EE 379 - CONTROL SYSTEMS

Week # 5 – Performance Specifications

• Steady State Error

• System Sensitivity
• Sensitivity Functions
• Effect of Disturbances
• Effect of Measurement Noise

Design of Feedback Control Systems 4th (Edition)

R.T. Stefani, C.J. Savant, B. Shahian, G.H. Hostetter
INSTRUCTOR: DR. KUNWAR FARAZ AHMED
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types

Steady State Error – System Types

• The steady-state error for a unity feedback system can be
determined easily by identifying the system type and input
signal
ess = lim e(t ) = lim sE ( s )
t → s →0
sR ( s )
ess = lim
s →0 1 + G ( s )

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

2
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
Steady State Error – Unit Step Input

s 1 where
ess = lim .
s →0 1 + G ( s ) s
K p = lim G ( s )
1 s →0
=
1 + lim G ( s ) Kp = position error constant
s →0
1
=
1+ Kp

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

3
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
Steady State Error – Unit Step Input
• For system of type 0,
K ( s + z1 )( s + z 2 )...( s + z m )
G (s) =
( s + p1 )( s + p2 )...( s + pn )
K ( s + z1 )( s + z 2 )...( s + z m )
K p = lim
s →0 ( s + p )( s + p )...( s + p )
1 2 n
K .z1.z 2 .....z m
=
p1. p2 ..... pn
1
 ess =
1+ Kp
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
4
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
Steady State Error – Unit Step Input

• For system of type 1 and above,

K ( s + z1 )( s + z 2 )...( s + z m )
G (s) = N for N 1
s ( s + p1 )( s + p2 )...( s + pn )

K ( s + z1 )( s + z 2 )...( s + z m )
K p = lim N
s →0 s ( s + p )( s + p )...( s + p )
1 2 n
=
1
 ess = =0
1+ Kp

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

5
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
Steady State Error – Unit Step Input
Conclusion
Those systems from type 1 and above will have zero
steady-state error for step input.

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

6
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input Steady State Error – Unit Ramp Input

s 1
ess = lim . 2 where
s →0 1 + G ( s ) s

1
= lim K v = lim sG ( s )
s →0 s + sG ( s ) s →0

1
= Kv = velocity error constant
lim sG ( s )
s →0
1
=
Kv

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

7
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input Steady State Error – Unit Ramp Input
• For system of type 0,
K ( s + z1 )( s + z 2 )...(s + z m )
G( s) =
( s + p1 )( s + p2 )...(s + pn )

K ( s + z1 )( s + z 2 )...( s + z m )
K v = lim s
s →0 ( s + p1 )( s + p2 )...( s + pn )
=0
1
 ess = =
Kv

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

8
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input Steady State Error – Unit Ramp Input
• For system of type 1,
K ( s + z1 )( s + z 2 )...( s + z m )
G (s) =
s ( s + p1 )( s + p2 )...( s + pn )

K ( s + z1 )( s + z 2 )...( s + z m )
K v = lim s
s →0 s ( s + p1 )( s + p2 )...( s + pn )
K .z1.z 2 .....z m
=
p1. p2 ..... pn

1
 ess =
Kv

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

9
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input Steady State Error – Unit Ramp Input

• For system of type 2 and above,

K ( s + z1 )( s + z 2 )...( s + z m )
G (s) = N for N 2
s ( s + p1 )( s + p2 )...( s + pn )

K ( s + z1 )( s + z 2 )...( s + z m )
K v = lim s N
s →0 s ( s + p )( s + p )...( s + p )
1 2 n
=
1
 ess = =0
Kv

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

10
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input Steady State Error – Unit Ramp Input

• Conclusion
– Systems of type 0 will have infinity steady-state error
– Systems of type 1 will have finite steady-state error
– Those systems from type 2 and above will have zero
steady-state error for ramp input

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

11
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input
+ Unit parabolic Input
Steady State Error – Unit Parabolic Input
s 1
ess = lim . 3 where
s →0 1 + G ( s ) s

1
K a = lim s G ( s )
2

= lim 2 s →0
s →0 s + s 2 G ( s )
Ka = acceleration error constant
1
=
lim s 2G ( s )
s →0
1
=
Ka

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

12
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input
+ Unit parabolic Input
Steady State Error – Unit Parabolic Input
• For system of type 0 and 1,
K ( s + z1 )( s + z 2 )...( s + z m )
G (s) = for N 1
s N ( s + p1 )( s + p2 )...( s + pn )

