Chapter 5 The Performance of

Feedback Control Systems

The ability to adjust the transient and steady-state performance

is a distinct advantage of feedback control system.
The performance of the control system is specified in terms
of both the transient response and stead-state response.
-

The transient response is the response that disappears

with time.
- The steady-state response is that which exists a long time
following any inputs signal initiation.
The design specifications for control systems normally include
several time-response indices for a specified input command
1
as well as a desired steady-state accuracy.

Engineering, or the Art of Achieving

the Optimal Compromise
The specifications of the system are often results of
a compromise.

Engineering, or the Art of Achieving

the Optimal Compromise
The specifications of the system are often results of
a compromise.

Test Input Signals

The time-domain performance specifications are important
because the control systems are inherently time-domain system.
In general, a standard test input signals are used. The standard
test input signals commonly used are:

Y(s)=T(s)R(s)
Ec ( s ) = (1 T ( s )) R( s )

Step input

r (t ) = A
A
R( s) =
s

Ramp input

r (t ) = At
A
R( s) = 2
s

Parabolic input

r (t ) = At 2
2A
R( s) = 3
s

A Unit Impulse Function

A unit impulse function (t ) is also useful for test signal purpose.
For unit impulse function: r (t ) = (t )

R( s) = ( s ) = 1

The unit impulse function (t ) has the following properties:

(t ) dt = 1

(t a ) g (t )dt = g (a )

Why?

A Unit Impulse Function

The unit impulse function is based on a rectangular function

1
, t ; ( 0, t 0)
f (t ) =
2
2
0, otherwise, f (t ) (t )

(t )dt

= lim f (t )dt
0

-/2

1/

/2

/2 1
= lim / 2 dt = 1

1
, t a ; ( 0, t a )
f (t a ) =
2
2
0,
otherwise, f (t a ) (t a )

f
a

-/2

1/
/2

t
6

A Unit Impulse Function

1
, t a ; ( 0, t a )
f ( t a ) =
2
2
0,
otherwise , f (t a ) (t a )

-/2

(t

1/
/2

a ) g (t )dt = lim f (t a ) g (t )dt

0

= lim [ a / 2 0 g (t )dt + aa+ //22 f (t a ) g (t )dt + a+ / 2 0 g (t )dt ]

0

= lim

a + / 2

0 a / 2

g (t ) dt = g (a ) lim

( t

a + / 2

0 a / 2

dt = g (a)

a ) g (t ) dt = g ( a )
7

A Unit Impulse Function

y (t ) =

g (t ) r ( ) dt = 1{G ( s ) R ( s )}.

r ( ) ( )
y (t ) =

g (t ) ( ) dt.

= t g (t ) ( )dt = g (t ) {if = 0}

For R(s)=(s)=1

Y ( s) = G ( s) R( s) = G ( s)

Higher order test input signal

r (t ) = t n ,

n!

R( s) =
s

n +1

Performance of a Second-Order System

2
/ s( s + 2n )

G( s)
n
R( s)
Y ( s) =
R( s) =
2
1 + G( s)
1 + n / s( s + 2n )

n2
= 2
R( s)
2
s + 2n s + n
n2
1
= 2
s + 2n s + n2 s

for a step input

R(s)=1/s
9

Response for second order systems

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

Transfer function:
Characteristic equation: s 2 + 2n s + n2 = 0
Poles:

Im(s )

s1, 2 = n jn 1 2

n 1 2

Time domain response

requirements can be translated
into pole positions

Re(s )

cos 1

time constant
oscillation
frequency
10

Transient Response for Second Order

Systems

n2
Transfer function:
T ( s) = 2
s + 2n s + n2
Laplace transform of step response (assuming zero states at t=0):
n2
1
Y ( s) = T ( s) =
s
s ( s 2 + 2n s + n2 )

By partial fractions decomposition, taking the inverse Laplace

transform and cancelling imaginary terms one can get the step
response in time: ( or we can use the Table 2.3, directly)
y (t ) =
1 e nt sin(n 1 2 t + cos 1 ) / 1 2

natural frequency oscillation/vibration

frequency
11
damping ratio

12

Influence of Damping Ratio on Step

Response

y (t ) = 1

e nt sin( n t + ),

Figure 5.5 (a) Transient response of a second-order system for a step input.
(b) The transient response of a second-order system for a step input as a
13
function of and n t .

