You are on page 1of 30

ELEC ENG 4CL4:

Control System Design


Notes for Lecture #5
Wednesday, January 14, 2004
Dr. Ian C. Bruce
Room: CRL-229
Phone ext.: 26984
Email: ibruce@mail.ece.mcmaster.ca

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Chapter 4

Continuous Time Signals



Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Specific topics to be covered include:


linear high order differential equation models
Laplace transforms, which convert linear differential
equations to algebraic equations, thus greatly simplifying
their study
methods for assessing the stability of linear dynamic
systems
frequency response.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Linear Continuous Time Models


The linear form of this model is:
dn y(t) dn1 y(t) dn1
n
+ an1 n1
+ . . . + a0 y(t) = bn1 n1 u(t) + . . . + b0 u(t)
dt dt dt

Introducing the Heaviside, or differential, operator o:


df (t)
f (t) = f (t) 
dt
 n1  df n (t)
f (t) = f (t) =
n n
f (t) =
dtn
We obtain:
n y(t) + an1 n1 y(t) + . . . + a0 y(t) = bn1 n1 u(t) + . . . + b0 u(t)

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Laplace Transforms
The study of differential equations of the type
described above is a rich and interesting subject. Of
all the methods available for studying linear
differential equations, one particularly useful tool is
provided by Laplace Transforms.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Definition of the Transform


Consider a continuous time signal y(t); 0 t < .
The Laplace transform pair associated with y(t) is
defined as


L [y(t)] = Y (s) = est y(t)dt
0
 +j
1 1
L [y(s)] = y(t) = est Y (s)ds
2j j

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

A key result concerns the transform of the derivative


of a function:
 
dy(t)
L = sY (s) y(0 )
dt

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Table 4.1: Laplace transform table


f (t) (t 0) L [f (t)] Region of Convergence
1
1 >0
s
D (t) 1 || <
1
t >0
s2
n!
tn n Z+ >0
sn+1
1
et C > {}
s
1
tet C > {}
(s )2
s
cos(o t) >0
s2 + o2
o
sin(o t) >0
s2 + o2
(sin )s + o2 cos sin
et sin(o t + ) > {}
(s )2 + o2
2o s
t sin(o t) >0
(s2 + o2 )2
s2 o2
t cos(o t) >0
(s2 + o2 )2
1 es
(t) (t ) || <
s

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Table 4.2: Laplace transform properties. Note that


Fi ( s ) = [ fi (t )], Y ( s ) = [ y (t )], k {1,2,3,...}, f1(t ) = f 2 )( t ) = 0 t < 0.
f (t) L [f (t)] Names

l 
l
ai fi (t) ai Fi (s) Linear combination
i=1 i=1
dy(t)
sY (s) y(0 ) Derivative Law
dt 
dk y(t)  di1
y(t) 
k
sk Y (s) i=1 ski  High order derivative
 t dt
k dti1 t=0
1
y( )d Y (s) Integral Law
0 s
y(t )(t ) es Y (s) Delay
dY (s)
ty(t)
dsk
d Y (s)
tk y(t) (1) k
 t
dsk
f1 ( )f2 (t )d F1 (s)F2 (s) Convolution
0
lim y(t) lim sY (s) Final Value Theorem
t s0
lim y(t) lim sY (s) Initial Value Theorem
t0+  s
+j
1
f1 (t)f2 (t) F1 ()F2 (s )d Time domain product
2j j
eat f1 (t) F1 (s a) Frequency Shift

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Transfer Functions
Taking Laplace Transforms converts the differential
equation into the following algebraic equation
sn Y (s) + an - 1 sn - 1 Y (s) + . . . + a0 Y (s)
= bn- 1 sn - 1U (s) + . . . + b0 U(s) + f(s, xo)

This can be expressed as Y (s) = G(s)U (s)

where G(s) =
B(s)
A(s)
and A(s) =sn + an1 sn1 + . . . + a0
B(s) =bn1 sn1 + bn2 sn2 + . . . + b0

G(s) is called the transfer function.



Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Transfer Functions for Continuous


Time State Space Models
Taking Laplace transform in the state space model
equations yields
sX(s) x(0) = AX(s) + BU (s)
Y (s) = CX(s) + DU (s)
and hence
X(s) = (sI A)1 x(0) + (sI A)1 BU (s)
Y (s) = [C(sI A)1 B + D]U (s) + C(sI A)1 x(0)

Y (s) = G(s)U (s)


G(s) = C(sI A)1 B + D

G(s) is the system transfer function.



Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Often practical systems have a time delay between


input and output. This is usually associated with the
transport of material from one point to another. For
example, if there is a conveyor belt or pipe
connecting different parts of a plant, then this will
invariably introduce a delay.
The transfer function of a pure delay is of the form
(see Table 4.2):
H(s) = esTd
where Td is the delay (in seconds). Td will typically
vary depending on the transportation speed.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Example 4.4 (Heating system). As a simple


example of a system having a pure time delay
consider the heating system shown below.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

The transfer function from input (the voltage applied


to the heating element) to the output (the temperature
as seen by the thermocouple) is approximately of the
form:

KesTd
H(s) =
( s + 1)

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Summary

Transfer functions describe the input-output


properties of linear systems in algebraic form.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Stability of Transfer Functions


We say that a system is stable if any bounded input
produces a bounded output for all bounded initial
conditions. In particular, we can use a partial
fraction expansion to decompose the total response
of a system into the response of each pole taken
separately. For continuous-time systems, we then
see that stability requires that the poles have strictly
negative real parts, i.e., they need to be in the open
left half plane (OLHP) of the complex plane s .
This implies that, for continuous time systems, the
stability boundary is the imaginary axis.
Consider an individual pole of a system transfer function
at s = s1 = 1+j1:

The impulse response of this pole is:

Note that if:


1. 1 < 0, then e1t 0 as t stable pole
2. 1 > 0, then e1t as t unstable pole
3. 1 = 0, then e1t = 1 for t unstable pole

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Impulse and Step Responses of


Continuous-Time Linear Systems

The transfer function of a continuous time system is


the Laplace transform of its response to an impulse
(Diracs delta) with zero initial conditions.

The impulse function can be thought of as the limit


(0) of the pulse shown on the next slide.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Figure 4.2: Discrete pulse



Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Steady State Step Response


The steady state response (provided it exists) for a
unit step is given by

1
lim y(t) = y = lim sG(s) = G(0)
t s s

where G(s) is the transfer function of the system.



Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

We define the following indicators:


Steady state value, y: the final value of the step response
(this is meaningless if the system has poles in the CRHP).
Rise time, tr: The time elapsed up to the instant at which the
step response reaches, for the first time, the value kry. The
constant kr varies from author to author, being usually
either 0.9 or 1.
Overshoot, Mp: The maximum instantaneous amount by
which the step response exceeds its final value. It is
usually expressed as a percentage of y

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Undershoot, Mu: the (absolute value of the) maximum


instantaneous amount by which the step response falls
below zero.
Settling time, ts: the time elapsed until the step response
enters (without leaving it afterwards) a specified deviation
band, , around the final value. This deviation , is
usually defined as a percentage of y, say 2% to 5%.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Figure 4.3: Step response indicators

y+Mp

y +
y

k y y
r

Mu

t t tp t
u r s
Time

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Poles, Zeros and Time Responses


We will consider a general transfer function of the
form
m
i=1 (s i )
H(s) = K n
l=1 (s l )

1, 1,, m and 1, 2, ,,, n are the zeros and poles


of the transfer function, respectively. The relative

degree is nr = n m .

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Poles
Recall that any scalar rational transfer function can
be expanded into a partial fraction expansion, each
term of which contains either a single real pole, a
complex conjugate pair or multiple combinations
with repeated poles.

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

First Order Pole


A general first order pole contributes
K
H1 (s) =
s + 1

The response of this system to a unit step can be


computed as
   
K K K
y(t) = L1 =L 1
t
= K(1 e )
s( s + 1) s s + 1

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Figure 4.4: Step response of a first order system

Step Response

0.632 y

Time

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

A Complex Conjugate Pair


For the case of a pair of complex conjugate poles, it
is customary to study a canonical second order
system having the transfer function.

n2
H(s) = 2
s + 2n s + n2

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Step Response for Canonical


Second Order Transfer Function
1 s + n n
Y (s) =
s (s + n )2 + d2 (s + n )2 + d2


1 1 s + n d
=
1 2
s 1 2 (s + n )2 + d2 (s + n )2 + d2

On applying the inverse Laplace transform we


finally obtain
en t
y(t) = 1
sin(d t + )
1 2

Chapter 4 Goodwin, Graebe, Salgado , Prentice Hall 2000

Figure 4.5: Pole location and unit step response of a


canonical second order system.

j y+Mp
d
n
y

T
d

n

j
d
tr tp