K ( s + z1 )( s + z 2 )...( s + z m )
K a = lim s 2 for N 1
s →0 s N ( s + p1 )( s + p2 )...( s + pn )
=0

1
 ess = =
Ka

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

13
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input
+ Unit parabolic Input
Steady State Error – Unit Parabolic Input
• For system of type 2,
K ( s + z1 )( s + z 2 )...( s + z m )
G (s) = 2
s ( s + p1 )( s + p2 )...( s + pn )
K ( s + z1 )( s + z 2 )...( s + z m )
K a = lim s 2 2
s →0 s ( s + p1 )( s + p2 )...( s + pn )
K .z1.z 2 .....z m
=
p1. p2 ..... pn
1
 ess =
Ka
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
14
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input
+ Unit parabolic Input
Steady State Error – Unit Parabolic Input

• For system of type 3 and above,

K ( s + z1 )( s + z 2 )...( s + z m )
G (s) = for N 3
s N ( s + p1 )( s + p2 )...( s + pn )
K ( s + z1 )( s + z 2 )...( s + z m )
K a = lim s N 2
s →0 s ( s + p1 )( s + p2 )...( s + pn )
=

1
 ess = =0
Ka

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

15
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input
+ Unit parabolic Input
Steady State Error – Unit Parabolic Input

• Conclusion
– Systems of type 0 and 1 will have infinity steady-state
error
– Systems of type 2 will have finite steady-state error
– Those systems from type 3 and above will have zero
steady-state error for parabolic input

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

16
 Performance Specifications
Steady State Error – Summary
• The steady-state error for a system with unity feedback can be summarized as
given below:

Steady Type 0 Type 1 Type 2

state Static Static Static
Error Error Error Error
Input Formula Constant Error Constant Error Constant Error
Step 1
1
u(t) Kp= Kp= ∞ 0 Kp= ∞ 0
1+ K p 1+ K p
constant
Ramp 1 1
tu(t) Kv Kv= 0 ∞ Kv= Kv= ∞ 0
Kv
constant
Parabola 1 1
1 2 Ka
Ka= 0 ∞ Ka= 0 ∞ Ka=
t u (t ) Ka
2 constant
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
+ System Types
+ Unit Step Input
+ Unit Ramp Input
+ Unit parabolic Input
Steady State Error – Non-Unity Feedback Systems
+ Non-Unity Feedback • The steady-state error with non-
Systems E ( s ) = input − output
unity feedback system can be
determined in two ways: = R( s) − C ( s)
– By solving the problem ess = lim sE ( s )
using the fundamental s →0
definition of steady-state
error
– By changing the block R(s)
+-
Ea(s)
G(s)
Y(s)

diagram into the equivalent

unity feedback system, and
the respective formula to H(s)

calculate the respective

steady-state errors
R(s) E(s) G (s) Y(s)
+
- 1 + G (s) H (s) − G (s)

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

18
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
+ Plant Variations
System Sensitivity – Plant Variations
Plant Y(s)
• Feedback can be used to R(s) + Amplifier 4
make the response of a - 100 s+2
system relatively
Output
independent of certain Sensor
type of changes or 1
inaccuracies 400
• Suppose we have a ( s + 2) 400
T (s) = =
system 1 + 400 s + 402
( s + 2)
• Now, suppose one of
R(s) Plant Y(s)
the plant parameters + Amplifier 4
change or is - 100 s + 2k1
inaccurately modelled
Output
• For k1 = 1 the plant is Sensor
the same; other values 1
of will cause 400
perturbations from the ( s + 2k1 ) 400
T (s) = =
nominal plant 1 + 400 s + 400 + 2k1
( s + 2k1 )
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
19
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
+ Plant Variations
System Sensitivity – Plant Variations
Plant
• Even 50% changes in R(s)
+ Amplifier 4 Y(s)
the parameter from k1 - 100 s + 2k1
=1/2 to k1 =3/2
Output
results in a relatively Sensor
minor change in the 1
T(s)
400
• Even negative value ( s + 2k1 ) 400
k1 =-1 gives the same T (s) = =
1 + 400 s + 400 + 2k1
stable overall T(s) ( s + 2k1 )
• The systems steady
state error for a unit 1 s + 2k1
lim s  [1 − T ( s )] = lim
step input is s →0 s + 400 + 2k
s →0
s 1
this is dominated by 2k1
the factor of 400 and =
400 + 2k1
is proportional to k1

