gan

lara
rsiri
ada GONTENTS
7115,

enl,
logi

Preface

CHAPTER 1 INTRODUCTION TO CONTROL SYSTEMS 1

1.1 \What is a Control System?
1
1.2 SystemRepresentation 2
1.3 Terms and Definitions 2
1.4 Examples of Control Systems J
1.5 Control System 5
1.5.1 OpenJoop System 5
1.5.2 Closed-loop System 6
1.5.3 Comparison berween Open-loop and Closed-loop Systems 8
1.6 Control Engineering Design 8
Further Readings 9

CHAPTER 2 MATHEMAIICAL MODELLING 11

2.1 MathematicalModels 11
2.1.1 Mathematical Model of a Physical System 11
2.2 Block Diagram Representation t2
2.2.1 Block Diagram 12
2.2.2 Translational Mechanical System r3
2.2.2.1 Linear spring t3
2.2.2.2 Translational motion of rigid bodies t5
2.2.2.3 Mass-springsystem L6
2.2.2.4 Damper 17
2.2.2.5 Spring-damper sysrem 17
2.2.2.6 Mass-spring-dampersysrem 1B
vi J Corrrnr

2.2.3 Rotational System r9

2.2.4 Gear System 20
2.2.5 Hydraulic System 22
2.2.5.1 Position control system: A servo-mechanism 24
2.2.5.2 \Talking lever 25
2.2.6 Liquid Level System 27
2.2.6.1 Interacting liquid level system 28
2.2.7 Thermal System 30
2.2.8 ElectricalSystem 3t
2.2.9 Other Systems 32
12
L.J Block Diagram Manipulation Method 35
2.3.1 Concept 36
2.3.2 Blocks in series 36
2.3.3 Blocks in Parallel 37
2.3.4 Negative Feedback 3B
2.4 Signal Flow Graph 42
2.4.1 Concept 42
2.4.2 Algebra of Signal Flow Graph 45 u f----:
- r-_- I _
---i
2.4.3 Definitions 44
2.4.4 Mason's Gain Formula 45
Further Readings 5r

CHAPTER 3 TIME RESPONSE OF A CONTROL SYSTEM 53

3't Time Response: An Introduction 53
3.1.1 Concept 53
3.1.2 Time Response 54
3.1.3 Mathematical Model 56
3.2 Laplace tansform 5B
3.2.1 Concept 58
3.2.2 Property of Laplace tansform 5B
3.2.3 Laplace Thansform of Derivatives 59
3.2.4 LaplaceTiansform of an Integral 59
3.2.5 Laplace tansform of a Unit
Step Function 50
3.2.6 LaplaceTlansform of a Ramp Function 61
3 .2.7 LaplaceThansform of an Impulse Function 52
3.2.8 LaplaceThansform of a Sinusoidal Function 62
-i]TE]
3.2.9 Differential Equation using LaplaceTlansform
Solr.ing 63
3.2.10 Tiansfer Function 69
3.3 Classification of Dynamic Systems 69
 Contents i
"il
19 3.3.1 System Classification 70
20 3.4 Step Response of a First Order System 72
22 3.4.1 First Order System 72
24 3.4.2 Step Response of a First Order System 73
)5 3.5 Step Response of a Second Order System 77
)7 3.5.1 Second Order System 77
28 3.5.2 Step Response of a Second Order System 7g
30 3.6 Effectiveness of a Feedback System g5
3t 3.6.1 Feedback Control System 86
32 3.5.2 Measure of Effectiveness BB
36 3-6.3 Step Input Test 89
Jb 3.6.4 Ramp Input 92
36 3.5.5 Unit Parabolic Input 92
37 3.7 Controller 94
38 3.7.1 Control Action 94
42 Further Readings 97
42
.+)
CHAPTER 4 STABILITY AND ROOT LOCUS 99
44 4.1 Stability of a Control System 99
45 4.1.I Concept 99
it 4.1.2 Definition 99
4.2 Routh's Stabiliry Criterion lO4
53 4.2.1 Method rc4
)3 4.3 Root Locus 109
i3 4.3.1 Concept 110
54 4.3.2 Angle and Magnitude Criteria ll2
56 4.4 Root Locus Construction ll5
iB 4.4.1 Construction Steps ll5
)B 4.4.2 Tlansient Response 124
i8 4.4.3 Determination of Damping Ratio, (, Natural Frequency,
59 (Dn, Damped Natural Frequency, @7, andTime
Constant,
)9 T, from Root Locus Diagram 125
60 Further Readings 131
6t
62 CHAPTER 5 FREQUENCY RESPONSE ANAIYSIS r33
62 5.1 Frequency Domain 133
o, 5.1.1 Concept 133
69 5.I.2 Graphical Representation of Frequency Response Data 137
69 5.2 Nyquist Diagram r37
viii I Crorrnt,

5.2.1 Concept 138

5.2.2 Differentiator 138
5.2.3 Integrator t40
5.2.4 First Order Lag 141
5.2.5 Second Order System r44
5.2.6 tansfer Operators in Series t47
5.3 Stabiliry Analysis using Nyquist Diagram 151
5.3.1 Nyquist Stabiliry Criterion 152
5.3.2 Degree of Stabiliry r53
5.3.3 Gain and Phase Margins 153
5.4 Bode Diagram 158
5.4.1 Concept r59
5.4.2 Straight-line Asymptotic Approximation for Gain Plot 160
-.
5.4.3 Straight-line Asymptotic Approximation of Phase Shift
i._-:
Plot t62
L-
5.4.4 First Order Lead 164
5.4.5 Bode Plot of Combination of Several Lead and Lag
Elements r65
5.4.6 Effect of Gain K 156
5.4.7 Integrator t67
5.4.8 Differentiator 168
5.4.9 Second Order Lag t69
5.4.n Modelling using Bode Diagram 170
5.5 StabiliryAnalysis using Bode Diagram 17r
5.5.1 Method 172
Further Readings 175

CHAPTER 6 CONTROL SYSTEM DESIGN 177

6.1 Design Basis 177
6.1.1 DesignSpecification 177
6.1.2 Controller Configuration 17B
6.1.3 Basic Design Principles 180
5.2 P- and Pl-Controllers 180
6.2.1 Design Procedure 181
6.3 PD-Controller 185
6.3.1 Design Procedure 185
Further Readings 1BB

Index 189
ig
r8

o
1

4
'7

1
PREFACE
2
3
)
8
9
0
This book is intended to serve as a text for a first course in control sysrem engineering for
Z
::;helor degree engineering students. The text has grown out of academic staff with in-depth
4
srox'ledge and experience in the subject matter. The book has been intentionally kept short so
--;rat students would be able to appreciate the fundamental contenrs to be
understood and the
:larerial can be covered in a one semester course.
This book emphasizes particularly on the design and analysis of feedback control sysrem.
Control systems are found in a broad range of applications within various engineering disciplines,
namely electrical, mechanical, chemical, or aerospace engineering. Many illustrated examples and
:xercises are included in the book to assist students to undersrand better the fundamental conceprs
of control system analysis and design. Every section in the rext is headed by learning outcomes so
that the students may evaluate their understanding of the subject matrer after completing each
section.
The pre-requisites for the students using this book are that they should have completed
courses in physics and mathematics particularly through differential equations. They should Jrerdy
be well versed in applying the physical laws of physics, such as equilibrium equarion, Newton's
Iaws of motion, conservation of mass and energy, Kirchorfft law, and other iommon physical
laws. Other required subjects are Laplace transformation and linear algebrawhich are incorporated
in this book.
This book consists of six chapters. Chapter 1 introduces studenrs the basic conceprs of
control system as well as definitions and terminologies. Examples of open- and closed-loop
systems are also included. The general tasks of control engineers are highlighted in this chapter.
Chapter 2 addresses the representation of various physical ,yrt.-r, namely mechanical,
rotational, hydraulic, liquid-level, thermal, and electrical systems into mathematical models. These
mathematical models are developed into input-output relationships or rransfer functions. Two
methods of determining the transfer function of a dynamics system are introduced which are
block diagram manipulation and the use of Mason's rule and signal flow graph.
xl Preface

Chapter 3 provides an introduction to system analysis during transient and steady-state

responses. The chapter starts with finding and describing the output response of first and second
order systems in time domain using the Laplace transformation technique. This chapter also
introduces the definition of order, rank, and class of dynamic system followed by the transient
and steady-state response analysis ofboth the first- and second-order systems. Finally, it introduces
briefly on different rypes of controllers namely proportional, integral, and derivative controllers.
In Chapter 4,the stabiliry of a feedback control system is investigated and the relationship
between system stabiliry and characteristic equation is studied with the use of Routh stabiliry
criterion. Chapter 4 also deals with determining the locus of the roots of a characteristic equation
on the s-plane graphically. The performace ofthe control system in evaluated based on the location
of the roots of the characteristic equation on the s-plane.
In Chapter 5, the steady-state response of a system is examined using sinusoidal input signal
as the frequency of the input signal is varied. Frequency response analysis based on Bode and
Nyquist diagrams are considered. System stability and relative stabiliry utilizing frequency response
methods are introduced.
Chapter 6 discusses several approaches in designing and compensating a control system with
various rypes of controller in time domain. Students may refer to the design of control system in
frequency domain in any major control system text books.
Outcomer

Hishamuddin Jamaluddin
Mohd ShafiekYaacob
Robiah Ahmad
Faculty of Mechanical Engineering
Universiti Teknolo gi Malaysia
20t1
t,il
s$lflt

JtrUt il L.i ,L*,].-l.

'::rrr.
@[llg!i]i[]l
-''rio
liilulrulrg[,,

i l5i,i]f: lLl:iil:
irifrilWtrUjr'i lrin l
ll:f l[]ir:]

'lfiullrlnrfl 'nrry* ;
'frur,,urult "r,,' ',LY:

,l[r"ifi
illunmi lyTTii

m[ ftt /ri iIT:r

.te
rd
so
nt
CS

's.

ip
CHAPTE,R 1

ry
rn

n
INTRODUCTION TO
al GONTROL SYSTEMS
d

h
n

Outcomes:

,ifier completing this chapter, the reader should be able to:

ll define the objective of a control sysrem,

l) define and explain what are open-loop and closed-loop systems, and
,j ) describe the tasks of a control engineer.

1.1 WHAT IS A CONTROL SYSTEM?

In our daily life, there are numerous objectives need to be fulfilled. For instance, we need ro
regulate the temperature and humidiry of homes or building so that we will have a comfortable
living. The temperature of an iron needs to be controlled such that it is not too hot or too cold
so that the ironing of clothes can be done properly and safely. Meanwhile, washing machines
need to be designed such that rhey can do the washing effectively.
In the field of transportation, automobiles, airplanes, and trains need ro be controlled so
thlt they can go from one point to another timely, saf.ly and economically. \fhile in industrial
field, boilers' power generation stations, and robots need to be controlled so that the machines
and systems can work properly and economically.
A controller is a device for controlling a source of power in which the output is controlled
to be some function of the input. A controller is designed to fulfill .o.r,rol system's design
2 I Introduction to Contol Engineering

objective. The design objective is to ensure the actual output of the system is equal ro the desired
output or the reference signal. Formally, a system can be defined as any collection of interacting
elements for which there are cause-and-effect relationships among the variables. Furthermore, a
control system is an interconnection of components forming a system configuration which will
provide the desired system response. In short, a control system should be able to produce the
value of an output variable similar to the desired value.

1.2 SYSTEM REPRESENTATION

A system is normally represented using the block diagram representation but its signals (or variables)
are given in the form of unidirectional arrows, as shown in Figure 1 . 1. Every block has an. input
that affects the output of the components. A component or system may have more than one
input. An input signal is represented by an arrow pointing into the block, while an ourput signal
is represented by an arrow pointing away from the block. Signais are also known as variables. In
-:
Figure 1.1, the variable u(t) is an input signal that can be manipulated or controlled. The
:
disturbance , n(t), is also an input signal but it cannor be manipulated.
--: .:

Disturbance, z(r)

Output,y(r)

Figure 1.1 Representation of a system by a block diagram

!t .+ ffi,
. tn -- -

. i - --
1.3 TERMS AND DEFINITIONS iIT*j" rll'-f
.l- llT I1/r. !l:r
State variable is a quantity that changes with time. It refers to the past, presenr, and future
l[lr L] ]r. !i
conditions of the svstem. The state variables of a system are defined as a minimal set of variables
such that knowledge of these variables at any time /6, and information on the input excitation
subsequently applied, are sufficient to determine the state of the system ar any future time / > /s.
Examples of state variables for some systems are as follows:

(1) For a mechanical system, the state variables might be the displacement, velociry accelerarion,
force, angular displacement, moment, etc.
(2) For an electrical system, the state variables might be the current, voltage, and power.
 Introduction to Control Systems I S

red i r For an economic system, the state variables might be the monetary exchange rate, GDB
ing stock index, etc.
er2
vill Output variable is,the quantiry to be controlled that is of concern to the user. Examples
of
*re ourput variables are as follows:

1) For an engine, the output variables might be the angular velociry or the shaft
torque.
l) For a car, the output variables might be the displacement, ,r.lo.iry or acceleration.
,l ) For an iron, the output variable is the temperarure.
es)
)ut Input variable is a reference quantity or a desired value for a feedback control system.
For a
ne in the control system, input variable is also known as rhe forcing function. It is a
'omPonent
quantiry that can be manipulated. The input variables for selected control ,yr,J-,
ra-l are as follows:
In
he
1) The reference temperature for an oven.
2) The reference angular velocity of an engine.
3) The reference warer level in a boiler.

\X4rile, examples of input

variables for components in a conrrol sysrem include:

(1) The water flowrate into a tank.

'l) The electric currenr supplied to the proportional valve.
-1) The acceleration measured by an accelerometer.

1.4 EXAMPLES OF CONTROL SYSTEMS

A motor cycle can be considered as a system and can be shown as a block
diagram represenration
in Figure 1.2. The throttle angle , 0(t), andthe vehicle speed, u(t), maybe
con"siderejas the input
and outputvariables, respectively. in a block diagram representatio.r,
,h. variables are represented
by arrows and components are represented by bLcks. Th. ortptrt variable
€ of a component can be
the input variable for another componenr.
5
n

).

Figure 1.2 Block diagram representation of a motor cycle

4 ] Introduction to Control Engineering

Figure 1.3 shows a schematic diagram of a liquid level control system. The reference input
for the system is the reference level, h,^1. This level is the desired level of the control system.
The output variable is the liquid level, h(t); the actual level is measured by a sensor. The objective
of the designed control system is that the outputvariable, h(t),is desired to be the same as the
reference variable, ,,4i, even though there exists a disturbance or a change in the reference value.
The signal r(r) is the difference berween the reference variable and the output variable, and it is an
input variable for the controller. The output of the controller is u(t) and it will instruct the
acruator to increase or decrease the valve opening thus forcing the output variable to be equal to
the reference variables. The block diagram representation of this system is shown in Figure 1.4.

4;(r)-

q,(.t)

t
Figure 1.3 A liquid level control system
+: dJ
E rl^ilF
tyi|ir

trLr][ iilr ilj-:1

"il[@Hllff']r ilIlillil
,j0m[il;"lf''r]!, i"1[t

1m5.* &
,Sh: rrfiJrfl,ax1i"-aru

Figure 1.4 Block diagram representation of a liquid level system iqulmqq iilrr, :r
*''"]l;:
,Il0l[8tr0][r[ril.

illltr[lo]iidlllLrri,mr. .il

Exercise 1.1
A speed control system is shown in Figure 1.5. The mechanical goyernor is acting as the sensc,:
element that measures the angular velocity of the engine. The hydraulic valve allows the spooi -
be displaced by a small force produced by the governor to control the fuel control valve. Expl-:
how this system works and list all the variables in the system. \Mhat are the input and ourp-:
variables of the system? Draw the block diagram representation of the sysrem.
 Introduction to Control Srystems I S

Put ---' ]
:m.
ive
the
ue.
an
lre
to f-L.:lt Lt--t:'r..ta,
.4.
<_
Oil under pressure ---+

Hydraulicvalve

Fuel -------+
Control valve

Figure 1.5 An engine speed control system

1,5 CONTROL SYSTEM

The objective of a control system as explained earlier can be achieved by employing rwo possible
control strategies. These control strategies are open-loop control sysrem (also known as feedforward
control system) and closed-loop control system (also known as feedback control sysrem).

1.5.1 Open-loop System

,{n open-loop control system utilizes an actuating device to conrrol the process directly without
using any feedback. In an open-loop control system, the output has no effect on rhe control
action. The block diagram contains one signal path from the input ro the output without any
junctions, as shown in Figure 1.6.

Actuating
v signil, u(t)
io

n
rt
Figure 1.6 An open-loop control system
6 I Introduction to Control Engineering

A domestic orren shown in Figure 1.7 rs an example of an open-loop control system. The L- ;:.: , :

block diagram represenration of the system is shown in Figure 1.8. The temperature of the oven x.- t.-

depends on rhe setting as well as other quantities beyond the control of the operator, such as --: r- r;
ambient remperarure and efficiency of the heating element. The heating unit is controlled by a
switch with several settings. The switch position controls the current flow through the heating
element. This position is the input to the system since it will cause the system to produce an
[jl;I'L::.
output in term of the oven temperature. Since the current flow to the heating element is rh. J{ li

quantiry that is varied, it is known as the manipulated variable or the controlled signal.

Switch position

*/
Power source
Temperature

T::tln
Heating element ililm/[!flui]lH lrl[ilrmui

fiffilmmflinlr ]rror

,fimx$muHs ffi!

Figure 1.7 Schematic diagram of a domestic oven WlI0S@ilU. mi

!fifrii*

For a certain switch position, the oyen temperature depends on the current flo..- r,: --. - ruuEfimm

heating element and the heat loss through the oven wall. If the oven temperature is not the .,::-. UOr@
as the desired temperature, this open-loop system cannot automatically change the input curr..,.
If there is a change in the ambient temperature, the oven temperature will also change. Theret,::. ru il@fmrI fi

an open-loop sysrem cannot make the desired correction in the presence of disturbance.

Input setting Current Temperature

Figure 1.8 Block diagram representation of a domestic oven

1.5.2 Closed-loop System

A closed-loop control system maintains a prescribed relationship between the output and rh.
reference input by comparing them and using the difference as a mean of control. The controlle;
 Introduction to Control Systems I Z

variable,y(r), is measured and fed back for comparison with the reference input, and an actuating
signal is sent through the system to correct the error. This system is also known as a feedback
control system. Figure 1.9 shows the block diagram representation of a closedJoop system.

Desired
output,
u(r)
Controller

Measurement

Figure 1.9 A closed-loop controlsystem

\fr/ith a thermometer in the oven, a human operator can act as a controller to change the
switch position so that the oven temperature approaches the desired temperarure. This is an
example of feedback control system. The controller action depends on the error or the difference
a
benveen the desired temperature and the actual temperature. Since the closed-loop system is
operated by a human controller, the result may not be satisfactory. The temperature may oscillate
around the desired temperature when the position of the switch is changed. The performance
may deteriorate further due to fatigue and other human weaknesses.
In order to improve the performance, the human controller may be replaced by a mechanical
or electrical device to act as the comparator and controller, as shown in Figure 1 .10. This system
is known as an automatic feedback control system. Now the temperarure is measured by a

Voltage from thermocouple

Reference
voltage

Amplifier

Figure 1.10 An automatic feedback oven temperature control system

B I Introrluction to Control Engineering

thermocouple . The voltage is fed back and compared with the reference voltage rhar represenrs
the desired temperature. The error signal is amplified electronically thus controlling the currenr
flow to the heating element. Since the feedback voltage is subtracted from the reference voltage,
this system is also known as a negarive feedback system.

Exercis
1.5.3 Comparison between Open-loop and Closed-loop Systems L.

The main advantage of a feedback system is that it does not require precise design for the controller.
The problem that may arise from feedback system is stabiliry if the controller and its parameters
are not properly tuned. The comparisons between openJoop and closed-loop systems are given
Erercis
in Thble 1.1.
:rr-:aJ.:

Thble 1.1 Comparisons between open-loop and closed-loop sysrems

Exercis
1

_ f, t
Open-loop system ClosedJoop system -*
: -:---:o1 ri
(1) The system components are simple (1) The system components are more complex a \t1
(2) Require precise design of the controller (2) Controller design is simple
(3) Require a dedicated controller (3) General controller may be used
Exercis
(4) Sensitive to disturbances (4) Not sensitive to disturbances
: .- L\\L
(5) The system is always stable (5) Stability problem needs to be tackled -'=.-ribe
(6) Control action usually based on time or cycle (5) Control action is based on error signal

Exercis
r--,-e m-C
Exercise 1.2 ]"-.;ribe
Give five examples of an open-loop system and five examples of a closed-loop system. Describe
the control action of each svstem. Exercis
l-t :ar art

Exercis
1.6 CONTROL ENGINEERING DESIGN ,i1ar ar,

Control engineers are required to identi$, components to be used in control systems, such as the Exercis
controllers, actuators, and sensors. The controller needs to be designed so that the response of the --i-.'c
nr-r
output variable is equal to the desired value with a satisfacrory speed of response. Before the
conffol system can be implemented, a mathematical model olthe sysrem i,, componenrs
must be derived. Analysis will be conducted to ensure that the sysrem "rrd response meets the
specification requirements using certain techniques such as time response analysis, frequency response Furthet
analysis, stabiiiry analysis, etc. These specifications include transienr response and steady-512s6
specifications. Usually the system response and its performance are evaluated by compurer {uo. B.
 Introrluction to Control Sy*ems I g

3nts simulation. Several computer aided design sofiwares for control system design are available in the
rent market. One of the most popular compurer packages is MATLAB.
€e,

Exercise 1.3
Draw the basic block diagram for a domestic air conditioning unit. Determine the functions of
each component in the system. Show all the variables involved in the system.
Ier.
ers
ren Exercise 1.4
Draw the basic block diagram for a steer control system of an automobile. Assume that the
desired path is the reference input and the driver is acting as the sensor and the controller.

Exercise 1.5
Build a basic block diagram of a position control system where an electric moror is used to
control the angular position of a shaft. A potentiometer is used to measure the angular displacement
x of the shaft.

Exercise 1.6
-ir-e rwo examples of a domestic system which are controlled by an open-loop control sysrem.
lescribe how these control svsrems work.

Exercise 1.7
Give two examples of a domestic system which are controlled using a closed-loop control system.
Describe how these control systems work.

Exercise 1.8
what are the advantages and the disadvantages of an open-loop control sysrem.

Exercise 1.9
\X/hat are the advantages and the disadvantages of a closed-loop control sysrem.

Exercise 1.10
Give two examples of a closed-loop system where human controller are used,.

Further Readings
Kuo, B. C. 1995. Automatic Control Systems. E nglewood Cliffs, NJ: Prentice-Hall International.
10 | Introtluction to Control Engineering

Mason, S. J. 1953. FeedbackTheory: Some Properties of Signal Flow Graphs. Proc. IRE. 4l(9),
1144-tt56.
Mason, S. J. 1956. Feedback Theory: Further Properties of Signal Flow Graphs. Proc. IRE.
44(7):920-926.
Ogata, K.2004. system Dynamics. Upper saddle River, NJ: Pearson Prentice-Hall.
Vu, H. V and R. S. Esfandiari. 1997. Dynamic Systems: ModelinganclAnalysis. NewYork:
McGraw-Hill.
Voods, R. L. and K. L. Lawrence . 1997. Modeling and Simulation of Dynamic Systems. Upper
Saddle River, NJ: Prentice-Hall.

2.1 M,

Outcome

.-.-.._-

- uf \Li l
- ..-T--
..l-
- t-'.'

2.1,1 Me
I (e):

[ork: CHAPTER 2
pper

MATH EMATICAL MODELLI NG

2.1 MATHEMATICAL MODELS

Outcomes:

--er completing this secrion, the reader should be able to:

1) describe the importance of a mathematical model, and

t2) apply the principles used in developing mathematical models

2.1.1 Mathematical Model of a Physical System

Control engineers analyse physical systems based on mathematical models. These models are
derived from physical laws. From the models, the subsequent analysis such as rime response,
lrequency response, stabiliry and controller design that satiifies some design specification, .6 b.
performed. The mathematical model used in this book is in the for* o? differential equation.
From the differential equation, a representation in the form of a transfer function is obtained
to
describe the behavior of the system. The assumptions used in the derivation of the mathematical
model include linear relationship between the variables, ignoring small effects in the
sysrem, rhe
system is not influenced by the surroundings, using lump parameter estimation,
the system is
time invarianr, and ignoring uncertainties and noise.
Some of the commonly used models are:
t2i Introduction to Contro I Engineering

(1) Scaled model

(2) Prototype model
(3) Mathematical model

Scaled and prototype models can be expensive and time consuming to develop and modifr.
Any modification may involve expertise, fabrication, additional componenrs, etc. Mathematical
model is cheap and can be developed quickly. Once the designer is satisfied with the performance
of the model, a scaled or prototype model can be developed. This will ensure minimal
modifications. L2"2"1 L,
A1l dynamic systems, such as engineering, economic, biological, or ecological systems, can
be described by differential equations. These differential equations have the same characteristics
even though theywere derived from different types of physical systems. Control engineers use
the same techniques to analyse these systems. Therefore the theory and analysis studied by the
control engineers are the same for all dynamic systems.

2.2 BLOCK DIAGRAM REPRESENTATION

Outcomes:

After completing this section, the reader should be able to:

(1) derive the mathematical model and draw the block diagram for translational mechanical
system, rotational mechanical system, gear system, hydraulic system, liquid level system,
thermal system, and electrical system.

2.2.1 Block Diagram

The derivations of the mathematical models in this book include the study of a translational
mechanical system, rotational mechanical system, gear sysrem, hydraulic sysrem, liquid level
svstem, thermal system, and electrical system. The derivation of the model is initiated by using
the laws of physics that govern the relationship berween the input and output variables of all the
elements in the system. These relationships are then represented by the block diagram
rePresentation. From the block diagram, the transfer function or the marhemarical model can be
obtained using block diagram manipulation method. The transfer function can also be obtained
by transforming the block diagram or representing the dynamic equations into signal flow graph
representation and applying Masonk gain formula or Mason's rule.
 Mathematical Modelling I tg

2.2.2 Translational Mechanical System

There are three basic elements that are used to model translational mechanical sysrems. These
elements are spring mass, and damper. The constants that represent the elements are known as
the parameters. The state variables are the quantities of interest ro rhe designers and users and in
w this case are displacement, x(t), velocity, u(t), acceleration, a(t), and force,
cal f(t).The dynamics
tce
describing these elemenrs can be derived by applying Newtons law.
ral
2.2.2.1 Linear spring
an
Figure 2.1 shows the symbol for a spring. The parameter, b, is the spring consranr or the spring
ics
stiffness and the variable, x(r), is the displacement. The relationship between the force,
se f(t), and
the displacement, x(r), is assumed to be linear as shown in Figure 2.2.Therefore, we can write:
he

f(l = kx(r). (2.1)

Figure 2.1 Symbol for a spring

Slope = *

x(r)

Figure2.2 Relation between force and displacement

Figure 2.3 is the block diagram representation of Equation (2.1). The input
of the block is
the displacement, x(t), andthe output is the force, fQ).Theexpression in the block is known as
the transfer operator. If the force,
f(t),it defined as input "nd th. displacemenr as ourpur, rhe
equation can be written as:

x(t):)rr,l (2.2)
14 | Introduction to Contol Engineering

x0\ d-,,/- Lt
/ln
-:-- -
;- - -
Figure 2.3 Block diagram representing Equation (2.1)

thus the block diagram representation is as shown rnFigure 2.4.