Natural Frequencies
For an under-damped system, the step response will be:
y (t ) = 1

1 nt
sin( n t + ),
e

1 2 ,

0 < <1

= cos 1 cos =

sin = 1 cos = 1 =
2

n 1 2

(time constant)
(oscillation/vibration frequency)

dy (t ) n nt
=
e
sin n t
dt

14

Unit Impulse Response

For an under-damped system, the unit impulse response (R(s)=1) will
be:
n2
Y ( s) = T ( s) R( s) = 2
s + 2n s + n2

y (t ) =

n nt
sin n t , =
e

1 2 ,

which is simply the derivative of the response to a step input.

1 nt
y (t ) =
e
1
sin(n t + ),

dy / dt = n e nt sin n t

Next, we will consider the transient response of y(t):

1
y (t ) = 1 e nt sin( n t + ),

15

16

Second-Order System
y (t ) = 1

1 nt
sin( n t + ), (0 < < 1)
e

Percent Overshoot =

M p fv
fv

100

(usually corresponding
response within 2%)

17

Second-Order System
y (t ) = 1

1 nt
sin( n t + ), (0 < < 1)
e

Percent Overshoot =

M p fv
fv

100

(usually corresponding
response within 2%)

18

Setting Time
y (t ) = 1

e nt sin( n t + ),

corresponding to 2% of input

y (t ) 1

nTs

1 nt
e
=
sin(n t + ) 0.02

sin(n t + )

0.02 = 1 2 = 1 cos 2 = sin

0,sin 0, 0

sin x
sin(n t + ) 0
1
=
lim
=
x

x
0

e nTs < 0.02

e nTs < 0.02 nTs 4.
Ts = 4 =

. { = 1 /( )}
n
n

19

Peak Time and Peak Value

y (t ) = 1
dy (t )
dt

e nt sin( n t + ), =

n nt
=
e
sin n t 0 t =T
p

1 2 ,

dy (t ) n n t
=
e
sin(n t + ) n e n t cos (n t + )
dt

n nt
[ sin(n t + ) cos (n t + )]
e

= cos 1 cos =
sin = 1 cos 2 = 1 2 =
=

dy (t ) n n t
=
e
[ sin(n t + )cos sin cos (n t + )]
dt

n nt
e
[ sin(n t + )]

= n e n t sin n t

20

Peak Time and Peak Value

n nt
sin n t 0 t =T
e
=
p

dy (t )
dt

sin n T p = 0 n T p = n , n = 0,1,

for
=
n 1 =
Tp

n 1 2

1 nT p
Mp =
y (T p ) =
1 e
sin(n T p + ),

= 1 e

nT p

sin ( + )

M pt = 1 + e

nT p

= 1+ e

1 2

= 1+ e

nT p

sin

= cos 1 cos =
,

sin = 1 cos 2 = 1 2 =

M pt fV
P.O. =
100%
fV

P .O . = 100e

1 2
21

Second-Order System
Swiftness of response: rise time and peak time
Tp =

n 1

Tpn =

1 2

Closeness of response: overshoot and settling time

M pt = 1 + e

/ 1 2

4
=

=
T
4
. Note e - 4 0.02
With = 0.02 f v then s
n

Thus the damping determines the

overshoot, and the real part of the
root sets the settling time.
22

Second-Order System
T pn =

M pt = 1 + e

/ 1 2

23

Second-Order System

Fixed damping ratio

0.2

24

Second-Order System

The step response for n = 5 with = 0.7 and = 1.

25

Second-Order System
Normalized rise time

26

Step Response of Second-Order System

1
y (t ) =
1 e nt sin(n t + cos 1 ),

=
1 2

Percent Overshoot =

M p fv
fv

100

27

Response For Second Order

Systems -Summary
(S-domain)
(Time-domain)

n2
1
Y ( s) = T ( s) =
s
s ( s 2 + 2n s + n2 )

1
1 ent sin(n t + cos 1 ),
y (t ) =

Ts = 4 =

( for = 2% )

PO = 100e /
Tp =

Tr1

1 2

1 2
=

Settling time
Percentage
overshoot
Peak time

n 1 2
2.16 + 0.6

Rise time (10% 90%)

28

Dominant Closed-loop Poles

Im

Re

Ts = 4 =

Which poles are more

important than others?

