Lecture 8 - Specification and Limitations

1. Introduction
Important to specify and evaluate systems

K. J. strm

1. Introduction
2. Classical Specifications and the Gang of Six
3. Properties of Simple Transfer Functions
4. Limitations
5. Summary
6. Appendix - Examples Simple Transfer Functions

Follow reference signals

Reduce load disturbances
Do not inject too much measurement noise
Sensitivity to modeling errors
Many classical specifications were geared towards
response to reference signals. Important to consider
all issues.

Important to understand fundamental limitations

Theme: To specify and evaluate. Important particularly if you

are a customer of control.

Skill in judging properties of transfer functions

Some Key Issues

Load Disturbances

r
F

u
C

x
P

r
F

Transient responses (steps)

u
C

x
P

1
Six transfer functions should be considered

Load disturbances drive the process variable away from it

desired value. The relevant transfer functions are
X ( s)
Y ( s)
P
=
=
D ( s)
D ( s)
1 + PC
U ( s)
PC
=
D ( s)
1 + PC

Frequency responses
Bode and Nyquist plots

Poles zeros plots

c K. J. strm August, 2001
&

Measurement Noise

Robustness to Process Variations

n

d
r
F

u
C

r
F

u
C

x
P

1
1 + PC
PC
T ( s) =
1 + PC

Response to Reference Signal

2. Classical Specifications

u
C

It is essential that the closed loop system is not too sensitive

to variations in the process dynamics. This is captured by the
sensitivity functions

U ( s)
P
=
N ( s)
1 + PC
X ( s)
PC
=
N ( s)
1 + PC

S( s ) =

d
F

Measurement noise is fed into the system by the feedback

and generates creates variations in the control signal and the
process output. The relevant transfer functions are

x
P

Focused on one a few of the transfer functions, typically

response to step in reference signal
Specifications in both time and frequency domains
Some robustness measure, e.g. gain and phase margins
Remember that a few transfer functions only give a limited
view. You must look at all six transfer functions to really
know how the system behaves

1
The process variable should follow the reference signal and the
control signals required for this are reasonable. The relevant
transfer functions are

Still useful to know about this because of the tradition.

This is what many persons you will encounter will talk
about.

X ( s)
Y ( s)
PC
=
=
R( s )
R( s )
1 + PC
C
U ( s)
=
R( s )
1 + PC
c K. J. strm August, 2001
&

Response to Step Reference Signal

Response to Ramp in Reference Signal

10

10

0
0

10

0
0

10

Useful to describe tracking of slow signals.

Particularly important in motion control
Overshoot M

Solution time Ts

Rise time Tr

Final error e0

Response to Step in Load Disturbance

Frequency Response to Reference Signal

Control error for unit step load disturbance

0.4
0.2
0
0

10

How to measure it?

Max error
Max time
Solution time
Integrated absolute error IE

Useful parameters
Bandwidth
Resonance peak

Integrated error IE
c K. J. strm August, 2001
&

Nyquist Curve of Loop Transfer Function L = PC

Bode Plot of Loop Transfer Function L = PC

Useful parameters
Gain margin gm

Gain

10

Gain crossover frequency g c

10

Phase margin m

10

Phase crossover frequency 180

10

10

Phase
100

Max sensitivities Mt , Ms

120

Frequencies mt, ms

140
160
180
200
1
10

Poles and Zeros of Closed Loop Transfer

Function from r to y

10

More Complete Specifications

Essential to give specifications on all six transfer functions
This was not done classically and is still often neglected
Specifications can be given in terms of time responses
and frequency responses.
Several choices, one possibility:
Response to reference signals is given by transfer functions F PC/(1 + PC), FC/(1 + PC)
Measurement noise is captured by transfer function
C/(1 + PC)
Load disturbances captured by P/(1 + PC), PC/(1 + PC)
Robustness captured by P/(1 + PC), PC/(1 + PC)

c K. J. strm August, 2001

&

3. Properties of Simple Systems

Second Order System with a Zero

It is necessary for you to have a working knowledge

of properties of simple systems. We will give a short
discussion here.

s 02
02
1
s+a
=
+
a s2 + 2 0 s + 02
a s2 + 2 0 s + 02
s2 + 2 0 s + 02
Notice that G (0) = 1. Let h0 be the step response of
G ( s) =

02

Practice this in the home-works.

Use Matlab and ICtools. (Start Matlab in CAD Lab, type
ICtools, team up with a friend and interact.)
Many results are collected in an appendix to this lecture.
See slides at the end.

G0 =

02
s2 + 2 0 s + 02

. The step response of G is then

h(t) = h0 (t) +

1 dh0 (t)
a dt

Implications for the shape of the response

Effect of different values of a
What happens if a is negative?

Second Order System with a LHP Zero

Step responses for 0 = 1 and = 0.5 a = 0.25, 0.5 1, 2, 5
and 10.

