Solution 7.9.9.13
ΣΣ
+ +
R C
K
r
G GG
c 1 2
θ
Figure 1: Maglev block diagram
For the system of Figure 1 wehave
G
1
(s)=
1
Js
and G
2
(s)=
1
s
:
Werst note that the proposed design turns a type 2 system into a type
1. The reason is that if we nd the equivalenttransfer function for the
subsystem consisting of G
1
and K
r
wehave
G
eqiv
(s) =
1=J
s + K
r
=J
:
Then the new blockdiagarm is shown in Figure 2 Wehavetwochoices. One
is to movethepole at s = ;K
r
=J to the left using K
r
and let G
c
= K
c
.In
that case wehaveanequivalentforward loop transfer function
G
c
G
p
(s)=
K
c
K
r
=J
s(s + K
r
=J)
=
K
s(s + p)
;;
where
K = K
c
K
r
=J and p = K
r
=J:
Toachieve the speed of reponse we require
=
1
p
0:15 s:
The rationale is that the time to reach90% of nal value is specied to be
00.2 s. If welet =0:15 then for critical damping we should reach63%of
1
Σ
+
R
C
GG
c2
G
equiv
Figure 2: Equivalentblock diagram
X
X
-p
-p
2
Im(s)
Re(s)
Figure 3: Calculation of gain for critical damping
nal value in about 0.15 s, whichshould assure that wereach90%ofnal
value in 0.2 s. However, this will not be sucienttogiveusthe required
steady state accuracy to a ramp input. Figure 3 shows why.Theoverall
gain required top achievecritical damping is
K =
p
2
p
2
=
p
2
4
:
The
K
v
=lim
s!0
sK
s(s + p)
=
p
4
:
Suppose welet p = 20 so that the compensated system has twopoles at
s = ;10, clearly meeting the specication on rise time. Then
K
v
=5;;
2
XX
X
-p
Im(s)
Re(s)O
-a
Figure 4: Root locus for K
r
=J small
XX
X
-p
Im(s)
Re(s)O
-a
p-a
-c -b
a - b
a - c
p
Figure 5: Calculation of gain for critical damping K
r
=J small
and we certainly do no meet the specication on steady state accuracy.
Therefore, we consider making the ratio K
r
=J smnall, and adding a lead
compensator as shown in Figure 4.The idea is to approximate an integrator.
As pointed out earlier this is somewhat contradictory since byadding rate
feedback, weeliminated an integrator whichwenoware trying to approx-
imate. In any case, we can now nd the gain that will critically damp the
system, namely
K =
(p; a) a (a; b)
a;c
;;
as shown in Figure 4. Since weareplacing both the pole at ;K
r
=J and
the zero of the lead close to the origin, wecangetaroughtidea of the
3
magnitude of p by simply letting
a; c = p and a;b = pand a
p
2
:
Then,
K =
p
2
4
:
If wenowletp=20, wehave
K
v
=
(p
2
=4) c
pb
=
p
4
)
c
d
If we let p =20thenforK
v
=50,wemust have
5
c
d
=50;;
or
c
d
=10
Thus, for convenience wecould make b =0:1andc =1. Then, letting
p = 24, we can now use the short MATLAB program:
gcgp = zpk([-1],[0 -0.1 -24],1)
rlocus(gcgp)
print -deps rl79913.eps
to generate the root locus shown in Figure 6. As can be seen from the gure
the break-out is very closetos = ;12. Wetherefore assume that a =12,
can compute the gain K
c
to place twopoles at s = ;12.
K
c
=
jsjjs +0:1jjs+24j
js +1j
j
s=;12
= 155:78;;
and
K
v
=
K
c
1
24 0:1
=64:9:
Thus, wemeet our specications. The following MATLAB program gener-
ates the unit step response shown in Figure 7.
Weseethat we are close to meeting the specications specications.
Therefore, weletp = 50, and then try to nd the gain that yields critical
damping with a little more precision. That is we solve the equation
K
c
=
jsjjs +0:1jjs+50j
js +1j
4
-25 -20 -15 -10 -5 0 5
-15
-10
-5
0
5
10
15
Real Axis
Imag Axis
Figure 6: Calculation of gain for critical damping K
r
=J small
Time (sec.)
Amplitude
Step Response
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Figure 7: Unit step response of compensated system
5
s ;24:7 ;24:9 ;25 ;24:8 ;24:85
K
c
629.50 629.545 629.536 629.533 629.541
Table 1: Locating break-out pointonreal axis
Time (sec.)
Amplitude
Step Response
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Figure 8: Unit step response of compensated system
for values around s = ;25 to nd the break-out point.
Table 1 summarizes that search. Thus the gain for critical damping is
K
c
= 630. Then we can generate the step response of the compensated
system, shown in Figure 8, using the MATLAB program:
gcgp = zpk([-0.2],[0 -0.02 -50],630)
tc = feedback(gcgp,1)
step(tc,2)
print -deps sr79913b.eps
Wecheck
K
v
=
6300:2
0:02 50
=126:
By examining this last
6