Control Systems

Lecture 1
Introduction

1
 Course information
 Required Textbook:
 Modern Control Systems, Richard C. Dorf and Robert H.
Bishop, Prentice Hall, 12th edition, 2010, ISBN-10: 0-13-
602458-0

2
 Math Prerequisites
 Complex Numbers
 Add, Subtract, Multiply, Divide
 Linear Algebra
 Matrix Multiply, Inverse, Sets of Linear Eq.
 Linear Ordinary Differential Equations
 Laplace Transform to Solve ODE’s
 Linearization
 Logarithms
 Modeling of Physical Systems
 Mechanical, Electrical, Thermal, Fluid
 Dynamic Responses
 1st and 2nd Order Systems of ODE’s

3
 Prerequisites: Complex Numbers
 Ordered pair of two real numbers

 Conjugate
 Addition

 Multiplication

4
 Complex Numbers
 Euler’s identity

 Polar form
 Magnitude
 Phase

5
 Logarithm
 The logarithm of x to the base b is written
 The logarithm of 1000 to the base 10 is 3, i.e.,


 Properties:

Why?

6
 Summary & Exercises
 Prerequisites
 Complex numbers, Linear Algebra, Logarithm,
Laplace transform
 Dynamics
 Next
 Introduction
 Exercises
 Buy the course textbook and keep it!
 Review today’s slides on complex numbers and
logarithm
 Read Chapter 1 and 2 of the textbook.

7
 What is “Control”?
 Make some object (called system, or plant)
behave as we desire.
 Imagine “control” around you!
 Room temperature control
 Car/bicycle driving
 Voice volume control
 “Control” (move) the position of the pointer
 Cruise control or speed control
 Process control
 etc.

2017-2018 8
 What is “Control Systems”?
 Why do we need control systems?
 Convenient (room temperature control, laundry
machine)
 Dangerous (hot/cold places, space, bomb removal)
 Impossible for human (nanometer scale precision
positioning, work inside the small space that human
cannot enter)
 It exists in nature. (human body temperature control)
 Lower cost, high efficiency, etc.
 Many examples of control systems around us

2017-2018 9
 Open-Loop Control
 Open-loop Control System
 Toaster, microwave oven, shooting a basketball

Signal Input input output

Controller Plant
(Actuator)

 Calibration is the key!

 Can be sensitive to disturbances
2017-2018 10
 Example: Toaster
 A toaster toasts bread, by setting timer.

Setting of timer Toasted bread

Toaster

 Objective: make bread golden browned and crisp.

 A toaster does not measure the color of bread during the
toasting process.
 For a fixed setting, in winter, the toast can be white and in
summer, the toast can be black (Calibration!)
 A toaster would be more expensive with sensors to
measure the color and actuators to adjust the timer based
on the measured color.
2017-2018 11
 Example: Laundry machine
 A laundry machine washes clothes, by setting a
program.
Program setting Washed clothes
Machine

 A laundry machine does not measure how clean

the clothes become.
 Control without measuring devices (sensors) are
called open-loop control.

2017-2018 12
 Closed-Loop (Feedback) Control
 Compare actual behavior with desired behavior
 Make corrections based on the error
 The sensor and the actuator are key elements of
a feedback loop
 Design control algorithm
Signal Input Error output

Controller Actuator Plant

+
-

Sensor
2017-2018 13
 Ex: Automobile direction control
 Attempts to change the direction of the automobile.
Steering
Desired Error
wheel Direction
direction
Brain Hand angle Auto

Eye

 Manual closed-loop (feedback) control.

 Although the controlled system is “Automobile”, the
input and the output of the system can be
different, depending on control objectives!

2017-2018 14
 Ex: Automobile cruise control
 Attempts to maintain the speed of the automobile.
Disturbance
Error
Desired Acceleration
speed Speed
Controller Actuator Auto

Sensor

 Cruise control can be both manual and automatic.

 Note the similarity of the diagram above to the
diagram in the previous slide!
2017-2018 15
 Basic elements in feedback control
systems
Error Disturbance
Reference Input Output
Controller Actuator Plant

Sensor

Control system design objective

To design a controller s.t. the output follows
the reference in a “satisfactory” manner
even in the face of disturbances.
2017-2018 16
Systematic controller design process
Disturbance
Reference Input Output
Controller Actuator Plant

Sensor

4. Implemenation
1. Modeling

Controller Mathematical model

2. Analysis
3. Design
2017-2018 17
 Goals of this course
To learn basics of feedback control systems
 Modeling as a transfer function and a block diagram
• Laplace transform (Mathematics!)
• Mechanical, electrical, electromechanical systems
 Analysis
• Step response, frequency response
• Stability: Routh-Hurwitz criterion, (Nyquist criterion)
 Design
• Root locus technique, frequency response technique,
PID control, lead/lag compensator
 Theory, (simulation with Matlab), practice in laboratories
2017-2018 18
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• mechanical • Bode plot
• electrical
PID & Lead-lag
• electromechanical Stability
• Routh-Hurwitz
Design examples
Linearization • (Nyquist)

(Matlab simulations &) laboratories

2017-2018 19
 Goals of this course
To learn basics of feedback control systems
 Modeling as a transfer function and a block diagram
• Laplace transform (Mathematics!)
• Mechanical, electrical, electromechanical systems
 Analysis
• Step response, frequency response
• Stability: Routh-Hurwitz criterion, (Nyquist criterion)
 Design
• Root locus technique, frequency response technique,
PID control, lead/lag compensator
 Theory, (simulation with Matlab), practice in laboratories
Fall 2008 20
 Laplace transform
 One of most important math tools in the course!
 Definition: For a function f(t) (f(t)=0 for t<0),

(s: complex variable)

f(t)

t F(s)
0

 We denote Laplace transform of f(t) by F(s).

