Math 5
Math 5
Math 5
stable or even makes the system unstable.Therefore, the values of must be chosen care-
fully to ensure a proper transient response. By properly designing the PI controller, it is possible
to make the transient response to a step input exhibit relatively small or no overshoot.The speed
of response, however, becomes much slower. This is because the PI controller, being a low-pass
The PD controller is a simplified version of the lead compensator. The PD controller has the
The value of Kp is usually determined to satisfy the steady-state requirement. The corner
frequency is chosen such that the phase lead occurs in the neighborhood of the gain
crossover frequency. Although the phase margin can be increased, the magnitude of the com-
pensator continues to increase for the frequency region (Thus, the PD controller is a
high-pass filter.) Such a continued increase of the magnitude is undesirable, since it amplifies
high-frequency noises that may be present in the system. Lead compensation can provide a
sufficient phase lead, while the increase of the magnitude for the high-frequency region is
very much smaller than that for PD control. Therefore, lead compensation is preferred over
PD control.
Because the transfer function of the PD controller involves one zero, but no pole, it is not
possible to electrically realize it by passive RLC elements only. Realization of the PD controller
using op amps, resistors, and capacitors is possible, but because the PD controller is a high-pass
filter, as mentioned earlier, the differentiation process involved may cause serious noise problems
in some cases.There is, however, no problem if the PD controller is realized by use of the hydraulic
or pneumatic elements.
The PD control, as in the case of the lead compensator, improves the transient-response
characteristics, improves system stability, and increases the system bandwidth, which implies fast
rise time.
The PID controller is a combination of the PI and PD controllers. It is a lag–lead compensator.
Note that the PI control action and PD control action occur in different frequency regions. The
PI control action occurs at the low-frequency region and PD control action occurs at the high-
frequency region. The PID control may be used when the system requires improvements in both
A–8–2. Show that the transfer function U(s)/E(s) of the PID controller shown in Figure 8–44 is
U(s)
E(s) = K0
T1 + T2
T1
c1+
AT1 + T2Bs +
T1T2s
T1 + T2
1Td 6 v.
1Td
Gc(s),
Kp and Ti
616 Chapter 8 / PID Controllers and Modified PID Controllers
Z1 Z2
C1 C2
R1
R2
R4
R3
R5
Figure 8–45
Modified PID
controller.
Solution
A–8–3. Consider the electronic circuit involving two operational amplifiers shown in Figure 8–45. This
is
a modified PID controller in that the transfer function involves an integrator and a first-order
lag term. Obtain the transfer function of this PID controller.
Solution. Since
and
we have
Also,
Eo(s)
E(s) = - R5
R4
E(s)
Ei
(s) = - Z2
Z1
= - AR2C2s + 1BAR1C1s + 1B
C2sAR1 + R3 + R1R3C1sB
Z2 = R2 +
C2s
Z1 = 1
R1
+ C1s
+ R3 = R1 + R3 + R1R3C1s
1 + R1C1s
= K0
T1 + T2
T1
c1+
AT1 + T2Bs +
T1T2s
T1 + T2
= K0 a 1 +
T1s + T2s +
T2
T1
= K0 a 1 +
T1s
b A1 + T2sB
T1s
Ka1
K0
T1s
1 + T1s
1 + T2s
b
U(s)
E(s) = K
1+Ka1
K0
T1s
1 + T1s
1 + T2s
Openmirrors.com
R(s) 1 C(s)
Tds
+–
Figure 8–46
Approximate
differentiator.
R(s) C(s)
D(s)
K(as + 1) (bs + 1)
s2 + 3.6s + 9 + – +
Figure 8–47
PID-controlled
system.
Consequently,
Notice that R1C1 and R2C2 determine the locations of the zeros of the controller, while R1, R3, and
C1 affect the location of the pole on the negative real axis. R5/R4 adjusts the gain of the controller.
A–8–4. In practice, it is impossible to realize the true differentiator. Hence, we always have to
approxi-
One way to realize such an approximate differentiator is to utilize an integrator in the feedback path.
Show that the closed-loop transfer function of the system shown in Figure 8–46 is given by the pre-
ceding expression. (In the commercially available differentiator, the value of g may be set as 0.1.)
Solution. The closed-loop transfer function of the system shown in Figure 8–46 is
Note that such a differentiator with first-order delay reduces the bandwidth of the closed-loop
control system and reduces the detrimental effect of noise signals.