The closed-loop poles nearest the j axis will dominate

in the transient-response behavior because these poles
correspond to transient-response terms that decay slowly.
Those closed-loop poles are called dominant closed-loop
poles occur in the form of a complex-conjugate pair.
They are most important among all closed-loop poles.
For higher order system if it have a pair of dominant roots
(or poles), it can be approximated as a second order system.

29

Effect of a Third Pole

T ( s) =

n2

( s 2 + 2n s + n2 )(s + 1)

3rd system

(n = 1)
Experiments show as far as percent
overshoot and settling time are
concerned, it can be approximated
by second-order when

1 / 10 n

1/

n
30

Effect of a Third Pole

T ( s) =

n2

(5.15)

( s 2 + 2n s + n2 )(s + 1)

R ( s ) = 1 / s = 0.45 n = 1

For 2rd system:

| n |= 0.45

P .O . = 100e /

If the complex poles are

dominant, we will have

1 2

P .O . = 20%

, Ts =

0.45

Ts = 8.9 sec .

31

Effect of a Third Pole

T ( s) =

n2

(5.15)

( s 2 + 2n s + n2 )(s + 1)

R ( s ) = 1 / s = 0.45 n = 1

For 2rd system:

P .O . = 100e /

If the complex poles are

dominant, we will have

| n |= 0.45
1 2

P .O . = 20%

, Ts =

0.45

Ts = 8.9 sec .

32

Example 5.1 Parameter Selection

Ts =

( for = 2% )

Find K and p such that

overshoot<5%,
2% settling time<4s.
P .O . = 100e /

1 2

= 0.707 gives overshoot of 4.3%

Ts =

4
<4
n

n > 1

= cos1 = cos1 0.707 = 450

?
?

33

Example 5.1 Parameter Selection

Find K and p such that
overshoot<5%,
2% settling time<4s.
Ts =

( for = 2% ) P .O . = 100e /

1 2

= 0.707 gives overshoot of 4.3%

Ts =

4
<4
n

n > 1
= cos 1 = cos 1 0.707 = 450

34

Second-Order System
PO = 100e

T pn =

overshoot<5%
35

Example 5.1 Parameter Selection

K=? p=?

Choosing the poles at 1j , means

overshoot of 4.3%.

= 1 / 2 and n = 1 / = 2

4 and

1
n =
2

2 =1

The closed-loop transfer function is then

K
G( s)
n2
= 2
T ( s) =
= 2
1 + G ( s ) s + ps + K
s + 2n s + n2
The specifications are then met by choosing
K

= n2

= 2 and p = 2n = 2

Proposed design

36

S-Plane and Transient Response

Y ( s ) Pi ( s ) i ( s )
=
T (s) =
,
R( s)
( s)

1
R( s ) =
s

N
Bk s + C k
1 M Ai
Y (s) = +
+ 2
,
2
2
s i =1s + i k =1 s + 2 k s + ( k + n )
M

y (t ) = 1 + Ai e
i =1

i t

+ Dk e k t sin(k t + k )
k =1

37

S-Plane and Transient Response

38

Steady State Error

The steady state error is the steady state value of the
difference between the desired and the actual output
of the control system:
ess = e( ) = lim e(t )
t

Using the final value theorem of Laplace transforms

(see The Laplace transform table), we can calculate
the steady state error based on the Laplace transform
of the error:
ess = lim sE (s )
s 0

39

Steady State Error

ess = lim sE (s )
s 0

Steady State Error of an Open Loop System

E ( s ) = R( s ) Y ( s ) = (1 G ( s )) R( s )

If the input was a unit step:

1
ess = lim sE ( s ) = lim s(1 G ( s )) = 1 G (0)
s 0
s 0
s
G (0)

is the dc gain.
40

Steady State Error

Steady State Error of a Closed Loop System
The error signal:
E ( s ) = (1 T ( s )) R ( s ) =

For

(1 + GH G )
R( s)
1 + G (s) H (s)