21
 Examples of Laplace transform
 Unit step function f(t)
1
t
0
(Memorize this!)

f(t)
 Unit ramp function
t
0

(Integration by parts)
22
 Integration by parts

EX.

23
Examples of Laplace transform (cont’d)
f(t) Width = 0
 Unit impulse function Height = inf
Area = 1

t
0
(Memorize this!)

f(t)
 Exponential function
1
t
0

24
Examples of Laplace transform (cont’d)
 Sine function
(Memorize these!)

 Cosine function

Remark: Instead of computing Laplace

transform for each function, and/or
memorizing complicated Laplace transform,
use the Laplace transform table !

25
Laplace transform table

26
Inverse Laplace Transform
 Properties of Laplace transform
1. Linearity

Proof.

Ex.

27
 Properties of Laplace transform
2.Time delay

f(t) f(t-T)
Proof.

0 T

Ex.

t-domain s-domain 28
 Properties of Laplace transform
3. Differentiation

Proof.

Ex.

t-domain s-domain

29
 Properties of Laplace transform
4. Integration

Proof.

t-domain s-domain

30
Properties of Laplace transform
5. Final value theorem
if all the poles of sF(s) are in
the left half plane (LHP)

Ex.

Poles of sF(s) are in LHP, so final value thm applies.

Ex.

Some poles of sF(s) are not in LHP, so final value

thm does NOT apply.
31
Properties of Laplace transform
6. Initial value theorem
if the limits exist.

Remark: In this theorem, it does not matter if

pole location is in LHS or not.

Ex.

Ex.

32
Properties of Laplace transform
7. Convolution

Convolution

IMPORTANT REMARK

33
 Summary & Exercises
 Laplace transform (Important math tool!)
 Definition
 Laplace transform table
 Properties of Laplace transform
 Next
 Solution to ODEs via Laplace transform
 Exercises
 Read Chapters 1 and 2.
 Solve Quiz Problems.

34
 ME451: Control Systems

Lecture 3
Solution to ODEs via Laplace transform

35
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
Transfer function • Transient
• Steady state
Root locus
Models for systems
• electrical Frequency response
Frequency domain
• mechanical • Bode plot
• electromechanical
PID & Lead-lag
Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

36
 Laplace transform (review)
 One of most important math tools in the course!
 Definition: For a function f(t) (f(t)=0 for t<0),

(s: complex variable)

f(t)

t F(s)
0

 We denote Laplace transform of f(t) by F(s).

37
An advantage of Laplace transform
 We can transform an ordinary differential
equation (ODE) into an algebraic equation (AE).
t-domain s-domain

ODE AE
1
2

Partial fraction
Solution to ODE expansion
3

38
 Example 1
ODE with initial conditions (ICs)

1. Laplace transform

39
 Properties of Laplace transform
Differentiation (review)

t-domain

s-domain
40
 Example 1 (cont’d)
unknowns
2. Partial fraction expansion

Multiply both sides by s & let s go to zero:

Similarly,

41
 Example 1 (cont’d)
3. Inverse Laplace transform

If we are interested in only the final value of y(t), apply

Final Value Theorem:

42
 Example 2

 S1

 S2

 S3
43
 In this way, we can find a rather
complicated solution to ODEs easily by
using Laplace transform table!

44
 Example: Newton’s law
M

We want to know the trajectory of x(t). By Laplace transform,

(Total response) = (Forced response) + (Initial condition response)

45
 EX. Air bag and accelerometer
 Tiny MEMS accelerometer
 Microelectromechanical systems (MEMS)

(Pictures from various websites) 46

Ex: Mechanical accelerometer

47
Ex: Mechanical accelerometer (cont’d)
 We would like to know how y(t) moves when unit
step f(t) is applied with zero ICs.
 By Newton’s law

48
Ex: Mechanical accelerometer (cont’d)
 Suppose that b/M=3, k/M=2 and Ms=1.
 Partial fraction expansion

0
 Inverse Laplace transform
-0.1

-0.2

Amplitude -0.3

-0.4

-0.5
0 2 4 6 8 10
Time [sec]
49
 Summary & Exercises
 Solution procedure to ODEs
1. Laplace transform
2. Partial fraction expansion
3. Inverse Laplace transform
 Next, modeling of physical systems using
Laplace transform
 Exercises
 Derive the solution to the accelerometer problem.
 E2.4 of the textbook in page 135.