Figure 2.4 Block diagram representing Equation (2.2)

In general, the block diagram represents the equation in the form of,
Output uariable = trdnsf€r o?eruttor x input uariable
If both ends of the spring can be displaced as shown in Figure 2.5, the force in the spring is
proportional to the effective elongation of the spring and the relationship can be wrimen as:

fO=k(x1Q)-xz@). (2.3)

l-r ,,(r) j nG)

k
ffi
Figure 2.5 Spring with displacements at both ends

The block diagram for Equation (2.3) is shown in Figure 2.6.

Figure 2.6 Block diagram for spring with displacements at both ends
 Mathematical Modrlling I tS

2.2.2.2 Translational motion of rigid bodies

Figure 2.7 shows a bodywith a mass, m, translates in a straight line subjected to aforce,f(t).
From Newton's equation of motion, we can write:

f (r)=ma(r)=*'':\').
dt
(2.4)

n
Using D-operator, that is D =;, we have a(t) = D2x(t). Thus we can write:

f (r) = mD'x (t). (2.5)

The block diagram for Equatio n (2.5) is as shown in Figure 2. 8 . Note ,tr", IDI represents an

-,-.qrator. Figure 2.8 can be manipulated to become an equivalent block diagram as shown in
- igure 2.9.

is

,)

Figure 2.7 Force acting on a mass

dxO = a(t) Dx(t) = u(t)

Figure 2.8 Block diagram for the system in Figure 2.7

Figure 2.9 The simplified block diagram

16 1 Introdrrtion to Control Engineering

2.2.2.3 Mass-spring system l-22"4 Dt

Figure 2.10 shows a mass-spring system. The left end of the spring is displaced by x;(t).This will :'-i--- -
-
cause the mass to translate with the displacement, xo(t).The equation for the spring is: -:u: ::r i

f(t)=k(x;(t)*x,(t)). (2.6)
The equation for the mass is:
mD2x,(t) = f@. (2.7)
The block diagram representation of Equations (2.6) and (2.7) can be drawn as shown in
Figure 2.1 1.

j x,G)

///////////////t

Figure 2.10 A mass-spring system

Figure 2.11 Block diagram of the mass-spring system z.2.2,5 Sp

Exercise 2.1
Draw the block diagram for a transladonal mechanical system represented by the equarions given
below. The input variable is x;(t) and the output variable is x,(t). F(r) is the force and k and m are
the system paramerers.

F, (r) = k, (x, (t) * * (t))

"
F,(t) = k, (x, (t) - * (t))
"
F (t)= F,(t)+ F,(t)
F (t): mD'x"(t)
 Mathematical Modzlling I tZ

2.2.2.4 Damper
ril I Figure 2.12 shows the symbol of a damper. The characteristic of a damper is that the force, F(r),
that resists the motion, is proportional to the relative velocity of the piston ro rhe cylinder. If the
cvlinder is stationary, then we can write:
6)
F(t) = cDx(t) (2.8)
where r is a constant known as coefficient of viscous friction. The corresponding block diagram
7)
is shown in Figure 2.13.
in

Ftgure2.12 Adamper

Figure 2.13 Block diagram for a damper

2.2.2.5 Spring-damper system

Figure 2.14 shows a spring-damper system. Here, we assume that the masses of the moving
parrs,
such as spring and damper, are negligible. The equations that completely describe this system
are:
The spring force, F,(t) = k(x;(t) - x,(r)). (2.e)

The damper force, FoQ) = cDx,Q). (2.10)

Since the spring and the damper are connecred, therefore:

F,(t) = Fo(t). (2.1r)

The block diagram for Equations (2.9) to (2.11) can be drawn as shown in Figure 2.15.
18 | Introduction to Control Engineering

Figure 2.14 A spring-damper system

Erercisr

: ,:_._ - ':
::: _;:-

Figure 2.15 Block diagram for the spring-damper system

2.2.2.6 Mass-spring-damper sysfem

Figure 2.15 shows a mass-spring-damper system. The equations that completely describe this
system are:

The spring force, F,(t)= k(x;(t) - x,(l). (2.t2)

The force on the mass, F,Q) - FoQ) = mD2x,(t). (2.t3)

The damper force, FoQ) = cDx"(t). (2.14)

The block diagram representation of Equations (2. 12) to (2.14) can be drawn as shown in
Figure 2.l7.It should be noted that the appearance of a block diagram depends on how the
variables are defined and it is not unique.

;e F. xo(n

oo
777)
Figure 2.16 A mass-spring-damper system
 Mathematical Modelling I tg

+'\

FoG)

Figure 2.17 Block diagram for the mass-spring-damper system

Exercise 2.2
Derive a set of equations that completely describe the system shown in Figure 2.18 and draw the
biock diagram to represent this set of equations. The input variable is the displacemen t x;(t) and
ihe output variable is the displacement x,(r).

| *,1t1

ts
oo
ts r,(rl 777777',/
Figure 2.18 Schematic diagram for Exercise 2.2

2.2.3 Rotational System

Rotational system is similar to a translational system. Basic elements for the rotational system are
shaft stiffness, Z, which has the same characteristic as the spring stiffness, k, the fluid viscous
friction coefficient, B, which is similar with the coefficient of ,rir.ou, friction, c, and.the mass
moment of inertia, /, which is similar with the mass, m. The observed variables are the angular
displacement, 0(r), which is similar with the displacemen t x(t), angular velociry
al(r), which is
similar with the linear velocity, u(t), and the acceleration , i(r),which is simiiar with the
"ng,rlrr
linear acceleration, a(t).The torque, t(t), actingon the momenr of inertia for the rotational
system has the same characteristic as the force, F(t). For example, ler us
study the rotational
system shown in Figure 2.19. Note that Figures z.L9(a) and Z.iqft)
ar. equiufl.nt. Similar to
translational system, the set of equations that completely describe the ,yrr.-
are as follows:
20 1 Introdurtion to Control Engineering

Rotational stiffness, r,(t) = k(9i(l- O,(il. (2.t5)

Rotational damper, ,lt = BD9,(t). (2.16)

Moment of inertia, r,(t) * ,lt) = 1D2 e"1t1. (2.t7)

The block diagram representation of these equations is as shown in Figure 2.20.This diagram
is similar with the system in Figure 2.17 with the variables and parameters are replaced as follows:

0(t) :- x(t), r(t) = F(t), J = *, B= c

d-here ro
-\l is the
-ie angu

o;(t)
*...., 3v diffe:

Figure 2.19 An inertia-stiffness-damper system. Systems (a) and (b) are equivalent
- tr t(

+-, - :irrn:

TAD

- :i- for(
Figure 2.20 Block diagram for the inertia-stiffness-damper system

2.2.4 Gear System

The observed variables for a gear system are torque, angular acceleration, angular velociry and r'.irere ?

angular displacement. \Mhile the system parameters are the number of gear ieeth or radii that 'r'.-r-ite:

produces the gear ratio.

Let us study the gear system shown in Figure 2.21 with gear A the driver and gear B as the
driven. We define the gear ratio, n, as:
The bic
rN knorvn
"^_ d _ 4
ft--
(2.18) amplifir
rb Nb
 Mathematical Modelling I Zt

15)

r5)

L7)

am
MS:
Figure 2.21 A gear system

''rhere roistheradius of gearA,rgistheradiusofgear B,Noisthenumberofteeth forgearA,and

-\7 is the number of teeth for gear B. Let the angular velocities of the gears are ar(t) anl ase) and
fie angular displacemen ts are )o(t) and. eb(t).fusuming there is no sli"pping o..i,rr, we can wrire:
rn9,(t) = r506Q). (2.19)
Bv differentiating 0(r) with respecr to time, we ger:

do Q\ do,()
_l
I
n;-=oA-' (2.20)

That is, in terms of angular velocities, we ger:

r,AnO = r5(D6(t). (2.2t)

[n terms of gear ratio, n, the variables are related by:

e,Q) _a,(t) rN ____L-,.fl.

_ _!_-
0,(r) aLQ) rb Nb
- (2.22)

Ihe force acting on the tooth of both gears is:

r' (t)
F (t) =r'(')
rr,
- (2.23)
uo

*l:* r,(t) is the torque acting on gear A and. tse) is the torque acting on gear B. Then, we can
wilte:

rr(r)=!',(,). (2.24)
n
The block diagram for the considered gear is as shown in
Figure 2.22.If n > 1, the gear system is
known as a reduction gear. This systJm will cause the reduction
in the angular velociry and
amplification of the torque.
 Introduction to Control Engineering

-',f the desigr

3.24. The sp
-oad is7(r).'
side of the p
'.-olumetric I

Figure 2.22 Block diagrams for a gear system

.'here * is a
:nd P1(r) is r

Exercise 2.3 a-rite:

The equations that describe the dynamic system in Figure 2.23 are given as:

c (t) = L',(')
rd

r (t) - r, (r) - r, (t) = P'e (t) It P1(r) is as

r, (t) = ke (t)
tu(t)= eo1 (r) We can also

where r(r) is the torque, r is the radius, and 0(t) is the angular displacement. Draw the block
diagram representation for this set of equations. Take coQ) as the input variable and 0(t) as the
output variable. 'n here ,4 is t

ba-sed on th
orifice, and

Figure 2.23 System for Exercise 2.3

2.2.5 Hydraulic System

The observed variables in a hydraulic system are the valve displacement manipulated by an operator
and the piston displacement that moves heavy load. System parameters are the physical properties
 MathematicalModelling I ZZ

of the designed valve and piston load. Schematic diagram of a hydraulic valve is shown in Figure
).24.The spool displacement is representedby x(t) and the piston displacement that moves the
load is7(r). tWhen the spool is displaced to the right, the high pressure fluid flows into the right
side of the piston and fluid on the left side of the piston will flow back through the capillary. Th.
r-olumetric flow rate, q(t), is:

q(t)= t.@"!ir@ (2.25)

1= P,(t)+zLP (t) (2.26)

n'here Z is a constant, LP(t) is the pressure drop across the valve , P, is the supply pressure source,
and 21(r) is the pressure difference across the piston. From Equations (2.25) and (2.26),we can
,.r-rite:

q(t)= bx(t) F:,N

!2 (2.27)

Ii PL(l is assumed to be small compared with P,, then:

q(t) = K' x(t). (2.28)

We can also write:

q(t) = ADyQ) (2.2e)

rvhere,4 is the cross sectional area of the piston and, 17'= k^E.These equations
I are derived
Vz
based on the following assumptions: zero nonlineariry incompressible fluid, no leaks, annular
orifice, and a constant P,. The block diagram is shown inFigure 2.25.

It,
rrll ,*," ,t,

LP(t) LpG)

Figure 2.24 Schematic diagram of a hydrauilc valve

24 I Introduction to Controi Engineering

Figure 2.25 Block diagram of a hydraulic valve

2.2.5.1 Position control system: A seruo-mechanism

The system shown in Figure 2.24iehaves as an integrator. -When a consranr input, x(r), is applied,
L2.5.2
the position of the load represented by the output, y(t), wlll moves ar a consrant speed until it
hits the end of the cylinder. Controlling the position of the load is difficult since ihe operaror :'--.--
- -:J1L
-) -),
-.-(
needs to move the spool position in opposite directions conrinuously. This position conr;l (also -,4-:o aar, aL
known as servo-mechanism) can be achieved by introducing a walking lever as shown in Figure
2.26.In this system, the load is controlled by manipulating the end of the lever, z(t).
The equation rhar relates the displacement x(r), y(t), andz(r) is given by:

x(t)=*+9 (2.30)
- -,: rhe lev

if the length \ and l2are assumed equal. The block diagram of this hydraulic servo system is given -:,:m the S

inFigtre 2.27 .

\ o1r\ -'.,l
a,u

-,
*(t)

+Y@
<-
s,G)

Figure 2.26 A position control system

 Mathemarical Modelling I ZS

Figure 2.27 Block diagram of a hydraulic servo-mechanism

[,
2.2.5.2 Walking lever
,t
r Figure 2.28 shows a general case of a walking lever. The equation that describes the relationship
D between the variables is:
e
tt
x(r)=;:I rlr)-;lt+l)'
; tQ ). (2.3r)
lt+l)
For the lever shown in Figure 2.2.9, it can be seen thar:

(2.32)
From the similar triangle relationship ,*r:::*:
'O) = lr+ a (2.33)
*(t) d
ard

tG) -!:1,-o
*(t) a d (2.34)
Substituting Equations (2.33) and (2.34) into Equation (2.31), we ger:
I t- I t
x(t)=-L- n
--:-t- .'r- *(t)= *(r).
4+ l) d l1+l) d
=.e)_

,J7"
Figure 2.28 A walking lever
26 | Intoduction to Control Engineering

ivhereflz
Draw the
variable i

{ x(

,I 2.2.6 r
The obsr
|4
y(t)
paramete
t-iowing i
liquid flc
required
Figure 2.29 Displacements of the walking lever

Exercise 2.4
The equations that completely described the dynamics of the system in Figure 2.30 are given as:
Let
rhe tank
f ,, (r) = k, (* , (t) - x, (t)) out of th,
*nQ)=L*"(r)
d

f,, (r) : k, (*, (t) * x, (t))

Usir
f,r(r): *D'*,(t) ofincreas
bf,,(r)= af,,(t) ;an write

Figure 2.30 System for Exercise 2.4

 Mathematical Motlelling I ZZ

,.hereflr) is the force

andx(t) is the displacement, whrle a, b, k, and m arethesysrem parameters.
Draw the block diagram representation for this system. The input variable is x;(t) anJ the outpur
rariable is x"(t).

2.2.6 Liquid Level System

The observed variables in a liquid level system are level (head), h(t), and flowrare, q(t).The
larameters for the system are the pipe resistance, R, and tank capacitance, C. Lett look at a liquid
:orving in a short pipe that connects two tanks. For a laminar flow, the pipe resistance, R, onthe
:quid flow is defined as the change in the difference of liquid levels berween the tanks that is
:equired to produce the change of one unit flowrate, thar is:

o _ the change in liquid level, m

the change in flowrate, mr/s (2.35)

Let us study the liquid level system shown in Figure 2.3l.The flowrate of the liquid into
:he tank is q;(t) , h(t) is the liquid level in the tank, and q,(t) is the flowrate of the liqui j flowing
out of the tank. The pipe resistance is:

h(t)
n= (2.36)
q, (t)
Using the principle of conservation of mass, fluid flow rate into the tank is equal to
-
of increase of the liquid in
the rare
the tank plus the flowrate of fluid going out of the t".,k. Th.r.fore, we
can write:

q,(t)=, q"(t)
ry+

Figure 2.31 A one tank system

28 | Introduction to Control Engineering

or using D-operator:

q, (t) = CDh (t)+ q, (t) (2_37)

If th
where Cis the tank capacitance which is equivalent to the cross-secrional area of the tank. From r-ariable, tl
Equations (2.36) and (2.37), the block diagram can be drawn as shown inFigure2.32.

qO

Figure 2.32 Block diagram for the one tank system

2.2.6.1 lnteracting liquid level system

Let us study the interacting liquid level system as shown in Figure 2.33.Theequations that relate
the variables are:
Exercise
h,(r)- h,(r) , Derive thr
--s "=q,Q) (2.38) 1.35. Dra'
i-ariable ar
ClDhtQ)= q(t)- q,(t) (2.3e)

h,G) /\
=q'lt) (2.40)
R

t
hr(t) t
hr(t)
I +

Figure 2.33 An interacting liquid level system

 MathematicalModelling I Zg

C2Dh2Q)= q,(r)- qr(r). (2.41)

)7)
If the flowrate into the tank, 4(r), is defined as the input variable and h2lt) is the output
)m rariable, then the block diagram representation is as shown rnFigure 2.34.

Figure 2.34 The block diagram of the interacting liquid level system

Exercise 2.5
Jerive the set of equations that completely describe the dynamics of the sysrem shown in Figure
3) 1"3>. Draw the block diagram representation for this system by assuming thar.
q(t) is the iriput
.ariable and h2Q) is the ourput variable.
))

Figure 2.35 A non-interacting liquid levelsystem

30 | Introduction to Control Engineering

2.2.7 Thermal System L2.8 Elt

The observed variables for a thermal system are the temperature , 0(t), and the heat fl,owrate, h(t). la.-ra:ic e

The system parameters are the thermal resistance, R, and thermal capacitance, C.Letus study the ru:-illes ar
thermal system shown in Figure 2.36.The mathematical model is derived by assuming that there :tr: -.ne-.e el
is no heat loss to the surrounding, no heat is stored by the wall, and the fluid in the tank is well
i..st.ror:
mixed with uniform temperature distribution. The thermal resistance is:
--
i.uctor:
e (t\ .*racitor:
R_ (2.42)
h"(r)
The b
where 0(r) is the fluid output temperature and b,(t) is the rate of heat flowing out of the tank. :a*-n;ftr fur
From the principle of conservation of energy, that is the heat flowing into the tank is equal to the For n
heat stored in the tank plus the heat flowing out of the tank, we get:

h,(,)=cff{,)* h,(,). (2.43)

From Equations (2.42) and (2.43), the block diagram representation of the thermal system
can be drawn as shown inFigrre 2.37.

Figurt

D Hotfluid

Coldfluid---}

Figure 2.36 A thermal system

Figur,

Figure 2.37 Block diagram for the thermal system

 Mathematical Modelling I gt

2.2.8 Electrical System

). The basic elements for a passive electrical system are inductor, L, capacitor, C, andresistoq R. The
le variables are the voltage, \t), andcurrent, i(t).The relationships between the voltage and currenr
re for these elements are given as follows:
:ll
Resistor: V(t): P;1r7. (2.44)
Inductor: V(t) : LDi(t). 12.4s)
)) Capacitor: CDW = i(t). (2.45)
The basic principles of a passive electrical circuit used to develop the block diagram and its
k. transfer function are as follows:
IE For two elements in series as shown in Figure 2.38, the voltage relationship is:

VG)=V1Q)+V2Q). (2.47)

3)
*<v,i)- fryt-

Figure 2.38 Two elements in series

Figure 2.39 shows two electrical elements in parallel. The voltage relationship is:

VQ) = vr(t) (2.48)

Figure 2.39 Two elements in parallel

Figure 2.40 shows a currenr divider at a node. The current relationship is:

i(t)=ilt)+i2Q).

Figure 2.40 Current flows through a node

32 | Introdoction to Control Engineering

Figure 2.41 shows an electrical circuit consists of an inductor, L, capacitor, Cand resisror, &
in series. The equations that describe the dynamics of the sysrem are: Exercisr
Figttre 2.2
VQ) =V,(t)-V,(t)-u, (t) (2.50)

"lu,*'i'
describes
is the inpr

'*',,',',',
CD

\(t)= Ri(t). (2.53)

Equations (2.50) to (2.53) relate the voltage across and the current through the inductor,
capacitor, and the resistor. The block diagram for this system is shown inFigure 2.42.

Exercisr
Frgure2.4
describe tl
rhe input

Figure 2,41 An electrical circuit

Exercisr
Figrue 2.'
equations
-\sume tl

Figure 2.42 Block diagram for the electrical circuit

2.2.9 Other Systems

For other engineering sysrems and their hybrid and other dynamic sysrems such as economic,
biological, ecological systems, etc. the block diagram representation can be obtained using the
same mentioned method.
 Mathemariral Modelling I gg

Exercise 2.6
Figure 2.43 shows a translational mechanical system. Derive a set of differential equations that
describes the dynamics of this system. Draw the block diagram by assuming that the force, F(t),
is the input and the displacement, x1(l), is the output.

2)

3)
)r'

Figure 2.43 System for Exercise 2.6

Exercise 2.7
Figure 2.44 shows a translational mechanical system. Derive the set of differential equations that
describe the dynamics of this system. Draw the block diagram by assuming that the force, F(r), is
rhe input and the displacement, x1(t), is the output.

Figure 2.44 System for Exercise 2.7

Exercise 2.8
Figure 2'45 shows a dynamic vibradon damper to control vibration.
Derive the differential
equations describing the dynamics of this system and draw
its block diagram ,.pr.r..r,r,iorr.
-{ssume that the force, F(t), is the input and the displacemen t, x1(t),i,
th."o,r,p,,t.'

T*,

Tno
Figure 2.45 System for Exercise 2.g
34 | Introduction to Control Engineering

Exercise 2.9 -quatlo

The shaft that supports the disk in Figure 2.46 is made-up of two secrions with the stiffness
of ,{ssumr
$ yd r(2 respectively. Derive the differential equations describing the dynamics of this sysrem
and draw its block diagram representation. Assume that the ,orq"r., rO, is the input and
the
angular displacement, 0(t), is the ourput.

Exerci
Derive
Figure 2.46 System for Exercise 2.9 assumir

Exercise 2.10
Figure 2.47 shows a mo disks torsional system. Derive the differential equations
describing the
dynamics of the system and draw its block diagram representation. Assume that the
torqu., r(r),
is the input and the angular displacement,0/t),is the output.

Figure 2.47 System for Exercise 2.10

Exercise 2.11
For the system shown is Figure 2.48, the rotational motion of the lever is small
such that the
motion of the top end and the middle point are essentially horizontal. Derive the
differential
 Mathematical Mofulling I gS

equations describing the dynamics of the system and draw its block diagram represenrarion.
)f Assume that the force, F(t), is the input and the displacement, x(t), is the output.
n
le

Figure 2.48 System for Exercise 2.11

Exercise 2.12
Derive the differential equations for the RLC circuits in Figure 2.49. Draw the block diagrams by
assuming that the voltage, v;(t), is the input and the voltage, vo(t), isthe output.

"--,-3_,--li-51
vt,)l ,3
lv"rrt
(4

14(dl

Figure 2.49 Circuits for Exercise 2.'12

36 | Introrlurtion to Control Engineering

Exercise 2.13
Figure 2.50 shows an armature control DC motor with a resistor, R, added in the armature loop.
The voltage across the resistor is fed back to the input of the amplifier. Derive the differential
equations for this system and draw its block diagram. Assume that the voltage, V;(t), is the input From Eqr
and the angular displacement, 0(t), is the output.

or

Ir.u,
* \ ,A,t
This ratic

Figure 2.50 The system for Exercise 2.13

2,3 BLOCK DIAGRAM MANIPULATION METHOD

Outcomes:

After completing this section, the reader should be able to:

(1) obtain the transfer function of any dynamic systems represented by block diagrams using
block diagram manipulation method.
2.3.3 Bl

2.3.1 Concept Ftgure 2.5

equivalent
Block diagrams are often used to represent control systems. Block diagram represenrs rhe
relationship between all the state variables for all components in a system. Usually it is a complex
diagram. In general, control engineers like to use a simplified block diagram rhat concerns only but
with the important variables need to be observed. The complex block diagram can be simplified
using certain rules. The simplest form of a block diagram only relates the input variable with the
output variable. The ratio of these two quantities is known as the transfer function. arld

2.3.2 BIocks in Series

Therefore
Figure 2.51 shows a block diagram in series. Two or more blocks may be connected in series. Two
blocks in series can be simplified into a single block as shown in Figure 2.52.This can be shown
as follows:
 Mathematical Modelling I Zl

e^t = Fr(D)9r(t) (2.54)

qo = Fr(D)0.r(t). (2.55)
From Equatrons (2.54) and (2.55), we ger:

e^t = Fz@) Fr(D) )r(t) (2_56)

0^(t\
#0,(r) = F,(D)F,(D).
1\ ' t' (2.57)

This ratio is known as the transfer function.

Figure 2.51 Blocks in series

FJD)F2@)

Figure 2.52 A simplified block diagram

2.3.3 Blocks in Parallel

Figure 2.53 shows rwo blocks in a parallel connection. These blocks can be simplified to be
equivalent to a block diagram shown in Figure 2.54.This can be shown as follows:
fie
lex eM=e^t)+q@ (2.58)
rly but
ed
he 0^d = Fr(D)}r(t) (2.5e)
and

0z(l = Fr(D)1r(t). (2.60)

Therefore
MO

vn
9aQ) = FID) 0{t) + Fr(D) 1r(t)
= \Fr@) + F2(D)) 01Q) (2.6r)
38 | Introduction to Con*ol Engineering

or, the transfer function is: - :lis rel

-;nctior
enQ)
= r\(D)+ F,(D). (2.62)
o,(t)

Figure 2.53 Blocks in parallel

FID)+Fz(D)

Figure 2.54 The simplified block diagram

2.3.4 Negative Feedback

Examp
Figure 2.55 shows two blocks connected in a negative feedback configuration. This block diagram Obtain
can be simplified into an equivalent single block as shown in Figure 2.56. This can be shown as in Figur
follows:
9a(t) = FlDq(l (2.53) Solutit
From Sr
e^t = F^De(t (2.54)
Jiagram
erOl=?Jt)-qzQ). (2.65) simplifi,
Eliminating 02Q) and fu(t) from Equations 12.63) to (2.65), we get:

+9
Ft(D)
=0,(t)- F,(D)O,(t). (2.66)

Rearranging yields:

0,(r)
_ Ft(D) (2.67)
0,(t) r+ F1@)F,(D)'
 Mathematical Modelling I gg

This relationship is known as the negative feedback law. Therefore, the ciosed-\oop ttansfer
function is given by:

enQ) _ foward path transfer function

er(t) 1 + open loop transfer function (2.68)

Figure 2.55 A negative feedback system

t+ Ft(D)F,(D)

Figure 2.56 The simplified btock diagram

Example 2.1
m Obtain the transfer function for the mass-spring sysrem represented by the block diagram shown
as
in Figure 2.11.

7) Solution
From Section 2.3.2, rwo blocks in series can be simplified resulting into the intermediate
4) block
diagr.lT in Figure 2.57. From Section 2.3.4, the negative feedback relationship gives the
5)
**i
simplified block diagram as shown in Figure 2.58.The t."rrsf.r function for this system
is:
*" (t)
*,(t) mD2 + h

Figure 2.57 lntermediate block diagram for the mass-spring system

{0 \ Introduction to ConnoL Engineering

Figure 2.58 The simplified block diagram

Table 2.1 shows the equivalent block diagrams that can be used for simplificarion or
manipulation of block diagram to obtain the transfer function of a control sysrem.

Table 2.1 Equivalent block diagram

_ It
x#A-B+C
-t Exer
ln Obta
the rr
A C
Exer
Obta

C
Exer
Obtai
D

Exer
Obtai
A
a Exer
Al-
Obtai
t, B
Exer
Obtai
 Mathematical Modelling I +t

Thble 2.1 Equivalent block diagram (cont.)

OI

Exercise 2.14
Obtain the transfer function for the interacting Iiquid level system shown in Figure 2.34 using
rhe manipulation of block diagram merhod.

Exercise 2.15
Obtain the transfer function for the sysrem in Exercise 2.5.