A–8–5. Consider the system shown in Figure 8–47. This is a PID control of a second-order plant G(s).
As-
sume that disturbances D(s) enter the system as shown in the diagram. It is assumed that the ref-
erence input R(s) is normally held constant, and the response characteristics to disturbances are
C(s)
R(s) =
1+
gTds
= Tds
1 + gTds
Tds
1 + gTds
Tds
= R5R2
R4R3
as+
R1C1
bas+
R2C2
b
sas+
R1 + R3
R1R3C1
Eo(s)
Ei
(s) = Eo(s)
E(s)
E(s)
Ei
(s) = R5
R4AR1 + R3BC2
AR1C1s + 1BAR2C2s + 1B
s a R1R3
R1 + R3
C1s + 1 b
Design a control system such that the response to any step disturbance will be damped out
quickly (in 2 to 3 sec in terms of the 2% settling time). Choose the configuration of the closed-loop
poles such that there is a pair of dominant closed-loop poles. Then obtain the response to the
unit-step disturbance input. Also, obtain the response to the unit-step reference input.
For the disturbance input in the absence of the reference input, the closed-loop transfer function
becomes
(8–14)
The specification requires that the response to the unit-step disturbance be such that the settling
time be 2 to 3 sec and the system have a reasonable damping. We may interpret the specification
as and vn=4 radsec for the dominant closed-loop poles.We may choose the third pole
at s=–10 so that the effect of this real pole on the response is small. Then the desired charac-
(s+10)As2
+2*0.5*4s+42 B=(s+10)As2
+4s+16B=s3
+14s2
+56s+160
s3
+(3.6+Kab)s2
+(9+Ka+Kb)s+K=0
Hence, we require
3.6+Kab=14
9+Ka+Kb=56
K=160
which yields
ab=0.065, a+b=0.29375
The PID controller now becomes
=s
(s + 10)As2 + 4s + 16B
D(s)
Cd(s) = s
= 160A0.065s2 + 0.29375s + 1B
z = 0.5
=s
Cd(s)
D(s) = s
Openmirrors.com
Example Problems and Solutions 619
Clearly, for a unit-step disturbance input, the steady-state output is zero, since
The response to a unit-step disturbance input can be obtained easily with MATLAB. MATLAB
Program 8–9 produces a response curve as shown in Figure 8–48(a). From the response curve, we
see that the settling time is approximately 2.7 sec.The response damps out quickly.Therefore, the
sCd(s) = limsS0
s2
(s + 10)As2 + 4s + 16B
s=0
numd = [1 0];
dend = [1 14 56 160];
t = 0:0.01:5;
[c1,x1,t] = step(numd,dend,t);
plot(t,c1)
grid
xlabel('t Sec')
denr = [1 14 56 160];
[c2,x2,t] = step(numr,denr,t);
plot(t,c2)
grid
xlabel('t Sec')
The response to a unit-step reference input can also be obtained by use of MATLAB Program 8–9.
The resulting response curve is shown in Figure 8–48(b).The response curve shows that the max-
imum overshoot is 7.3% and the settling time is 1.2 sec.The system has quite acceptable response
characteristics.
Cr(s)
A–8–6. Consider the system shown in Figure 8–49. It is desired to design a PID controller such that
the dominant closed-loop poles are located at For the PID controller,
choose a=1 and then determine the values of K and b. Sketch the root-locus diagram for the
designed system.
Solution. Since
Gc(s)G(s) = K (s + 1)(s + b)
s
s2 + 1
s = -1 ; j13 .
Gc(s)
14
–4
12
10
–2
t Sec
(a)
Output to Reference Input
1.2
0.6
0.8
1.0
0.4
0.2
t Sec
(b)
Figure 8–48
(a) Response to
unit-step disturbance
unit-step reference
input.
Openmirrors.com
–5 0 –4 –3 –2 –1 1
Imag Axis
–2
–3
–1
Figure 8–50
Root-locus plot of
the compensated
system.