50
ME451: Control Systems

Lecture 4
Modeling of electrical systems

51
 Course roadmap
Modeling Analysis Design
Laplace transform
Time response
Transfer function Design specs
• Transient
• Steady state
Models for systems Root locus
• electrical
Frequency response
• mechanical Frequency domain
• Bode plot
• electromechanical
PID & Lead-lag
Stability
Block diagrams
• Routh-Hurwitz
Design examples
• Nyquist
Linearization

(Matlab simulations &) laboratories

52
Controller design procedure (review)
Disturbance
Ref. Input Output
Controller Actuator plant

Sensor

4. Implemenation 1. Modeling
Controller Mathematical model
2. Analysis
3. Design
 What is the “mathematical model”?
 Transfer function
 Modeling of electrical circuits
53
 Mathematical model
 Representation of the input-output (signal)
relation of a physical system
Physical Output
Input
system
Modeling

Model

 A model is used for the analysis and design of

control systems.

54
 Important remarks on models
 Modeling is the most important and difficult task
in control system design.
 No mathematical model exactly represents a
physical system.

 Do not confuse models with physical systems!

 In this course, we may use the term “ system” to
mean a mathematical model.
55
 Transfer function
 A transfer function is defined by

 A system is assumed to be at rest. (Zero initial

condition)

56
 Impulse response
 Suppose that u(t) is the unit impulse function
and system is at rest.

System

 The output g(t) for the unit impulse input is

called impulse response.
 Since U(s)=1, the transfer function can also be
defined as the Laplace transform of impulse
response:
57
 Models of electrical elements:
(constitutive equations)
Resistance Inductance Capacitance
i(t) i(t) i(t)

v(t) R v(t) L v(t) C

Laplace
transform

58
 Impedance
 Generalized resistance to a sinusoidal
alternating current (AC) I(s)
I(s) V(s)
 Z(s): V(s)=Z(s)I(s) Z(s)

Element Time domain Impedance Z(s)

Resistance

Inductance

Capacitance

59
Memorize!
 Kirchhoff’s Voltage Law (KVL)
 The algebraic sum of voltage drops around any
loop is =0.

60
 Kirchhoff’s Current Law (KCL)
 The algebraic sum of currents into any junction
is zero.

61
 Impedance computation
V1(s) V2(s)
 Series connection

Z1(s) Z2(s)
I(s)
V(s)
 Proof (Ohm’s law)

62
 Impedance computation
 Parallel connection
I1(s)
I(s) Z1(s)

Z2(s)
 Proof (Ohm’s law) I2(s)
V(s)

 KCL

63
 Modeling example
i(t) R1

Input v1(t) R2 v2(t) Output

C

 Kirchhoff voltage law (with zero initial conditions)

 By Laplace transform,

64
 Modeling example (cont’d)
i(t) R1

Input v1(t) R2 v2(t) Output

C

 Transfer function

(first-order system)
65
 Example: Modeling of op amp
If(s)
Zf(s) Rule1: i-=0
I(s) Rule2: vd=0
i-
Zi(s) -
Input Vi(s) vd +
Vo(s) Output

 Impedance Z(s): V(s)=Z(s)I(s)

 Transfer function of the above op amp:

66
 Modeling example: op amp
R2 C
i-=0
i(t) R1 i- Vd=0
-
vd +
Input vi(t) vo(t) Output

 By the formula in previous two pages,

(first-order system)
67
 Modeling exercise: op amp

C1 C2

R2 i-=0
R1
i- Vd=0
-
Input vi(t) vd +
vo(t) Output

Find the transfer function!

68
 More exercises
 Find a transfer function from v1 to v2.

 Find a transfer function from vi to vo.

69
 Summary & Exercises
 Modeling
 Modeling is an important task!
 Mathematical model
 Transfer function
 Modeling of electrical systems
 Next, modeling of mechanical systems
 Exercises
 Do the problems in page 19 of this lecture note.

70
ME451: Control Systems

Lecture 5
Modeling of mechanical systems

71
 Course roadmap
Modeling Analysis Design
Laplace transform
Time response
Transfer function Design specs
• Transient
• Steady state
Models for systems Root locus
• electrical
Frequency response
• mechanical Frequency domain
• Bode plot
• electromechanical
PID & Lead-lag
Stability
Block diagrams
• Routh-Hurwitz
Design examples
• Nyquist
Linearization

(Matlab simulations &) laboratories

72
 Time-invariant & time-varying
 A system is called time-invariant (time-varying)
if system parameters do not (do) change in time.
 Example: Mx’’(t)=f(t) & M(t)x’’(t)=f(t)
 For time-invariant systems:

Time shift Time shift

Sys

 This course deals with time-invariant systems.

73
 Newton’s laws of motion
 1st law:
 A particle remains at rest or continues to move in a
straight line with a constant velocity if there is no
unbalancing force acting on it.
 2nd law:
 : translational

 : rotational

 3rd law:
 For every action has an equal and opposite reaction
74
 Translational mechanical elements:
(constitutive equations)
Mass Spring Damper
f(t) f(t)
M x2(t) B x2(t)
K

f(t) f(t) f(t) x1(t)

x(t) x1(t)

75
Mass-spring-damper system

K B

x(t)

76
 Free body diagram
Direction of actual force will be
automatically determined by the
relative values!

K B

 Newton’s law: F=ma

77
 Mass-spring-damper system

K B

x(t)

 Equation of motion

 By Laplace transform (with zero initial conditions),

(2nd order system)

78
 Gravity?
M
K
K

 At rest,
 y coordinate:
 x coordinate:

79
Automobile suspension system
automobile M1
x1(t)
suspension K1 B

wheel M2
x2(t)
f(t)
tire K2

80
Automobile suspension system

Laplace transform with zero ICs

Block diagram
F X2 X1
G2 G1

G3

81
 Rotational mechanical elements
(constitutive equations)
Moment of inertia Rotational spring Friction
torque
B
K
J

rotation angle

82
Torsional pendulum system Ex.2.12

J B

friction between
bob and air

83
 Torsional pendulum system

K friction between
bob and air
J B

 Equation of Motion

 By Laplace transform (with zero ICs),

(2nd order system)

84
 Example

 By Newton’s law

 By Laplace transform (with zero ICs),

85
 Example (cont’d)
 From second equation:

(2nd order system)

 From first equation:

(4th order system)

Block diagram
G2 G1

86
 Rigid satellite Ex. 2.13

Thrustor

• Broadcasting
• Weather forecast
• Communication Double
• GPS, etc. integrator

87
 Summary & Exercises
 Modeling of mechanical systems
 Translational
 Rotational
 Next, block diagrams.
 Exercises
 Derive equations for the automobile suspension
problem.

88
 Exercises (Franklin et al.)
 Quarter car model: Obtain a transfer function
from R(s) to Y(s).
y(t)
M2 Answer

Ks B

M1
x (t)
Kw
Road surface

89
 ME451: Control Systems

Lecture 7
Linearization, time delays

90
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

91
 What is a linear system?
 A system having Principle of Superposition

System

A nonlinear system does not satisfy

the principle of superposition.
92
 Linear systems
 Easier to understand and obtain solutions
 Linear ordinary differential equations (ODEs),
 Homogeneous solution and particular solution
 Transient solution and steady state solution
 Solution caused by initial values, and forced solution
 Add many simple solutions to get more complex
ones (use superposition!)
 Easy to check the Stability of stationary states
(Laplace Transform)

93
 Why linearization?
 Real systems are inherently nonlinear. (Linear
systems do not exist!) Ex. f(t)=Kx(t), v(t)=Ri(t)
 TF models are only for linear time-invariant (LTI)
systems.
 Many control analysis/design techniques are
available for linear systems.
 Nonlinear systems are difficult to deal with
mathematically.
 Often we linearize nonlinear systems before
analysis and design. How?
94
 How to linearize it?
 Nonlinearity can be approximated by a linear
function for small deviations around an
operating point
 Use a Taylor series expansion

Linear approximation

New coordinate
Nonlinear function

Operating point

Old coordinate 95
 Linearization
 Nonlinear system:
 Let u0 be a nominal input and let the resultant
state be x0
 Perturbation:
 Resultant perturb:
 Taylor series expansion:

96
 Linearization (cont.)

notice that ; hence

N. sys L. sys.
97
Linearization of a pendulum model
 Motion of the pendulum

 Linearize it at
 Find u0
 New coordinates:
98
Linearization of a pendulum model (cont’)

 Taylor series expansion of

99
 Time delay transfer function
 TF derivation

(Memorize this!)

 The more time delay is, the more difficult to

control (Imagine that you are controlling the
temperature of your shower with a very long
hose. You will either get burned or frozen!)

100
 Summary and Exercises
 Modeling of
 Nonlinear systems
 Systems with time delay
 Next
 Modeling of DC motors
 Exercises
 Linearize the pendulum model at /4

101
ME451: Control Systems

Lecture 8
Modeling of DC motors

102
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

103
 What is DC motor?
An actuator, converting electrical energy into
rotational mechanical energy

(You will see DC motor during Lab 1 and 4.)

104
 Why DC motor?
 Advantages:
 high torque
 speed controllability
 portability, etc.
 Widely used in control applications: robot, tape
drives, printers, machine tool industries, radar
tracking system, etc.
 Used for moving loads when
 Rapid (microseconds) response is not required
 Relatively low power is required

105
How does DC motor work?

Armature

(from Dorf and Bishop book)

106
 Model of DC motor
La Output
ia(t) Ra

ea(t) eb(t) J
Input Bm

Armature circuit Mechanical load

107
Modeling of DC motor: time domain
 Armature circuit

 Connection between mechanical/electrical parts

 Motor torque
 Back EMF
Load torque

 Mechanical load

 Angular position

108
Modeling of DC motor: s-domain
 Armature circuit

 Connection between mechanical/electrical parts

 Motor torque
 Back EMF

 Mechanical load

 Angular position

109
DC motor: Block diagram
Tachometer
(Page 46)

Encoder
(Page 44)

Feedback system

110
Useful formula for feedback
Negative feedback system
Ea

Memorize this!
111
Ex: Derivation of transfer functions
Compute transfer functions from R(s) to Y(s).
Ea Ea

Ea
Ea

112
DC motor: Transfer functions (TF)

2nd order system

113
DC motor: Transfer functions (cont’d)
Note: In many cases La<<Ra. Then, an
approximated TF is obtained by setting La=0.

2nd order system 1st order system

114
 Summary and Exercises
 Modeling of DC motor
 What is DC motor and how does it work?
 Derivation of a transfer function
 Block diagram with feedback
 Next
 Stability of linear control systems, one of the most
important topics in feedback control
 Exercises
Go over the derivation for DC motor transfer functions
by yourself. Obtain T(s)/Ea(s).