Exercise 2.16
Obtain the transfer function for the sysrem inExercise 2.7.

Exercise 2.17
Obtain the transfer function for the sysrem in Exercise 2.g.

Exercise 2.18
Obtain the transfer function for the system in Exercise 2.9.

Exercise 2.19
Obtain the transfer function for the system in Exercise 2. 10.
+2 | Intoduction to Control Engineering

Exercise 2.20 2.4.2

Obtain the transfer function for the system in Exercise 2.1 1.
Additic
Exercise 2.21 As an e:

Obtain the transfer function for the sysrem in Exercise 2.12.

Exercise 2.22
Obtain the transfer functions for the sysrems in Exercise 2. 13.

2.4 SIGNAL FLOW GRAPH

Outcomes:
After completing this section, the reader should be abie to:

(1) Signal
draw the signal flow graph from differential equations,
(2) draw the equivalent signal flow graph from a block diagram, and
(3) use Mason's gain formula to obtain the transfer function.
Based r

2.4.1 Concept
Besides the block diagram manipulation method, the transfer function of a complex sysrem may
be obtained using a simple formula. It is done by redrawing the block diagramtr transforming
directly from the differential equations relating all the variables into a signal flo* graph
representation. From the signal flow graph, the transfer function is obtained using Masons gain
formula' The signal flow graph is another graphical form that can be used to ,.pr.r.r-r, dyrrrlrrri.
"
system. It represents the flow of signals in the system. Figure 2.59(b) is a signal flow graph for the
system represented by the block diagram in Figure 2.59(a). Variables in the signal flow graph are
represented by nodes. Tiansfer operarors are represented by branches.

Produ<
;Node diagrar
z(t) are
xri)#xz(t)
\
\. grr,r.1,
Elimin
(a) (b)

Figure 2.59 A equivalent system: (a) btock diagram, (b) signal flow graph
The eq
 Mathematical Mode lling l+s
2.4.2 Algebra of Signal Flow Graph
Addition: The variable at a node is equal to the sum of all signals coming into the node.
As an example, based on Figure 2.60, thevariable at the output node is:

y(t) = mx(t) + c(t).

Figure2.60 Addition

Signal transfer: The variable represented by a node is transferred to all outgoing branches.

Based on Figure 2.51, the equation for the variables at the outpur nodes are:

t(t) = 3x(t) and z(t) = -4x(t)

may
ning
aph
gain
rnic
r the
I are Figure 2.61 Signal transfer

Product: The transfer operators in series can be simplified into a product, similar to the block
diagram manipulation method. As shown in Figure 2.52(a), the variables at the nodes
z(t) are:
/t) and
y(t) = 10x(t), z(t) = *20yQ)
Eliminating the variable, T(r), we get:

z(t) = -200x(t).
The equivalent signal flow graph is shown in Figure 2.62(b).
44 | Introduaion to Control Engineering

Forward
10 -20 -200
# #
x(t) tG) z(t) x(t) z(t) Feedbacli

(a) (b)
Self loop
Gain:
Figure 2.62 Signal flow graph (b) is simplification of (a)
Path gair

Loop gai
Example 2.2
Draw the signal flow graph for the mass-spring-damper system represented by the block d'iagram
inFrg'tre2.l7.

Solution
The signal flow graph for this system is shown in Figure 2.63.The arrows must be shown on the
branches to indicate the directions of signal flow. The negative sign for the transfer operator is
used when the effect of summation is negative.

F-(t)

2.4.4 n

The trans
expressior
Figure 2.63 Signal flow graph for Example 2.2

where
Exercise 2.23
Draw the signal flow graph for the sysrem in Exercise 2.1. !@
u(t)
Pi
A
2.4.3 Definitions
Li
In general, there are several terms used to describe signal flow graph. Consider the signal flow
graph shown in Figtre 2.54. Jwo
determinz
 Matbematical Mode/ling I +S

Forward path: The path from the input node to the ourput node. That is from xlt) to x2Q) to
4(t) ro xaQ).
Feedback path: A path that starts from a node and ends at the same node, that is from x2(t) to
4(t) and back to x2Q).It is also called a loop.
Self loop: A feedback loop with one branch, that is E
Gain: It is the transfer operator for a branch, that is Efor the self loop.
Path gain: The product of the gains for all branches that formed the path. The path gain
for the path x1(r) to x2Q) to 4Q) to xaQ) ts ABC.
Loop gain: The product of the gains for all branches that formed the loop, that is for the
loop from x2Q) to 4Q) and back to x2Q) is BF.

Figure2.64 A signalflow graph

2.4.4 Mason's Gain Formula

The transfer function of dynamic system can be obtained from the signal flow graph using the
a
expression known as rhe Mason's gain formula. The expression is:

v(*) =I AIn, (2.69)

"(t)
rvhere
y(t = output variable
u(t) = input variable
Pi = the gain for the i'h forward path
A = 1 - (the sum of all loop gains) + (the sum of the producrs of wo non-touching
loops) - ...
Li = A evaluated by dropping the loop gains that touch the im path.

. Two loops or paths are said to be non-touching if they do not share a common node.The
determinant of the signal flow graph or the .har".t.iistic equation is A.
16 | Intoduction to Control Engineering

Us
Example 2.3
Figure 2.65 is a block diagram of a dynamic system. Draw the signal flow graph for this system
and obtain the transfer function using Mason's Gain Formula.

Figure 2.65 Block diagram

Solution
The signal flow graph for this system is shown in Figure 2.66.There are two forward paths with
the following gains: Exerci
Draw tl
h = Gr(D)
P2= G^D).
There are two feedback loops with the loop gains of:
Lt = -Gt(D)Ht@)
L2 = -G2(D)HJD).

Therefore,
A = 1 +G/D) H1(D)+Gz(D) HID).
Since both loops touch the first forward path P1, then A1 is obtained by dropping Ll and
L2 from A. Therefore,
Ar=1.
Similarly, both loops touch the second forward path P2. So,
Az=1'
 Mathematical Modelling 1 +Z

Using Masont Gain Formula, we get;

tu)
"(t)
G,(D)+G,(D)
r+ G,(D) H,(D)+ G,(D) H,(D)

Gz(D)

-Ht@)

Figure 2.66 Signal flow graph for Figure 2.65

vith
Exercise 2.24
Draw the equivalent signal flow graph for the sysrem inFigure 2.67 .

Figure 2.67 Block diagram for Exercise 2.24

48 | Introdrction to Control Engineering

Exercise 2.25
Draw the signal flow graph for the system in Figure 2.58. Exam;
Consid
slt) flow grr

There a

Figure 2.68 Block diagram for Exercise2.25

Exercise 2.26
Using Masont Gain Formula, obtain the transfer functions for the systems shown in Figure 2.69.

Then,
Gr(D) Gr(D)

Since all

Gr(D)

Similarl'
Gt@) Gz@)

Then

Gs@)

Gr(D) Gz@) q(D)

where (f
Figure 2.69 Signal flow graphs for Exercise 2.26
 Mathematical Modelling I ng

Example 2.4
Consider a dynamic system represented by the block diagram shown in Figure 2.7O.The signal
flow graph for this system is shown in Figure 2.71.There are two forward paths. These are:

4 = GID)G2(D)q@)
Pz = GID)G^D)
There are five loops, namely:

Lr = GJD)G2(D)H.(D)

L2 = G2(D)q(DHr(D)

4 = -GJD)G2@)q@)
L4 = G4(D)HiD)

L5 = -GJD)G{D).
Then,

A= 1 - (L1 + L2 + L3 +La + L5).

Since all the loops touch the first forward path, ?1, then:

Ar=1'
Similarly,

Lz=l'
Then

y(t) _ 4A,,+
p,a,,
A
"(t)
G.G.G3+GlG4
l-G.G2H.- G2G3H2+ G.G2G3 - GnHr+ GtG4

where (D)t are dropped from the equarion for simpliciry.

50 | Introduction to Control Engineering

Draw the
.-,-stem in

Exercisr
Draw the
.r-Stem in

Exercisr
Draw the
sr-stem in
Figure 2.70 Block diagram for Example 2.4
Exercisr
H,(D) Draw the
rystem in

Further
Kuo, B. (
\Iason, S,
Figure 2.71 Signal flow graph for Figure 2.7 1 114,
\fason, S

at(t
Exercise 2.27 Ogata, K.
Draw the signal flow graph and using Mason's gain formula, obtain the transfer function for the \-u, H. V
system in Exercise 2.6. Mc(
Voods, R
Exercise 2.28 Sadc
Draw the signal flow graph and using Mason's gain formula, obtain the transfer function for the
system in Exercise 2.7.

Exercise 2.29
Draw the signal flow graph and using Mason's gain formula, obtain the transfer function for the
system in Exercise 2.8.

Exercise 2.30
Draw the signal flow graph and using the Mason's gain formula, obtain the transfer function for
the system in Exercise 2.9.
Exercise 2.31
 MathematicalModelling I St

Draw the signal flow graph and using Mason's gain formula, obtain the transfer function for the
system in Exercise 2.10.

J*\ Exercise 2.32

Draw the signal flow graph and using Masons gain formula, obtain the transfer function for the
system in Exercise 2.1 1.

Exercise 2.33
Draw the signal flow graph and using Mason's gain formula, obtain the transfer function for the
:a-stem in Exercise 2.12.

Exercise 2.34
Draw the signal flow graph and using Mason's gain formula, obtain the transfer function for the
$'stem in Exercise 2.13.

Further Readings
Kuo, B. C. L995. Automatic Control Systems. Englewood Cliffs, NJ: Prentice-Hall.
-\4ason, S. J. 1953. Feedback Theory: Some properties of Signal Flow Graphs. Proc. IRE. 4I(9):
tt44-1155.
-Vason, S. J. 1956. FeedbackTheory: Further properties of Signal Flow Graphs. Proc. IRE.
44(7):920-926.
ogata, K. 2004. s)rtem Dynamics. upper Saddle River, NJ: Pearson Prentice-Hall.
\ru, H. V and R. S. Esfandiari. 1997. Dynamic Systems: Modeling andAnalysis. NewYork:
McCraw-Hill.
Woods, R. L. and K. L. Lawrence.1997. Modeling and Simulation of Dynamic Systems. Upper
Saddle fuver, NJ: Prentice-Hall.
 CHAPTER 3

TIME RESPONSE OF
A CONTROL SYSTEM

3.1 TIME RESPONSE: AN INTRODUCTION

Outcomes:
After completing this section, the reader should be able to:

(1) define time response, and

(2) identify standard input signals.

3.1.1 Concept
It has been stated in Chapter 2 thatthe first step in analysing a dynamic or a conrrol system is ro
derive a mathematical model of the system. The model is needed by a designer ro evaluate the
performance of the system before a prototype model is developed. This wilt drastically save the
cost of development. If the model does not perform as desired, the system can be re-modelled or
new components of different parameters can be introduced.
Once the model has been obtained, the performance of the system has to be evaluated.
Various methods are available. In analysing the performance of a model, conrrol engineers need
to consider what particular standard input signals should be used and compare the response of the
mode,l with the design specification. Some of the standard input signals that can be used are step,
impulse, ramp, and sinusoidal inputs. The rype ofinput signal to be used for perform"rr..
depends on the application of the system and the narure of input signal used by the system."nrlyri,
If a
 54 | Introduction to Control Engineering

i
rypical input signal to the system is gradually changing with time, a ramp input may be a good
test input. Similarly if a system is subjected to a sudden load disturbance but the output signal is Exercise
required to be constant, a step input may be a good test signal.
Similar analysis is also applicable to control system design. Once a control sysrem is designed Consider:
and it satisfies the standard input test signal, the performance of the real sysrem is generally vertical dis
satisfactory. The usage of these standard test signals enables one to compare the performance of the car acc
the designed systems. In short, time response analysis is a study on how the output of a system
changes with time when it is excited with some standard input signals.
An e>

3.1.2 Time Response system as s

the earth s
Let us consider a feedback control system of the oven discussed in Chapter i. fu an example, lett is always cl
say initially the oven temperature is 200'C. Due to some reasons, the operator needs to change and outpu
the oven temperature to 250"C. This sudden need to change rhe oven remperature from 200"C The r
to 250"C represents a step input as shown in Figure 3.1. As soon as the value of the input is by a transft
changed, the controller will produce a signal to the heater as a function of the error berween rhe (oDE). Tr
desired output with the actual output and this will increase the voltage or currenr ro the hearer. can be obt
The increase of voltage or current to the heater will cause the actual temperature in the oven ro engineers,
increase until it reaches a steady-state value as shown in Figure 3.1. This figure shows several
system responses with different overshoots and delay. These differences may be due to different
controller parameters used. Input sig
the desirt

(,
o
250
O

d
u
o"
225
-U overshoots and speeds
F

200
45
Time (second)

Figure 3.'1 Step input and system response Figr

 Time Response of a Con*ol System I SS

lod
Ll is Exercise 3.1

Consider a car moving horizontally. If vertical displacement of the car body is the output and
red
vertical displacement of the wheel is the input, sketch the possible responses of the car bodywhen
,1ly
the car accidentally hits a road divider.
eof
tem

An example of the usage of a ramp function as a test signal can be seen in aircraft tracking
system as shown by a block diagram in Figure 3.2(a). This aircraft tracking system is placed on
the earth surface and designed to track the position of a moving aircraft. Thus, the input signal
lett is always changing with time and it can be written as u(t) = at, where a is a constant. The input
mge and output signals are shown in Figure 3.2(b).
0'c The mathematical model of a dynamic system, as discussed in Chapter 2, can be represented
nt is by a transfer function. The transfer function can be re-written as an ordinary differential equation
Lthe (ODE). Thus, given any input signal represented by any time function, the output of the system
ater. can be obtained by solving ODE. Various methods are available to solve ODE. For control
nto engineers, Laplace transformation method is used for solving linear differential equations because
.eral
rent
Output signal,
the desired position theactual position

Time, second

(b)

Figure 3.2 Aircraft tracking system: (a) block diagram, (b) input and output signals
56 | Introrluction to Control Engineering

standard Laplace transform table can be used to find the Laplace transform, F(s), of a given
function,flt), and the solution can be obtained directly.
In this chapter, we will consider the usage of Laplace transformation method to solve ODE
and the time response of a dynamic system, how a control system is classified, the characteristics
of first and second order systems, the transient response and steady-state specifications for a
feedbackcontrol system, and introducdon to PlD-controller (Proportional + Integral + Derivative). Example
Consider a
3.1.3 Mathematical Model assuming a

Figure 3.3 shows a dynamic system represented by a block diagram whereT(r) is the output and time respor
u(t) is the input to the system. Both of these quantities are function of time and are known as
variables. Time response analysis is the study of the output response of a system, T(r), when the
system is excited with some input signals, u(t). As seen in Chapter 2, the mathematical model
developed is in the form of an ODE. Therefore the time response of the model can be obtained
by solving the ODE.

Solution
The differe
Figure 3.3 Representation of a dynamic system

To obtain the output response of a dynamic system, it is easier if we re-write the transfer
function of the system as an ODE. In general, an ODE is:
Time respc
this case is
4*
dt,
* o,-,4
n-t -, * o,-,ry+...
4-z
h4-
dt,
+ ao!(r) = with given
equation.
t
s

d*-'u(r\ t...tt^,
,',d*uQ) *, h,-,
.d-'u(r) *, .-r-:7#J+"'+ braQ) G'l)
dt* dt*-'
Exercise
with the initial condition y(t)= lr,ff{rr)= !r, ,ffi{rr)= !,-, and. n > m. The
Derive the
damper is a
corresponding D-operator transfer function model is: form of OI
 Time Response of a Control Sltstem t5/

y(t) _ b*D- + b*_rD--t + b_-rD*-2 + ...+ bo

(t.z)
)E "(t) D" + dn ,D"-'+ d,_rD"-' +...+ ao
ics
ta
,.). Example 3.1
Consider a mass-spring system as shown in Figure 3.4. \fi4rat is the ODE model of the sysrem
assuming u(t) as the input andyQ) as the output? \X/ith reference to this system, describe what a
time response analysis is and explain how the time response of the system can be obtained.
rrd
las
the r(t i

del ,(t) t---)

I

ned
F)
oo
777777,
Figure 3.4 A mass-spring displacement system

Solution
The differential equation model representing this system is:

d'tU)
sfer
*-;i'+ kyQ)= ku(r).

Iime response analysis is the study of behaviour of the ourpur response variable, y(t), which in
this case is the mass displacement when some input signal, u(t), is applied at the end of the spring
with given some initial condidons. The output response can be obtained by solving this differentiJ
equation.

). 1)

Exercise 3.2
Derive the mathematical model in the form of ODE for the system shown in Example 3.1 if a
lhe
damper is added to the system in parallel to the spring. Compare this equation with ih. g.r,.r^l
form of ODE given by Equation (3.1).
58 | Introduaion to Control Engineering

3.2 LAPLACE TRANSFORM 3.2.3

Outcomes: The La1

After completing this section, rhe reader should be able to:

(1) define Laplace transform,

(2) derive Laplace transform of a time function, and
(3) solve differential equations using Laplace transform method.

Hence,
3.2.1 Goncept
In the previous section, we have seen the general form of an ODE and how ir represents a
dynamic system. The solution of the ODE is the output response of the dynamic system in time
domain. There are many methods of solving differential equations and the method considered
in this book is the Laplace transformation method.
wheref
Consider a function in time domain,flr), where:

f@= o, tr<0.
The Laplace transform offr) is:
Using tl
Llf @l= J*, e f (t) dt
.
(3.3)

where-s= o+ ja is acomplexnumber.TheLaplacerransform of f(t) canbesimplifiedas

follows: Therefo

rlf @l= r(,) (t.+7

Therefore, if f(r) is given, F(s) can be obtained and similarly if f(s) is given,flr) can be obtained
For zero
using the inverse of Laplace transform:

f (') = 11 [F (r)]. (3.5)

3.2.2 Property of Laplace Transforrn

Lett consider two functions in time domainf (r) andf2G),and scalars aand
3.2.4 r

B. Laplacetransform
is a linear operator that satisfies:
The Lap
Ll"f,(t)+ Ff,(r)] =
"cLf,(r)]* frrlf,@] G.e1
 Time Response of a Control System I Sg

3.2.3 Laplace Transform of Derivatives

The Laplace transform of a derivati ur, t ('), can be derived using Equation (3.3), thus:
dr

'1ry)=fiu"4P "
=lr"f (r)]. *rj*o{',f (t)dt.
Hence,

1{21= sLlf ()l-lror

ld, l (3.7)
: sF(s)-/(0)
*.herefl0) is the initial condition. If the initial condition is zero,flO) = 0, rhus:

d{Jf)=
Ld, J "
sF(s) (3.8)

Using the same approach, the Laplace transform of a second derivative is obtained as:

(3.3)
,l'' {9]I =,,r,(,) - y 1oy{-(o)-
-L dr' d,
(3.e)
ied as
Iherefore, in general,

(3.4)
Ll
d'
f \,)1 = r df-g- d'' f (0)
0) - ;-,
- s., Jr(\-' -.. . -
I dr' I ^s" dt
) . (3. t o;
dt"
ained
For zero initial condidons,

{3.5)
,lryJ=,',1,; (3. t t;

3.2.4 Laplace Transform of an lntegral

form

The Laplace transform of an integr il,,

I'rf O)dt , canbe derived by assuming:
(3.5)
60 | Introduction to Control Engineering

with magnir
7Q)=Lrtln. (3.12) 3.2, it canb

By manipulating the equation, it is found that:

tP
dt
= f (t), ,r (o) = o. (3.t3)

Using the definition of Laplace transform:

,1ry)=F(,)-/(o) (3.r+S

Therefore

LU:r o)a,)=lr 61 (3.15)

3.2.6 LaF

A ramp fun<
Example 3.2
Derive the Laplace transform of an exponential functron, f (t)= r"'
The graphic,
Solution is:

From Equation (3.3):

Llr"'f = J: t "eo'dt

l* ,,o ,\,dt
-Jo" -t41-l- -;'
= 1

= "'-1"-,_1.

3.2.5 Laplace Transform of a Unit Step Function

A unit step function is defined as:

H(t)={i::: (3.15)

and its graphical representation is shown in Figure 3.5. One way of obtaining the Laplace transform
of a unit step function is using the method shown in Example 3.2. The exponential function
 Time Response of a Control Srystem I Ot

with magnitude a = 0 is equal to H(t).Therefore by replacing a = 0 into the solution in Example

(3.r21 it
3-2, can be shown as:

LIH (4)=! (3.17)

(3.13)

(3.t4)

Figure 3.5 A unit step function

(3.t5)
3.2.6 Laplace Transform of a Ramp Function
A ramp function is defined as:

f (r): r, t> o. (3. t a;

The graphical representation ofa ramp function is as shown in Figure 3.5. Thelaplace transform
is:

clf Ol= J*or-"

rdr

=ly:)- -l-La,
Jo (3.1e)
-,L -s_1.

= [g-l- = t '
L-r'l. 12

(3.16)

rsform
nction
Figure 3.6 A ramp function
62 | Introduction to Control Engineering

3.2.7 Laptace Transform of an lmpulse Function

An impulse function is defined as:

f (t)= 6 (t -a) (3.201

where a is the instant when the impulse occurs that satisfies:

6 (t -a)=9, t * a

where Re( '

Jj -, e -a)* =t )

for both €r > 0 and e2 > 0. 3.2.9 Solv

The graphical representation of an impulse function is shown in Figure 3.7 . The Laplace
Solving differ
transformation of an impulse function for t = 0 can be obtained by replacing a = 0 and using
table. The tab
Equation (3.3) to obtain:

c[6 (/)] = 1 (3.2t1

t=d+tt t=d l=d-€t

Figure 3.7 An impulse function

3.2.8 Laplace Transform of a Sinusoidal Function

A sinusoidal function is given as follows:

f (r): cos Bt (3.221

where B is the frequency of the oscillation. From Equation (3.3), it can be shown that:
 Tirue Response of a Control System I 0g

.C[cos Btl= e-" ReleiP'fdt

Il
:BrJ* ,{io t'5,
(3.20)

=*.[ r l
l'-jB)
t
=;+B' (3.23)

n'here R.( ' ) is the real part of a complex number.

3.2.9 Solving Differential Equation using Laplace Transform

-aplace Solving differential equation using Laplace transform is easy with the aids of Laplace transform
I using
table. The table is shown in Thble 3.1. The procedure starts
with obtaining the Laplace rransform

(3.2r)
Thble 3.1 Laplace transform table

Time functions Laplace transform

Unit impulse, 6(r) 1

Unit step, HQ) = l, t) I 1

J

1
Unit ramp, /
2
.f

Polynomial, At Ant
.t "*1

1
r
Lxponent, e-at
s+a

Sine, sin atr

a
s'+a'
il

Cosine, cos Ct)t .;, J

s'+a'
r ''' sln r0l
r.-dt. 0)
O.22) L/amPecl slne,
, ,2 )
\s+d) +o'
s*a
Damped cosine, lt cos at .-------i .
ls+a)-+a'
64 | Introduction to Control Engineering

of derivatives in the ODE by taking into consideration all the initial conditions followed by the Therefore,
Laplace transform of the input function. This will lead to an ordinary algebraic equarion in
Laplace domain. The algebraic equation is then decomposed into terms that are available in the
Laplace transform table. The solution of the ODE in time domain is obtained by taking the
inverse Laplace transform by comparing the corresponding functions in the table. Rearrangin

Example 3.3 In ord

Solve: transform t
into terms:
4G)
dr
+ t(r)= '(')
where u(r) =H(t) is a unit step input and the initial condition /(0) = 0. This equation is the
mathematical representation of the same system represented by a block diagram as shown in
Figure 3.8. To use the I

The values
methods. Tl
Figure 3.8 A block diagram representation
Let,

Solution
Firstly, the Laplace transforms of each term in the ODE are obtained as follow:
thus,

tlry)=,r (,) - / (o) =,r (,)

rly(')l= r(,)
For this exan
cla 1t11=L.

Thus,
and,

)'@)= Llu(t))
'Iry.
,lryl+ rly(t)f= Llu(t))
 Time Response of a Control System | 0S

the Therefore,
nin
. the sr (s)+ r(s; = 1.
.t
;the
Rearranging this equation will give:

)z(s)=.+
's(s+1)
In order to ohtainT(r), the inverse Laplace transform of fG) is obtained by using the Laplace
transform table (Thble 3.1). This will require us to re-write the right hand side of ih. equation
into terms as a function of s that are available in the Laplace rransform table. That is,

is the
t(t)= c'ly(,)l = r'[4,ir]
wn in
To use the Laplace rransform table, re-write IZ(s) as:

Y(s)=ilb=+.*
The values of constants Al and 42 can be calculated using partial fraction method or other
methods. The partial fraction method can be briefly described as follows:
Ler,

)'(s) =
l/ (r) :s4
D(r) o (, -,,)' (3.24)

rhus,

e,:N(')(r-,'
,., _
D (,
,, _ r,)1,=,, .
o.25)
For this example,

/',_
=
1 I
=
,11 11)'1,=n ''
and,

rl
,4, =,G = -,.
+1)(,*,)1,__,
66 1 Introduction to Control Engineering

Therefore,
Exercise
(t) =
J\/
y . !l = r- e-,,
,,1!- (s+t)] r > o. Using Exar
[s unit step fi.

Exercise
Example 3.4 A different

sow,
ff . 4ry + 3 y Q) = zu G\ if the given initial conditions areT(0) - o,
ff @) = o,
and u = H(t) is a unit step input.
If a unit ste
Solution
obtain the c
For a zero initial condition, the Laplace transform of the above equation is:

(s' + 4s+ 3))z(r) =zU (s)

with
Example I

U (s)= 1
J
Solu.
d'1(
\7e obtain,
dt'
r(i)=
,(r'-r as+3) Solution
2 The Laplace
s(s+1)(s+3)'
Using partial fraction,

I'(s)= ?,-L*- L ..
3 s+1 3(s+3)
From the Laplace transform table, T(r) is:

!(t)=l-n' *1r''' .
and,
3)
 Time Response of a Conrrol System I Al

Exercise 3.3
Using Example 3.1, if k = 2 and m = l, obtain the time response of this sysrem if the input is a
unit step function. Assume all initial conditions are zero.

Exercise 3.4
A differential equation describing a dynamic system is:

d'tG) . . dt,(r\
=0, -;*t;+zyQ)=,(r)
If dy(o) _r.
a unit step input is used ro excire the system with initial conditions 7(0) = o and
dr
obtain the output response of this sysrem.