H (s) = 1

E (s) =

(1 + GH G )
1
R( s) =
R( s)
1 + G (s) H (s)
1 + G (s)

the steady state error of the close-loop system

sR( s )
s 0 1 + G ( s )

ess = lim sE ( s ) = lim

s 0

41

Steady State Error

Steady State Error of a Closed Loop System
For H (s) = 1
1
E ( s ) = (1 T ( s ))R( s ) =
R( s)
1 + Gc ( s )G ( s )

the steady state error of the closedloop system

sR( s )
lim e(t ) = ess = lim sE ( s ) = lim
t
s 0
s0 1 + G ( s )G ( s )
c

42

Relationship between Steady

State Error and System Type
sR( s )
s0 1 + G ( s )G ( s )
c

ess = lim sE ( s ) = lim

s 0

( for H ( s ) = 1)

The general form of the open loop transfer function of a

M
system is given by
K ( s + zi )
Here N is known as the
i =1
Gc ( s )G ( s ) =
Q
N
system type, and corresponds
s ( s + pk )
k =1
to the number of integrators in
the system.

We can calculate
ess for different
types of standard
signals

System type

Type zero

Type 1

Type 2
43

Relationship between Steady

State Error and System Type
sR( s )
s0 1 + G ( s )G ( s )
c

ess = lim sE ( s ) = lim

s 0

A) Step Input

R( s ) =

A
s

s( A / s )
A
=
s0 1 + Gc ( s )G ( s ) 1 + lim Gc ( s )G ( s )

ess = lim

s 0

Gc ( s )G ( s ) =

K ( s + zi )
i =1
Q

s ( s + pk )
N

For N=0

ess =

k =1

A
=
1 + Gc (0)G (0)

A
M

i =1

k =1

1 + K zi / p k

A
1+ K p

Here K p is known as the position error constant.

M

i =1

k =1

K p = lim Gc ( s )G ( s ) = Gc (0)G (0) = K zi / pk

s 0

A
ess =
1+ Kp
44

Relationship between Steady

State Error and System Type
A) Step Input

R( s ) =

A
s

sR( s )
s0 1 + G ( s )G ( s )
c

ess = lim sE ( s ) = lim

s 0

s( A / s )
A
=
ess = lim
s0 1 + Gc ( s )G ( s ) 1 + lim Gc ( s )G ( s )
s 0

Gc ( s )G ( s ) =

K ( s + zi )
i =1
Q

s N ( s + pk )

For N1 ess = lim

s 0

i =1

k =1

= lim

1 + K zi / ( s p k )
N

k =1

As N

s 0 N

i =1

k =1

=0

s + K zi / ( p k )

45

Relationship between Steady

State Error and System Type
B) Ramp Input

R( s) =

A
s2

s( A / s 2 )
A
A
= lim
= lim
ess = lim
s0 1 + Gc ( s )G ( s )
s0 s + sGc ( s )G ( s )
s0 sGc ( s )G ( s )
ess = lim

s 0

A
M

sK ( s + zi ) / [ s ( s + pk )]
N

i =1

= lim

s 0

k =1

A
M

K ( s + zi ) / [ s
i =1

N 1

( s + pk )]

k =1

For a type 0 system,

ess = lim

s 0

A
M

i =1

k =1

sK zi / ( s 0 pk )

46

Relationship between Steady

State Error and System Type
B) Ramp Input
ess = lim

s 0

A
M

sK ( s + zi ) / [ s ( s + pk )]
N

i =1

= lim

s 0

k =1

A
M

K ( s + zi ) / [ s

N 1

i =1

( s + pk )]

k =1

For a type 1 system,

ess = lim

s 0

A
M

i =1

k =1

sK ( s + zi ) / [ s ( s + pk )]

i =1

k =1

K zi / pk

i =1

k =1

K v = lim sGc ( s )G ( s ) = K zi / pk
s 0

For N2 system,
ess = lim

s 0

A
M

sK ( s + zi ) / [ s ( s + pk )]
i =1

k =1

= lim

s 0

A
Kv

velocity constant
As N 1

i =1

k =1

K zi / pk

=0

47

Relationship between Steady

State Error and System Type
C) Parabolic Input
R(s ) =

s( A / s 3 )
A
ess = lim
= lim 2
s0 1 + G ( s )G ( s ) s0 s G ( s )G ( s )
c
c

A
(r (t )=At 2 / 2)
3
s

s( A / s 3 )