115
Main message until this point

Many systems
can be represented
as transfer functions!

Using the transfer functions, ….

(to be continued)
116
ME451: Control Systems

Lecture 9
Stability

117
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • (Nyquist)

(Matlab simulations &) laboratories

118
 Simple mechanical examples
 We want mass to stay at x=0, but wind gave some
initial speed (F(t)=0). What will happen?
f(t) K f(t)
M M

x(t) x(t)
K
B
f(t) f(t)
M M
B
x(t) x(t)

 How to characterize different behaviors with TF?

119
 Stability
 Utmost important specification in control design!
 Unstable systems have to be stabilized by
feedback.
 Unstable closed-loop systems are useless.
 What happens if a system is unstable?
• may hit mechanical/electrical “stops” (saturation)
• may break down or burn out

120
What happens if a system is unstable?
Tacoma Narrows Bridge (July 1-Nov.7, 1940)

Wind-induced vibration Collapsed!

2008…

121
 Mathematical definitions of stability
 BIBO (Bounded-Input-Bounded-Output) stability :
Any bounded input generates a bounded output.
ICs=0 y(t)
u(t)
BIBO stable
system

 Asymptotic stability :
Any ICs generates y(t) converging to zero.
ICs y(t)
Asymp. stable
u(t)=0 system

122
 Some terminologies
Ex.

 Zero : roots of n(s)

 Pole : roots of d(s)

 Characteristic polynomial : d(s)

 Characteristic equation : d(s)=0

123
Stability condition in s-domain
(Proof omitted, and not required)
For a system represented by a transfer
function G(s),
system is BIBO stable

All the poles of G(s) are in the open left

half of the complex plane.

system is asymptotically stable

124
 “Idea” of stability condition
Example

Asym. Stability:
(U(s)=0)

BIBO Stability:
(y(0)=0)

Bounded if Re0
125
Second order impulse response-
Underdamped and Undamped

… Overdamped
… Critically damped
… Underdamped

… Undamped

126
 Second order impulse response –
Underdamped and Undamped
Changing / Fixed

Impulse Response
5

2
Amplitude

-1

-2

-3

-4
0 2 4 6 8 10 12
6
Time (sec)
4

-2

-4

-6 127
-5 0 5
 Second order impulse response –
Underdamped and Undamped
Changing / Fixed

Impulse Response
5

2
Amplitude

-1

-2

-3

-4
0 2 4 6 8 10 12
6
Time (sec)
4

-2

-4

-6 128
-5 0 5
 Second order impulse response –
Underdamped and Undamped
Changing / Fixed

Impulse Response
5

2
Amplitude

-1

-2

-3

-4
0 2 4 6 8 10 12
6
Time (sec)
4

-2

-4

-6 129
-5 0 5
 Remarks on stability
 For a general system (nonlinear etc.), BIBO
stability condition and asymptotic stability
condition are different.
 For linear time-invariant (LTI) systems (to which
we can use Laplace transform and we can
obtain a transfer function), the conditions
happen to be the same.
 In this course, we are interested in only LTI
systems, we use simply “stable” to mean both
BIBO and asymptotic stability.

130
 Remarks on stability (cont’d)
 Marginally stable if
 G(s) has no pole in the open RHP (Right Half Plane), &
 G(s) has at least one simple pole on -axis, &
 G(s) has no multiple poles on -axis.

Marginally stable NOT marginally stable

 Unstable if a system is neither stable nor

marginally stable.

131
 Examples
 Repeated poles

 Does marginal stability imply BIBO stability?

 TF:

 Pick
 Output

132
Feedback Technique

133
 Positive Feedback

K will depends on the distance between the guitar and the amplifier.

134
 Stability summary
Let si be poles of G.
Then, G is …
 (BIBO, asymptotically) stable if
Re(si)<0 for all i.
 marginally stable if
 Re(si)<=0 for all i, and
 simple root for Re(si)=0
 unstable if
it is neither stable nor
marginally stable.
135
Mechanical examples: revisited
f(t) K f(t)
M M

x(t) Poles= x(t) Poles=

stable? stable?

K
B
f(t) f(t)
M M
B
x(t) Poles= x(t) Poles=
stable? stable?

136
Examples
Stable/marginally stable
/unstable

???

137
 Summary and Exercises
 Stability for LTI systems
 (BIBO and asymptotically) stable, marginally stable,
unstable
 Stability for G(s) is determined by poles of G.
 Next
 Routh-Hurwitz stability criterion to determine stability
without explicitly computing the poles of a system.
 Exercises
 Solve examples in the previous slide.

138
ME451: Control Systems

Lecture 10
Routh-Hurwitz stability criterion

139
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
• Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

140
 Stability summary (review)
Let si be poles of
rational G. Then, G is …
 (BIBO, asymptotically) stable if
Re(si)<0 for all i.
 marginally stable if
 Re(si)<=0 for all i, and
 simple root for Re(si)=0
 unstable if
it is neither stable nor
marginally stable.
141
 Routh-Hurwitz criterion
 This is for LTI systems with a polynomial
denominator (without sin, cos, exponential etc.)
 It determines if all the roots of a polynomial
 lie in the open LHP (left half-plane),
 or equivalently, have negative real parts.
 It also determines the number of roots of a
polynomial in the open RHP (right half-plane).
 It does NOT explicitly compute the roots.