Example 3.5

solu,
d't(') *s4'\ :2 4@) -1.
if the initialconditions arey(0) and
dr, d;+6yQ)=l2e' dt

Solution
The Laplace transforms for the terms in this equation are

,lryl= s,Y (,) -,y (q -

ry = szY (s) - 2s -l
,\ry)= sY (s) - y(o) =,). (,) - z

rLy@l=y(,)
arid,

LL4=*
68 | Introduction to Control Engineering

Therefore the equation becomes:

(2) G(r)
[,'r1,; - 2s -r] + ;[,r(, )- z)* 6Y (s) :
*
Rearranging this equation will give:
(3) c(,)
2s2 +9s+1 (4) G(r),
^r(-')=
\"
' (s-1)(s +2)(s+3)'
Using partial fraction method, we obtain: Exercise I
Obtain the
)z(s)=*.** equations. (
excited with
From the Laplace transform table, the solution of this equation is:
(1) 4I)
Y(t)= e' +3e " -2e-3' '
dt

Q\ II9
dr'

Exercise 3.5
Obtain the solutions of the following differential equations with the given initial conditions.
3.2.10 Tn
u) +P.6++t3y(t)=5, y(o):r. ffrct:o Tiansfer func
Laplace trans
(z) transform of
+y.3ry+4y(t)=7s; -y(o)=1, fft )=,
(3)
+P+2ff+at(,)=6sin@t; y(o)=4, !@=,
This is the for
(4) y(o)=r
ry.zy(t)=1,
Exercise 3.6
3.3 CLA
Obtain the output response of the following dynamic systems excited with a unit step input:
Outcomes:
(1) G(,)=
#, /(o):1 After completi

(1) define or,

 Time Response of a Control S)rstem | 0g

(2) G(r)= t@)=r,!-p1=o

-=,
J +)J+4 dt'

B) G(s)=6j.ffi , t@)=0.ff{D=o
(4) G(,):
#*r, /(o)= 0,ff{0=t.

Exercise 3.7
Obtain the transfer function, G(s), for the system represented using the following differential
equations. Obtain the output response of the system with the given initial conJitions when
excited with a unit step input:

(1) y(o)=r
ry+6y(t)=u(t);
(2) u(t); y(o)=
ry.?ry.rcy(t)= 1,
ff{o=n

3.2.10 Transfer Function

Tlansfer function is defined as the ratio of the Laplace transform of the ourput variable to the
Laplace transform of the input variable assuming zero initial conditions. Thus taking Laplace
transform of Equation (3.1) assuming zero initial condition results in:

r(r)_b*s- +b*_rs*-t +b__rs--, +...+bo

U(r) sn +dn_rs'-l +d,_rs"-2 +...+ao G.ze)

This is the form of transfer function used throughout the book.

3.3 CLASSIFICATION OF DYNAMIC SYSTEMS

Outcomes:
After completing this section, the reader should be able to:

(1) define order, rank, and class of dynamic systems.

70 | Innoduction to Control Engineering

3.3.1 SystemGlassification (2) G(r)

A dynamic sysrem is classified according to certain definitions. From this classification, control
engineers will be able to predict the behaviour of the system. Consider a feedback control system
as shown in Figure 3.9, where GG) i. the process or plant and I is the controller gain. (3) G(,)

Solution
(1) order.
(2) order =

(3) order=
Figure 3.9 A block diagram of a feedback control system

The process G(s) can generally be written as:

k' (s + a,)(s + dr)"'(s' + bs* r)"' Exercise I

\{/hat are th
6(s) = (3.27)

(1) G(r)=
where b' ts aconstant. This transfer function can also be written as:

2nr'r (2) G(,) =

G(s)= J*t , n)m. (3.2s;

/\an'r Exercise i
h=0
A control sys
The classification of a process or plant, G(s), is according to the following definitions: function of r

(1) Order, z, is defined as the highest power of s at the denominator.

(2) Class or Vpe, l, is defined as the power of the factor of s at the denominator.
(3) Rank, 72 - 771, is defined as the difference between the highest power of s at the denominator
and the highest power of s at the numerator.

Example 3.6
t{/hat are the order, rank and class of the following systems?
-,4
S.T L
(1) G(s) =
so +3s3 + 3s2 + s
 Time Response of a Control System I Zt

(2) G(s) =
rtrol s3 (s+2)(s+1)
;tem
s2 +s+1
(3) G(r)=
(s+2)(s'z+s+4)

Solution
(1) order = n=4, rank=n-m=4-l= 3,class=l=l
(2) order = n=5, rank=n-m=5-0= 5,class=l=3
(3) order = n=3, rank= ru-m=3-2= l,class=l=0

Exercise 3.8
What are the order, rank, and class of the following sysrems?
.3.27)
(1) G(s)= 4 o1 t)
-f +JJ +J-t

(2) s*1
G(s)=
,' (r' + 2s +1)
o.za1
Exercise 3.9
A control system is represented by a block diagram in Figure 3.10. Obtain the open-loop transfer
ts: function of this sysrem and find the order, rank, and class of the open-loop sysrem.

inator L(s+ 40)

s(s+ 10)

Figure 3.10 Block diagram of a dynamic system for Exercise 3.g

72 | Introduction to Control Engineering

3,4STEPRESPONSEOFAFIRSTORDERSYSTEM
Outcomes: where Kis th
two paramet,
After completing this section, the reader should be able to:

(1) define a first order transfer function, 3.4.2 Stel

and initial conditions,
(2) derive the output response of a first order system given any input signals
The transfer
and
of time constant and gain of the system on the time response of the
as:
(3) describe the effect
system.

3.4.1 First Order SYstem Here we wot

1 1' The block
An example of a first order system is a spring-damper ty*t- shown in Figure 3'
that the transfer function is:
diagram Lf ,h. system is shown in Figuie 1.12. ltcan be shown
x,(r)= 1
(3.2e) \7ith
x,(') 1+(c t k)s is:
zero in

function of
This is a first order system as defined in Section 3.3.1.Ingeneral, the transfer
a

frrst order system is:

since the La1

a
u(r)=1. s
J

l--"
Figure 3.11 A spring-damper system
Using partial

F,(t) = Fo(l

Iherefore th

Figure 3.12 Block diagram of the spring-damper system Figure 3.13 s

 Time Response of a Control Slsrem I lZ

r(,) K
,A=,*r=' (3'lo)
where Kis the gain of the system and lis the time consranr. These paramerers, Kand T, arcthe
two parameters that characterise the response of a first order system.

3.4.2 Step Response of a First Order System

litions,
The transfer function of a first order system can be written in the form of a differential equation
of the AS:

, y(r): r<,(t). (s.tr)

ry+
Here we would like to study the response of the system when excited with a unit step input:
e block
ln is: [t.
,lt):lo, r>o
(3.32)
r<0.
(3.2e)
!7ith zero initial conditionT(0) = 0, the Laplace transform of the ordinary differential equation
is:
Lon of a

TsY (s)+Y (s): L (3.33)

.t

since the Laplace transform of a unit step function obtained from Laplace transform table is
1

U (s)=:. Solring for )z(s), we found that:

J

\/ s(7s+1) (3.34)

Using partial fraction,

Y lsl=
KK
(3.35)
s s*llT
Therefore the output response of the system is:

"f.l
t(t)= n('-"11 I
'l

o.36)

Figure 3.13 shows the step response of the sysrem for K = 1.

74 | Introduction to Control Engineering

The gai
u(t)
n'stems are s
value of the
/\ r_7( /
0.8
0.632
I
0.6
I
I
I
0.4
I
I v(t) 0.6:
I
0.2 I
I

- T

Iime (second)

Figure 3.13 Step response of a first order system

The effect of tirne constant, 4 on the time response can be analysed as follows. Let K = 1.
Then the output response is:

y(t)=r- r-(+) (3.37)

Let say we have three first order systems with the time constanrs of T, = l, T2 = 2, and Tt = 3
seconds. tW4ren time, /, is equal to the time constant, the outpur is: I
1.5
y(r):r-,(+Y =0.632. (3.38)
!(t)
The output responses for these three systems are shown in Figure 3.14. Nlthe outputs will
reach the value of 0.532 when the time, ,, is equal to the time constants. Thus, the larger the
value of the time constant, the slower is the outpur response.
The effect of gain, ( of the system can be studied by considering three first order sysrems
0.5
with the same value of time constant, 7i say equal to one but having three different values of
gains, i.e., Kt= 0.5, Kz= l, and K3= 2.The speed of response of each of this sysrem is the same
as shown earlier, since the time constants of these systems are the same. The output response is
given by:

t(r)= x(r-e'). (3.3e)

 Time Response of a Control Slstem I Z S

The gain, ( has the effect of amplification or atrenuarion

and the ourpur responses of these
systems are shown in Figure 3.15. The bigger the value
of the gain, the larger is tih. rt."dy-rt"t.
value of the output response.

t(l 0.632 / -.-,".''

I
t,:
li
T1 72734 r4
Time (second)

Figure 3.14 The effect of time constant on the speed of output response

37)
K3

=3

t.5
38) /
v@
n ill / K
the
i

ir'
:InS K1
0.5
sof
[ne
eis
f;:-
10
Time (second)
39)
Figure 3.15 The effect of gain on the output response
76 | Introduction to Control Engineering

3.5
Exercise 3.10
A dynamic system is known to be a first order process. If the time constant is 0.5 and the gain is Outcom
10, obtain the transfer function of this system. If the input to the system is a unit step function
and the initial condition is zero, obtain the time response of this system. Sketch the output -{fter com
response.
1) defir
,2) desc
Exercise 3.11
the c
The relation bewveen input voltage, u(t), with output torque, r(t), for a DC motor can be
r3) sket<
represented by a first order transfer function. A step input test of 6 V with zero initial condition
of th
was performed. A steady state output torque of 20 N-cm was obtained. It is observed that the
t4) deri
time taken to reach output torque of 12.5 N-cm is 0.4 sec. Derive the transfer function of the
|)) deri'
DC motor.
syst€

Exercise 3.12
A first order system is represented by the following transfer function: 3.5.1 S

l'(r) 1 An exami
U(r) s-t2 block diq

If the input signal, zz(r), consists of two impulses as shown in Figure 3.15, obtain and sketch the
time response, y(t), of the system for zero initial conditions Qtlease note tltat the initial conditions
dre no longer zero for the second impulse input signal).

Figure 3.16 lnput signal

 Time Res?onse of a Control Srysrem I lZ

3.5 STEP RESPONSE OF A SECOND ORDER SYSTEM

rin is Outcomes:
ction
Jtput After completing this section, the reader should be able to:

(1) define the transfer function of a second order system,

(2) describe the effect of damping ratio, (, natural frequency, @n, andgain of the system, .( on
rn be the output response,
lition (3) sketch the output response if the values of damping ratio, (, natural frequenry, a),, andgain
rt the of the system, K, are given, and vice versa,
rf the (4) derive the transfer function if a step response data is given, and
(5) derive the transfer function if damping ratio, $, natural frequency, a),, and gain of the
system, K, are given.

3.5.1 Second Order System

An example of a second order system is a mass-spring-damper system shown in Figure 3.l7.The
block diagram is shown in Figure 3.18. Its transfer function can be shown as follows:

:h the
litions

Figure 3.17 A mass-spring-damper system

Figure 3.18 Block diagram of a mass-spring-damper system

 78 | Introduction to Control Engineering

I
x"(') klm 3.5.2 S
X,(t)-
i

(3.40)
s'+(clm)s+klm' In general
This transfer function indicates that the system is a second order system where the highest zero initia
power of s at the denominator is two. The general transfer function of a second order system is:

r(,) Kai
= (3.4t1
U (t) s' +2(a,s+a) where

where, at, is the undamped natural frequency, ( is the damping ratio, and Kis the gain of the
system. These are the three parameters that will characterise the response of the system. The
is the darr
system can exhibit three types of responses that depend on the value of the damping ratio. These
are undamped response if e = 0, overdamped response if 4> t, and underdamped response if
0. 6.1.
The type of response can also be identified based on the values of poles of the system. The
poles are the roots of the characteristic equation. The characteristic equation of a second order Ther
system is: the poles :

+2(ro,s + d, =0. (3.421

s2
(1) un
The poles are given by:
The gener
-2(a, t (21r.)'
(3.43)
2
If € = 0, the poles are:
The chara
s =Xon.j , O.44)
i.e. rwo poles with only imaginary parts. Thus if any pair of the poles are only imaginary pams,
and, thus
the system will be undamped.
Conr
If 1> l, the term under the square root is positive. Thus we will have two real number
poles. Thus if all poles of the system are real numbers, then the output response is overdamped.
If 0 < 6 <1, the term under the square root is negative. Square root of a negative number is
a complex number leading to a pair of complex conjugate poles. Thus if the poles of a system is
complex conjugate, the output response is underdamped. If this svst

Exercise 3.13 Using par

A second order dynamic system is found to have the gain of two, damping ratio of 0.5, and
undamped natural frequency of 2 radls. Derive the transfer function of the system.
 Titne Response of a Control System I Zg

3.5.2 Step Response of a Second Order System

i.40)
In general, the output response of a second order system when excited with a unit step input and
zero initial conditions can be shown as follows:
Shest
m is: 1.
y (t\ :1 - (' 6'' ,in (alrr + a)
l, c, (3.45)
v'-9
\.41) where

rf the a, = ,,t[l- ( (3.46)

. The
is the damped natural frequency and the phase shift is:
fhese
if
A:tan .--
rnse
,,h-t,
(3.47)
The
r. e
order There are three possible rypes of responses that depend on the value of the damping ratio or
the poles as discussed earlier.

"3.42)
(1) Undamped response, 6 = 0
The general transfer function becomes:

(3.43) r (r)
u (') (3.48)

The characteristic equation is:

{3.44)
s'+c,fi=9. (3.4e)
Parts, and, thus the poles are,r = t@nj, rwo poles with imaginary parts only.
Consider as an example, a system represented as follows:
rmber
mped. v(,) _ 4
rber is (3.50)
, (t)- t'+4
tem is
If this system is excited with a unit step input and the inidal condidons are zero, rhen:

r(i)= +-.!=A *u'u!c, (3.5r)

s-*4,r .r s']-4
Using partial fraction method or otherwise yields:
i, and
),(s)=1-,' . (3.52)
s s" *4
a
80 | Ir*oduction to Contol Engineering

The time response can be obtained by taking the inverse Laplace transform by referring to The pair o{
the Laplace transform table, thus: that can be
are:
I Q)=l- cos2t' (3.53)

Figure 3.19 shows that the output is continuously oscillating with the frequency equals to
the natural frequency, o)n, and in this case is 2 radls.

(.)
Thking the
right hand
b t.5
o
o.
q9l
-E Thusa= 1

2 0.5

2 4 6 8 10
Thking invr
Time (second)

Figure 3.19 An undamped output response

In general, ,

in Figure 3
(21 Underdamped response, 0 < e < 1

Consider a dynamic system with the following transfer function:

r(,)
- 2
U (t) s2 +2s +2'
(3.54)

Comparing with the general second order transfer function (i.e., Equation (3.41)), we obtained

d:Z and 2(a,:2. Thus the damping ratio is0.707 which ir 0 . 6 < 1. The values of the
poles are -l , j, which is a pair of complex conjugate poles.
If the system is excited with a unit step input and the initial conditions are zero, then:

2IABs+L
Y(sl=--;-'-=-* . . (3.55)
s' l2s -f 2 .r J s' *2s 12
Using partial fraction method or otherwise, we have:

r(s)= L- (3.55)
s s'+2s*2
"*2
 Time Response of a ControL Slstem I gt

to The pair of the corresponding time and s domain in the Laplace transform table (i.e., Thble 3.1)
that can be used to obtain the time response for the second term on the right side of the equation
are:
i3)
0)
Damped sine, e
o'sin ;-- ., .
o)/
ito ls+a)- +@'
s+a
o'
Damped cosine, e cos o)t
(s+a)'+a'

Thking the denominator of the s function in the table and compare with the second rerm on the
right hand side of Equation (3.56) to be solved, we have:

(s+ a)2 +a)2 = s2 +2as+ a2 +a)2 = s2 +2s+2. (3.57)

Thus a = 1 and @= l. Therefore, Equation (3.55) becomes:

s*l 1
-)r(.")=l-
\- /
s (s+t)'+t- (s+1)'+t' (3.58)

Taking inverse Laplace transformation yields:

I (t) = 1 - e-' cos r - e-' sin t. (3.5e)

In general, the unit step response ofa general second order sysrem (i.e., Equati on (3.45)) is shown
in Figure 3.20.

1.6

7T-
1.4

'; 1)
.54)

ned
-o
O
Y]
a
Y L: 2o/o

e
U
tthe 0.8
O

-d 0.6
I

zd
0.4

5s) 0.2

0
/ tt t2 t,
-1
t, I
Time (second)
56)
Figure 3.20 An underdamped response of a second order system
82 | Introduction to Con*ol Engineering

The following observations from the output response will enable conffol engineers ro obrain First peak oc
the values of the gain of the system, damping ratio, and the natural frequency and vice-versa.
These observations are:

G) The first peak occurs at time, /1, given by: The second

@,,tt=P.rd. (:.e01
.\lr- 4'
(b) The second peak occurs at time, t2, givenby: Settling tim,

(3.6t)

(.) Settling time, /r, is the time the output response is considered to reach its steady-state value Maximum ,

and if it is assumed 2o/o eruor of the steady-srare value, then:

4
(3.62)
q(0,
(d) Ma,ximum overshoot, M, rs:
(3) Over
Is^
l--i I (3.63) Consider a
M =e \vr-s /
and for a unit step input, the percentage overshoot is given by 100 x Mu
These expressions can be obtained by differentiatingy(t) ofEquation (3.45) and equating it
to zero. Rise time, /,, is the time taken by the ourpur ro rise from 10 to 9oo/o,5 to 95o/o, or 0 to If the chara
l00o/o of the steady-state value. For an underdamped system, normally the rise time is measured
of a second
from 0 to 100%o, while for overdamped sysrem is from 10 to 90%.
The p
than one a
overdampe
Example 3.7 The u
A second order system has an undamped natural frequency, c0,, of 12 rudls and the damping
ratio, (, of 0.2. Calculate the times when the first and second peaks occur, settling tim.^r.rJ
maximum overshoor when excited with a unit step input.
The consta

Solution
Damped natural frequenry:
Taking the

@d = anfj =D$-g2'= 11.8 radls.

 Time Response of a Control Sytem I gg

)tatn First peak occurs at time, ,1:

'ersa.
t nfi l?-
-: -n
o)nJr -g' ,rJ, -0.2)
The second peak occurs at time, t2:

r -- 3n 3n
(3.60)
' *,Jr-E' tzJt-l.2)
-na.

Settling time, /,:

4
(3.61)
'
t- =1.67 s.
(ro, 0.2x12
e value Maximum overshoot:
O.2T
-r-
tt u2
MP-- ( =0.527.
3.52)

(3) Overdamped response, 6 > 1

(3.63) Consider a dynamic system given by the transfer function:

r(,)
- 2
U (t) s2 +3s +2'
(3.64)
it
rring
or0to If the characteristic equation of this system is compared with the general characteristic equation
asured ]
ofa second order system, then we have d = Zand, 2(o1,,=3.Thus I- = -2x.,12- rt.
The poles are .t = -1 and -2. Thus we find that the damping ratio of this system is greater
than one and the corresponding poles are two real numbers. Thus this system will exhibit
overdamped response.
The unit step response of this system with zero initial conditions can be obtained as follows:
mprng
Y(s\= ^ ' .!=A* B * C (3.65)
te and s'+3s+2 s J J+l s+2
The constan ts A, B, and C can be calculated using partial fraction method or orherwise, thus:

l'(s)= L- ? .++. (3.56)

s s*\ s*2
Thking the inverse Laplace transform using the Laplace rransform table yields:

lG)=l-2e-'+e-2' (3.57)
U I Introduction to Control Engineering

The output response is shown in Figure 3.2i. There is no overshoot and the response is Exercise
slower than an underdamped system. The larger the value of (, the slower is the output response. The block,
and param,

(1) Overs
(2) Settlir

1.5 -- -'t

I
I

10
Time (second)

Figure 3.21 An overdamped response

Exercise
Exercise 3.14 The open-l
A feedback control system is shown in Figure 3.22.If the controller gain, K, is 100, obtain the
damping ratio and natural frequency of the system. Calculate the rise time, the time first peak
occurs, maximum overshoot, and settling time of the output response when excited with a unit
step input with zero initial conditons and sketch the output response. The contrc
Percentage

(1) Can t.
(2) Deter
values

Figure 3.22 Block diagram of the system in Exercise 3.14

3.6 EF
Exercise 3.15 Outcome
The transfer function of a dynamic system is:
After comp
r (r) 18
=_
u(r) s2 +3s +9 (1) descril
Obtain the damping ratio and natural frequency of the system. Calculate the rise time, the time (2) derive
first peak occurs, maximum overshoot, and settling time of the output response when excited (3) use th
with a unit step input with zero initial conditons and sketch the output response. (4) define
 Time Response of a Contol Sqstem I gS

)nse ls Exercise 3.16

Ponse.
The block diagram of a feedback control system is shown in Figure 3.23. Obtain the values of K
and parameter, p, that satisSz the output srep response specification:

(1) Overshoot less than 57o.

(2) Settling time less than 4 sec.

Figure 3.23 The block diagram for the system in Exercise 3.'16

Exercise 3.17
The open-loop transfer function of a uniry feedback control system is:
ain the
st peak KG(s\- K
ra unit s(s+2)
The controller gain, ( is to be designed such that the first peak occurs ar rime,
b = 1 s and
percentage overshoot is 57o when excited with a step input.

(1) Can these two specificarions be fulfilled simultaneously?

(2) Determine thevalue of Kthatwould compromise the given specification. tMhat are the
values of tuand maximum overshoot?

3.6 EFFECTIVENESS OF A FEEDBACK SYSTEM

Outcomes:
Afler completing this section, the reader should be able to:

(1) describe rhe difference berween regulator sysrem and tracking system,
e time (2) derive the transfer function relating error and input signal,
xcited (3) use the final value rheorem, and
(4) define and calculate position error, velocity error, and acceleration error constants.
86 | Introduction to Control Engineering

3.6.1 Feedback Gontrol System (21 Tra

There are two types of feedback systems that are based on their applications: regulator and tracking A tracking
systems. varying in1
system anc
(1) Regulator system
A regulator system is a feedback control system where the set point (input signal) is set constant.
The controller is designed to ensure that the output variable equals to the set point value although
the system is subjected to load variations or disturbances. This system is common in process
control system. As an example, the water level in a boiler is kept constant although the steam
generated by the boiler increases or decreases. A schematic diagram of a boiler is shown in Figure
3.24 and the block diagram of a regulator system in general is shown inFigure 3.25.

N7ater in
It
tp.

Figure 3.24 A boiler regulating system

For b<
the transier
Load or
response sF
disturbance
resPonses a
Constant
Output For a highe
reference
order syster
input
to the imag
approximar
The sr

Figure 3.25 Block diagram of a regulator system time tends

 Time Response of a Con*ol System I gZ

(21 Tracking system

acking A tracking system requires a controller that will ensure the output of the system to follow a
varying input signal. An example is a robot arm position control. The schematic diagram of the
system and its general block diagram are shown in Figures 3.25 and 3.27 respectively.

nstant. Output
hough potentiometer

Process
: steam
Figure Amplifier

Input Robot arm

potentiometer

Figure 3.26 A robot arm control system

9lt + e"@
Varying
input signal

Figure 3.27 Block diagram of a tracking system

For both systems, the controller is designed such that the ourput response should fuifil both
the transient response specification as well as the steady-state response ipecification. tansient
response specification is normally defined in terms of overshoot and speed of response. These
responses are determined by the values of damping ratio and the undamped natural frequency.
For a higher order system, the response may be analysed by approximating the system as a second
order system by considering only the dominant poles. Dominant poles are the poles that are close
to the imaginary axis. Once the two dominant poles have been identified, the responses can be
approximated as a second order system and can be analysed as discussed earlier.
The steady-state response specification requires the steady-state error to be zero, that is, as
time tends to infiniry the output response tends to the reference value.
BB I Introduction to Control Engineering

3.6.2 Measure of Effectiveness effectiven

using van
The effectiveness of a feedback system is measured based on the error signal as time tends to
infiniry. Let's consider a feedback system shown in Figure 3.28.
3.6.3 S

\We have r

If this sys

Figure 3.28 A block diagram of a feedback system if we are

theorem,

(1) Error
Since in this analysis we are interested in the error when the system is excited with some given test
signals, it will be easier ifwe relate the error directly to the input signal. The transfer function can wherc k,
be obtained as follows. From the block diagram: a system I

E(i)= u (')-11(i)r(r) (3.68)

)'(s)= t<c(s)a(s). (3.6e)

Since we want to relate E(s) directly to [.I(s), eliminating ($ from Equations (3.68) and (3.69)
and re-arranging the equation yields:

E(s): u('). (3.70)

t+ KG(s)H(s)
In this analysis, we are interested in the steady-state value, that is, when the time, t, tends to
infiniry thus the complete solution of the error signal is not necessary. The final value of any time
function can be obtained by applying the final value theorem given by:

\*f @= I'q'[rr(r)]. (3.7t)

Consider
(21 Test signals
the error r
Test signals are used to test the effectiveness of feedback systems. The common test signals are
unit step input, unit ramp input, and unit parabolic input. Step input is normally used to test the
 Time Response of a Connol System I gg

effectiveness of a regulator system while for tracking system, this test is insufficient. Further tests
using varying inputs are required and these include ramp and parabolic inputs.
nds to

3.6.3 Step lnput Test

\7e have seen the transfer function relating error ro the input signal is given by Equation (3.70).

If this system is excited with a unit step input, U (s)=1, ,nd zero initial conditions, then:
J

E(s)=-o."fu"O
i (372)
If we are interested in the steady-state value, this value can be obtained using the final value
theorem, thus:

r'' =lim., I .]= 1

'-o 1+ KG (r)H(j) s 1+ k, (3.73)

Yen test
ion can where ko =li1;KG (i)11(i) is known as position error constant. Figure 3.29 shows the error of
a system when excited with a unit step input.
(3.68)

rI nput signal
(3.6e)
I
r (3.6e) 1+h p

0.5
(3.70)

:endsto
nytime 0 i0 20 30 40
Time (second)

(3.7r) Figure 3.29 System response and steady-state error

Consider a simple system shown in Figure 3.30. If this sysrem is excited with a unit step input,
the error response is given by:
nals are
test the
90 | Introduction to Control Engineering

Thus,

Iherefore,
Figure 3.30 A feedback control system

E(r): In genr
r+ KG(s)H(s) s conclude th
11s+101 zefo.
(3.74)
;: i+10+ 1( ;
-K-s+ 10 Example I

The steady-state error can be obtained using the final value rheorem, rhus: Calculate pc

r'' =lim, s*lo f= lo

(3.75)
to a unit ste
r >o s+10+K s 10+K

Thus this system has a steady-state errot .f when excited with a unit step input signal.
#
This error can be decreased by increasing the value of K The error cannot be eliminated completely Solution
unless Kis infinity but this may cause some other practical problems. Position errr

Example 3.8 Thus, the st

Calculate the steady-state error of a Class 1 system.

Solution
The open-loop transfer function of a Class 1 system is:

KGH (s\=kN (s) Exercise I

,D(r) Calculate th
where tV(s) and D(s) are the numerator and denominator of the open-loop transfer function when it is e>
respectively. \We have seen position error constant is given by:

hu =\gKG (s)H (s).

 Time Response of a Control System I gt

Thus,

ro=*Pl
,a(,)
:-
L_.
Therefore,

e":t+l:Q.
k?