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

20
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
+ Plant Variations
+ Amplifier Variations System Sensitivity – Amplifier Variations
R(s) Plant Y(s)
• Suppose the +
Amplifier 4k 2
inaccuracy is of the - 100 s+2
form k2
Output
• The Transfer function Sensor
becomes 1
• The systems steady 400k 2
state error for a unit ( s + 2) 400k 2
T (s) = =
step input is 1+
400k 2 s + 400k 2 + 2
this is dominated by ( s + 2)
the factor of 400k2
and is inversely 1 s+2
lim s  [1 − T ( s )] = lim
s →0 s + 400 k + 2
proportional to k2 s →0
s 2

• Changes in amplifier 2
gain of 400 will =
400k 2 + 2
produce the same
effect on T(s) and its
step response

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

21
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
+ Plant Variations
+ Amplifier Variations
+ Sensor Variations
System Sensitivity – Sensor Variations
• Suppose the inaccuracy is in
the senor gain i.e., k3
Plant
• The Transfer function R(s) + Amplifier 4
Y(s)
becomes - 100 s+2

• The systems steady state Output

error, for a unit step input is Sensor
k3
this can become quite large
for comparable percentage 400
( s + 2) 400
parameter changes T ( s ) = =
1+
400k3 s + 400k3 + 2
• Is this result expected ?? ( s + 2)
why?
1 s + 400(k3 − 1) + 2
• Error by the sensor in the lim s  [1 − T ( s )] = lim
s →0
s s →0 s + 400k3 + 2
perceived output is
indistinguishable from 400(k3 − 1) + 2
=
actual output error 400k3 + 2

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

22
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
Sensitivity Functions
• A process, represented by G(s), • However, a closed-loop system
is subject to a changing senses the changes in the output
environment, aging, ignorance due to process changes and
of the exact values of the attempts to correct the output
process parameters.
• In the open-loop system, all • Consider a change in the
these errors and changes result process as: G ( s ) +  G ( s )
in a changing and inaccurate
output.
• In the open-loop case, the Y ( s ) = G ( s ) R ( s )
change in the output is:
• In the closed-loop system:
G ( s ) + G ( s )
Y ( s ) + Y ( s ) = R( s)
1 + (G ( s ) + G ( s )) H ( s )
G ( s )
Y ( s ) = R( s)
(1 + G ( s ) H ( s ) + G ( s ) H ( s ))(1 + G ( s ) H ( s ))
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
23
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
Sensitivity Functions
G ( s ) + G ( s )
Y ( s ) + Y ( s ) = R( s)
1 + (G ( s ) + G ( s )) H ( s )
G ( s )
Y ( s ) = R( s)
(1 + G ( s ) H ( s ) + G ( s ) H ( s ))(1 + G ( s ) H ( s ))
G ( s ) H ( s )  ΔG ( s ) H ( s )
ΔG ( s )
ΔY ( s ) = R( s)
(1 + G ( s ) H ( s )) 2

The change in the output of the closed-loop system is reduced by a

factor [1+G(s)H(s)]2.

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

24
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
Sensitivity Functions
• Consider the effect of plant or controller parameter variation on
the closed-loop characteristics.

• The sensitivity function of T with respect to changes in a

parameter  is defined as
T
T % change in T (due to  ) T 
S = lim
T
= = 
 →0  % change in   T


© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

25
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
Sensitivity Functions
• Assume that the function G depends upon a parameter  ,
then for the closed loop system,
G
T (s) =
1 + GH
The sensitivity function of T with respect to  is
T  T G G 
ST =  =    = SGT SG
 T G T  G
where
T G 1 G 1
SGT =  =   S T
=
G T (1 + GH ) 2 G G
(1 + GH )
(1 + GH )
As the loop gain GH(s) becomes large, the sensitivity of the closed-
loop characteristics is reduced
Result: Sensitivity of the system may be reduced below that of the
open-loop system by increasing G(s)H(s) over the frequency range of
interest.
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
26
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
Sensitivity Functions
The sensitivity of the feedback system to changes in the feedback
element H(s) is
T H  G  −H − GH
2