142
 Polynomial and an assumption
 Consider a polynomial

 Assume
 If this assumption does not hold, Q can be factored as

where
 The following method applies to the polynomial

143
Routh array
From the given
polynomial

144
 Routh array
(How to compute the third row)

145
 Routh array
(How to compute the fourth row)

146
Routh-Hurwitz criterion

The number of roots

in the open right half-plane
is equal to
the number of sign changes
in the first column of Routh array.

147
 Example 1

Routh array

Two sign changes Two roots in RHP

in the first column

148
 Example 2

Routh array
If 0 appears in the first column of a
nonzero row in Routh array, replace it
with a small positive number. In this
case, Q has some roots in RHP.

Two sign changes Two roots

in the first column in RHP

149
 Example 3

Routh array
If zero row appears in Routh array, Q
has roots either on the imaginary axis
or in RHP.

No sign changes No roots

in the first column in RHP
But some
Take derivative of an auxiliary polynomial roots are on
(which is a factor of Q(s)) imag. axis.
150
 Example 4

Find the range of K s.t. Q(s) has all roots in the left
half plane. (Here, K is a design parameter.)

Routh array
No sign changes
in the first column

151
Simple & important criteria for stability
 1st order polynomial

 2nd order polynomial

 Higher order polynomial

152
Examples
All roots in open LHP?

Yes / No

Yes / No

Yes / No

Yes / No

Yes / No

153
 Summary and Exercises
 Routh-Hurwitz stability criterion
 Routh array
 Routh-Hurwitz criterion is applicable to only
polynomials (so, it is not possible to deal with
exponential, sin, cos etc.).
 Next,
 Routh-Hurwitz criterion in control examples
 Exercises
 Read Routh-Hurwtiz criterion in the textbook.
 Do Examples.

154
 ME451: Control Systems

Lecture 11
Routh-Hurwitz criterion: Control examples

155
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
• Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

156
 Stability summary (review)
Let si be poles of
rational G. Then, G is …
 (BIBO, asymptotically) stable if
Re(si)<0 for all i.
 marginally stable if
 Re(si)<=0 for all i, and
 simple root for Re(si)=0
 unstable if
it is neither stable nor
marginally stable.
157
Routh-Hurwitz criterion (review)

The number of roots

in the right half-plane
is equal to
the number of sign changes
in the first column of Routh array.
158
 Example 1

 Design K(s) that stabilizes the closed-loop

system for the following cases.
 K(s) = K (constant)
 K(s) = KP+KI/s (PI (Proportional-Integral) controller)

159
 Example 1: K(s)=K
 Characteristic equation

 Routh array

160
 Example 1: K(s)=KP+KI/s
 Characteristic equation

 Routh array

161
 Example 1: Range of (KP,KI)
 From Routh array,

3.5

2.5

1.5

0.5

0
-1 0 1 2 3 4 5 6 7 8 9

162
Example 1: K(s)=KP+KI/s (cont’d)
 Select KP=3 (<9)
 Routh array (cont’d)

 If we select different KP, the range of KI changes.

163
Example 1: What happens if KP=KI=3
 Auxiliary equation
2

1.8

1.6

1.4

1.2

1
 Oscillation frequency 0.8

0.6

0.4
 Period 0.2

0
0 2 4 6 8 10 12 14 16 18 20

Unit step response

164
 Example 2

 Determine the range of K and a that stabilize the

closed-loop system.

165
Example 2 (cont’d)

166
 Example 2 (cont’d)
 Characteristic equation

167
 Example 2 (cont’d)
 Routh array

 If K=35, oscillation frequency is obtained by the

auxiliary equation

168
 Summary and Exercises
 Control examples for Routh-Hurwitz criterion
 P controller gain range for stability
 PI controller gain range for stability
 Oscillation frequency
 Characteristic equation
 Next
 Time domain specifications
 Exercises
 Read Chapter 6 again.
 Redo Examples 1 and 2
 Do Problem 6.6-(a) and 6.7-(b)-Find the range of K for which the
system is stable.

169
 More example 1

Routh array
Derivative of auxiliary poly.

2
(Auxiliary poly. is a factor of Q(s).)

No sign changes
in the first column No root in OPEN(!) RHP

170
 More example 2

Routh array
Derivative of auxiliary poly.

4 4

2
No sign changes
in the first column
No root in OPEN(!) RHP
171
 More example 3

Routh array
Derivative of auxiliary poly.

4 0

One sign changes

One root in OPEN(!) RHP
in the first column
172
ME451: Control Systems

Lecture 12
Time-domain specifications

173
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

174
 What we do next
 We learned stability.
 Definition in time domain
 Condition in s-domain
 Routh-Hurwitz criterion to check the condition
 Stability is a necessary requirement, but not
sufficient in most control problems.
 Specifications other than stability
 How to evaluate a system quantitatively in time domain?
 How to give specifications in time domain?
 What are the corresponding conditions in s-domain?