R(s) C(s)
Gc(s) G(s)
(s + a ) (s + b)
sK
s2 + 1 + –
Figure 8–49
PID-controlled
system.
the sum of the angles at one of the desired closed-loop poles, from the zero at
90°-143.794°-120°-110.104°=–283.898°
Hence the zero at s=–b must contribute 103.898°. This requires that the zero be located at
b=0.5714
or
K=2.3333
From this equation a root-locus plot for the compensated system can be drawn. Figure 8–50 is a
root-locus plot.
s2 + 1
2K (s + 1)(s + 0.5714)
s2 + 1
s= -1+j13
=1
s = -1 + j13 ,
The closed-loop poles are located at and s=–0.3333. A unit-step response curve
is shown in Figure 8–51. The closed-loop pole at s=–0.3333 and a zero at s=–0.5714 produce
A–8–7. Consider the system shown in Figure 8–52. Design a compensator such that the static velocity
, phase margin is 50°, and gain margin is 10 dB or more. Plot unit-step and
unit-ramp response curves of the compensated system with MATLAB.Also, draw a Nyquist plot
of the compensated system with MATLAB. Using the Nyquist stability criterion, verify that the
Solution. Since the plant does not have an integrator, it is necessary to have an integrator in the
where is to be determined later. Since the static velocity error constant is specified as
4 sec−1
, we have
Kv = limsS0
sGc(s)s + 0.1
s2 + 1 = limsS0
sK
s Gˆ
c(s)s + 0.1
s2 + 1 = 0.1K = 4
Gˆ
c(s)
Gc(s) = K
s Gˆ
c(s), limsS0
Gˆ
c(s) = 1
s = -1 ; j13
C(s)
Time (sec)
0 8 2 4 6 10 12
Amplitude
0.4
0.8
1.2
0.6
1
0.2
Figure 8–51
Unit-step response of
the compensated
system.
Gc(s) s + 0.1
s2 + 1 +
Figure 8–52
Control system.
Openmirrors.com
MATLAB Program 8–10 produces a Bode diagram of G(s) as shown in Figure 8–53.
s(s2 + 1)
Gc(s) = 40
s Gˆ
c(s)
We need the phase margin of 50° and gain margin of 10 dB or more. Let us choose to be
Then Gc(s) will contribute up to 90° phase lead in the high-frequency region. By simple MATLAB
trials, we find that a = 0.1526 gives the phase margin of 50° and gain margin of dB. +q
Gˆ
c(s) = as + 1 (a 7 0)
Gˆ
c(s)
bode(num,den)
Figure 8–53
Bode diagram of
G(s)=
40(s+0.1)/[s(s2
−200
−50
−100
−150
−100
300
200
100
sys = tf(num,den);
w = logspace(-2,2,100);
bode(sys,w)
[Gm,pm,wcp,wcg] = margin(sys);
GmdB = 20*log10(Gm);
[GmdB,pm,wcp,wcg]
ans =
Figure 8–54
Bode diagram of
G(s)=40(s+0.1)
(0.1526s+1)/
[s(s2
−200
50
−50
−100
−150
100
−50
100
50
Gc(s) = 40
s Gˆ
c(s) = 40(0.1526s + 1)
See MATLAB Program 8–11 and the resulting Bode diagram shown in Figure 8–54. From this
Bode diagram we see that the static velocity error constant is 4 sec−1
Openmirrors.com
We shall next check the unit-step response and the unit-ramp response of the designed system.
s=3.0032+j5.6573
s=3.0032-j5.6573
s=0.0975
MATLAB Program 8–12 will produce the unit-step response curve of the designed system.The re-
sulting unit-step response curve is shown in Figure 8–55. Notice that the closed-loop pole at
s = −0.0975 and the plant zero at s = −0.1 produce a long tail of small amplitude.
C(s)
s3 + 6.104s2 + 41.6104s + 4
= 6.104s2 + 40.6104s + 4
s(s2 + 1)
= 40(0.1526s + 1)
s + 0.1
s2 + 1
t = 0:0.01:10;
step(num,den,t)
grid
Figure 8–55
Unit-step response of
C(s)/R(s)=(6.104s2
40.6104s+4)/(s3
6.104s2
0 1 2 3 4 5 6 7 9 8 10
Amplitude
1.4
0.8
1.2
0.4
0.2
1
0.6
Step Response
MATLAB Program 8–13 produces the unit-ramp response curve of the designed system. The
t = 0:0.01:20;
c = step(num,den,t);
plot(t,c,'-.',t,t,'-')
title('Unit-Ramp Response')
xlabel('t(sec)')
text(13.8,11.2,'Output')
Figure 8–56
Unit-ramp response
of C(s)/R(s)=
(6.104s2
+40.6104s+
4)/(s3
+6.104s2
+
41.6104s+4).
t (sec)
0 2 4 6 8 10 12 14 16 18 20
20
12
18
16
10
14
Unit-Ramp Response
Nyquist Plot. Earlier we found that the three closed-loop poles of the designed system are
all in the left-half s plane. Hence the designed system is stable. The purpose of plotting Nyquist
diagram here is not to test the stability of the system, but to enhance our understanding of Nyquist
stability analysis. For a complicated system, Nyquist plot will look complicated enough that it is
not easy to count the number of encirclements of the −1+j0 point.