S HT =  =   =
H T  1 + GH  G 1 + GH
(1 + GH )

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

27
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances Effect of Disturbances
• Another major advantage of D (s) Inaccessible
a feedback system is that it Disturbance Inputs
can be used to reduce the
effects of disturbance inputs Plant
upon system response 1
W(s) ++ Y(s)
s+2
• In this system the Accessible Plant
disturbance signal D(s) Control Inputs
affects the plant but is not
accessible to the designer 1
TD ( s ) =
s+2
• The transfer function
relating Y(s) to D(s) is 1 1
Y (s) = 
• For a unit step disturbance s s+2
input the final value of the 1
output due to the lim y (t ) = lim sY ( s ) =
t → s →0 2
disturbance is

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

28
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances Effect of Disturbances
D (s)
• If the plant is driven in
the feedback
arrangement as shown Plant
+ Y(s)
+ Amplifier W(s) 1
• The transfer function R(s) K + s+2
-
relating Y(s) to D(s) is
• For a unit step
disturbance input to the
1
feedback system, the ( s + 2) 1
TD ( s ) = =
resulting steady state
1 − − K  s+2+ K
output is  ( s + 2)
• This error can be made
1 1 
arbitrarily small by Y (s) =  
making K sufficiently s  s+2+ K 
large 1
lim y (t ) = lim sY ( s ) =
t → s →0 2+ K

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

29
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances Effect of Disturbances
• Another type of system is as shown
• The two system transfer functions are
• Routh Array shows that the system is stable D (s)
• Unit step disturbance produces
Plant
+
+ s+3 W(s) 10
R(s) Y(s)
-
s + s + 2 s + 10
2

Y (s) 10( s + 3) s ( s 2 + 2 s + 10) 10( s + 3)

TR ( s ) = = = 3
R ( s ) 1 + 10( s + 3) s ( s 2 + 2 s + 10) s 3 + 2 s 2 + 20 s + 30 s 1 20
Y (s) 10 ( s 2 + 2 s + 10) 10 s s2 2 30
TD ( s ) = = = 3
D( s ) 1 + 10( s + 3) s ( s + 2 s + 10) s + 2 s 2 + 20 s + 30 s1
2
5
1 10 s s0 30
lim s TD ( s ) = lim 3 =0
s →0
s s → 0 s + 2 s + 20 s + 30
2

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

30
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances Effect of Disturbances
• Now consider the open loop
(non-feedback) system as shown
• This system has the same
transfer function relating Y(s)
and R(s)
Y (s) 10
• Relationship between output and D T ( s ) = =
D ( s ) s 2 + 2 s + 10
disturbance is not modified by
feedback
1 10
lim s TD ( s ) = lim 2 =1
• Unit step disturbance produces s →0  s  s →0 s + 2 s + 10

D (s)

Plant
+
+ ( s + 3)( s + 2 s + 10) W(s)
2 10
R(s) s + 2s + 10
2 Y(s)
s 3 + 2 s 2 + 20 s + 30 +

© Dr Systems,
EM 412 – Control Kunwar Faraz
EMEAhmed http://www.kfaraz.com/#!__controls
College, NUST 31
31
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances Effect of Disturbances
• Effect of D(s) for an open-loop system.
Y ( s ) = Gc G p ( s ) R ( s ) + G p ( s ) D( s )
• Effect of D(s) for the closed-loop system.
Gc G p ( s ) G p (s)
Y (s) = R( s) + D( s)
1 + Gc G p H ( s ) 1 + Gc G p H ( s )
• For large loop gain, GcGpH(s)
1 1
Y (s)  R( s) + D( s)
H (s) Gc H ( s )
High gain controllers Load disturbances
D(s)
can significantly or plant noise
reduce the effect of +
R(s) E(s) + Y(s)
disturbance inputs Gc(s) Gp(s)
+
while maintaining the – cascade
desired Y(s)/R(s) compensator
relationship. H(s)
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
32
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances
- Effect of Measurement
Effect of Measurement Noise
Noise
• Measurement noise may be represented by a
signal injected in the feedback path.
• For the closed-loop system
Gc G p ( s ) Gc G p H1 ( s )
Y (s) = R( s) + N (s)
1 + Gc G p H1 H 2 ( s ) 1 + Gc G p H1 H 2 ( s )
• For large loop gains
1 1
Y (s)  R( s) + N (s)
H1 H 2 ( s) H 2 (s)
R(s) E(s) Y(s)
Gc(s) Gp(s)
+
Note: that a high loop –
gain does not help in
H1(s) H2(s)
reducing the effect of +
measurement noise +
N(s) Measurement Noise
© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls
33
 Performance Specifications
Ziegler Nichols Compensation
• A typical process control plant has real poles and zeros and is type 0
• The compensator to the plant is generally either Proportional (P), Proportional Integral
(PD) or Proportional Integral Derivative (PID) having the following form
Ki K d s 2 + K p s + Ki
Gc ( s ) = K p + + Kd s =
s s
• The Ziegler Nichols Method has the following two steps
Step I : Set the true plant under proportional control, with a very small gain so that
Gc(s) = Kp. This gain is adjusted until the system becomes marginally stable
Step II : The compensator is defined by
 1 