175
 Time response

System

 We would like to analyze a system property by

applying a test input r(t) and observing a time
response y(t).
 Time response is divided as

Transient response Steady-state response

(after yt dies out)
176
 Example of transient &
steady-state responses

Step response
Step Response
3

2.5
 Transient response 2
Amplitude

1.5

 Steady-state resp. 1

0.5

0
0 2 4 6
Time (sec)
8 10 12
Time (sec) 177
 Usage of time responses
 Modeling
 Some parameters in the system may be estimated by
time responses.
 Analysis
 Evaluate transient and steady-state responses
(Satisfactory or not?)
 Design
 Given design specs in terms of transient and steady-
state responses, design controllers satisfying all the
design specs.

178
 Typical test inputs

Step function Ramp function

(Most popular)

Parabolic
Sinusoidal input function
will be dealt with
later.

179
Steady state value for step test signal

G(s)

 Suppose that G(s) is stable.

 By the final value theorem:

 Step response converges to some finite value,

called steady state value .
180
Typical unit step response

181
Steady-state error for reference us(t)

182
Peak value, peak time, and
percent overshoot

183
Delay, rise, and settling times

Delay time: time to reach 0.5 yss

Rise time: time to rise from 0.1yss to 0.9yss
Settling time: time to settle within 5% of
yss

184
 An example revisited
 For the example in a previous slide,
 Steady-state error : -2
 Delay time around 1.5 sec
 Rise time around 5 sec 3
Step Response

 Settling time around 6 sec 2.5

Amplitude
1.5
Remark: There is no peak in
this case, so peak value, peak 1

time and percent overshoot 0.5

cannot be defined.
0
0 2 4 6
Time (sec)
8 10 12

185
Remarks on time-domain responses
 Speed of response is measured by
 Rise time, delay time, and settling time
 Relative stability is measured by
 Percent overshoot
 In general ….
 Fast response  Large percent overshoot
 Large percent overshoot  small stability margin
 We need to take trade-off between response
speed and stability.

186
 Summary and Exercises
 Time response and time domain specifications
 Time response can be used for
• Parameter estimation
• Design specification of the feedback system
 Time response is difficult to compute analytically,
except 1st and 2nd order systems (we’ll study later).
 Next
 When does steady state error become zero?
 Exercises
 Read about performance of feedback control
systems.
187
ME451: Control Systems

Lecture 13
Steady-state error

188
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

189
Performance measures (review)
 Transient response (From next lecture)
 Peak value
 Peak time
 Percent overshoot
Next, we will connect
 Delay time these measures
 Rise time with s-domain.
 Settling time
 Steady state response
 Steady state error (Today’s lecture)

190
Steady-state error: unity feedback
We assume that the
CL system is stable!
Unity feedback!

 Suppose that we want output y(t) to track r(t).

 Error
 Steady-state error

Final value theorem

(Suppose CL system is stable!!!)
191
 Error constants
 Step-error (position-error) constant

 Ramp-error (velocity-error) constant

 Parabolic-error (acceleration-error) constant

 Kp, Kv, Ka : ability to reduce steady-state error

192
Steady-state error for step r(t)

Kp

193
Steady-state error for ramp r(t)

Kv
194
Steady-state error for parabolic r(t)

Ka
195
 System type
 System type of G is defined as the order
(number) of poles of G(s) at s=0.
 Examples

type 1

type 2

type 3

196
 Zero steady-state error
 If error constant is infinite, we can achieve zero
steady-state error. (Accurate tracking)
 For step r(t)

 For ramp r(t)

 For parabolic r(t)

197
 Example 1
 G(s) of type 2
G(s)

 Characteristic equation

 CL system is NOT stable for any K.

 e(t) goes to infinity. (Don’t use today’s results if
CL system is not stable!!!)

198
 Example 2
 G(s) of type 1 G(s)

 By Routh-Hurwitz criterion, CL is stable iff

 Step r(t)

 Ramp r(t)

 Parabolic r(t)
199
 Example 3
 G(s) of type 2 G(s)

 By Routh-Hurwitz criterion, we can show that CL

system is stable.
 Step r(t)

 Ramp r(t)

 Parabolic r(t)

200
 A control example

 Closed-loop stable?
 Compute error constants

 Compute steady state errors

201
 Summary and Exercises
 Steady-state error
 For unity feedback (STABLE!) systems, the system
type of the forward-path system determines if the
steady-state error is zero.
 The key tool is the final value theorem!
 Next, time response of 1st-order systems
 Exercises
 Go over the examples in this lecture.

202
ME451: Control Systems

Lecture 14
Time response of 1st-order systems

203
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

204
Performance measures (review)
 Transient response (Today’s lecture)
 Peak value
 Peak time
 Percent overshoot
Next, we will connect
 Delay time these measures
 Rise time with s-domain.
 Settling time
 Steady state response
 Steady state error (Done)

205
 First-order system
 A standard form of the first-order system:

 DC motor example

206
 DC motor example (cont’d)
 If La<<Ra, we can obtain a 1st-order system

 TF from motor input voltage to

 motor speed is 1st-oder
 motor position is 2nd-order

207
Step response for 1st-order system
 Input a unit step function to a first-order system.
Then, what is the output?
u(t) y(t)

0 0

(Partial fraction expansion)

208
How to eliminate steady-state error
 Make a feedback system with a controller having
an integrator (copy of Laplace transform of a unit
step function):
Controller
u(t)
1

One has to select controller parameters

to stabilize the feedback system.
Suppose K=T=1, and obtain such parameters!
209
 Meaning of K and T
 K : Gain
 Final (steady-state) value 1
K=1,T=1

0.8

 T : Time constant 0.6

Amplitude
 Time when response
0.4
rises 63% of final value
 Indication of speed of 0.2

response (convergence)
0
 Response is faster as T 0 1 2 3 4 5 6
Time
becomes smaller.
210
 DC gain for a general system
 DC gain : Final value of a unit step response
 For first-order systems, DC gain is K.
 For a general stable system G, DC gain is G(0).

