Solution: 7.9.2.10
R
+
C
G
c
G
p
Σ
-
Figure 1: Cascade Compensation with UnityFeedback
For the system shown in Figure 1 wehave
G
p
(s)=
10
(s +4)(s+20)
Two compensators are being considered
G
c1
(s)=
K
c
(s +0:5)
s
and G
c2
(s)=
K
c
(s +4)
s
Using G
c1
(s)wehave
G
c1
(s)G
p
(s) =
K
c
(s +0:5)
s
10
(s + 4)(s+20)
=
10K
c
(s +0:5)
s(s + 4)(s+20)
:
Since the pole and zero of the PI compensator are very close together and
distant from the desired closed loop pole locations, wecanassume that the
closed loop poles will be at s = ;12j9. Note with =0:8wehavea3-4-5
righttriangle. The situation is shown in Figure 2. That is weignore the net
angle contribution of the PI compensator and compute the gain required to
place the poles at s = ;12j9as
K =
jV
1
jjV
2
j
10
=
p
8
2
+9
2
p
8
2
+9
2
10
=14:5
For the second compensator wehave
G
c1
(s)G
p
(s) =
K
c
(s +4)
s
10
(s +4)(s +20)
=
10K
c
s(s +20)
:
1
Re(s)
Im(s)
-12
9
j 9
j 6
Closed Loop Poles for G
c1
(s)
-4-20
V
1
V
2
V
3
V
4
Figure 2: Closed Loop System
Re(s)
j Im(s)
Closed Loop Poles for G
c1
(s)
V
1
V
2
7.5
-20 -10
j 7.5
Figure 3: Closed Loop Poles Using G
c2
(s)
This is exactly T
N2
(s)asshown in Figure 3.
K =
jV
1
jjV
2
j
10
=
p
10
2
+7:5
2
p
10
2
+7:5
2
10
=15:625
2
(a)
For G
c1
(s)wehave
K
v
= lim
s!0
sG
c
(s)G
p
(s)
= lim
s!0
s
10K
c
(s +0:5)
s(s +4)(s +20)
= lim
s!0
145(s+0:5)
(s +4)(s +20)
= 0:90625
Thus
e
ss
=
1
K
v
=1:103:
For G
c2
(s)wehave
K
v
= lim
s!0
sG
c
(s)G
p
(s)
= lim
s!0
s
10K
c
s(s +20)
= lim
s!0
156:25
s +20
= 7:8125
Thus
e
ss
=
1
K
v
=0:128
Thus the second compensator has muchbetter steady state error character-
istics.
3
t
c(t)
Figure 4: Closed Loop System
t
c(t)
Figure 5: Step Response Using G
c2
(s)
(b)
For G
c1
(s), the response to a step will be
c(t)=
h
1+be
;t
+2jMje
;12t
cos(9t+ )
i
:
It is the term be
;t
that will delay the settling time. As discussed in the
text, the response is likely to look like that shown in Figure 4.
For G
c2
(s) the response can be determined exactly from the formulas
for T
N2
(s) and will look likeFigure 5. Thus G
c2
will yield both a shorter
settling time and less steady state error to a ramp input.
4