Second Order System with a RHP Zero

Step responses for 0 = 1 and = 0.5 a = 0.25, -0.5 -1, -2,
-5, and -10

Step Response

Step Response

1.5

2.5

0.5

Amplitude

Amplitude

1.5

0.5

1.5

0.5
2

10
Time (sec.)

15

c K. J. strm August, 2001

&

10

15

Time (sec.)

4. Limitations

Minimum Phase Systems

Saturation and rate saturation

There are many factors that limits the achievable performance

r
F

Actuator friction and resolution

x
P

Measurement noise
Difficult dynamics, time delays, poles and zeros in the right
half plane (non-minimum phase systems)
Philosophy:

1
Essential limitations:

Understand the limitations

If at all possible modify the process
Never pose unrealistic specifications

Sensor noise and resolution

Actuator saturation and quantization
Actuator friction

Phase Curve of Bode Plots should Look Like This

Factor process transfer function as P(s) = Pmp(s) Pnmp(s) such

that h Pnmp(i )h = 1 and phase of Pnmp is negative, then
arg L( i g c ) = arg Pnmp ( i g c) + arg Pmp( i g c ) + arg C( i g c )

+ m

Phase curve

50

arg G (i )

Non-minimum Phase Systems

100
150

Bodes relations give arg Pmp(i g c) + arg C(i g c )  n /2 where

n is the slope of the gain curve of the loop transfer function at
the gain crossover frequency. Hence
arg Pnmp(i g c) + m n

200
250
300
1
10

10

10

Dashed line is phase curve for non-minimum phase equivalent

Simple Rule of Thumb m = 45 and ng c = 0.5 gives

arg Pnmp(i g c)

2
c K. J. strm August, 2001
&

System with RHP Zero

Assume

System with Time Delay

Pnmp(s) = esT

z s
Pnmp(s) =
z+ s

Hence

Hence
arg Pnmp(i g c ) = 2 arctan

arg Pnmp(i g c) = g c T
Cross over frequency inequality

g c T m + ng c

Cross over frequency inequality

arg Pnmp(i g c) = 2 arctan

Hence
Choosing m

gc

z
= /4 and ng c

+ m ng c

Choosing m = /4 and ng c = 1/2 gives

gcT

+ ng c )
2
2
4
= 1/2 gives g c < z.

tan(

gc

A time delay limits the response speed of the system.

A RHP zero limits the response speed of the system.

System with RHP Pole

Pnmp(s) =

Hence

s+ p
s p

arg Pnmp(i g c ) = 2 arctan

Pnmp(s) =
p

Cross over frequency inequality

p
2 arctan
+ m ng c
gc
2
Hence

gc

System with RHP Pole and Zero Pair

For z > p we have

arg Pnmp(i ) = 2 arctan

g c / z + p/ g c
1 p/ z

Cross over frequency inequality

gc

tan( /2 m /2 + ng c /4)

Choosing m = /4 and ng c = 1/2 gives g c p

( z s)(s + p)
( z + s)(s p)

gc


p

m
(1 ) tan

+ ng c
z
2
2
4

Simple calculations give

z
2 + 2 1 + 2
1+
p
2

A RHP pole requires a high crossover frequency.

c K. J. strm August, 2001

&

Example - The X-29

RHP Pole-Zero Pair

z
2 + 2 1 + 2
1+
p
2
Assume a phase margin m = /4 and a slope at the
crossover frequency of ng c = 1/2, then

Advanced experimental aircraft. Much design effort was done

with many methods and much cost. Specifications m = 45
could not be reached. Here is why!
Non-minimum phase part of the transfer function

z 5.83p

Pnmp(s) =

Phase margin

m < + ng c
z/ p

-6.0

2.24 3.86
0

30

r
4 arctan

5
38.6

45

The zero pole ratio is z/ p = 4.33 with ng c = 1/2 we get

p
z

5.83 8.68
60

m = 32.4
10

20

Bicycle with rear wheel steering

a = 0.3m

s= p=

Q = 1.2 m
b = 0.7 m
3

J = 120 kgm
V0 = 5 ms1

arg Pnmp(i g c) + m ng c

RHP zero and pole at

s= z=

V0
a
r

Summary of Dynamics Limitations

For non-minimum phase systems the limitations can be expressed by the crossover frequency inequality

( s)
amQ V0 s + V0/a
=
( s)
bJ s2 mgQ/ J

m = 70 kg

A phase margin of 45 cannot be achieved!

64.8 84.6

Example - Kleins Rear Wheel Steered Bike

Typical values:

s 26
s6

Simple Rule of Thumb:

arg Pnmp(i g c) /2

RHP zeros and time delays give upper bound on g c

Long time delays are bad
Slow unstable zeros are bad

mgQ
J

mgQ V0
mgQ V0
z
=
=
p
aJ
a( J cm + mQ2 )
z
= 2.74, ; m = 10.4
p

RHP poles gives a lower bound on g c

Fast unstable poles are bad
RHP poles and zeros cannot be too close
The product of a RHP pole and a time delay cannot be too

large
c K. J. strm August,
&
2001

Rules of Thumb for Limitations - Part 1

A RHP zero z

gc
z

A time delay T

0.5

for Ms , Mt < 2

0.2

for Ms , Mt < 1.4.

gcT
A RHP pole p

gc
p

0.7

for Ms , Mt < 2

0.37

for Ms , Mt < 1.4.

for Ms, Mt < 2

for Ms, Mt < 1.4.