Final value theorem

 Examples

211
Settling time of 1st-order systems

 Relation between time and exponential decay

5% settling time is about 3T!

2% settling time is about 4T!

212
Step response for some K & T
K=1,T=1 K=1,T=2
2 2
Amplitude

Amplitude
1 1

0 0
0 5 10 0 5 10
Time Time
K=2,T=1 K=2,T=2
2 2
Amplitude

Amplitude
1 1

0 0
0 5 10 0 5 10
Time Time
213
 System identification
 Suppose that we have a “black-box” system

Unknown

 Obtain step response

 Can you obtain a transfer function? How?

214
Ramp response for 1st-order system
 Input a unit ramp function to a 1st-order system.
Then, what is the output?
u(t)=t y(t)

0 0

(Partial fraction expansion)

215
Ramp response for 1st-order system
K=1,T=1
5
K=1,T=1
4 y(t)
u(t)=t
Amplitude
Amplitude

slope
2

0
0 1 2 3 4 5
Time
Time

 Steady state response

 We may want to modify the system s.t.
216
How to eliminate steady-state error
 Make a feedback system with a controller having
a double integrator (copy of Laplace transform of
ramp function):
u(t)=t Controller

One has to select controller parameters

to stabilize the feedback system.
Suppose K=T=1, and obtain such parameters!
217
 Summary and exercises
 Time response for 1st-order systems
 Step and ramp responses
 Time constant and DC gain
 System identification
 Next, time response for 2nd-order systems
 Exercises
 Review examples in this lecture.

218
ME451: Control Systems

Lecture 15
Time response of 2nd-order systems

219
 Course roadmap
Modeling Analysis Design

Laplace transform Time response

Design specs
• Transient
Transfer function • Steady state
Root locus
Models for systems Frequency response
Frequency domain
• electrical • Bode plot
• mechanical
PID & Lead-lag
• electromechanical Stability
Block diagrams • Routh-Hurwitz
Design examples
Linearization • Nyquist

(Matlab simulations &) laboratories

220
Performance measures (review)
 Transient response (Today’s lecture)
 Peak value
 Peak time
 Percent overshoot
Next, we will connect
 Delay time these measures
 Rise time with s-domain.
 Settling time
 Steady state response
 Steady state error (Done)

221
 Second-order systems
 A standard form of the second-order system

 DC motor position control example

Amplifier Motor

Closed-loop TF

222
Step response for 2nd-order system
 Input a unit step function to a 2nd-order system.
What is the output?
u(t) y(t)

0 0

DC gain

223
Step response for 2nd-order system
for various damping ratio
 Undamped 2

 Underdamped 1.5

 Critically damped 1

 Overdamped 0.5

0
0 5 10 15

224
Step response for 2nd-order system
Underdamped case
 Math expression of y(t) for underdamped case

Damped natural frequency

225
Peak value/time: Underdamped case

1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
0 5 10 15

226
Properties of 2nd-order system

(5%) (2%)
227
 Some remarks
 Percent overshoot depends on , but NOT n.
 From 2nd-order transfer function, analytic
expressions of delay & rise time are hard to
obtain.
 Time constant is 1/(n), indicating
convergence speed.
 For >1, we cannot define peak time, peak
value, percent overshoot.

228
P.O. vs. damping ratio

229
 Pole locations of G
 Poles (0<<1)

 Damping ratio

Next, we clarify the influence of

pole location on step response.

230
 Influence of real part of poles
 Settling time ts decreases.

ts

231
 Influence of imag. part of poles
 Oscillation frequency d increases.

232
 Influence of angle of poles
 Over/under-shoot decreases.

233
 An example
 Require 5% settling time ts < tsm (given):

Im

Re

234
 An example (cont’d)
 Require PO < POm (given):

Im

Re

235
 An example (cont’d)
 Combination of two requirements

&

Im

Re

236
 Summary
 Transient response of 2nd-order system is
characterized by
 Damping ratio  & undamped natural frequency n
 Pole locations
 Delay time and rise time are not so easy to
characterize, and thus not covered in this course.
 For transient responses of high order systems,
we need computer simulations.
 Next, Root locus

237
 Exercises
(Use a calculator if necessary.)
 Read the related topics from the textbook.
1. For the system below with =0.6, n=5
(rad/sec), obtain
• Percent overshoot ?
• 5% settling time ?

238
 Exercises
2. For the system below, design K1 and K2 s.t.
 Percent overshoot is at most 20%?
 Peak time is at most 1 sec.?
 With designed K1 and K2, what is 5% settling time?

239