ess = lim

= lim

s 0

1+

s 0

K ( s + zi )
i =1
Q

s N 2 A ( s + pk )
k =1

k =1

i =1

s N ( s + pk ) + K ( s + zi )

s N ( s + pk )
k =1

For type 0 and type 1 systems, the steady state error is infinite.
For type 2 systems the steady state error is given by

ess =

A
M

i =1

k =1

K zi / pk

A
Ka

acceleration error
constant
M

i =1

k =1

K a = lim s Gc ( s )G ( s ) = K zi / pk
2

s 0

48

Relationship between Steady

State Error and System Type
System Type
(number of
pure
integrators)

Input
step

ess =

A
1+ Kp

ess = 0

ess = 0

ramp

parabola

A
Kv

A
Ka

Conclusions:
1.Adding integrators (increasing system type) eliminates steady state error.
2. If steady state error is finite and not zero, then increasing the system dc
gain (increasing controller proportional gain), reduces steady state error.

49

Error Constants For Unit Feedback

System
K p = lim Gc ( s )G ( s ) K v = lim sGc ( s )G ( s )
s 0

s 0

K a = lim s 2Gc ( s )G ( s )
s 0

Then for
Step input

Ramp input

Acceleration input

ess =

A
1+ Kp

A
ess =
Kv
A
ess =
Ka

50

Mobile Steering Robot

Gc ( s ) = K1 + K 2 / s

K p = lim Gc ( s )G ( s )
s 0

Gc ( s )G ( s ) =

K2 K
K1K
+
s + 1 s(s + 1)

K v = lim sGc ( s )G ( s )
s 0

When K 2 = 0 , we have a type-zero system and

for a step input,

ess =

A
A
=
1 + K p 1 + KK1

When K 2 > 0, we have a type-one system and

for a step input,
A
A
ess = 0
ess =
=
1+ K p 1+
for a ramp input,

ess =

A
A
=
K v KK 2

51

Mobile Steering Robot

A
A
ess =
=
K v KK 2
52

Steady-State Error For Non-Unity

Feedback

E ( s ) = R ( s ) Y ( s ) = [1 T ( s )]R ( s )

1
For a step input ess = lim s[1 T ( s )] = [1 T (0)]
s 0
s
G( s)
K ( s + 4)
T ( s) =
=
1 + G ( s ) H ( s ) ( s + 2)( s + 4) + 2 K
4K
T ( 0) =
8 + 2K
This is obtained when K=4.
Achieving e = 0 requires T(0)=1.
ss

53

Steady-State Error For Non-Unity

Feedback

T ( s) =

where

Y ( s)
Gc ( s )G ( s )
Z (s )
=
=
R( s ) 1 + H ( s )Gc ( s )G ( s ) 1 + Z (s )

Z (s ) =

Gc (s )G (s )
1 + Gc (s )G (s )(H (s ) 1)

The loop transfer function of the equivalent unity feedback system is Z(s).
The error constants for non-unity feedback system are given by:
K p = lim Z ( s ) K v = lim sZ ( s ) K a = lim s 2 Z ( s )
s 0

s 0

Noted that: H ( s ) = 1 Z ( s ) = Gc ( s )G ( s )

s 0

K p = lim Z ( s ) = lim Gc ( s )G ( s )
s 0

s 0

54

Performance Indices For Elevator

A simplified description of a control system for a elevator

When the fourth floor button is pressed on the first floor, the
elevator rises to the fourth floor with a speed and floor level
accuracy designed for passenger comfort.

55

Performance Indices For Elevator

Push of the fourth-floor button is an input that represent a

desired output, shown as a step function.

56

Performance Indices For Elevator

Transient response
Passenger comfort and passenger patience are dependent
upon the transient response.
If this response is too fast, passenger comfort is sacrificed;
if too slow, passenger patience is sacrificed.