In general, Class 1 system has zero steady-state error when excited with a step input. \7e can
conclude that at least one integrator must exist in the sysrem for the steady-state error to be
zefo.
o.74)
Example 3.9
Calculate position error constant and steady-state error of a uniry feedback system when subjected
to a unit step input. The openJoop transfer function is:
{3.75)
\/ \/
KG (s) H (s) =
-+-
(s+20)
signal.
rpletely Solution
Position error constant, ku, is:

ho=tj1;KG(s)11(s)=-9[ =*=o.t
s* 201, ,n 20
Thus, the steady-state error is:

e" =
11
=o'567
r+ 1= 1*oj

Exercise 3.18
Calculate the position error constant and steadr5s2ls error of a unity feedback
conrol sysrem
when it is excited with a unit step input, given the open-loop tr"nrf.i function:
nctl0n

KG(s)H(s) = -+-
(s+ 20)
92 | Introduction to Control Engineering

3.6.4 Ramp lnput The stead;

tWe have seen the transfer function that relates the error with the input signal for a feedback
Er,
systemisgivenbyEquation (3.lOl.ifafeedbacksystemisexcitedwitharampinput, E(s)=L, -
J
then:
where kn i
*E(-,)=
I ..l. In ger
\'t r + KG(s) H (s) i' (3.76) determine,
steady-stat
Applying the final value theorem yields the steady-state error:

€u =r'
r
fl _
=
r | =
1 I
3]7)
t.KG(dHO 71.-. r+rK6i(rl16)1,_. roc1ryr1r1;= 4
where *, is known velocity error constant. Figure 3.31 shows the steady-state error when
as the
excited with a unit ramp input.

r-
ltr
Output resbonse ,/ Input signal
'' 1: ----l--------
40
I ti ? '/

30 ,
i l _.r_;t- 1

20

10
2{r-'-'-:'
). -4-T'-
i I
k

Exercise
0 -q{-"1 i :

Determine
10 20 30 40 and parabo
Time (second)

Figure 3.31 Steady-state error

(1) KG(

3.6.5 Unit Parabolic Input (2) KG(

A unit parabolic input is defined as:

t)
(3) KG(s
u(t)= 3.78)
,
where the Laplace transform is:
Exercise
Obtain the
Y(,)=1. (3.7e) and parabol
r

J
 Time Response of a Connol System I gS

The steady-state error can be obtained as:

)ack

.;,
1
"
1
KG(s)H(s)
il
s'l,_.
r
s' + s'KG(s)II(r)
l_ = + (3.80)
I+
l*.- s'KG(s) H(r)1,.,0 k,

where hn is the acceleration error constant and I I b, is known as the acceleration lag.
In general, the steady-state error depends on the test signal and the class of the system is
',.76)
determined by the open-loop transfer function, KGG)I{s). Table 3.2 shows the summary of
steady-state errors for different classes of systems with respect to the standard test signals.

Table 3.2 Steady-state errors

,. / / )
Class 0 Class I Class 2

yhen 1

Step input *r, 0 0

1
Ramp input 6 , 0
K,

1
Parabolic input @
h,

Exercise 3.19
Determine the class and order of the system below. Obtain the error constanrs for step, ramp,
and parabolic inputs.

(t) KG (s) H (s) = -L-

s(s+ 2)

(2) KG (s) H(i) = ----1?!-

(s+1)(s+5)
(3) KG(s)H(s)=
i.7B) (s+1)(s'z+10s+6)

Exercise 3.20
of the system shown in Figure 3.32 if theinputs are unit srep, rarnp,
.7e)
*ffiffi:;#f.:f:::rror
94 | Introduction to Contol Engineering

The trans

Figure 3.32 A feedback system for Exercise 3.20 where Koi

when excir
the value r
be elimina

3.7 CONTROLLER not be abl

Outcomes: (21 lnte

After completing this section, the reader should be able to: The input

(1) define the transfer functions of proportional, integral, and derivative acrions or their
combinations, and
(2) describe the effect of control acdons on the response of a feedback system. and the tr:

3,7.1 Gontrol Action

A control action or controller is used by engineers to design a feedback control system to satisfy
design specifications mentioned earlier. These are transienr response specification as well as steady- where K; is

state specification. Figure 3.33 shows a typical feedback control sysrem wher e G,(l is the controller
transfer function. (3) Der
Derivative
domain, tl
Output, (s)
Reference
input

and the tre

Figure 3.33 A feedback control system

(1) Proportional (P) action where Ksi:

In time domain, the input-output relationship is: In genr
proportior
 Time Response of a Control System I gS

m(t) = Kue(t). (3.s t;

The transfer function is:

G.(,)=ffi=uu (3.s2)

where rl is the proportional gain. If the process, G(s), is Class 0, there will be a steady-stare error
when excited with a step input. The magnitude ofthe steady-state error can be reduced by increasing
the value of the controller gain, K, but this may lead to orher pracrical problems. This error can
be eliminated by introducing an integral action. The output response of a Class 0 system would
not be able to follow ramp and parabolic inputs.

(21 lntegral(l) action

The input-output relationship is given by:

their *(t) = K;J'oe(t)dt. (3.83)

and the transfer function is given by:

G,(,)=#=+ (3.84)
rrisS,
where K; is the integral constant.
zdy-
roller
(3) Derivative (D) action
Derivative action predicts error before it occurs, thus pre-emptive action can be taken. In time
domain, the input-output relationship is:

(3.85)

and the transfer function is:

(3.86)

where KTis the derivative gain.

In general, derivative actions cannot be used independently. It is commonly combined with
proportional acdons. The combination results with the controllers will be discussed next.
96 | Introdurtion to Control Engineering

(4) Proportional plus derivative (PD) action

The transfe r function of PD conroller is:

M(s\
c.(r)=ffi=Ko(t+Trs) (3.87)

Kd
where T, =
K,
Obtain
(5) Proportional plus integral (Pl) action output r

The transfer function of PI controller is: (1) G(

(2) Gc
Gr(')=H ="'['.*) (3.88) (3) Gc
K Discuss
-_J.
where T,K,

Furthe
(6) Proportional plus integral plus derivative (PID) action
Close, (
The transfer function of PiD controller is: Bo,
Dorf, R
G,(,)=H= *,(r.r,.+) (3.8e)
Evans,
Sac
\
Franklin
The problem faced by control engineers is to choose suitable values of Ko, Ts, and I that Up
can satisfy the design specifications. The problem of determining the controller parameters is Kalman,
called tuning. Various methods have been proposed. For a regulator system, one of the suitable D(r
methods is the Ziegler-Nichols method. Ogata, 3
Ziegler, .
Exercise 3.21 7ha

Describe the effect of proportional, integral, and derivative actions on the output response of a
feedback conffol system.

Exercise 3.22
A feedback control sysrem of a first order process is shown inFigure 3.34.
 Time Response of a Control Sryaem I gZ

(3.87)

Figure 3.34 A feedback system for Exercise 3.22

Obtain the unit step response of the sysrem if the initial conditions are zero and sketch the
output responses for the respective controllers used as given below;

(1) Gdt = 2, a P-controller.

(2) Gls) = 9, a P-controller.
(3.a4; (3) Grll =z+?, a PI controller.
.t

Discuss the speed of response and the steady-state error of the sysrem due to these controllers.

Further Readings
Close, C. M. and D. K. Frederick. 1993. Modeling and Analysis of Dynamic Systems. 2"d ed.
Boston: Houghton Mifflin.
Dorf, R. c. and R. H. Bishop. 2001. Modern contol system.International Edition. Upper
(3.8e) Saddle River, N|: Prentice-Hall.
Evans, \fl R. 1954. Control System Dynamics. New York: McGraw-Hill.
Franklin, G. F., J. D. Powel, and A. Emami-Naeni. 2002. Feedback Control of Dynamics Slsteml
T;that Upper Saddle River, NJ: Pearson Prentice-Hall.
reters iS Kalman, R. E. 1964. \7hen Is a Linear Control System Optimal. ASME J. Basic Engineering.
;uitable D(82): 5r-60.
Ogata, K.2002. Modern Con*ol Engineering. Upper Saddle Riveq NJ: Prentice-Hall.
Ziegler, J. G. and N. B. Nichols. 1942. Optimum Setting for Automatic Controllers. ASME
Ti,ans. 54:759-768.
nse of a
 CFIAPTE,R 4

STABILITY AND ROOT LOCUS

4.1 STAtsILITY OF A CONTROL SYSTEM

Outcornes:
After completing this section, the reader should be able to:

(1) define stabiliry of a control sysrem,

(2) identify bounded and unbounded signals,
(3) define characteristic equation, and
(4) describe the effect of a closed-loop systemt poles to stabiliry.

4.1.1 Goncept
\'X/hen designing a closed-loop control system, the problems of stabiliry may arise if the controller
is not properly designed. A stable open-loop control system may become an unstable closed-loop
control system. In certain conditions, an unstable openJoop control system may become a stabli
closed-loop control system. Therefore a controller or a compensaror must be designed so that
the closed-loop system is stable and its response meers the specification.

4.1.2 Definition
A system is stable if when input signal is used as an input signal, rhe ourpur signa)
a bounded
produced by the system is also bounded. Figure 4.1 shows some examples of bounded iriput
signals with bounded and unbounded outpur signals.
100 | Introduction to Control Engineering

Consider a closed-loop system as shown in Figure 4.2. The transfer function for this The
system is: format as:
r(,) G(,)
u(,) 1+G(s)r1(s)' (4.r)

The expression 1 + G(tH(t = 0, the denominator of the transfer function equated to 0 is where p is
known as the characteristic equation. Generally, a closed-loop transfer function can be written in and poles r
the form of ratio of polynomial equations: (a bounde,

r(r) ry(r) *b*s- +b*-rs*-t +...+bo

= (4.2)
U(t) O(r) sn +d,,r'-'*...*do
where fl ) ffi, and n and m are the order of polynomials in the transfer function. Therefore, the The time r

expression D(s) : 0 from Equation (4.2) is the characteristic equation for the system.

The respor
astable or ,

equarion.
The r
'"' Bounded be stable,

I /\ ,*0,,.,, output signal -

stable system
numeratol
stabiliry. L
condition
must have

Unbounded
output signai -
unstable system

Figure 4.1 lnput and output signals, stable and unstable systems

Figure 4.2 A closed-loop system

 Stability and Root Locus I tOt

The transfer function for the feedback system can also be written in poles and zeros
format as:

Y(r) _ (s+ z,)(s+2,)...(s+ z-)

u(,) Q+ p,)G* p,)...(,+p,) (4.3)

) is where p is the system's poles and z is the system's zeros, while m and n are the number of zeros
rrn and poles respectively. If all the poles are distinct and the input signal is impulse signal, UG) = 1
(a bounded signal), then:

(s + z,)(s + zr)... (s + z-)

Y (s)=
l._
G+ p,)('* p,)..'('+ p")
. rlt The time response for the system can be shown as:

y(t)=cre-?'t +cre P" +"'+ c,€ P" (4.5)

The response is bounded if the roots of the denominator or poles are negatives. It is observed that
a stable or unstable response is determined by the position ofthe poles or roors of the characteristic
equation.
The same goes to repeated poles or complex roots. For all cases, in order for the sysrem ro
be stable, the real part of the roots must be less or equal to zero. Although the root of the
numerator, the zeros, has effect on the response of the system, however, it does not have effect on
stabiliry. It effects the transient and the steady-state responses. Therefore, the requirement and
condition for a feedback system to be stable is that all the poles of closed-loop transfer function
must have negative real parts.

Unstable region

x
Critically stable

Figure 4.3 Stability region on s-plane

102 I Introduction to Control Engineering

Figure 4.3 shows the positions of poles or roots of the characteristic equation on.r-plane. Inverse La1
The system is stable if all the poles are located on the left-hand side of the imaginary axis. System
is critically stable if there is a pole located on the imaginary axis.

lVhen / J
Thus the s
Example 4.1
Derive the time response with impulse input and zero initial conditions for a control system
represented by the transfer function below. Determine the stabiliry of the system. Example
Y(t) * s*l Show the s

u(') (s+3)(s+2)

Solution
This system has two poles and both of them are negatives: \ The Laplace Solution
= -2 and s2 = -3.
ffansform for impulse input is UG) =1.Thus, The charac

(s+3)(s+2) s+3 s-t2

Inverse Laplace transformation gives: The roots {
2''
Y(t)=2e-3' -e
2'
t$7hen t ) 6,
rhe terms a3' and e -+ O. Therefore, y(t) -+ 0, which shows that 7(r) is Since all th
bounded. Thus, the system is stable. Note that all poles are negative. is stable.

Example 4.2
Derive the time response with impulse input and zero initial conditions for a control system Exercise
represented by the transfer function below. Determine the stabiliry of the sysrem. Define the
are the reqr
_ .'+1
Y(t)
Y(r) (s+3)(s-2) Exercise
Determine
Solution determinin

The system has two poles which 2re 11 = +2 and sz= -3, with one positive pole. For impulse
(1) Q(,)
input, UG) =1: (2) Q(,)
s*1 (3) Q(,)
Y(s)=
(s+3)(s-2) 5(s+3) 5(s-2)
 Stability and Roor Lorus I tOS

le. Inverse Laplace transformation gives:

:m
tU):1'-'' *?"'
55 '

V4ren t -+ @, e 3' -s O but /' --; -, there fore, y(t) -+ -, which shows that y(t) is unbounded.
Thus the system is unstable. Note that positive pole leads to unbounded time function.

Example 4.3
Show the stabiliry of a control sysrem with the characteristic equation given by:

Q(')= s3 +4s'+5s+4

lace
Solution
The characteristic equation can be rewritten as:

Q(r)=(s+2)(s'+2s+2).
The roots for this characteristic equation are:

Jr = -1, sz = -7-t j, and sz = -l- j.

.r) is Since all the real parts ofthe roots ofthe characteristic equation are negative, therefore the system
is stable.

stem Exercise 4.1

Define the characteristic equation of a control system. Explain the meaning of stabiliry and what
are the requirements for absolute stabiliqy for a control system.

Exercise 4.2
Determine the stabiliry of the control systems with the characteristic equations given below by
determining their roots.
(1) Q(r)= s3 +3s'+4s+2
pulse
(2) Q(r)=s'+s'+2s+24
(3) Q(r)= s3 +2s'+4s+8
104 | Introrluction to Control Engineering

Exercise 4.3 where dr, d,

Determine the roots of the characteristic equation of a control system with a negarive feedback
having open-loop transfer functions as given below. Then, determine the stability of the sysrem.

(1) K"(s)H(s)=
(s+ 2)(s+ 1)

5(s+ 3)
(2) K,(s)u(s)=
s (s+ 3)(s + 8) The co
ones are zer(
affecting thr
parts if and
4,2 ROUTH'S STABILITY CRITERION Otherwise,
changes.
Outcomes:
After completing this section, the reader should be able to:

(1) construct Routh array, Example,

(2) determine stabiliq, from Routh array, and Figure 4.4 sl
(3) determine the range of design parameter for a stable sysrem. of three. De

4.2.1 Method
Consider the characteristic equation of a dynamic system, in general given by:

ans" +d,_rs'-t +...+ do=0 (4.6)

As stated earlier, the stabiliry of a dynamic system depends on the value of the poles or roots of
the characteristic equation. For a stable system, all poles must be negative. Routh stability criterion
enables a designer to determine whether all poles are negative without the needs to solve for the
values of the poles. This can be done by consrrucing the Routh array defined as:
Solution
J dn d^n-z a,
n1 The charact
dd do-5
: 4-L n-)

4b2 b3

C,tt C^ C.3
: The Routh
0
::
J
 Stabiliry and Root Locus I tOS

where dn, dn-:, .. ., do are the coeffi.cients of the characteristic equation, and:
back
tem. dn-t -dn-t --aan n-1^ l^ d, rdr-o - drdr-5
br= - / -
-
an-l d nt

^ _b,d, r-d,-rb, brd,_, - A,_,b.

'r- h
'4 (4.7)

-, using Equ ation (4.7) until the remaining

The coefficients in all rows in the table are computed
ones are zeros. Any row can be multiplied by a constant before the next row is computed without
affecting the properties of the table. A11 the roots of the characteristic equation have negative real
parts if and only if all the elements in the first column of the Routh array have the same sign.
Otherwise, the number of roots with positive real parts is equal ro rhe number of times the sign
changes.

Example 4.4
Figure 4.4 shows a feedback control system using a proporrional controller with a controller gain
of three. Determine the stabiliry of the feedback sysrem.

(4.6)

rrs of
erion
rr the Figure 4.4 Feedback system for Example 4.4

Solution
The characteristic equation of the feedback system is:

s3 +5s']_7s*3=0.
The Routh array is:
106 | Introduction to Con*ol Engineering

J3 1 7 0
2
J 5 3 0

J
1 5x7 -1x3
_-b.4 5x0-lx0 0
5
0 6.4x3*5x0
J
6.4

All elements in the first column have the same sign (i.e., positive), therefore the feedback In th,
system is stable.
replaced w

If an element in the first column is zero, it may be replaced with a small positive number, t,
and allowed to approach zero a{aer completing the array.

Example 4.5
Figure 4.5 shows a feedback control system using proportional controller with a controller gain
Inspe
system is sr
critical staL
oscillating

If the
then the sy,
use Routh's
Figure 4.5 Feedback system for Example 4.5
Example
Solution
Figure 4.6 r

The characteristic equation of the Feedback system is: values of ,(

st +3s'*4sf 12=0.
The Routh array is then:
 Stabili4t and Root Locus I tOl

-f3 1 4 0
J
2
) t2 0

-t
1 3x4-l2xl 3x0- 1x 0
0
3
0 0x12-3x0

In the first column of row s0, a,ralue of 9 i, obtained. In this case, rhe value 0 in row s1 is
0
replaced with a small positive figure, e. Thus:
:r, a
.t
3
I 40
J
2
3 t20
.t
1
t 00
0
.t t2
galn
Inspecting the first column, we find that all the elements are positive, rhus the feedback
system is stable. \7e can show that the poles of the system are -3 and x2j.Thus, this is a case"of
critical stabiliry where if the system is excited with a bounded signal, the output will continuously
oscillating with a bounded magnitude.

If there is any coefficient of the characteristic equation equals zero or has negative value,
then the system is unstable. Hence only characteristic equation with positive coefficient needs to
use Routh's stability criterion.

Example 4.6
Figure 4.6 shows a feedback control system using a proportional controller. Obtain the range of
values of K for the feedback system to be stable.

s(sr +6s'+1ls+6)

Figure 4.6 A closed-loop system for Example 4.6

108 | Introdoaion to Control Engineering

Solution (3) s' +,

The characteristic equation of this system IS:
(4) s'+
sa + 6s3 +11s2*6s*K=0.
(5) su +
Routh array is built as follows: Exercise
The open-J
J
4
1 11 KO
.,, 6 6 0
J
2
10 K 0
Determine
1
50-6K
J 0
10
Exercise
-f
0
K 0
The charac
For stabiliry all elements in the first column must have the same sign. Therefore:

K>0 Determine
and
Exercise
6o-6K A control s
> o.
10 positive nu
Solving for K we obtain the range for stability as follows:

0<K<10.

Although Routh's stabiliry criterion will allow designers to determine absolute stabiliry
relatively fast, nevertheless it does not provide information on the degree of stabiliry and show
how to improve system behaviour. The biggest problem to utilise this criteria is the need for the
designer to know the transfer function of the system. In most practical cases, transfer functions of
many dynamic systems are complex and difficult to derive. This problem can be overcome by
using experimental methods, such as the frequency response method, where data can be plotted
as Nyquist diagram or Bode diagram.
4.3 Rr

Outcome
Exercise 4.4
Using Routht stability criterion, determine the stability ofthe control systems having characteristic After comp
equations as follows:
(1) sketcl
(1) s'+.r' + Zs + 24 : o (2) show
(2) so+2s3+8s2+4s*3=0 from I

(3) utilise
 Stability and Root Locus I tOg

(3) s5 +.s4 + 3s3+9s2 + 16s + 1o = 0

(4) ss +2sa +2s3+4s2 +1ls+10=0
(5) s6 +3s5 +5sa +9s3+8s2 +6s +4 = 0

Exercise 4.5
The open-loop transfer function of a unity feedback control system is:

KG(s)H(s)=
s(s+ 4)(s+ 8)'
Determine the range of the constant controller gain, K, for the closed-loop sysrem to be stable.

Exercise 4.6
The characteristic equation of a feedback control sysrem is given as:

s3+(4 +K)s2 * 6s * 16+ 8K : 0.

Determine the range of Kfor the system to be stable.

Exercise 4.7
A control system is represented by a block diagram as shown in Figure 4.7, where K and a are
positive numbers. Determine the range of Kand a for the system to be stable.

lin-
rO\\'
the
sof
Figure 4.7 Control system for Exercise 4.7

b,v
ted
4.3 ROOT LOCUS

Outcomes:
After completing this section, rhe reader should be able to:

(1) sketch the root locus diagram,

(2) show mathematically how the locations of the poles change as the design parameter is varied
from 0 to - for a second order system, and
(3) utilise the magnitude and angle criteria.
1 10 | Introduaion to Control Engineering

4.3.1 Concept The value

The stabiliry and performance of a closed-loop control system depend on the pole locations. If
any of the coefficients in the characteristic equation changes the values of the poles will also
change. The study of the movements of the poles or roots of the characteristic equation in the
If (= 0, tl
s-plane, as a coefficient of the characteristic equation is varied, is known as the root locus method.
*1 and-1
Root locus was introduced by\fl R. Evans in 1948.
r

the locus o
Consider a feedback system shown in Figure 4.8 where the closed-loop transfer function is:
4.10, this I
r(,) KG (s)
u(r)- r+ KG (i)11(r) (4.8)

where 1(is the controller gain assumed to be positive throughout and KG(lH(s) is the openJoop
transfer function.

Figure 4.8 A feedback control system

The dynamic behaviour of the closedJoop system depends on the roots of the characteristic
equation:
Obser
1+ KG(s)H(, = 0. (4.e) zeros for K
Consider a feedback system shown in Figure 4.9.
Consi

and define

Figure 4.9 A second order system

Then,
The characteristic equation of the feedback system is:

s2 +2s*K:0, (4.10)
 Srabitity and Roor Locus I ttt

The value of the poles can be obtained from:

s. If -2a,{a-ay
also (4.rt)
the
If K= 0, the poles are 0 and -2,if K= 0.5, the poles are -0.29 and-I.7,if K= l the poles are
rod.
-1 and -1 and if K= 2 the poles are -1 + j and-l -j. if these values are plotted on an s-p1ane and
the locus of the pole locations is drawn as the value of Kis varied from 0 ro € as shownin Figure
a is:
4.10, this locus is known as roor locus diagram.

4.8)

ooP

K= 0 K= 0.5

Figure 4.10 A root locus diagram with the arrows indicate the movement
of the pole location when the value of K is increased
lStlc

Observe that the loci start at the open-loop poles for K = 0 and terminare ar the open-loop
4.9) K = * ('x' denores poles) . The proof is as follows:
zeros for

Consider the general closed-loop transfer function:

v (,) KG (s)
u (') t+ KG (i)11(i) (4.12)

and define the openJoop transfer function as:

KG (s)u (s)=
m/(r)
D(r) (4.13)

Then,
y (,) KG (s)D(s)
_
10) u (') D(s)+x-al(,) (4.14)
ll2 1 Introrluction to Control Engineering

where ,n/G) and D(s) are the numerator and denominator of the open-loop transfer function
respectively. Therefore the closed-loop poles are the roots of the characteristic equation DG) *
m/G) = o.
Show the c
If K = 0, the characteristic equation becomes:

D(s)=6 (4.15) Solution

and if K= -, the characteristic equation is: The closed
I/ (s) = 6. (4.t5)
Hence the loci of the roots of the characteristic equation originate at the open-loop poles (.K= 0)
and terminate at the open-loop zeros (K= -). and the chr
In the example above, the root locus diagram is constructed by solving a quadratic equation.
If a system is of higher order, calculation of the values of the poles is more difficult. Rules have
been devised such that the locus of the closed-loop poles can be constructed from the knowledge Therefore,
ofthe open-loop poles and zeros.

4.3.2 Angle and Magnitude Criteria

All poles musr fulfil the angle and magnitude criteria. Therefore all points on the root locus must IfK=0,t
also fulfil these criteria. The characteristic equation is:
In this casr
diagram is
KG(s)H (s) = -t. (4.r7)
From this equation, we can infer that two conditions must be satisfied by every point on the root
.

locus. These are:

(1) l(rc(s)1{s)l = 1 l

(2) arg(KG(s)H(s)) = 180"(1+22), k:0,+1,+2, "'

t

It should be noted that ilgQo ( +n radians) are the same point on the s-plane and:

arg(Kc (s) 11 (,)) =

",,
=
".,
(,<a/ (, )) -,,g (a (,)) : 1 80" (z * + t) Example
[# )
Using ang
thus poles cause phase lag while zeros cause phase lead.
equation I

Example 4.7
The open-loop transfer function of a uniry feedback system is given by: ifK=4.
 Stabili7r and Root Locas I ttS

)n t('*^'.)
KG(s)H(s):
s(s+ 2)
, Hg)=1.
Show the change in the poles location when the parameter Kvaries from 0 to
-.
Solution
The closed-loop transfer function is:
6 r(,) r(s+l)
=
LI tz (r) s'+(2+K)s+K
and the characteristic equation is:
)n.
| -e s'+(2+K)s*K=0.
Therefore, the values of the poles are:

, _-(2+ K)ttlQ+ tc)'

2
L<t rf K= 0, thepolesare 0and-2.\when K-+*,theaboveequationyieldss-+ -1 ands-)--.
In this
case, the loci end at the open-loop zeros that is
-1 and implied zero at--. The root locus
diagram is shown in Figure 4.11.
I

ot

Figure 4.11 Root locus diagram (, = open-loop poles, O = open-loop zeros)

Example 4.8
Using angle and magnitude crireria, show that rr =
-1 +.7 is one of the roors of the characteristic
equation for a unity feedback control system with the open-loop transfer function given by:

KG (s)H (s)=
,(r'+ 4s+6)
ifK=4.
ll4 | Introduction to Control Engineering

Solution 4.4 F
If s1 is a root of a characteristic equation, it must fulfil the requirement:
Outcon
KG (s)H (r): -t.
After cor
Ifgiven K=4andoneoftherootsisrr =-1 *T,substitutingsl =-1* jandK=4intoopenJoop
transfer function yields:
(1) sket

(s): ,-.' K4
(2) ider
KG (s)H
,(r'+ 4s+6) (-r*y)[(- t* j)'+4(-r+7)+e] @6
(3) ider
4 usir
=
7-t* i17r*4=-' conl

This can also be shown graphically as in Figure 4.l2.The figure shows the position of the
poles of the open-loop transfer function (0, -2 t 1 .414) and root s1 . Angle criterion is given by:
4.4.1 (

ZKG(lH(l = 0t+ 02+ 03 = 135" + 67.49" -22.49o = 180o In genera

while magnitude criteria is given by: follows:

/, (1) Plor
IKG(s)a(,)l =
4h pole
absc
(2) On
(1.4r 4) (2.61 3 ) (1 .082 )
il
as
-1

(3) The
nuII
tran
nuII
(4) For
Figure 4.12 Positions of poles with respect to root s1
 Stability and Root Locus I t tS

4.4 ROOT LOCUS CONSTRUGTION

Outcomes:
After completing this section, the reader should be able to:
rcp

(1) sketch root locus diagram,

(2) identify the value of damping ratro, (, natural frequenry, @,, damped natural frequenry,
@1, and time constant, T, if given pole location on the root locus diagram or s-plane, and
(3) identi$, the value of proportional control gain, K, that satisfies the design specification
using damping ratio, (, naural frequency, a* damped natural frequency, o)7, or time
constant, z
the
br-:
4.4.1 Construction Steps
In general, the root locus can be sketched easily by following some simple rules. These rules are as
follows:

(1) Plot the locations of the poles and zeros of the open-loop transfer function using 'x' for
poles and' O ' for zeros. Note that it is important to use the same scale for the ordinate and
abscissa for a meaningful angle measuremenr.
()\ On the real axis, the locus lies to the left of an odd number of singularities (poles and zeros),
as illustrated in Figure 4.13.

