. Asymptotic stability is the ability of the system to return to equilibrium
after some arbitrary displacement.
2. Bounded input bounded output (BIBO) stability is the ability of the
system to produce a bounded output for every bounded input.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 8 of 32

2. GENERAL STABILITY THEORY

Stability Condition
Free response of the system is described by the following (lumped
parameter, homogeneous, linear) differential equation:
𝑎𝑛 𝐶 𝑛 𝑡 + 𝑎𝑛−1 𝐶 𝑛−1 𝑡 + 𝑎1 𝐶ሶ 𝑡 + 𝑎0 𝐶 𝑡 = 0 2.1

Performing Laplace transform gives the characteristic equation as:

𝑎𝑛 𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠 + 𝑎0 = 0 2.2

The necessary and sufficient condition for asymptotic stability of system in

a free state can be stated as follows:
 All roots of the system characteristic equation must be on the left
half plane of the s-plane.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 9 of 32

2. GENERAL STABILITY THEORY

S-Plane Roots

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 10 of 32

2. GENERAL STABILITY THEORY

Stability Criteria
Does this mean we have to find roots of high order systems as well?
Consider:
𝑠 2 + 7𝑠 + 12 = 0
and
𝑠 5 + 𝑠 4 + 3𝑠 3 + 9𝑠 2 + 16𝑠 + 10 = 0
There are two powerful methods of determining the presence of unstable roots
for continuous-time models, without actually finding their mathematical values:
 Routh-Hurwitz stability criterion, and
 Nyquist stability criterion.
Both methods give the same results, a statement of the number (but not the
specific numerical values) of unstable roots. This information is generally
adequate for design purposes.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 11 of 32

3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

 It works with the closed-loop system characteristic equation in an
algebraic function and requires that the characteristic equation be a
polynomial in the differential operator D.
𝑠2 + 𝑠 + 1
𝑇 𝑠 = 4 3.1
𝑠 + 2𝑠 3 + 3𝑠 2 + 𝑠 + 1
 It gives the number (not specific numerical values) of roots of the
system characteristic equation not in the LHP and does not distinguish
between real and complex roots.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 12 of 32

3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

Necessary Condition: Coefficients must be positive and non-zero
For
𝑎𝑛 𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + 𝑎𝑛−2 𝑠 𝑛−2 … + 𝑎2 𝑠 2 + 𝑎1 𝑠 + 𝑎0 = 0 3.2

 an > 0.
 We also assume that a0 is nonzero, otherwise the characteristic
equation has one or more zero roots which we easily detect and do not
correspond to stable systems.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 13 of 32

3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

Sufficient Condition: Elements in the first column of Routh’s Array must have the same sign.

Routh’s Array
 Arrange the coefficients of the characteristic polynomial into the following array:
𝑎𝑛 𝑎𝑛−2 𝑎𝑛−4 𝑎𝑛−6 𝑎𝑛−8 𝑎𝑛−10 ⋯
𝑎𝑛−1 𝑎𝑛−3 𝑎𝑛−5 𝑎𝑛−7 𝑎𝑛−9 𝑎𝑛−11 ⋯
 Then form a third row: 𝑏1 𝑏2 𝑏3 ⋯
𝑎𝑛 𝑎𝑛−2 𝑎𝑛 𝑎𝑛−4
𝑎𝑛−1 𝑎𝑛−3 𝑎𝑛−1 𝑎𝑛−5
𝑏1 = − 𝑏2 = −
𝑎𝑛−1 𝑎𝑛−1
𝑎𝑛 𝑎𝑛−6
𝑎 𝑎𝑛−7
𝑏3 = − 𝑛−1 ,⋯ 3.3
𝑎𝑛−1

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 14 of 32

3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

This is continued until no more rows and columns can be formed, giving a
triangular sort of array:
𝑠𝑛 𝑎𝑛 𝑎𝑛−2 𝑎𝑛−4 𝑎𝑛−6 𝑎𝑛−8 𝑎𝑛−10 ⋯
𝑠 𝑛−1 𝑎𝑛−1 𝑎𝑛−2 𝑎𝑛−4 𝑎𝑛−6 𝑎𝑛−8 𝑎𝑛−10 ⋯
𝑠 𝑛−2 𝑏1 𝑏2 𝑏3 𝑏4 𝑏5 ⋯
𝑠 𝑛−3 𝑐1 𝑐2 𝑐3 𝑐4 𝑐5 ⋯
⋯ ⋯ ⋯ ⋯ ⋯ ⋯
𝑠3 𝑑1 𝑑2 0
𝑠2 𝑒1 𝑒2 0
𝑠1 𝑓1 0
𝑠0 𝑔1 0

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 15 of 32

3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

Consider the polynomial:
s5 + s4 + 3s3 + 9s2 + 16s + 10 = 0
How many roots of the polynomial are in the right half plane of the complex plane?