57

Performance Indices For Elevator

Steady-state error

Passenger safety and convenience would be

sacrificed if the elevator is not properly level.
58

Performance Indices & Optimal

Control
A Performance index is a quantitative measure of the
performance of a system and is chosen so that emphasis
is give to the important system specifications.
A system is considered an optimal control when the system
parameters are adjusted so that the index reaches an extreme
value commonly a minimum value.
A performance index, to be useful, must be a number that is
always positive or zero. Then the best system is defined as
the system that minimizes this index.

59

Performance Indices-Example
A suitable performance index is the integral of the square
of the error, ISE, which is defined as
T

ISE = e 2 (t ) dt
0

The upper limit T is a finite time, usually convenient to choose T as the settling
time Ts .
A step input and response for ISE is shown here:

60

More Performance Indices

T

ISE = e 2 (t )dt

Integrated squared error

IAE = | e(t ) | dt

Integrated absolute error

ITAE = t | e(t ) | dt
0

Integrated time-multiplied
absolute error

ITSE = te 2 (t )dt
0

Integrated time-multiplied
squared error
61

Example 5.6 Performance Criteria

Type=?

T ( s) =

1
s 2 + 2s + 1

The ITAE tends to have better

selectivity (i.e. minimum more
discernible) than the others
T

ITAE = t | e(t ) | dt
0

For second order system

resulting a swift response and 5% overshot

For step input

62

Example 5.7 Attitude Control System

The values of ISE and IAE for this system are plotted in
the following figure.

IAE = | e(t ) | dt
0

IAE K3 = 4.2 , = 0.665.

Note that IAE cannot be analytically solved.

63

Optimum T(s) for Step Input ITAE

(characteristic equation)
b0
Y ( s)
= n
T ( s) =
R ( s ) s + bn 1s n 1 + bn 2 s n 2 + b1s + b0

64

Normalized Responses

65

Normalized Responses

66

Normalized Responses
The best
among
ISE,IAE
and ITAE

67

Design Example: Hubble Telescope

Positioning Control

Hubble is known for showing that

the recessional velocity of a galaxy
increases with its distance from the
earth, implying the universe is
expanding. Known as "Hubble's
law.
68

Design Example: Hubble Telescope

Positioning Control
The orbiting Hubble space telescope is the most complex expensive

scientific instrument that has ever been built

Launched to 380 miles above earth on April 24,1990.
The telescopes 2.4-mter mirror has the smoothest surface of any
mirror made, and its pointing system center it on a dime 400 miles
away. The mirror has a spherical aberration that was largely
corrected during space mission in 1993 and 1997.
Considering the model of the telescope-positing system shown in Fig (a)

Design Goal:
select K,K1,
so that:
1) Overshoot less than 10% for step input, 2) minimizing SS error for a ramp
input, and 3) reducing the effect of a step disturbance.
69

Design Example: Hubble Telescope

Positioning Control
Using block diagram reduction we have
Type=?
Td(s)=0
+

G(s)

R(s)

d ( s ) =0

R(s)=0

Td(s) +
-

Y ( s)
R( s ) T

G(s)

= T ( s) =

KG ( s )
L( s )
=
1 + KG ( s ) 1 + L( s )

Y ( s)
G( s)
G( s)
= TTd ( s) =
=
= T ( s) / K
Td ( s) R ( s )=0
1 + KG( s) 1 + L( s )

E (s ) = R(s ) Y (s ) = R(s ) T (s )R(s ) [T (s )/K ]Td (s )

=
E (s )

1
G (s )
R (s )
Td (s ).
1 + L (s )
1 + L (s )
70

Design Example: Hubble Telescope

Positioning Control
First, let us select K and K1 to meet the overshoot (<10%)
requirement for a unit step input, R(s)=1/s. Setting Td(s)=0,
we have
Y (s) =
=K

KG ( s )
R( s)
1 + KG ( s )

1
1
1
1
=K 2
.
s( s + K1 ) + K s
s + K1s + K s

(second order system: s

0
+ 2n s + n2 =

K1 = 2 n K = n2
/ 1 2

P .O . = 100e
= 0.6 O.P % = 9.5%
Goal: 1.

Overshoot less than 10% for step input,

P.O. = 10% = 0.6.