Figure 4.13 Locion the realaxis

(3) The root locus starts at the open-loop poles (-K= 0) and the number of loci is equal to the
number of open-loop poles, p. As K -+ *, zloci terminare at the z zeros of the op.rr-loop
transfer function, where z is the number of zeros. If the number of zeros is lesslhan the
number of poles, p - zlociwill approach implied zeros at oo.
(4) For large values of gain, the locus is asymptotic to the directions having the angles of:
1 16 | Introduction to Control Engineering

(2p+1)180" Calculate t
fuymptote angle = k - 0. I. 2,.'. (4.18) loop syster
;,
The loci approach the asymptotes when K--> *.The number of asymptote angles is /, with
d is the difference between the number of poles and zeros, i.e- d = p - z. For Solution
d = I to 4, the asymptotes are as shown in Figure 4.14.The asymptote angles are as follows: Rule l: On
zeros with
d = l: asymptote angle = 180o is 0. The v
d = 2: asymprote angle = 90", 270"
d = 3: asymprore angle = 50o, 180o, 300" Rule 2: On
d = 4: asymptote angle = 45", 135", 225", 315"
Rule 3:Th<
-) at eithe

Rule 4:

Figure 4.14 Asymptote angles for d = 1,2,3, and 4 producing

(5) The intersection of the asymptotes with the real axis is known as the centre of graviry (cg) o{ Rule 5:Th
the pole-zero configuration.

-
-2, -
I P'-\'' (4.1e)
and thus ti
wherep; andz;arethe values of open-loo :;"irrand zeros respectively.

Example 4.9
A feedback system with a proportional controller is shown in Figure 4.15.

Figure 4.15 A feedback system for Example 4.9

 Stabi/i1y and Root Locus I ttZ

Calculate the asymptote angle and the centre of graviry and sketch the root locus of the closed-
t.1S ioop system.

11-1Ih

For Solution
O\vS: Rule 1: On the root locus diagram, the openJoop poles are marked with ',.' while the openJoop
zeros with 'O '. The number of openJoop poles, p, is 3, and the number of openJoo p zeros, z,
is 0. The values of poles arei ?t= O, pz= _1, and?3=
-4.
Rule 2: On the real axis, the locus exists berween 0 to -1 and on rhe left of -4.

Rule3:Therootlocusstartsatopen-looppoles forK= 0 (startsar0,-1, and-4) andends (K-+

-) at either implied or open-loop zeros. Number of loci = number of open-loop poles, ? = 3.
( zk +t)r 80" _ (z[+r ;r ao"
Rule 4: Asymptore angle = : 60" (2k +t)
;= ff
producing the asymptote angles of 60o, 180o, and 300o.

g: of Rule 5: The point of intersection of the asymprore angle with the real axis is:

(o)+(-r)+(-4) _,_
'g= =*l'ltr
i.19)
3-o
and thus the sketch of the root locus diagram is shown in Figure 4.16.

Figure 4.16 The sketch of the root locus diagram for Example 4.9
1 18 | Introduoion to Control Engineering

Rule 2: O
Exercise 4.8 zeros -2.
Sketch the root locus diagram for a feedback system with the open-loop transfer function given
by,
Rule 3: T
K (s+4) at openJc
KG (s)H (s):
,(r'+ 2s+2)
Rule 4: Si
180o, ther
(6) A breakaway point, o6, is the point on the real axis where the locus leaves or meets at the real from the <

axis. The location of the breakaway point, 06, is obtained by solving: and impli
The
$':tr
(o,- p,) (4.20)
using Rul,
7 7 (on-",)
where p; and z; are the values of open-loop transfer function poles and zeros respectively. Rule 6:
A breakaway point, C6, CZn also be determined by rewriting the characteristic equation in
the following form: that prodr

t<=f(') (4.2r)

wherefs) is a function of s. On the real axis, a breakaway point, 06,is atmaximum ( with Therefore
in Figure,
_-u
dK
(4.22) Alter
ds
in the forr

Example 4.10
The open-loop transfer function of a feedback control system is given by:

KG(s)H(sl:ffi
Sketch the root locus of the feedback system. The chara,

Solution
Rule l: This system has one real zero (q = -2) on the real axis and a pair of complex conjugate

poles (p1 = -1 + J5 j ^ra pr= -r-Jl j ).

 Stabili4t and Root Locus i ttg
Rule 2: On the real axis, the loci are on the left of odd number of singularities, i.e. on the left
of
zeros -2.

Rule 3: The root locus srams from open-loop poles (p,= -t *.,|-li andp2= _1_ r[37') and ends
at open-loop zeros (q = -2) or implied zeros (oo).

Rule4:Since?=2andz= l,therefore d=p-z= I whichmakestheasymptote angleequals

180o, therefore cgin Rule 5 does not need to be calculated. For this exampie,
the locus starus
2-l from the complex conjugate poles and joints the real axis and the locus ends at the
zeros zt = -2
and implied zeros.
The point where the locus meets at the real axis is the breakaway point and can
be calculated
using Rule 6.

Rule 6: +__ 1
ob+r+J5j ob+l-"13i ob+2
that produces
t
o! + 4on =6
h Therefore ot = 0 or 4. The solution that fulfils this case is
-4.Theroor locus diagram is shown
inFigwe 4.17.
Alternatively, the breakaway point can be obtained by rewriting the characteristic
- equation
in the form of K=fls). The open-loop transfer function .an b. rewriten as:

KG(s\H(s',= K(s+2)
\/ \/
(s+l+J-A;Xr* r-rl-li)

K(s+2)
KG(s)H(s)=
(s') +2s+ 4)'

The characteristic equarion is:

r+ KG (s)rz(s; = s

(s'+2s+4)
(s+2)
120 | Introduction to Control Engineering

The breakaway point occurs *hen dry -

(7) The
Q, thus:
(8) The
^ 0s, r

(s + 2)(2s +Z)- (r' + 2s + 4)

(s+2)'
-0
or whe
valu
s2 + 4s =0. Exat
That yields the same solution as before.

Exampl,
The open
system.

Solutior
This syst,
p2=-3,i
conjugate

Figure 4.17 Root locus for Example 4.10

Rule 1: O
Exercise 4.9 zeros witl
Calculate the breakaway points of the root locus diagrams for a system of the following open-
loop transfer functions: Rule 2: O

K('+5) Rule 3: T
(1) KG(s)H1r;= and
s(s+1) -7) e

(2) KG (s) H1r; =

r ('+ 1) Rule 4:T
s(s+5)
 Stabiliyr and Root Locus I tZt

(7) The locus leaves or approaches the real axis at the angle of +90o.
(8) The angle the locus leaves a complex conjugate openJoop poles which is the angle of departure,
07, can be found by applying the angle criterion and using:

0a=-20+180 (4.23)

where 0 is the sum of angular distribution of poles and zeros with poles angles are positive
values while zeros angles are negative values. The application of this rule is illustrated in
Example 4.1 1.

Example 4.11
The open-loop transfer function of a feedback system is given below. Plot the root locus of the
system.

K (s+ 4)
KG(s)H(s)=
s(s+3)(s+5)(s'+ 2s+2)
Solution
This system has one real open-loop zeros (q = -4) and three real open-loop poles (pt = O,
p2 = -3, and p = -5) and a pair of complex conjugate open-loop poles. The pair of complex
conjugate open-loop poles is:

J+- +xz
?s, ?s =-zx = -r_j.
2
Rule 1: On the root locus diagram, the openJoop poles are marked wirh 'x' while the open-loop
zeros with 'O'.

Rule 2: On the real axis, the loci are berween 0 and -3 and betwee n -4 and -5.
Rule 3: The loci start from open-loop poles
@1 = O, ?z = -3,p3 = -5, ?q= -l + j, and p, = -1
- j) and end at open-loop zeros (21= 4) or implied zeros (-).
Rule 4:The asymptote angles are:

(2[+ 1) I 80'
_ (2k +l)180'
=(2k-1)45 = 45o'135'225'375"
?- z 5-l
122 | Introduction to Control Engineering

Rule 5: Centre of graviry =Z P'-L''

Z ?-
_ [(o)+ t-:l+ t-r)+ (-r- ;)+ (-r + ;)] -L(-+ t] _r 5 The sketci
=
5-r
Rule 6: Breakaway point is obtained from:

i1_+1
,r-Ot-Pi r,-Ot-Zi
thus
111111 I-J-I-

on-O ob-(-3) on-e5) ob-(-t-,r) ob-(-t*,r) o,-(-4)'

-I
The value oscanbe obtained by using the trial and error method. Observing the loci on the
real axis, the breakaway point must be between 0 and -3. Thus by trying a few values, it is found
that os= -2.5.

Rule B: The angle of departure of the locus from the complex pole can be o'otained by referring to
Figure 4.18. Thus:
(9) The i
\e :0,+or+q+o^-0, Kat r
be ap
= 90o+ 135 +27" +14" -l8o =248".
obtair

Example
..,,,,,,+""'X'
'":,... ; Consider tl
:
the imagin;
i
:< 0,
..x
!\
.\. Solution
The charac
Figure 4.18 The complex- conjugate pole with respect to other poles and zeros

In the above calculation, the angles contributed by poles are positive while zeros are negarive.
Therefore:
The Routh
 Stability and Root Locus I tZZ

0,uLl=-Fo+l8o
= -248 + 1B0o = -68o.
The sketch of the root locus using the above rules is shown in Figure 4.19.

Figure 4.19 Root locus for Example 4.11

(9) The intersections of locus with the imaginary axis can be located by calculating the values of
Kat the intersection. This is a critical stabiliry situation where Routh stabiliry criterion can
be applied. Auxiliary equation is used for this Kvalue and the point of intersecion can be
obtained by solving the auxiliary equation. This is illustrated inExample 4.12.

Example 4.12
Consider the same problem as Example 4.1 1. Obtain the point of intersection of the locus with
the imaginary axis.

Solution
The characteristic equation of the closed-loop system is:

s5 +10s4 +33s3 +46s2 +(30+ K)s+4K =0.

The Routh array is shown below.
124 | Irtrodrrtion to Con*ol Engineering

,t 1 33 30+K 0
.t
4
10 46 4K0
.t
3
28.4 30+0.6K 00 I
i

4K 00
k
2
.t 35.4 -0.2| K I

1
to52-98]2K -0.t226K2
.t 0
35.4-0.2t1 cor
lComple*
0
4K 0 wtth posrttl

At the point of intersection of the locus with the imaginary

1062-98]2K -0j226K2
axis:

-0
te
l

35.4-0.211 | *.*,,,,.
This equation is satisfied tf K = 10.6. The coefficient of the auxiliary equation is obtained from
the elements of row s2 in the Routh array, thus for K = 10.5, the auxiliary equation is: I

(z;.+ - o.ztl (10.6)) s'z + 4 (10.6) = o

yields r = +1.137. Thus the locus intersects with the imaginary axis at
K = t0.6.
i.l3j when the gain
t_
I

The value of Kcorresponding to any point s6 on the locus can be found by measuring the
s6 to all the poles and zeros of the open-loop transfer function
length of each vector drawn from 4.4.3 Dt
and then evaluated using: N;

As discussr
r,,r- - P ll'. - P'\"'l', -,',1
t,'t-ffi. lro
(4.24) defined in
order syste
frequency,
The application of this equation is illustrated in Example 4.13.
system can
poles. The
4.4.2 Transient Response represent z

are selecter
The transient response of a dynamic system is closely related to the location of the poles. Poles
located on the right side of s-plane will result with an unstable system. If all poles are negative real
A dif
difFerent n
numbers, the output response will be overdamped. A complex conjugate poles will result with a
frequencl.,
damped oscillating system. Poles on the imaginary axis will result in sustained oscillation. Figure
specificatir
4.20 shows the location of the poles on the s-plane and the corresponding types of response.
@1, the de
 Stabiti4t and Root Locus I tZS

Imaginary
axis r Stabiliryboundary

t+
Region ofslow response

ln ^-
V\ \
mplex conjugate poles \
with positive real parts Complex conjugate poies

lr-.
X
with negative real parts

Real a-xis

Negative real poles

:d from
Positive real poles

tl
he gain

Figure 4.20 The location of the poles on s-plane and the types of response

ing the
mction 4.4.3 Determination of Dampinq Ratio, (, Natural Frequen cy, qtn, Damped
Natural Frequency, @a, and rime constant, r, from Root Locri's Diagram
As discussed in Chapter 3, the design transient response specifications of a dynamic system are
(4.24) defined in terms of time constant, T, if thesystem is a first order system or approximated as a firsr
order system, while the values of damping ratio, $, natural frequency, an, inddamped natural
frequency, tos, if the system is a second order sysrem or ahigher order sysrem. Ah-igher order
system can be approximated as a second order system. For example, a third order system has
three
poles. The two dominant poles that are close to the imaginary axis are assumed as the values
that
represent a second order system. Similarly with a higher order system, the rwo
dominant poles
are selected for the approximation.
;. Poles
ive real A different values of poles located on the s-plane as shown in Figure 4.20 willresult with
rvith a
different rypes of responses and these values correipond to the yalues oid"mping ratio, (,natural
Figure
frequency, a)n, and damped natural frequenry, al). Th,rr if we are girr.., tt. tirrri.nt
response
nse.
specification in terms of damping ratio, (,natural frequency, a* anldamped natural f..qrr.rr.y,
@7, the desired values of the dominant poles can be locaied on the ,-pi".r.
or the root locus
126 | Introduction to Control Engineering

diagram and correspondingly the value of Kparameter can be calculated. Similarly, with known The pole
pole position of the closed-loop system on the.r-plane, the behaviour of the system can be deduced
from the values of T, $, 0)n, and a)7.

Referring to Figure 4.2la,a vertical line (real part of a pole) corresponds to constan J il = I
, 4a, .
T'
A horizontal line (imaginary part of a pole) corresponds to constant damped natural frequency,
a1.The constant undamped natural frequency, cD, is represented by a circular line as shown in
Figure 4.21b and a radial line represents a constant damping ratio, (.

Constant line, ro7

The va.lt
earlier, th
Constantline, 1/I

The dam

Figure 4.21a Values of T, (, an, and ar4 on the s-plane

\7e can s

Constant line, (
Im The time
-'--l ja,
"
,,"''t.'..... |
-6=o thus I=
The
general tr

we have:
Figure 4.21b Values of T, (, an, and ar6 on the s-plane

Consider a dynamic system with the following characteristic equation:

and
s *2s*2=0.
2
 Stabili4t and Root Locus I tZZ

The poles of this system is -1 +,r.Thus the pole position on the s-plane is shown inFigure 4.22.

Figure 4.22 Position of poles on the s-plane

The values of (, @n, and a4 can be deduced from the pole position on the s-plane as discussed
earlier, the undamped natural frequency, 0)r, is:

,, = ,fu *1, = I .41 radls.

The damped natural frequency, a,7; tSi

@a =lrudls'
\(/e can show that F = 45", then the damping ratio is:

6 = .or 0 = .or 45" =0.707.

The time constant is given by:

!=t,
T
thus 7= 1s.
These values can be validated as follows. Comparing the characteristic equation with rhe
general transfer function of a second order system:

sz +2(a,s+o)z =0,
we have:

d, =z
and

2(co,=2.
I2B I Introduction to Control Engineering

Thus, The r,

@,=l.4ltadls 0.5, we nee

and locus is shc

€= 0]0t .
shownbyli
In Chapter 3, we have seen:
let say this i
@r=A,rrll-(, by making
and ls + 10
thus

,, =t.4t [14.707'= l rad/s.

Example 4.13
Sketch the root locus of the feedback control system shown in Figure 4.23. Choose the value of
Ksuch that the damping ratio does not exceed 0.5 and the time constant is less than 1 second.
Therefore ,

s(s+4)(s+10)

Figure 4.23 Feedback controlsystem for Example 4.13

Solution
The open-loop transfer function is:

KG(s)H(s)=
s(s+ 4)(s+ 10)
Using the rules to construct root locus diagram, we have:

(1) Intersection of asymptote on real axis, rg= -4.67.

(2) Asymptote angles = 60o, 180", and 300o.
(3) Breakaway point, oo: -1.76.
(4) Intersection with the imaginary axis at t6.32j with the gain value of K= 560.
 Stability andRoot Locus I tZg

The root locus olot of this system is shown in Figure 4.24.If the required damping ratio is
0.5, we need B = cos-' 0.5 = 60o. This is shown by line OB.Theintersection of this line with the

locus is shown by point D. For the required time constant of 1 second, we need ]T : t . This is

shown bylineAC.The intersection with the root locus is E. Since D satisfies the design requiremenr,
let say this is the designed feedback pole location. The value of the controller gain can be obtained
by making measurements on the root locus diagram where lsl = OD = 2.85,ls + 4l = FD = 3.60,
and ls + 101 = GD = 8.90, thus:

r
Ircc 61a 1,)l =
K
-1
l{ls+alls+tol
K
of = I'
(235)(3.60)Geo)

Therefore K = 91.3 is the required controller gain that satisfies the design specification.

Figure 4.24 The root locus diagram for Example 4.13

 f
I

130 | Introduaion to Control Engineering

i

Further I
Exercise 4.10
Sketch the root locus of the feedback control system with the open-loop transfer functions
Evans, \fl .

Hurwitz, 1
given by:
Real l
K
(1) KG(s)H(s)= K
Khatwani,
(2) KG(s)H(s)=
(s+1)(s+6) (s+ 1)(s+ 3)(s + 5) 583.
Krishran, I
(3) K
KG (s)H (s)= (4) KG(s)H(s) -
-^-\-,t-^\-, AC(1
r(r'+4s+8) s(s+1), Lewis, P. F
Prent
(5) KG(s)H(s)= . K (6) KG (s)H (s)= Pillai, S. K
s2 +4s+8 s(s+2)(s+3) AC(1

Exercise 4.11
Sketch the root locus plot of a feedback control system shown in Figure 4.25 and determine the
value of Kso that the damping ratio is 0.5. Further, determine the natural frequency, damping
ratio, and time constant of the designed system.

Figure 4.25 Feedback control system for Exercise 4.1'l

Exercise 4.12
Sketch the root locus plot of the control system shown in Figure 4.26 and determine the value of
K if the required damping ratio is 0.707.

Figure 4.26 Feedback control system for Exercise 4.12

 Stabili4t and Root Locus I tgt

Further Readings

fi10ns
Evans, \fl R. 1950. Control System Synthesis by Root Locus Method. Trans. AIEE.69:66-69.
Hurwitz, A. 1895. On the Conditions underwhich an Equation has Only Roots with Negative
Real Parts. Mathematische Annalen. 46: 273-284.
Khatwani, K. J. 1981. On Routh-Hurwitz Criterion. IEEE Thans. Automatic Control. AC(6):
583.
Krishran, V. 1966. Semi-analytical Approach to Root Locus. IEEE Thanl Automatic Control.
AC(l1): 102-108.
Lewis, P H. and C. Yang. 1997. Basic Control Systems Engineering. Upper Saddle fuver, NJ:
Prentice-Hall.
Pillai, S. K. 1981 . The e Method of Routh-Hurwitz Criterion. IEEE Tians. Automatic Con*ol.
AC(10): 584.

ne the
mping

ilue of
 CHAPTER 5

FREQU ENGY RESPONSE ANALYSIS

5.1 FREQUENCY DOMAIN

Outcomes:
After completing this section, rhe reader should be able to:

(1) explain the objective of using frequency response method,

(2) show mathematically the application of this method, and
(3) explain rhe experimenral procedure of this merhod.

5.1.1 Concept
Frequency resPonse analysis is the study of steady-stare response of a dynamic system
when it is
excited with a sinusoidal signal of frequency, @. Theexcitation frequency i, urii.d from
0 to -
and frequency response data are acquired. These data represent th. characteristic of a dynamic
system. These frequency response data are the gain and phase shift of the system as function
of
frequency.
\7e can use frequency response method for two purposes. These are: to determine the
mathematical model or transfer function of a dynamic iystem and to analyse the stabiliry
of a
closedJoop sysrem. Consider a sinusoidal input signal:

u(t)= Asin(.,t) (;. t1

 T

tl

I34 | Introduction to Contol Engineering

where ,4 is the amplitude and o is the frequency of the input signal. For a linear system, the and the pl
output signal is also sinusoidal with the same frequency given by:

y(t)= Bsin(at +Q) (5.2)

where Z is
where B is the amplitude of the output signal and Qis the phase shift. As (0varies, it is found that
output sig
the amplitude B and Qvary.These quantities result in the frequency response data gain,
!- , of the outl
A ^nd,
phase shift, @. Let tl
Experimentally this data can be obtained as follows. Let say, a dynamic system is represented the transfr
by a bloik diagram as shown in Figure 5.1. The system is excited with a sinusoidal input of some output Yar
frequency, A), andlet the input and output responses be shown by Figure 5.2. to be zero.

The frequ

Figure 5.1 A dynamic system represented

by a block diagram

Equations
u(t)=Asin(ar) of G(s), it:

/t)=Bsin(at+Q)
Example
Obtain th

Time (second)
Solutior
Figure 5.2 lnput-outputfrequencyresponse
The trans
thus:
The gain is the ratio of the output to the input amplitudes, and measuring these amplitudes
results in:
B
Cain = (5.3)
A
 Frequencl Response Ana/ysis I 135

and the phase shift, @, is:

O= rad
<l i"2n 6.+)
_a:_ where Z is the length for_ one complete cycle and x is the length of delay benveen input
and
output signals, for example the length between peak of input signal and thl .orr.rpondirrg
-- l peak
of the outpur signal.
Let the transfer function of a dynamic system is GG). As discussed in the previous chaprers,
the transfer function of a linear system is defined as the ratio of the Laplace rransform of th.
output variable to the Laplace transform of the input variable wirh all initial conditions assumed
to be zero. The Laplace transformation for a time-invariant system of
f(t) is defined as:
F(s)= Il,"f (t)dt 5.5)
The frequency domain of the system is described by the Fourier transform given by:

F (a)=
J*. '*' f (t)*. (5.6)

Equations (5.5) and (5.6) show that by substituting , = j@, for a sysrem with a transfer function
of G(s), its transfer function in frequency domain ii given as,

c(jr)= G(s). (5 7\

Example 5.1
obtain the frequency response ffansfer function of a first order system given by:

G (s):
l+Ts

Solution
The transfer function of the system in frequency domain can be obtained by replacing
s with j6,,
thus:

G(jot)=
l+Qat
<l
 f
I

136 | Introduaion to Control Engineering i

The frequency response transfer function, G(ja), is represented by a complex number: Since u(t

c (jr) = R(ar)+ jI (ot) (5.s1

and
where R(al) and I(a) are the real and the imaginary parts respectively. Alternatively, it can be
represented in terms of the amplitude response, lc(at)|, and phase response, Q(ia1, in polar Therefor
notation as:

G (j @) =lc (i ql,ia@) - lc (i ql zQ @). (5.e)

Thus, th,
Thus, if a stable linear system is excited with an input, u(t) = Asin@t, and the output is
y(r) = Bsin( @t + Q), then the gain and the phase shift are:

lc(ir)l=X= refrn r t-f (5.10)

and

1(o) The steat

Q: zG (jr)=,rrr-' (5. t t;
n (r)
respectively.

Example 5.2
5.1.2 (
Obtain the steady-state response of the dynamic sysrem shown in Figure 5.3 lf the input is
The freq
u(t) = 4sin(2t).
represent
logarithn
the scopr
Thr
function
combina
Figure 5.3 A dynamic system for Example 5.2

Solution 5.2 t
The output signal is given by:
Outcon
y(t)=Bsin(at+Q)
After cor
where

n =lc(1at)lxa.
(1) deri
(2) plor
 Frequencl Response Analysis I tZl

Since zz(r) = 4sin2t, then:

A=4
and
(D=2.
Therefore,

c( jr) = G (2\ j)=

r/ .+..
2j+r
Thus, the amplitude ratio and phase shift of the sysrem are:

l"ei)l=, --l- A =
",Ei J'
Q = zG (jr)=-ra.,-'f3l= i.to7 rad..
(1,
The steady-state response of the sysrem:
4
7,, = ,' sin (zt -t.107 ).
V)

5.1.2 Graphical Representation of Frequency Response Data

The frequency response data of a dynamic system can be presented graphically. The graphical
representation which uses a polar graph paper is known as Nyquist diagram while with a semi-
logarithmic graph paper is known as Bode diagram. Other graphical represenrarions are beyond
the scope of this book.
The steps to sketch the Nyquist and Bode diagrams of common elemenrs in a transfer
function, such as integrator, differentiator, first and second order lags and leads and their
combinations are shown and analysed in the nexr secrion.

5.2 NYQUIST DIAGRAM

Outcomes:
After completing this section, the reader should be able to:

(1) derive the expressions to calculate the gain and phase shift from transfer function,
(2) plot Nyquist diagrams for differentiator, integ.ator, first and second order sysrems,
138 | Introduaion to Control Engineering

(3) determine the frequency range to calculare the gain and phase shift, By substi
(4) calculate the gain and phase shift, and
(5) draw polar plots or Nyquist diagram.
The abor
5.2.1 Goncept modulus
phase shi
The plot of frequency response data, i.e. the gain and phase shift, as a function of frequency, rrt.
that varies from 0 to rc on a polar graph paper is known as the Nyquist diagram. The polar plot
of the frequency response data is drawn using the complex number representation where the and
coordinates of the polar plot are the real and imaginary parts of G(ja) as shown in Figure 5.4.

0. 4 resPectlY(
\7e
0. zG(ja) expressio
phase shi

0" when a;

lc(ja,l
and a--

Using tht
is shown

Figure 5.4 Frequency response data plotted as a polar plot

In this chapter, we will start by presenting the calculation of frequency response data for
some common elements in a transfer function typically representing a dynamic sysrem. Then,
the plots of the Nyquist diagrams of these data are shown.

5.2.2 Differentiator
The transfer function of a differentiator is:

GG) = r. 6.12)
 r

fl Frcquenqy,RarponrAnz/ytz't I ISS
ii

By substituting s with ja, the frequency response rransfer function is:

G(jctt) = ja . (5.13)
The above expression is a complex number without the real part. The gain for the system is the
modulus of G(ja) while the phase shift is the argument of G(ja).Therefore, the gain and
phase shift are:

lc(1r,41= r,t (5.t+;

and

zG(jo)=9s" (5.r5)
respectively.
We will analyze the change in the gain and phase shift as rr) varies from 0 to
-. The above
expressions for gain and phase shift show that as the frequency varies, the gain varies while the
phase shift remains consranr with the value of 90o. For the gain,
when at -+ 0

lc (7a;)l -+ o (5.1e;

and a) -+ -

lc17a;;l-+ -. (5.r7)

Using the above information, the data is plotted on a polar graph paper and the Nyquist diagram
is shown in Figure 5.5.