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 16 of 32

3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

Let us consider two exceptional (special) cases can occur mathematically:
1. Left column zeros of the array.
 If a zero occurs in the first column, and the remaining elements in that row are not all zero,
replace the zero element with a small positive number ε, and continue forming the array.
 When completed, let ε → 0.

Example 2
Consider the polynomial s4 + s3 + 2s2 + 2s + 3 = 0. Forming Routh’s array shows that a snag
develops in the third row.

𝑠4 1 2 3
𝑠3 1 2 0
𝑠2 0 3 0
𝑠1
𝑠0
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 17 of 32
3. STABILITY CRITERION

Routh-Hurwitz Stability Criterion

Example 2 : s4 + s3 + 2s2 + 2s + 3 = 0.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 18 of 32

3. STABILITY CRITERION

Premature Termination of the Array

 If any derived row is identically zero, the system is not stable. It may be
unstable or neutrally stable.
 The zero row corresponds to roots located symmetrically about the
origin. These can be pairs of real roots with the same magnitude but
opposite sign (±r), purely imaginary complex pairs (±iω), or two pairs of
complex conjugate roots (a ± ib, -a ± ib).
 The auxiliary polynomial containing these roots is obtained from the row
above the zero row. The coefficients of the derivative of the auxiliary
polynomial are used to replace the zero row, and the rest of the array is
then constructed.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 19 of 32

3. STABILITY CRITERION

Example
Consider the polynomial s4 + s3 + 5s2 + 3s + 6. This polynomial has a Routh’s
array which begins as follows:

𝑠4 1 5 6
𝑠3 1 3 0
𝑠2 2 6 0
𝑠1 0 0
𝑠0

 The s1 row consists solely of zeros, so the s0 entry would otherwise become
indeterminate. When a row of zeros occurs, there exist an even or odd
polynomial divisor of the original polynomial. The coefficients of this divisor
polynomial are given by the previous nonzero row of the array.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 20 of 32

3. STABILITY CRITERION

Example
Consider the polynomial s5 + 3s4 + s3 + 5s2 + s + 3 = 0.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 21 of 32

3. STABILITY CRITERION

Nyquist Stability Criterion

 The Nyquist stability criterion is a valuable tool that determines the
degree of stability, or instability, of a feedback control system.
 It is the basis for other methods that are used to improve both the
steady-state and transient response of a feedback control system.
 Application of the Nyquist stability criterion requires a polar plot of the
open-loop transfer function Go(s), which is usually referred to as
Nyquist plot.
 The Nyquist criterion determines the number of roots of the
characteristic equation that have positive real parts from a polar plot of
the open-loop transfer function, Go(iω), in the complex plane.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 22 of 32

3. STABILITY CRITERION

Advantage of Nyquist vs Routh-Hurwitz

 It uses the open-loop transfer function to determine the number of unstable roots of the
closed-loop system characteristic equation.
 If some components are modelled experimentally using frequency response
characteristics, these measurements can be used directly in the Nyquist criterion.
 Nyquist criterion also gives some useful results on relative stability, i.e., gain margin
and phase margin.
 The polar plot used keeps the effects of the individual pieces of hardware more
apparent making needed design changes more obvious. It is easier to use Nyquist
criterion to design a compensator than Routh’s criterion.

Reference Control
Controller Process Output
Input Effector

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 23 of 32

3. STABILITY CRITERION

Frequency Response Nyquist Plot

Nyquist Plot
Two Methods:
 Cartesian Plot using Real and Imaginary
Components
𝑇 𝑗𝜔 = 𝛼 𝜔 + 𝑗𝛽 𝜔 3.4
 Polar Plot using Magnitude and Phase angle
𝑀 𝜔 = 𝑇 𝑗𝑤 = 𝛼2 + 𝛽 2 3.5

𝛽
𝜑 𝜔 = tan−1 3.6
𝛼

Consider 4 points to plot

1) ω = 0
2) ω = ∞
3) Imaginary Axis Intercept
4) Real Axis Intercept

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 24 of 32

3. STABILITY CRITERION

Example
Produce a Nyquist Plot for the following Transfer Function:

1
𝐺 𝑠 =
𝑠2 + 3𝑠 + 2

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 25 of 32

3. STABILITY CRITERION

Example
Produce a Nyquist Plot for the following Transfer Function:

1
𝐺 𝑠 =
𝑠2 + 3𝑠 + 2

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 26 of 32

3. STABILITY CRITERION

Example
Produce a Nyquist Plot for the following Transfer Function:

1
𝐺 𝑠 =
𝑠2 + 3𝑠 + 2
0.4

0.3

0.2

0.1
Re

0
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6

-0.1

-0.2

-0.3

-0.4
Im
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 27 of 32
3. STABILITY CRITERION

Example
Produce a Nyquist Plot for the following Transfer Function: (Example 5 pg 176 Lecture Notes)