71

Design Example: Hubble Telescope

Positioning Control
Goal: 2) minimizing SS error for a ramp input,
KG ( s )
Y ( s) =
R( s)
1 + KG ( s )
Next, for ramp input requirement,
r (t ) = Bt ,t 0 , R ( s ) = B / s 2

we have

K v = lim sG ( s )
s 0

B
B
B
= lim
=
.
ess =

s 0
Kv
sKG ( s ) ( K / K1 )

72

Design Example: Hubble Telescope

Positioning Control
Goal: 3) reducing the effect of a step disturbance.
To reduce the step disturbance, by setting R(s)=0, and
Td(s)=1/s, we have
E ( s) =

G( s)
G( s)
Td ( s ) =
Td ( s )
1 + L( s )
1 + KG ( s )

G (s ) 1
=
eTd () lim=
sE (s ) lim s
s 0
s 0 1 + KG (s ) s
1
1
=
lim
=
s 0 s (s + K ) + K
K
1
73

Design Example: Hubble Telescope

Positioning Control
In summary:
ess =

Y ( s) =

K
1
.
2
s + K1s + K s

B
B
B
= lim
=
.

s
0
Kv
sKG ( s ) ( K / K1 )

= 0.6 O.P % = 9.5%

eTd ( ) = lim sE ( s ) =
s 0

G ( 0)
1
=
1 + KG (0) K

We seek a large K and a large K/K1 to obtain a low

steady-state error for the ramp input and a step
disturbance.
74

Design Example: Hubble Telescope

Positioning Control
We seek a large K and a large K/K1 to obtain a low steady-state error for
the ramp input and a step disturbance.

s 2 + 2 n s + n2 = s 2 + 2(0.6)n s + K = s 2 + k1s + K .

K1 = 2(0.6) n n = K
K
K
K
=
=
.
K1 1.2 K
1 .2

K
K1
K
K1

K = 25

25
=
= 4.17 ,
1.2

K =100

ess =

K1 = 1.2n = 1.2 K

B
B
B
= lim
=
.

s
Kv
sKG ( s ) ( K / K1 )

B
B
ess |k = 25 =
=
= 0.24 B .
K / K1 4.17

B
B
100
=
=
= 0.12 B .
e
|
=
= 8.33. ss k =100 K / K
8.33
1
1.2

K=100 is selected as the final design.

75

Design Example: Hubble Telescope

Positioning Control
Realistically, we must limit K so that system operation
remains linear. Using K=100, K1=12 we obtain the system as:

The responses of the system to a

unit step input and a unit step
Disturbance are shown in Fig (d)
Note that the effect of the disturbance is
relatively insignificant and
B
ess |k =100 =
= 0.12 B.
8.33
This design, using K=100 is an excellent system

76

Sequential Design Example: Disk

Drive Read System
In chapter 4, we consider the
response of the closed-loop read
head control system. The results
are:
We will attempt to adjust the
amplifier gain Ka, in order to
obtain the best performance
possible.

77

Sequential Design Example: Disk

Drive Read System
Goal:

to achieve the fastest response to a step input r(t)

while (1) limiting the overshoot and oscillatory
(2) reducing the effect of a disturbance on the
output position of the read head.
Specifications:
Percentage overshoot
Less than 5%
Setting time
Less than 250ms
Max. value of response
Less than 0.005
to a unit step disturbance

G1 =

5000
5
=
1 = 1 / 1000 = 0.001
s + 1000 1s + 1

G1 5

Dominate poles

78

Sequential Design Example: Disk

Drive Read System
Type=?

Y ( s) =
=

5K a
R( s ),
s( s + 20) + 5K a
5K a

s + 20 s + 5K a
2

R ( s ),

n2
s + 2 n s
2

+ n2

R ( s ),

n2 = 5K a , and 2n = 20.
or

K a =20 5 K a =n2 n =10

2n =20 =1
= 1 2n = 20 n = 10 5K a = n2 K a = 20 K a > 20 < 1

79

Sequential Design Example: Disk

Drive Read System
Y ( s) =

5K a
R( s ) ,
s 2 + 20 s + 5K a

Kv

The best compromise (Ka=40) still does not meet all the specifications. (Ts=400ms)
80
What to do?