Rules of Thumb for Limitations - Part 2

A RHP pole-zero pair with z > p

z
6.5
for Ms , Mt < 2

for Ms , Mt < 1.4.

14.4
p
A RHP pole-zero pair with z < p

p
6.5
for Ms, Mt < 2

for Ms, Mt < 1.4

14.4
z
A RHP pole p and a time delay T

0.16
for Ms, Mt < 2
pT
0.05

5. Summary

for Ms, Mt < 1.4.

Appendix - Properties of Simple Systems

Classical specifications are restricted

Important to look at all six transfer functions (Gang of
Six)

In this Appendix we have collected properties of simple systems

First order systems

Important to be aware of limitations

Second order systems without zero

Non-minimum phase elements give severe limitations

Third order system without zeros

Time delays
RHP zeros
RHP poles

Second order system with zero

Examples of assessment of a control system

Properties of simple transfer functions

c K. J. strm August, 2001

&

Second Order System without Zeros

Step Response

Transfer function
G ( s) =

Introduce

+ 2 0 s + 02
2
0

s2

d = 0

Poles

p
0 i 0 1 2 = 0 i d
p
0 0 2 1 = 0 d

where d = 0

1 2 if 1 and d = 0

1 (1 + t) e 0 t,

1 cosh d t + p

where

<1

=1

2 1

12


s=

h(t) =

e 0 t

p
1

sin ( d t + ),

1 2


sinh d t e d t

= arccos

if 1
if > 1

p
2 1 if > 1

Frequency Response

Step Response

Bode plot for = 0.1, 0.2, 0.5, 0.707, 1, 2, 5 och 10.

= 0, 0.1, 0.2, 0.5, 0.707, 1, 2, 5 and 10.

Bode Diagrams

Step Response
2
20

1.8
0

1.6
20
Phase (deg); Magnitude (dB)

1.4

Amplitude

1.2
1
0.8

40

60
0

50

0.6
100

0.4
150

0.2
0

10

15

200
1
10

10

10

Frequency (rad/sec)

Time (sec.)

c K. J. strm August, 2001

&

10

Third Order System without Zeros

Step responses for = 0.5 and a/ 0 =0.25, 0.5, 1, 2 and 10.

Transfer function
G ( s) =

Third Order System without Zeros

a 02
(s + a)(s2 + 2 0 s + 02)

Step Response

Notice G (0) = 1.

What do we mean by a solution?

0.8

Amplitude

step([a],conv([1 a],[1 2*z*w0 w0^2]))

ICTools and SysQuake!

0.6

0.4

0.2

10

15

Time (sec.)

Third Order System without Zeros

Bode plots for = 0.5 and a/ 0 =0.1, 0.2, 0.5, 1, 2, 5, 10
Bode Diagrams

50

Second Order System with a Zero

02
s 02
1
s+a
=
+
a s2 + 2 0 s + 02
a s2 + 2 0 s + 02
s2 + 2 0 s + 02
Notice that G (0) = 1. Let h0 be the step response of
G ( s) =

02

G0 =

Phase (deg); Magnitude (dB)

50

. The step response of G is then

100

h(t) = h0 (t) +

150
0

1 dh0 (t)
a dt

Implications for the shape of the response

50
100

Effect of different values of a

150

What happens if a is negative?

200
250
300
2
10

02
s2 + 2 0 s + 02

10

10

10

10

Frequency (rad/sec)

c K. J. strm August, 2001

&

11

Second Order System with a LHP Zero

Step responses for 0 = 1 and = 0.5 a = 0.25, 0.5 1, 2, 5
and 10.

Second Order System with a RHP Zero

Step responses for 0 = 1 and = 0.5 a = 0.25, -0.5 -1, -2,
-5, and -10.

Step Response

Step Response

1.5

2.5

0.5

Amplitude

Amplitude

1.5

0.5

1.5

0.5
2

10

10

15

Time (sec.)

15

Time (sec.)

Second Order System with a LHP Zero

Bode plot for = 0.5 and a/ 0 =0.1, 0.2 0.5 1 2 5 10

Second Order System with a RHP Zero

Bode plot for = 0.5 and a/ 0 =-0.1, -0.2 -0.5 -1 -2 -5 -10
Bode Diagrams

20

20

20

20
Phase (deg); Magnitude (dB)

Phase (deg); Magnitude (dB)

Bode Diagrams

40

60
100
50

40

60
0
50
100

0
150
50
200
100
150
2
10

250
1

10

10

10

10

Frequency (rad/sec)

300
2
10

10

10

10

10

Frequency (rad/sec)

c K. J. strm August, 2001

&

12