Figure 5.5 Nyquist diagram for a differentiator

140 | Introduction to Control Engineering

5.2.3 lntegrator 5.2.4 Fi

The transfer funstion for an integtator is: The transl

G(i)=!. (5.18)
.t

By substituting s with j(0, the frequency response transfer function is: Bv substir

G (ir;.)= --l-' (5.19't

.1a
Thus
\(/e can si
;1
G(iro)--r =i-Z-90'. (5.20',t
aa
Equation (5.ZO1is a negative cornplex number without the real part, thus with a consranr By multil
phase shift independent of the frequency, i.e., (5'25) be,

zG(ja)= -90o 5.zt)

and gain,

lc(ir)l=* (5.zzt where R(

The value of the gain varies with frequency as shown below: As a com
As shown
when ar --> 0, point on
-, (5.23) lG(ja)1,
lc(7a;;l-+ be easier r
and () -+ -,
o, 6.24)
lc17a;)l-+
thus the Nyquist diagram for an integrator is shown in Figure 5.6.
and

for the ga

Figure 5.6 Nyquist diagram of an integrator

 Frequency Response Ana lysis I t+t

5.2.4 First Order Lag

The transfer function of a first order lag is:

G(s) =
K
(5.25)
1+Ts
By substituting s withTa;, then,

c (j')=
' -+-'
l+7j0) 6.2e1
\(/e can see that G(jtl) is a complex number and can be rewritten as:

c(jr)=R(ar)+ t(r)j. (5.27)

By multiplying the denominator and numerator with the denominator's coniugate, Equation
6.Zel becomes:

(\Jira\ r-lj@ K(r- jra)

G =
' l+Tjo l-Ua r+(rro)' 6.2a1

where R(a\:
\-- / ! - -and-- I^ (ro\=
\--, -ffi*, .
l+(Tro), 7+(T*)r'
As a complex number, G(ja) can be plotted on the Argand diagram as shown in Figure 5.7.
As shown above, the complex number is a function of ar. Point G(ja) varies as al varies. Any
point on the locus can be determined by the real and imaginary parts of G(ja) or the gain,
lc(ja)|, and phase shift, ZG(ia), of G(ja). For a first order lag sysrem, the calculationiwill
be easier using:

* (rat)' (5.2e)

ZG(ja)= -tan-ITa) (5.;o;

for the gain and phase shift respectively.

al

Figure 5.7 Argand diagram

142 | Introtluction to Control Engineering

The values of the gain and phase shift vary as the frequencyvaries. Firstly, lett determine the TI
gain and phase shift at two extreme frequencies , i.e . o) = 0 and ro = €, rhus: semicir<
is showr
when o = 0,

Gain = lc (ir)l = K and"phase shift @ = 0"

and when a) ) * ,

Gain
lG
(ir)l-0 andphase shift Q =-90".
Another important frequency needs to be investigated by a control engineer is the centre frequeno-
(also known as break frequency). For a first order system, the cenrre frequency is cD = ],T *h.r.
Iis the time constant. Substituting the valu e o) =l irr,o the expressions for gain and phase shift
" T I
yields From F
K
lc(i,)l= T (5.tr)
and
and
zc(ja)- -45" . (5.22)

Based on these data, the Nyquist diagram can be drawn and is shown in Figure 5.8. This
Iocus is a semicircle where the center is located atO.5Konthe real axis which is equivalent to the
radius of the semicircle. Using F

Therefc
00

Thus,

where -r

Figure 5.8 Nyquist diagram for first order lag

 Frequency Response Analysi I t+S

The plot can be shown to be a semi-circle as follows. For simpliciry let K = l. For a
semicircle, the radius, x, is a constant, equals to 0.5 independent of aL The geometry of a semicircle
is shown in Figure 5.9. Ve need to show r = 0.5, aconsranr independent of a.

Figure 5.9 Semicircle geometry

From Figure 5.9, Xis the real part of G( ja) while )'is the imaginary part, that is:

(5.33)

and
l
-:- (5.34)
Ll

h.
Using Pathagorous theorem:

*'=( x-t)'*r'. (5.35)

\ z)
Therefore, substituting Equations (5.33) and (5.34) into (5.35) yields:

,( I
.r.--l-__r.Lr_- 1)' ( r, .l.f
It*1rr;' 2) l,*trol'')
Thus,

x=0.5
where x is a constant, independent of r0, therefore the locus is a semicircle.
144 | Introduction to Control Engineering

Exercise 5.1
A frequency response test is performed on a dynamic system. The data is plotted on a polar graph
paper and the Nyquist diagram is shown in Figure 5.10. The locus is a semicircle. Derive the
transfer function for this system.

Th
importar
are 0, 0),
phase sh

at = 0.5

Figure 5.10 Nyquist diagram for Exercise 5.1

Based or

5.2.5 Second Order System

The transfer function of a second order system is:

rd'
G(s)=+2(a,s+0t'
s2
. (5.36)

By substituting s withT(0, then:

G(ja)= Kai
al-a' + j2(o4ot 6.37)

Thus, the gain is:

'
lc(,rtl=g
lt '-GII. *'(:'l
(5.38)
Res
rutro,
system is
maximur
$.

and the phase shift is

 Frequencl Response AndQsis | 145

Q =tan-' (5.3e)

'[;l
The change in gain and phase shift as frequency varies can be determined by studying some
important frequencies that can describe the characteristic ofa second order system. These frequencies
are 0, o)r, and *, where ro, is the centre frequency of a second order system. Thus the gain and
phase shifts for the frequencies of interesr are:

(D -+ 0, lc(1rrt)l= t<, Q -+ O" (5.40lt

a) ) a)n, lc tir)l= +, @ -+ -eo" (5.4r1

(o ) 6, o, -+ -180" (5.42lt
lc 11i,o11= @ .

Based on these data, the Nyquist diagram is plotted as shown in Figure 5.11.

Figure 5.11 Nyquist diagram of a second order system

Resonance may occur for a second order s)rstem. This depends on the value of the damping
ratio, (. If resonance occurs, the amplitude of oscillation of ,h. or,pu, is maximum when the
system is excited with the_ input signal of frequency equals to the resonance frequency. Let the
maximum gain occurs at frequency, 60, hence the expression for resonance frequency is:

1-24' (5.43)
146 | Introduction to Control Engineering

The derivation of this expression can be shown as follows. The gain is given by Equation (5.aS).
The maximum gain that gives o)rcan be determined by solving:

alc(io)l _0. (5.44)

da
The solution for a,can also be obtained by solving:

dlc(jat)|,
dto
=o (5.45)

since the maximums for lC(7ar)l ana lC(i{l' occur at the same frequency. Solving for
alc (iat)l'
= 0 is easier because the square roor rerm d.oes not exist in the expression.
dro
Differentiating:
Exerci
Figure I
system.
lc(iql'= (5.46)

[' [;ll.^r("*)
with respect to a) and equating to 0 yields the resonance frequency:

@, = @,ll-2c' . (5.47)

Figure 5.12 shows the Nyquist diagram of a second order system with resonance. From

Equation (5.43), resonance occurs if 1;zq is a real number, thus:

t-2e2 >o

therefore,
5.2.6
c < 0.707. It is corr

The gain at resonance frequency can be obtained by substituting the resonance frequency, the syste
Equation (5.43), into the gain expression given by Equation (5.aS) resulting with: 5.14.

lc {i')1.", -
n*=E@
1
5'48)
 Frequency Response Anallsis I t+l

Figure 5.12 Nyquist diagram of a second order system with resonance

Exercise 5.2
Figure 5.13 shows the Nyquist diagram of a dynamic sysrem. Derive the transfer function of this
system. Does resonance occur in this system? If so, determine the resonance frequency.

l2
Figure 5.13 Nyquist diagram for Exercise 5.2

5.2.6 Transfer Operators in Series

Itis common in block diagram presentation of a dynamic sysrem that the elemenrs or blocks in
the system are arranged in series. Let say a system consists of three elements as shown in Figure
5.14.

Figure 5.14 A dynamic system with three elements in series

 Frequencry Response Anafisis I t+g

Example 5.3
Figure 5.15 shows a dynamic system that consists of nvo elements in series. Obtain the frequency
response data and sketch the Nyquist diagram.

Figure 5.15 Block diagram for system in Example 5.3

Solution
Both transfer functions in the system can be rewritten as:

) 0.5
\ / =--t
GI (s)
1+0.5s
and G (r)=
l+0.25s
Substituting s with ja, the gains and phase shifts are given as

2
lc,(jdl= __+ (0.5a;)" and ZGt(j@) = -ran ' 0.5rr.r
,tt
and

lc,(jdl= and lcr(j@)= -ran t 0.25a.

t+(0.25ar')
The selection of frequency range to calculate the gain and phase shift is important. Since
both transfer functions are first order systems, the centre frequencies are related to time constant
I Since the respective time constants are T1= 0.5 s and T2= 0.25 s, rhus rhe centre frequencies are
the reciprocal of these values which are a)1 = 2 radls and a2 = 4 radls. These frequencies have to
be in the middle of the selected frequency range. The frequencies may be used for caiculating the
gains and the phases are 0.5, 2,3, 4, and 10. The calculated frequency response data is shown in
Table 5 .2 and the Nyquist diagram is shown in Figure 5. 1 5.
150 | Introduction to Control Engineering

Table 5.2 Frequency response data for Example 5.3 Or

characte
lc,l zG, lcrl ZG, lY/u1 /.YIU easier tc
pracdca
0 2 0 0.5 0 1.0 0
0.5 1.94 -t4.0 0.5 0 -7 .t 0.97 *2t.1
2 1.41 -45.0 0.45 -25.5 0.63 *7 t.6 Exercir
) 1.11 -55.3 0.40 -36.9 0.44 -93.2 Sketch t
4 0.8 9 -63.4 0.35 -45.0 0.31 - 108.4
10 0.39 -78.7 0.19 -68.2 0.07 -145.9
(1) K(

(2) K(

(3) K(

(4) K(

Exercis
The tran
follows:

Obain r

Figure 5.16 Nyquist diagram for Example 5.3

5.3 {

Exercise 5.3 Outcon

Figure 5.17 shows a dynamic system that consists of three elements in series. Obtain the frequency
response data and sketch the Nyquist diagram.
After cor

(1) defi
(2) defi
(3) anal
(4) detr
Figure 5.17 A dynamic system for Exercise 5.3 (5) calc
 Frequency Response Analysis I tSt

_ One advantage of using Nyquist diagram is that it represents the frequency response
characteristic of a system over the entire frequency range in a single plot and therefore ,n"kirrg it
easier to analyse. However, one disadvantage is rhat it is difficult to plot. It will be no lor!.,
practical when a system needs analytical modification and checking foi its performance.

Exercise 5.4
Sketch the Nyquist diagram for these open-loop transfer functions:

(1) KG (s) H (s) = ---l

(t+0.:s)(t+zs)

(2) (t+9';') -
KG(s)H (sr=

(t) (s+3)
KG (s) H (s) =
(s'z+4s+16)

(4) 30(s+8)
KG (s)a (s)=
s(s+2)(s+4)

Exercise 5.5
The transfer function of an open-loop control system with a negative unity feedback is given as
follows:

10
KG(s)H(s)=
s(s+ 5)(s + 2)
Obtain the frequency response data of the open-loop sysrem ancl sketch the Nyquisr diagram.

5.3 STABILITY ANALYSIS USING NYQUIST DIAGRAM

Outcomes:
After completing this section, rhe reader should be able to:

(1) define Nyquist stabiliry criterion,

(2) define gain and phase margins,
(3) analyse the stability of a feedback sysrem,
(.al determine the phase and gain margins from Nyquist diagram, and
(5) calculate the phase and gain margins mathematically.
152 | Introduction to Con*ol Engineering

5.3.1 Nyquist Stability Criterion (2) If th,

Stability analysis using Routh stabiliry criterion requires the mathematical model of a system to with
be known. However, in practice, to derive a mathematical model for a dynamic system can be (3) If th
difficult and if it is successful a complex model may be obtained. feedl
Frequency response method can be applied to predict stability of a feedback system based
on experimental method. Frequency response test is carried out on the open-loop system, and Figure 5.2
based on this data, by plotdng a Nyquist diagram, the stability of the closed-loop system can be
analysed. This method will not only predict the stability of a closed-loop system but it will also (1) The
provide the measure of relative stabiliry. The measures of relative stabiliry are the gain and phase (2) The
margins. (3) The
Lett study a feedback system shown in Figure 5.18. Assume the frequency response data of
the open-loop system KG(lH(s) is plotted as a Nyquist diagram as shown in Figure 5.19, where
Kis the controller, GG) ir the process, and 11(s) is the feedback element.
As frequency varies from fo = 0 to 0) -)6, the conditions are as follows:

(1) If the gain IKG(ja)H(ja)l < 1 when the phase shift IKc(jat)H(jat) : -180o, the
closed-loop system is stable.

Figure 5.18 A feedback system

5.3.2 Dr

The degre
response d
rGQa)tt(ja:) gain margi

5.3.3 Gi
Figure 5.19 Nyquist diagram of the open-loop system For a stabl
hand side
 Frequency Response Analysis I tSS

(2) If the gain IKG( ja)H(ja)l:1 when Z KG( ja)H(ic,t1:-rrrowhere the locus inrersects
with point -1, the closed-loop system is critically stable.
(3) If the gain IKG(ja)H(ja)l > 1 when the phase shift Z KG(ja)H(jot) = -180o , the
feedback system is unstable.

Figure 5.20 shows the possible plots for three different range of gain, ,(
(1) The closed-loop system is becoming unstable as Kincreases.
(2) The system is said to undergo critical stability as the locus passes through point
-1.
(3) The system is more stable for smaller value of K

KG(jat)u(

Unstable
Critically
stable Stable

Figure 5.20 stability of a closed-loop system for different values of K

5.3.2 Degree of Stability

The degree of stabiliry of a feedback control sysrem can be determined from the frequency
resPonse data of an open-loop system. The degree of stabiliry can be measured using phase
and
gain margins.

5.3.3 Gain and Phase Margins

For a stable closedJoop system, the locus of an open-loop control system must be at the right
hand side of point -1. The gain margin , GM, is defined as the ,.tiproc"l of the gain of lhe
154 | Introduction to Control Engineering

open-loop system, IKG(ja)H(ja)1, at the frequency in which the phase shift is -180o and is
given as:

GM =!x (5.54)

where

* =ltcc (1.,.,) u (ir)lru"r(ta)H(j(D)__180" .

The phase margin, 7 is defined as 180o plus the phase shift of the open-loop system,
Solutir
IKG(jot)H(jot), at the frequency in which the gain of the open-loop system is unity. The valr
is to obr
r:180" + ZKG(j@)U (ir)lKG4a)H( j,.,)=l. (5.55)
phase ar

Figure 5.21 shows the,gain and phase margins defined on a Nyquist diagram. by calcu
(5.55).

(1) Ur

The frec
v

The exp
-c
11a1u 11ro11=t

t<c (jct) a (jat)

Figure 5.21 Gain and phase margins and

For feedback system design, a typical value for phase margin, 7 is in between 40o to 60o
and gain margin, GM, is between two to three. If the gain and phase margins are in these ranges,
the damping for a closed-loop system is reasonable. Th,
constant
the frequ
expressic
These p<
Example 5.4
Figure 5.
Figure 5.22 shows a feedback control system. Determine the phase and gain margins for this
system.
 Frequency Response Analysis I tSS

Figure 5.22 A feedback system for Example 5.4

Solution
The values of phase and gain niargins may be determined
using two merhods. The first method
is. to obtain the frequency response data,
and these data plio,t.d as a Nyquist diagram. The
phase and gain margins are measured from the plotted "r.
Nyq,rir, diagram. Tir.'r..ord method is
by calculation based on the definition of phase and gain
(5.55). -".gi.r, given by Equations (5.54) and

(1) Using Nyquist Diagram

The frequency response rransfer function of the open-roop
sysrem is:

10
KG (j@)H (jr)=
ja(t+o.t jioo)')

The expressions for the gain and phase shift are:

lxc(p)a(i,)l=4ft,
and

zKG (j a) U (j r)= -90" - 2 tan-1 (0. tar).

The transfer function consists of an integrator and rwo
first order lags with the same time
constants of 0' 1' Thus the centre frequency is l0 radls.
Let the selected f.".q,r.r.i., to ."l..rlar.
thefrequencyresponsedatabe a=3,5,l0,and20rad"lsec.substitutingthevaluesof
rointothe
expressions for gain and phase shifts results with
the frequency response data shown in Table
These points are plotted on a polar graph paper ,.rrrting 5.3.
with'a Nyquist di";;r; shown in
Figure 5.23.
156 | Introduction to Control Engineering

Table 5.3 Frequency response data for Example 5.4

10 20

IKc(jat)a(ja)l 3.1 \.5 0.5 0.1 \7e need

ZKG(ja)H(ja) -123 -143 -180 -216

Measurements done on Figure 5.23 yield the gain margin:

thus,

GM=l =
|
=2
x 0.5
and phase margin, hence,

v-)1o

Therefor

Thr

The phas

\We need

Hence,

Figure 5.23 Nyquist diagram for Example 5.4 Solving d

(21 Mathematically Substituti

If the values of gain and phase margin are needed without plotting the Nyquist diagram, these
values can be determined mathematically. Gain margin is defined as:
 Frequenclt Response Analjtsb I tSl

GM=
It<c
(1 a) ru ( i r)1,
o",,.)H( io)=_r so.

Y/e need to identify the value of al when the phase shift of the open-loop is 80o, therefore:
- 1

zKG (jot) H (jr) :-90' - 2 tan 1

(0. lar) = -180o

thus,

ran ' (0. llo) = 45"

hence,

ar=10 rad/s.
Therefore, the gain at this frequency is:

10
* =lr<c(7ro)rr(7ro)l= = 0.5
ro(r+(o.rxro)')
Thus, the gain margin is:

GM:1 =J-:2.
x 0.5
The phase margin y canbe determined from the definition:

I = -180o + z.KG (1a) a et@)11*",,,t,t,.t1=,

tWe need
to identi$, the frequency, a.l, when the gain of open-loop sysrem is uniry i.e.:

(ir)l= 1;l
Ir<c (1a)a ;
' ,(1+ (0. ta;)')
:; r = 1.

Hence,

0.0k03 +ar-10 = 0.
Solving this equation yields:

a= 6.8 radls.
substituting the value of a = 6.8 rad/s into the phase shift expression:

z"KG ( j a) H ( j r) = -90' - 2 tan-1 (0. 1 x 6. 8) = -r 58.4"

158 | Introduction to Control Engineering

therefore, (1) de
(2) US
/ = I 800+ (-1 58.4) = 21.6" (3) de
(4) dr

5.4.1
Exercise 5.6
A feedback control system is shown in Figure 5.24.If K = 1, plot the Nyquist diagram for this Nyquisr
system and determine the gain and phase margins. Veri$ryour answer using mathematical method.
involr.ir
Determine the critical value of Kfor stability. ofa dyn
if the g;
much er
presentz

s (1 + 0. li)(1 + 0.02s)

Figure 5.24 A feedback control system for Exercise 5.6

Linear

Exercise 5.7
Consider the system in Exercise 5.5. Analyse the stabiliry of the feedback system by plotting a
Nyquist diagram. Determine the values for gain and phase margins. Verify these values
mathematically.

Exercise 5.8
A negative uniry feedback system has an open-loop transfer function given by:

10K (s + 0.5)
KG (s)H (s)=
s2 (s+2)(r+10)
Th
Sketch the Nyquist diagrams for K = 1 and K= 10. Compare the values of the gain and phase
margins and discuss the effect of Kon the stability of the feedback sysrem. superim
plotted i
gain in <
5.4 BODE DIAGRAM
Outcomes: th
logarithr
After completing this section, rhe reader should be able to: approacl
 Frequency Response Analysis I tSg

(1) define gain in decibel,

(2) use a straightJine asymptotic approach to plot gain curve,
(3) determine the frequency range to calculate and plot phase shift curve, and
(4) draw a Bode diagram on a semi-logarithmic graph paper.

5.4.1 Concept
Nyquist diagram is easy to plot but may not be easy ro use for control sysrem design when
involving two or more elements in a transfer function. The overall gain of the transfer function
of a dynamic system is multiplicative. However, multiplicarion can be trarrsformed into addition
if the gains are computed logarithmically. Thus, designing using logarithmic graph would be
much easier. Figure 5.25 shows a semi-logarithmic (is also known as semi-log) scaG graph. The
presentation of frequency response data using logarithmic plot is known as a Bode diagram.

Linear

Figure 5.25 A semi-logarithm graph paper

The gain
lrc (1co) a (1a)l
^"aphase
shift zKG (jro) a (jr)
are plotted separately but
superimposed on a same graph paper using linear against logarithmic axes. The gain curve is
plotted in decibel (dB) on the linear scale against frequency, ar, Ln the logarithmic sclle *here the
gain in dB is:

lxc (1r,t) u (1a)lav = zotoslKG (1la) n (1ra)1. 5.56)

The phase shift curve is plotted as phase shift on the linear scale against frequency, al,
on the
logarithmic scale. The Bode diagram plots can be simplified ,rring "" straight-line ,ry-ptoti.
approach. Consider a transfer function given by:
160 | Introduction to Control Engineering

K (s + z,)(s + zr)-..(s + z*) "nd,

G(i)= 6.57)
s/ (s+ a)(r+ p,)...(' + ?,)'
The gain of the system is: The gain

r l(, +,, )ll(, *,)1...11,+,. )l

lc(i,.)1= (5.58)
l,'ll(,*A)ll(, + ?)l. l(,*a)l l,_,,
Using a sr
which is the product of the frequency response gain of each term. Simplifying the expression can be apl
with logarithm of the magnitude, the numerator terms would be added and the denominator
would be subtracted, and the gain in dB is thus: when rr) <

2\loglGlja)l = 2}logK + 20logl(s + z1)l + 2Ologl(s + z2)l +... + 2Ologl(s + z-)l

-zologl/l - 20logl(s + ?r) -... - 20logl(, * ?,)ll,-jo, 6.5e)

when rD.

The phase frequency response is the sum of the phase frequency response of the numerator
terms minus the sum of the phase frequency response of the denominator terms. The algebraic
sum for magnitude and phase responses is the graphic addition of each term. Thus at 1c
\7e observe that the effect of each term on the gain of the overall system when using gradient c
logarithmic values is addition rather than multiplication when using absolute values. Therefore if Figu
a constant gain is used as a controller, using Bode plot the gain will shift the gain curve vertically.
1
This will ease the design of a proportional type controller. 0)-- rs
T
5.4.2 Straight-line Asymptotic Approximation for Gain PIot
Consider a first order system given by: Based on
and can b
G(i)= (5.60;
l+Ts
By substituting s withTo,

G ( jat)=.+-. (5.51)
t+ lJa
The gain and phase shift are:

lc(r,)l= r+(rot)' (5.62)

Fi
 Frequencry Response Anallsis I tOt

and,

zG(jot)= -ran-1 (Tr). (5.53)

The gain in dB is:

lc (iqldB = 2olos
t+ (rilo)' 5.eql
Using a straight-line asymptotic approximation, the gain at low frequency and high frequency
can be approximated as:

*h., al << !
T

lc(ir)ldB = -1otog1= o dB
NI
when a<<-
T

lc17ar;lau = -10 tog(Tat)' =*2otosT -2llos1l..

Thus at low frequency, the gain is 0 dB while at high frequenry the gain is a straight line
with a
gradient of -20 dB/decade. , .' ,

Figure 5.26 shows the Bode plot for gain response curves of a first order lag. Frequency

a = ! is known as break frequency.

1------rThe actual gain at * I
T b*"'-' ar =
T
is

lc17ar;lan = -1olog(1+ 1)= -3 de.

Bas-ed on this plot, 3 dB is small therefore a straightJine asymptotic
approximation is reasonable
and can be used in plotting the gain plot of the Bode diagram.

Asymptotic
- approximation
=trr
fa
0)

-20 dB/decade

Figure 5.26 Asymptotic and actual magnitude response curves of a first order lag
162 | Introdraion to Control Engineering

5.4.3 Straight-line Asymptotic Approximation of Phase Shift Plot

The phase shift for a first order lag is given by Equation (5.63), i.e., Exan
A dyn
tG(j@)=-tan-1 (rT)'
1
@ <<
'When -T
z.G(ja)-> 0o. Draw
1
\fi/lren o>>
T So/ui
zG(jro) -+ -90'.
The br
As for asymptotes of phase shift, the phase is approaching 0o at low frequency when ar --> 0 and
-90o at high frequency when o) --s @. A straight-line asymptotic approximation with a negative produ
o=5
slope can be used for plotting asymptote Oo at o)= + and asymptote -90o at a =* . O, Iine u'
T
break frequ ency (t)= I , ,h. onr* ,, - 45".ToO**,il:rrve, starts one decade below the break phase
T
frequency with 0o phase, draws line of slope -45oldecade passing through -45o at break frequency
a

and continues to -90o one decade above the break frequenry. Figure 5.27 shows the approximated
phase curve.
The straight-line asymptotic approximation for phase shift is more difficult to use because
there are many components in a transfer function such as integrators, differentiators, first order
Iags, first order leads, etc. as compared with the gain plot. It will be easier to calculate the phase
shifts for some frequencies that cover the whole range and plot these data and draw the phase
shift curve.

Figure 5.27 A straight-line asymptotic approximation of phase shift

 Frequenqr Response Ana[rsis I tOS

Example 5.5
A dynamic system is represented by the transfer function:

5
G(r)=
s+5
Draw the Bode diagram for this sysrem.

Solution

The break frequenry for this system is al = rad/s. A straightJine asymproric approximation
#=5
produces the gain curve as shown in Figure 5.28. Atlow frequency lower than the break frequency,
a = 5 radls, the gain is 0 dB. At high frequency, higher than 5 rad/s, the gain curve is a stiaight-
line with the slope of *2! dB/decade passing through point 5 rad/s at gain 0 dB. To ploJth.
phase shift curve, a few frequencies are selected. The frequencies of 0.5, 1, 5, 10, and 30 radls

::it::::
\r\: iiii:ii
rr::::
lGldB

rN
::::;:::
Frequency (radls)

-20

-3010
Frequency (rad/s)

Figure 5.28 Bode plots for Example 5.5

164 | Introduction to Control Engineering

may be selected so that the curve will cover the frequency range of the system. The expression for
phase shift is: For 6)<

Zc(jat)= -tan '(O.Za).

Thusthe respectivephaseshifts are-5.7o,-11.3", -45.0",-63.4,and-80.5o.TheBodeplots For (D >
for the system are shown in Figure 5.28. The straight-line approximation is shown by the bold
line.