Step Response For Second Order

Systems -Summary
1
1 ent sin(n t + cos 1 ),
y (t ) =

Ts = 4 =

( for = 2% )

PO = 100e /
Tp =

n 1 2

Tr1

2.16 + 0.6

(0 <
1 2
=

< 1)

Settling time
Percentage
overshoot
Peak time
Rise time (10%
- 90%)

81

Sequential Design Example: Disk

Drive Read System
The best compromise (Ka=40) still does not
meet all the specifications. (Ts=400ms)
What to do?
Y ( s) =

5K a
R( s ) ,
2
s + 20 s + 5K a

Kv
Using PD controller, or using velocity feedback to increase 2n

82

Summary
In this chapter, we have concerned with the definition and
usefulness of quantitative measures of the performance of
feedback systems.
Transient response: (for a unit step input signal:) in terms:
overshoot, peak time, and settling time
Steady state errors for different standard test signals.
Optimal control indices.

83

Test Input Signals

The time-domain performance specifications are important
because the control systems are inherently time-domain system.
In general, a standard test input signals are used. The standard
test input signals commonly used are:

Y(s)=T(s)R(s)
Ec ( s ) = (1 T ( s )) R( s )

Step input

r (t ) = A
A
R( s) =
s

Ramp input

r (t ) = At
A
R( s) = 2
s

Parabolic input
r (t ) = At 2 / 2

A
R( s) = 3
s

84

Second-Order System
y (t ) = 1

1 nt
sin( n t + ), (0 < < 1)
e

Percent Overshoot =

M p fv
fv

100

(usually corresponding
response within 2%)

85

Second-Order System
y (t ) = 1

1 nt
sin( n t + ), (0 < < 1)
e

Percent Overshoot =

M p fv
fv

100

(usually corresponding
response within 2%)

86

Relationship Between Steady

State Error And System Type
sR( s )
s0 1 + G ( s )G ( s )
c

ess = lim sE ( s ) = lim

s 0

( for H ( s ) = 1)

The general form of the open loop transfer function of a system is given
by
M
Here N is known as the
K ( s + zi )
system type, and corresponds
i =1
Gc ( s )G ( s ) =
Q
N
to the number of integrators in
s ( s + pk )
k =1
the system.
N

We can calculate
ess for different
types of standard
signals

System type

Type zero

Type 1

Type 2
87

Error Constants for unit feedback

system
K p = lim Gc ( s )G ( s )
s 0

K v = lim sGc ( s )G ( s )
s 0

K a = lim s 2Gc ( s )G ( s )
s 0

Then for

ess =
Step input

Ramp input
Acceleration input

A
1+ Kp

A
ess =
Kv
ess =

A
Ka
88

Relationship Between Steady

State Error And System Type
System Type
(number of
pure
integrators)

Input
step

ess =

A
1+ Kp

ess = 0

ess = 0

ramp

parabola

A
Kv

A
Ka

Conclusions:
1.Adding integrators (increasing system type) eliminates steady state error.
2. If steady state error is finite and not zero, then increasing the system dc
gain (increasing controller proportional gain), reduces steady state error.

89

Steady-State Error for Non-Unity

Feedback

T ( s) =

Y ( s)
Gc ( s )G ( s )
=
R( s ) 1 + H ( s )Gc ( s )G ( s )

T ( s) =

Y ( s)
Z ( s)
=
R( s) 1 + Z ( s)

where

Z ( s) =

Gc ( s )G ( s )
1 + Gc ( s )G ( s )( H ( s ) - 1)

The loop transfer function of the equivalent unity feedback system is Z(s).
The error constants for non-unity feedback system are given by:
K p = lim Z ( s ) K v = lim sZ ( s ) K a = lim s 2 Z ( s )
s 0

Noted that:

s 0

H ( s ) = 1 Z ( s ) = Gc ( s )G ( s )

s 0

K p = lim Z ( s ) = lim Gc ( s )G ( s )
s 0

s 0

90

More Performance Indices

T

ISE = e 2 (t )dt

Integrated squared error

IAE = | e(t ) | dt

Integrated absolute error

ITAE = t | e(t ) | dt
0

Integrated time-multiplied
absolute error

ITSE = te 2 (t )dt
0

Integrated time-multiplied
squared error
91