6𝐾
𝐺 𝑠 =
𝑠3 + 2𝑠2 + 2𝑠 + 40

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 28 of 32

3. STABILITY CRITERION

Example
Produce a Nyquist Plot for the following Transfer Function: (Example 5 pg 176 Lecture Notes)

6𝐾
𝐺 𝑠 =
𝑠3 + 2𝑠2 + 2𝑠 + 40

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 29 of 32

3. STABILITY CRITERION

Example
Produce a Nyquist Plot for the following Transfer Function: (Example 5 pg 176 Lecture Notes)

6𝐾
𝐺 𝑠 =
𝑠3 + 2𝑠2 + 2𝑠 + 40
20

15

10

5
Re

0
-5 0 5 10 15 20 25

-5

-10

-15

-20
Im

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 30 of 32

3. STABILITY CRITERION

Nyquist Stability
Nyquist stability criterion can be formulated
as follows:
1. If the open-loop system is asymptotically stable, then
the closed-loop system will be asymptotically stable, if
and only if the polar plot of G(iω) does not encircle
point (-1,i0) while the frequency ω increases from zero
to +∞.
2. If the open-loop system has P poles in the RHP (i. e.,
it is unstable), then the number of encirclements of
point (-1, i0) must be equal to P for a stable closed-
loop system.
Asymptotically stable: If as t → +∞, the trajectory of the describing
point reverts back to the equilibrium position.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 31 of 32

3. STABILITY CRITERION

Nyquist Stability
0.4
Nyquist stability criterion can be formulated
0.3 as follows:
0.2 1. If the open-loop system is asymptotically stable, then
the closed-loop system will be asymptotically stable, if
0.1
and only if the polar plot of G(iω) does not encircle
point (-1,i0) while the frequency ω increases from zero
Re

0
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 to +∞.
-0.1
2. If the open-loop system has P poles in the RHP (i. e.,
-0.2
it is unstable), then the number of encirclements of
point (-1, i0) must be equal to P for a stable closed-
-0.3 loop system.
Asymptotically stable: If as t → +∞, the trajectory of the describing
-0.4
Im 1 point reverts back to the equilibrium position.
𝐺 𝑠 =
𝑠2 + 3𝑠 + 2
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 32 of 32
3. STABILITY CRITERION

Nyquist Stability
20
Nyquist stability criterion can be formulated
15 as follows:
10 1. If the open-loop system is asymptotically stable, then
the closed-loop system will be asymptotically stable, if
5
and only if the polar plot of G(iω) does not encircle
point (-1,i0) while the frequency ω increases from zero
Re

0
-5 0 5 10 15 20 25
to +∞.
-5
2. If the open-loop system has P poles in the RHP (i. e.,
-10
it is unstable), then the number of encirclements of
point (-1, i0) must be equal to P for a stable closed-
-15
loop system.
-20
Asymptotically stable: If as t → +∞, the trajectory of the describing
Im
6𝐾 point reverts back to the equilibrium position.
𝐺 𝑠 =
𝑠3 + 2𝑠2 + 2𝑠 + 40
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 33 of 32
3. STABILITY CRITERION

Gain Margin and Phase Margin

Phase Cross-Over Frequency: ω c is the frequency at
which the Nyquist plot intersects the negative real axis - i.e.
the frequency at which the phase angle is 180°

Gain Cross-Over Frequency: is the frequency at which the

Nyquist plot has the magnitude of 1.

Phase Margin: φm is the phase angle where the magnitude

ratio or gain of Go(iω) is unity. This is determined by
𝑃𝑀 = 180° + 𝜑𝑔𝑐 3.7
Gain Margin: The factor by which the gain must be
multiplied at the phase cross-over to have a value of 1. i.e.
1
𝐺𝑀 = 3.8
𝑂𝐶
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 34 of 32
3. STABILITY CRITERION

Gain and Phase Margin Stability

The stability of the control system based on the relation
between the gain margin and the phase margin is listed
below.
 If the gain margin GM is greater than one and the
phase margin PM is positive, then the control system
is stable.

 If the gain margin GM is equal to one and the phase

margin PM is zero degrees, then the control system is
marginally stable.

 If the gain margin GM is less than one and / or the

phase margin PM is negative, then the control system
is unstable.

MECN5015A – Chapter 5: Stability of Feedback Systems Slide 35 of 32

3. STABILITY CRITERION

Gain and Phase Margin Specifications

The gain and phase margins can be stated as safety margins in the design
specifications. A typical set of such specifications is as follows:
 gain margin ≥ 8 db, and
 phase margin ≥ 30°.
In common design situations, only one of these equalities can be met, and the
other margin is allowed to be greater than its minimum value. It is not desirable to
make the margins too large, since this results in a low gain. The design will be
sluggish and might have a large steady-state error. We note that another
common set of specifications is:
 gain margin ≥ 6 db, and
 phase margin ≥ 40°.
MECN5015A – Chapter 5: Stability of Feedback Systems Slide 36 of 32