The pha

Exercise 5.9 For a; <

Draw the Bode plots for the system with a transfer function:
For 6p;
G(s)=.+-
' l+0.25s
Use a straight-line asymptotic approximation for plotting the gain curve and use calculation to
plot the phase shift curve.
Figure i
by calcu
straight
5.4.4 First Order Lead
Let's study a simple first order lead element. The transfer function is: 5.4.5 I

G(r) =t+Ti. 6.65) Let's cor:

By substituting s withTat, the frequency response transfer function is:

G(ja)=t+7jco. (5.55)
The gain and phase shift are: where.

lc(i.,)l= r+lT.t,)) (5.67)

Gr(:
and
Let 11>
zG(jrlo) = tan-1 (Tr). (5.68)
function
The gain in dB is: 5.30. Sir
adding a
lc (ir)ldB = 2o bs (.ff (rr,,f ). (5.6e) shift can
the gain
For a straight-line asymptotic approximation, the gain in dB at low frequency and high
frequency approximations are as follows:
 Frequency Response Anallsis I t0S

I
T

lc(iqldB=1olog1=odB.
1

T
lc (iqldB = 10 log(Tro)' = 2otosT +2otosa.
The phase is:
ZC (1ot) = ran-' (rT).
For r0 <<117,
tG (ja) -+ oo.
For 6 >>llT,
zG(jro)-+ 90o.

Figure 5.29 shows the gain and phase shift curves for the system. The phase shift curve is obtained
by calculation. Note that at low frequency, the gain is 0 dB while at high frequency the gain is a
straight line with slope of 20 dB/decade passes through ,= .
|
5.4.5 Bode Plot of Gombination of severar Lead and Lag Elements
Lett consider a system with a transfer function that consists of lead and lag elemenrs given by:

(t+{s)(1+1')
G(s) =
(t+ {s)(t +T,s)$+r,s)
= G,(s)xG,(r)"G, (r)*G, (s)xG, (s) 6.lo)
where,

G(,)= G,(s) = G, (r) = . i-, - G,(t)=t+Tns, G, (r) =t+\s.

1+Trs' l+Trs' I+T-s
1
"

I,et T1> T4> Tr> Ts> 73. The asymptote gain curves for individual components in the transfer
function, i.e. lG1l, lGrl, lGtl, lGal, and lG5l arc plotted and shown by th. thin lines in Figure
5.30. Since logarithmic scale is used, the overall gain of the sysrem
lC(jat)lcan be determinJ by
addJng all the comPonen_ts. The resultant plot for the gain is shown by the thick line. The phasl
shift can be plotted by calculating the phase shifts by choosing several frequencies in the ,r.rg. of
the gain plot. The phase shift data is obtained from:

zc(jro)=-ran-l Tra-tanl Zrao-uln' \o+ran Tna+tan' 1

rra (5.1t)
166 | Intyoduction to Control Engineering

100 The g

lcldB 50
?,
, a
:+ 20 B/decade

There
1o-1 1oo 1lr 101 102 103 104
gain,.
Frequency (rad/s)

5.4.7
90 * " "i"'-i ::

The l
60
z.Go
i/
30 ''./ /
.4 By sul
.tttttF

nr
|i ,r-.-li-t
T;;i:i:l l::
i i :ii:iit ; i ;:;:::r i ::::ii;t i :ll;:::l
io-' 1oo 101 102 103 104
Frequency (rad/s)

Figure 5.29 Bode plots for a first order lead Thus,

and ir

The p

Figure 5.30 Plotting the gain curve

5.4.6 Effect of Gain K

Let's consider the system in Section 5.4.5 havin* the gain of K i.e.:

vrl-c\ t_ K(1+Tos)(t+4r)
u /
l=
(r+4s)(r+4i)(t +r,s)
= t<G (s) (5.t21
 Frequency Response Analysis | 167

The gain response in dB:

le Ur)ldB = 2o toslKc (ja)l

=20logK +2\logc(ja)
$.73)
= 2atosK + lc 17ar;lan.
Therefore, gain Kwill shift the gainlG(ja) | dB by 2OlogKdB at all frequencies. However, the
gain, K, has no effect on the phase shift since Kis not a function of ar.

5.4.7 lntegrator
The transfer function for an integrator is:

1
G (r) =;. (:5.t+1

By substituting s withTal, the frequency response transfer function is:

c(ja,)=l (5.75)
J0)
Thus, the gain is:

lc (ir)l=*, 6.te)
and in decibel, it can be written as:

lc (iqldB = -2orosal. (5.77)

The phase shift is -90o for all frequencies and Figure 5.31 shows the Bode plots for an integrator.

lGldB

Gain plot

-20 dB/decade

Phase shift plot

Figure 5.31 Bode plots for an integrator

168 | Introduction to Control Engineering

The gai
Exercise 5.10
Sketch the Bode plots for the systems given by the following transfer functions:
and in <

(l) G(s):
' ,(l+l0s)
''O
The ph
(2) G(s)= , 'O
\ / s(o.ts+t)
5.4.9
(3) G(s)= 3oo(s+5)
\/ s(s+10)(s+30) The tra

Use straightJine asymptotic approximation for plotting gain curves and use calculation for plotting
phase shift curves.
The fre

5.4.8 Differentiator
The transfer function for a differentiator is:

G(s)= 5. (5.ts) The er;

By substituting s withTro, the frequency response transfer function is:

G(jr)= j@. (5.7e)

T]
Phase shift plot
where r

90"

oj For hie.

40 dB

20 dB

0dB At brea
at= | a= 10

Figure 5.32 Bode plots for differentiator

 Frequency Response Anallsis I tOg

The gain is:

lc (1i.o)l= co (5.s0)
and in dB, it can be written as:

lc (ir)ldB = 2o togia'. (5.s t1

The phase shift is 90o for all frequencies. Figure 5.32 shows the
Bode plots for an integrator.

5.4.9 Second Order Lag

The transfer function of a second order lag is:

nc G(s)= , Fd -.
s'+2$o,s+ol' 6.82)
The frequency response transfer function, for K 1, can be written
= as:

G(ja)=

'-(tl.',tE) (5.83)

The expression for gain in dB is:

lc 11,,,71an- -,r,"r[[ r-EI

I
. *el] (s.s<1

The centre frequency for a second order system is the natural frequenry, arr. For low frequency
where a << 0)n, the system gain is:

lc (iqldB = -1olog 1 = o dB.

For high frequency where a >> o)n, the system gain is:

lctiqldB=-40,*[g ]
\@" )
At break frequency, i.e. (o = an, theactual gain is:

lcuql,,,,=20'.r[* ]
\'5 )
170 | Introduction to Control Engineering

The phase shift is: paper

muld
( .,* ) frequ
proce
Q:,a,
1#l
I l'.JJ
(5.85)

Exer
Thus, as al -+ 0, the phase shift, @ -) 0o; as rD -+ cD,, the phase shift @ + -90'; and as o) ) 6, A frer
the phase shift @ -+ -180o. diagn
Figure 5.33 shows the gain curve of the Bode plot for a second order system with varied Derir
damping ratios. The gain curve breaks at the natural frequency and decreases at the rate of *40
dB/decade at high frequency. This is expected since a second order system consists of two first
order lags in series with the same time constanr.
a (rat

lG( jt
tG(

Exer
In orc
The r

Figure 5.33 Bode plot for gain of second order system

Exercise 5.11 lc(ja

Sketch the Bode diagrams for the sysrems given in Exercise 5.4. ZG(

Exercise 5.12
Repeat Exercise 5.5 using Bode diagram.
5.5
5.4.10 Modelling using Bode Diagram Outc
Tlansfer function of a dynamic system can be estimated using experimental frequency response After
data. It is easier to use Bode diagram rather than Nyquist diagram because asymptoticallywe have
shown the gains of common elements in a transfer function will result with gradients of +20 dB/ (1) (

decade or its multiples. If two or more elements are in the transfer functions, the gain plot wili (2) (

have gradient of 0 or multiple of +20 dBidecade. The gain dataare plotted on a semi-log graph (3) (
 Frequenqt Response Analysis I tZ t

paper' A ruler is used, to produce straight lines that fitted the data that
produce gradients in
multiples of +20 dB/decade or 0 dB/decade. The intersections
of these lines are the break
frequencies that provide the information on the time consrants
for the sysrem. This is the inverse
process of plotting the Bode gain plot using straight line
asymptotic approximation.

Exercise 5.13
A. frequency response data obtained from an experiment is shown in T"ble 5.4. Plot the Bode
of the sysrem. Prior knowledge indicates that the system is a second
{iagram order system.
Derive the transfer function of the sysrem from the gain plot.

Table 5.4 Experimental frequency response dara

a; (radls) 10 20 60 100 200

lG j@lde 20
( 20 19 t4 B
-11 -20 -32
zG (j(D)" -14 *28 -59 _r73
-90 -120 -r57 -166
Exercise 5.14
In order to determine the transfer function ofa plant, G(s), a frequency
response resr was conducted.
The results are as shown in Thble 5.5. Determine the transfer
fun.,ion of .hl, plant.

Thble 5.5 Frequency response dara for Exercise 5.14

l0 20 30 60 BO r00
lc(j..l0)l 7.847 3.719 1 .569 0.863 0.359 0.109 0.0550.020 0.013 0.01 1
zG(ja)" - 100 -109 *123 -132 -139 -r39 _r33 -1 19
-tt3 _109

5.5 STABILITY ANALYSIS USING BODE DIAGRAM

Outcomes:
After completing this section, the reader should be able to:

(1) determine the stability of a feedback system based

on frequency response dara,
determine the gain and phase *rrgin, from Bode dl"g.r;',
l?l
(3) determine the value of constant controller gain for
,riUitry "rd
from Bode diagram.
172 | Introduction to Control Engineering

5.5.1 Method
Bode diagram is easier to use for system design. \fhen a proportional controller is used to fulfil
the phase and gain margins' specifications or to determine the value of proportional gain for
critical stabiliry the value of the gain can be determined easily. The effect of proportional value
on logarithmic scale is the shift of 20log Kfor all frequencies. Therefore, 0 dB axis can be shifted
vertically by the value of 20log K
Consider a feedback system as shown in Figure 5.34.The stabiliry of the feedback sysrem
can be analysed by plotting the frequency response data of the open-loop sysrem, i.e.
KG(ja)H(7al). Based on the Bode diagram (Figure 5.35), we identifi. two frequencies, that is
the frequenry when the open-loop gain is 0 dB, let this frequenry be als6, and the other is the
frequency when the phase shift is *180o and let the frequency be (o_1s6..

Examp
Consider
determil
Figure 5.34 A feedback system

Solutio
A feedback system is stable if at6as < (D-ra6.. Conversely if at66s ) the feedback
o)-180.,
system is unstable and the Bode diagram is shown in Figure 5.35. Critical stability can be identified
Since the
if ahae = (0-180". Referring to Figure 5.35, the gain margin is shown by GM thatcorresponds to
-lKc(ja)H(ja)l dB when the phase shift of zKG(ja)H(jco) is -180o. The phase margin is and Tr=
indicated by yat frequency (r;66.
The gain
the phas<
Ga:nlKG(ja)A(jr,t)l
values for

which arr
Phase shift t KG(j'g) H(j a) shown in
The
-1 800

Figure 5.35 A stable closed-loop system

 Frequency Response Analysis I tZg

tbr
rlu"
ftr,i

[ern
Phase shili
i.e.
rrL
d-re

Figure 5.36 An unstable closed-loop system

Example 5.6
Consider the feedback system in Exercise 5 .6. If K = 1 , draw the Bode diagram for
the system and
determine the phase and gain margins. Determine the value of Kfor criJcal stabiliry.

Solution
rck
ied
Since the dme constants for the first order lags in the open-loop transfer
to function are T1= 0.1 s

ris and T2 = O.O2s, hence the break frequencies are a-1= : 10 radls and @r= = 50 radls.
# #
The gain curve can be determined using the straight-line asymptotic approximation.
To produce
the phase shift curve, a few frequencies are chosen, such as 4, I0,30,
60, and 200 radls. The
values for phase shift can be calculated from the phase shift expression:

ZKG (ja) U (jr) = -90 - tan-' 0.lar - tan-, 0.02a

which are of -1 16, -146, -192",
-220", .and..-253" respectively. Based on the Bode diagram
shown in Figure 5.37 , we can deduce that the phase margin is 43; and gain margin is 16 dB.
The maximum value for K can be determined direc-tly from the ,i,r.
of g;r, margin as:

20log K= 16,
of
l6
K =10i,
174 | Introduction to Con*ol Engineering

therefore, Exercir
K= 5.3.
Repeat I
Exercis
Repeat I

IGllldB

Further
Bode, H
-5 Bel
Brocken
77a
IKGH
Brocken
Tia
-90"
Ferreira,
Tra
*1 90"
Nise, N.
Yeunp I
-270" E(:
Frequency

Figure 5.37 Bode diagram for Example 5.6

Exercise 5.15
A feedback control system is shown in Figure 5 .38.If K = 1 , draw the Bode diagram and determine
the phase and gain margins for this system. Calculate the limit of Kfor stabiliry.

s(r+2.5s)(t+0.25s)

Figure 5.38 A feedback system for Exercise 5.15

 Frequencl Response Analysis I 17 5

Exercise 5.16
Repeat Exercise 5.7 using Bode diagram.

Exercise 5.17
Repeat Exercise 5.8 using Bode diagram.

Further Readings
Bode, H. \( 1940. Relations between Attenuation and Phase in Feedback Amplifier Design.
Bell System Technical Journal. 19: 421-454.
Brockett, R. \( and J. L. 'X/illems. 1965. Frequency Domain Stabiliry Criteria part I. IEEE
Trans. Automatic Control. AC(10): 255-26I.
-
Brockett, R. \M and J. L. \7illems. 1965. Frequency Domain Stabiliry Criteria Part II. IEEE
Trans. Automatic Control. AC(10): 407-413.
-
Ferreira, P. 1999. Concerning the Nyquist Plots of Rational Functions of NonzeroType. IEEE
Transaction on Education. 42(3) : 228-229 .
Nise, N. s. 2004. control systems Engineering. Hoboken, NJ:
John \x/iley & Sons.
Yeung, K. S. 1985. A Reformulation of Nyquist's Criterion. IEEE Tial.ns. Educ., Feb. 1985.
E(28): 58-60.
 CHAPTE,R 6

CONTROL SYSTEM DESIGN

6.1 DESIGN BASIS

Outcomes:
After completing this section, the reader should be able to:

(1) describe the design specification of a control system, and

(2) describe controllert configuration.

6.1.1 Design Specification

A feedback control system is designed to fulfil a unique specification based on the process and its
application. Normally the specificarion covers ,rp..t, of,

(1) relative stabiliry

(2) steady-state accuracy,
(3) transienr response, and
(4) frequenryresponsecharacteristic.

In other applications, the specification used is sensitivity toward,s paramerer variarions (that is,
robustness or disturbance rejection).
The design of a control system can be done in time domain or frequency domain. For
example, the steady-state accuracy is defined based on srep, ramp, or parabolic inputs.
The design
178 | In*oduction to Connol Engineering

to satisfy this specification is easier to be done in time domain. Other specifications such as
maximum overshoot, rise time, and settling time are also based on dme domain. The measures
of relative stabiliry are phase and gain margins. These specifications are evaluated in frequency
domain.

6.1.2 Controller Gonfiguration

Consider a dynamic system represented by a block diagram as shown in Figure 5.1. The design
objective is to ensure the controlled variable that is the output variable,/(r), follows the desired
response. This can be done by choosing suitable control signal represented by the input variable
to the process, z(r).

Input variable Output variable

Figure 6.1 A dynamic system

Normally, a conventional controller design uses a fixed controller configuration. The designer
decides the overall system configuration by positioning the controller relative ro the position of
the process. The next problem is determining the controller parameters that will satisfy the design

(a)

Figure 6

(b) specifica
characte.
Figure 6.2 Controller configurations: (a) series compensation, (b) feedback compensation comPens
 Control S\stem Design I tZg

Figure 6.2 (cont'd) Controller configurations: (c) state-feedback, (d) series-feedback compensation,
(e)forward and series compensation, (f) feedforward compensation

specifications. Since the design process involves modi$ring or

compensating the performance
characteristic of the sysrem, thus the design based on fii.d
corrfrg.,rrtiorr"i, 1* krro*., ,,
compensation. some compensation configurations are shown in
Figrlre 6.2.
180 | Introduction to Control Engineering

Compensators in Figures 5.2(a)to (c) have one degree of freedom. The weakness of this 6.2.1
configuration is that the design criteria that can be satisfied are limited. For example, if relative
stabiliry is satisfied, possibly the sensitivity towards parameter variation may not be satisfied. Steadr-
Configurations in Figures 6.2(d) to (f) have two degrees of freedom. an inter
The most widely used controller in industries is the PID (proportional plus integral plus If this ir
derivative) controller using series compensation configuration. Besides, lead, lag, and lead-lag of the s

controllers are also used. C{

6.1.3 Basic Design Principles

In most situations, designers will choose the simplest controller structure. A complex controller
is expensive, less reliable, and difficult to design. Once the controller configuration and strucrure
are chosen, its parameters have to be determined. Time domain design uses root locus method in
the s-plane. \X4rile designing in frequency domain, the phase and gain of the open-loop frequency
response data are manipulated to satisfy the specifications. In general, guidelines for controller Let the
design in time domain and frequency domain are as follows:

(1) Complex conjugate poles of the closed-loop transfer function results with underdamped
response. If the values of all the poles are real, the response is overdamp ed. Zeros of closed-
T1
loop system results with overshoot although the response is overdamped.
respoI1!
(2) The response of the system is dominated by poles nearest to the origin of the s-plane. This
is known as dominant poles. The transient response of poles further from the origin fades
faster.
(3) Dominant poles further from the origin result with faster response.
(4) If the poles and zeros are near cancellation, the effect of the poles on the response is
small.

The time domain specification is loosely related to the frequency domain specification. The
rise time is inversely proportional to the bandwidth. Phase margin, gain margin, and resonance
amplitude are inversely propotional to the damping ratio.

6.2 P. AND PI.CONTROLLERS

Outcomes:
After completing this section, the reader should be able to:

(1) describe the efFects of integral gain on the transient response based on roor locus, and
(2) design P- and Pl-controllers.
 Control System Design I tgt

6.2.1 Design Procedure

Steady-state error can be eliminated by inserting an open-loop pole at the origin, that
is inserting
an integrator to the open-loop system. This makes Class 0 ,yrt.- be.o*.", a Class 1
system.
If this is done, the steady-state error will be eliminated but it will change the transienr response
of the system.
Consider a feedback control system using P-controller as shown in Figure 6.3.

Figure 6.3 A feedback control system using p_controller

Let the transfer function of the process be:

G(s) =
(s+ 1)(s+ 2)g + Q' (5.1)

The root locus plot is shown in Figure G.4. Let say, K is chosen to satis$, the transient
response specified by the damping ratio, ( = 0.707, and take nore the corresponding
time consranr

Figure 6.4 Root locus for system in Figure 6.3

lB2 | Introduction to Control Engineering

and natural frequency. The dominant closed-loop pole shown in Figure 6.4 is denoted by,4. This
is a Class 0 system. Choosing an appropriate value of ( the response satisfizing ( = O.lOl can be
obtained but there exists steady-state error if excited with a step input. In Chapter 3, it has been
stated that if an integrator is inserted as shown in Figure 6.5, the steady-state errorwill be eliminated.

Figure 6.5 System with integralcontroller

Figure 6.6 shows the root locus of the system compensated with an integral controller. For
the same specification, e = 0.707, point" A' is located but this location indicates that the time
constant and the natural frequency are different, thus may not satisfy the speed of response T
specification. This problem can be solved using the Pl-controller as shown in Figure 6.7.The Comp;
transfer function for PI compensator can be written as: are a]rr
transi;:

Exerc
Consic

(1) L
1

r-,1

/1\ -:

itl
:-
(3: ^-
(_

j-.
lL (

Figure 6.6 Root locus for system compensated with l-controller

 Control System Design I t8g

Figure 6.7 Feedback system with pl-controller

G(,)=t[t.+)
_ K(s+ K,) (6.2)
-t

The root locus for the system compensated with PI controller is shown in Figure 6.8.
Comparing the root locus in Figure 6.8 with rhat in Figure 5.4 shows that pointr4 and.ioint A'
are almost located at the same position. Clearly Pl-conuoller is able to prod,r..
the desired
transient response as well as able to eliminate the steady-state error.

Exercise 6.1
Consider rhe sysrem in Figure 6.3.

(1) UsingMAILAB,obtainthevalue of Kfordamping ratio,(=O.TOT.Calculatethemaximum

\W4rat is
Percentage overshoot. the natural frequency of the system? Plot the rime response
of this sysrem if it is excited with a step input,ig"ri.
(2) If l-controller is used, what is thevalue of Kfor damping ratio, (= 0.707? calculate the
maximum percentage overshoot. \What is the natural frequency of ih. ryrt.m? plot
the time
response of this sysrem if it is excited with a step inpur ,lgrr"l.
(3) If Pl-controller is used, what are the values oirrr"d i,fo, damping ratio, ( 0.707?
=
Calculate the maximum percentage overshoot. \7hat i, th. rr*rri frequency
of the
system?
(4) Superimpose all the three output responses obtained above and make
comparison. Discuss
the effect of K;on the root locus and the transienr response.
LU I Introduction to Control Engineering

6.3
6
Outcc
4
After c,
1= 0.707
2 (1) dt
A7 ar
h
(n
'5o
(2) d.
E

6.3.1
-2
The rr;
-4 s-plane
determ
loop p,
-6 .2
-6 -4 -2 0 4 6 locus .',
Real axis
in the t

the ibr
Figure 6.8 Root locus for system compensated with Pl controller

Exercise 6.2
Consider a unity feedback control system with the process given by the following mansfer function:

s+6 g.here
G(s\=
\ / (s+1)(s+2)(s+4) -

(propc
follo",:
(1) If P-controller G.(s) = K is used:
C
tfan.--
G) Sketch the root locus diagram.
(b) Calculate the time constant and the controller parameter if the required damping ratio
is 0.866.
k) Show that steady-state error exists if the system is excited with a step input function.
T
1

respo:l
(2) If I-controll er, G, ={, i, used, show that:
-t
the dc,.
G) The speed of response is slower. ri rth ii
(b) The steady-state error is zero. The s:
br- a ze
(3) Show that Pl-controller will make the response faster. speci:-"
 Control Srystem Design I tAS

6.3 PD.CONTROLLER
Outcomes:
After completing this section, the reader should be able to:

(1) describe the effect of derivative gain on the transient response based on roor locus diagram,
and
(2) design PD controller.

6.3.1 Design Procedure

The transient response can be specified by choosing the position of the closed-loop poles on the
s-plane. If this position is on the locus, the transient response specification is easily satisfied by
determining the gain value that will place the poles at the desired position. If the desired .losej-
loop pole location is not on the locus, compensator is required to modi$, the locus so that the
locus will pass through the desired closed-loop pole. This can be done by adding poles and zeros
in the forward path. One successful method to increase the speed of response ii to add zeros in
the forward path. This can be done using:

G,=Krs+K

=r(,*L\
'l, K, ) (6.3)

where Kis the proportional gain and KTis thederivative gain. This compensator is known as pD
(p_roportional plus derivative)-controller. The effect of improved speed can be illustrated
by the
following example.
Consider a con&ol system with a proportional controller as shown in Figure 6.3 with the
transfer function of the process given by:

l-
G(s) =
(s+1)(s +2)(s+5)
The root locus is shown in Figure 6.9. Letsay the sysrem is designed to satisfy the transient
response specification_of damping ratio, ( = 0.4. From the root locus diagram, the position
of
the dominant closed-loop_pole is at point A.Thevalue of the dominarrt-pol.s r.. ! 2.lj
-0.9
with the value of K= 23. This value for Kis obtained using the same as in Examp le 4:-i.
-.thod
The speed of response is the time constant I10.9, that is 1. 1 s. Let say this system is compensated
by a zeros ofvalue IJKa= 3. This root locus is shown in Figure 6.10. For the same damping ratio
specification, the value of the dominant closed-loop pol.s ar.
-2.3 + 5.5j at A, with K) = 35.
186 | Introduaion to Control Engineering

Simila
the res
impro'

Exerr
Use \l
to r-ali
ur = d,.eo

Don-l
a1:
Settlr

For ea'
erro rs.
Figure 6.9 Root locus for system with P-controller
Exerr
Consi
;\

I rlT=2
.T= 1.43 s
and th

u'ith
desisr
effec:
r-

Figure 6.10 Root locus for the system with PD-controller

 Control Syaem Design I tAZ

Similarly, the value for Ksis obtained using the same method as in Example 4.l3.The speed of
the response is now 112.3 or 0.43 s. Clearly this example shows that the speed of response can be
improved by using PD-controller.

Exercise 6.3
Use MAfLAB to validate the data in the example discussed above. Use MATLAB and SIMULINK
to validate the data in Thble 6.1 for the system compensared with the given I{/Ka.

Thble 6.1 tansient response data

G.(s) K= 23.7t8 I{/K, = 2 I{/Kd = 3 K/Kd = 4

Dominant poles -0.939x2.15rj -3*5.874j -2.437 t5.583j -r.8694.282j
a)n 2.347 7.5 6.901 4.673
Settling time 4.26 r.33 1.64 2.14

For each case, calculate the gain Kand Ks, percentage overshoots, time constants, and steady-state
errors.

Exercise 6.4
Consider a process with the following transfer function:

- \-'l=
G(-,) , '
(s+1)(s +2)(s+5)'
and the controller:

C,(r)=Kd(s+KlKd),
with unity feedback. Using MAILAB or otherwise, obtain the time consranrs if the system is
designed to satisfy damping ratio of 0.4 for the values of l{/K;given in Thble 6.2. Discuss the
effect of l(/Kson the speed of response.

Table 6.2 Thble for Exercise 6.4

IlKa 2 2.5 3.0 3.5 4.0

1BB I Introduaion to Control Engineering

Further Readings
Astrom, K. J. and T. Haaglund. 1988. Auromatic Tuning of PID Controllers. Research Thiangle
Park, NC: Instrument Society ofAmerica.
Bailey, F. N. and S. Meshkat. 1983. Root Locus Design of a Robust Speed Control. Proc.
Incremental Motion Control Symposium, June 1983. 49-54.
Gopal, M.2003. Control Systems: Principles and Design. Singapore: McGraw-Hill.
Smith, H. \fl and E. J. Davison. 1972. Design of Industrial Regulators. Proc. IEE. 119: l2lO-
r2t6.
Van de Vegte, J. 1994. Feedback Con*ol Systems. Englewood Cliffs, NJ: Prentice-Hall.
\fillems, J. C. and S. K. Mitter. 197l. Controllabiliry, Observabiliry Pole Allocation, and State
Reconstruction- IEEE Thans. Automatic Control. AC(16): 582-595.
Ziegler, J. G. and N. B. Nichols. 1943. Process Lags in Automatic Control Circuits. ASME
Ti'ans. 65: 433-444.

A
absolur
acceler,
,
(

,a,l.rr,o
amplin
angle o
angle c
aslTnpi
autom;
aLrxiliar

B
block c
11.
E

:
r

Bode d
1-0