Gc ( s ) = K p 1 + + Td s  Design Equations for Ziegler Nichols
 Ti s 
Kp Ti Td
once values of Td and Ti have
been calculated P Kp= 0.5Kpo
Kp PD Kp= 0.45Kpo Ti= 0.83To
Ki = , K d = K pTd
Ti PID Kp= 0.6Kpo Ti= 0.5To Td= 0.125To
 Performance Specifications
Ziegler Nichols Compensation
• Consider the process control plant • At this value of Kpo the complex conjugate
64 roots are obtained from the row s2 i.e., 14(
G p (s) = 3 s2+56) which gives the roots as j7.483
s + 14 s 2 + 56 s + 64 • The characteristic polynomial is then
• First step is to let Gc(s) = Kp and find divided by( s2+56) which gives the
the value of Kp = Kpo such that 1+Gc(s) remaining root -14 .
Gp(s) is marginally stable • This gives T0 = 2/0 = 2 / 7.483 = 0.84
• The Routh Array is formed as and the remaining values can be easily
s3 1 56 found from the design table
s2 14 64(1+Kp)
s1 (1/14)[784-64(1+Kp)] 0 Kp Ti Td

s0 64(1+Kp) P Kp= 0.5Kpo

PD Kp= 0.45Kpo Ti= 0.83To
• The value of Kpo is found by setting the PID Kp= 0.6Kpo Ti= 0.5To Td= 0.125To
first term in row s1 to 0 this gives Kpo =
11.25
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances
- Effect of Measurement
Ziegler-Nichols Compensation
Noise
- Ziegler-Nichols • The Simulink model of a simple f(t)
Compensation
mass-damper-spring (mck)
c
system with a PID controller
m1
k1,L01

x1

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

36
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances
- Effect of Measurement
Ziegler-Nichols Compensation
Noise
- Ziegler-Nichols • Response of uncompensated System
Compensation

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

37
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances
- Effect of Measurement
Ziegler-Nichols Compensation
Noise • set up the system with only proportional control
- Ziegler-Nichols
Compensation and provide a step input f(t)

• start with a small value for KP and increase c

until the output gives constant amplitude m1

oscillations k1,L01

• this value of the proportional constant will be

called, Ku, the ultimate gain x1

T0

T0

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

38
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances
- Effect of Measurement
Ziegler-Nichols Compensation
Noise
• The controller equation can be written as Kpo = 6,000 N/m
- Ziegler-Nichols
Compensation
 1  To = 0.08 sec
Gc ( s ) = K p 1 + + Td s 
 Ti s 
select Kp, Ti, and Td according to
Kp = 3,600 N/m
Design Equations for Ziegler Nichols
Kp Ti Td Ti = 0.04 sec
P Kp= 0.5Kpo Td = 0.01 sec
PD Kp= 0.45Kpo Ti= 0.83To
PID Kp= 0.6Kpo Ti= 0.5To Td= 0.125To

Kp = 3,600 N/m
Ki = 90,000 N/(m sec)

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

Kd = 36 N sec/m
39
Lecture Sequence

• Performance Specifications
- Steady State Error Performance Specifications
- System Sensitivity
- Sensitivity Functions
- Effect of Disturbances
- Effect of Measurement
Ziegler-Nichols Compensation
Noise
- Ziegler-Nichols • Response of Ziegler-Nichols compensated System
Compensation

© Dr Kunwar Faraz Ahmed http://www.kfaraz.com/#!__controls

40