Openmirrors.com
Because the designed system involves three open-loop poles on the jw axis, the Nyquist dia-
Define the open-loop transfer function of the designed system as G(s). Then
Let us choose a modified Nyquist path in the s plane as shown in Figure 8–57(a). The modified
path encloses three open-loop poles (s=0, s=j1, s=j1). Now define s1=s+ . Then,
the Nyquist path in the s1 plane becomes as shown in Figure 8–57(b). In the s1 plane, the open-
A MATLAB program to obtain the Nyquist plot is shown in MATLAB Program 8–14. The re-
= 6.104s1
2 + 40.48832s1 + 3.5945064
s1
3 - 0.03s1
2 + 1.0003s1 - 0.010001
= 6.104(s1
2 - 0.02s1 + 1.0001)
s = s1 - s0 s , 0 = 0.01.
s0
s2 + 1 = 6.104s2 + 40.6104s + 4
s(s2 + 1)
Figure 8–57
(a) Modified
Nyquist path in
the s plane;
the s1 plane.
s plane s1 plane jv
(a) (b)
0s
s0
jv
0s
MATLAB Program 8–14
nyquist(num,den)
Figure 8–58
Nyquist plot.
Real Axis
Nyquest Diagram
Imaginary Axis
2500
2000
1500
1000
500
−500
−1000
−1500
−2000
−2500
Figure 8–59
Redrawn Nyquist
plot.
Im
Re v = 0+
v = 0−
v = −`
v = +`
Using the Nyquist plot obtained here, it is not easy to determine the encirclements of the −1+j0
point by the Nyquist locus. Therefore, we need to redraw this Nyquist plot qualitatively to show
the details near the −1+j0 point. Such a redrawn Nyquist diagram is shown in Figure 8–59.
From this diagram we find that the −1+j0 point is encircled counterclockwise three times.
Hence, N −3. Since the open-loop transfer function has three poles in the right-half s1 plane,
we have P 3.Then, we have Z N+P 0.This means that there are no closed-loop poles in
A–8–8. Show that the I-PD-controlled system shown in Figure 8–60(a) is equivalent to the PID-
controlled
Openmirrors.com
The closed-loop transfer function C(s)/R(s) of the PID-controlled system with input filter
The closed-loop transfer functions of both systems are the same.Thus,the two systems are
equivalent.
A–8–9. The basic idea of the I-PD control is to avoid large control signals (which will cause a
saturation
phenomenon) within the system. By bringing the proportional and derivative control actions to
the feedback path, it is possible to choose larger values for than those possible by the
Compare, qualitatively, the responses of the PID-controlled system and I-PD-controlled system
Solution. Consider first the response of the I-PD-controlled system to the disturbance input.
Since, in the I-PD control of a plant, it is possible to select larger values for than those
of the PID-controlled case, the I-PD-controlled system will attenuate the effect of disturbance
Next, consider the response of the I-PD-controlled system to a reference input. Since the
I-PD-controlled system is equivalent to the PID-controlled system with input filter (refer to Prob-
lem A–8–8), the PID-controlled system will have faster responses than the corresponding I-PD-con-
trolled system, provided a saturation phenomenon does not occur in the PID-controlled system.
Kp and Td
Kp and Td
Kp
Tis
Gp(s)
1 + Kp a 1 +
C(s)
R(s) = 1
1 + Tis + TiTds2
Kp a 1 +
1 + Kp a 1 +
C(s)
R(s) =
Kp
Tis Gp(s)
1 + Kp a 1 +
(b)
Gp(s)
R(s) C(s)
Kp(1 + + Tds) 1
Tis
1 + Tis + TiTds2 + –
(a)
Kp
Tis Gp(s)
R(s) C(s)
Kp(1 + Tds)
+–+–
Figure 8–60
(a) I-PD-controlled
system;
(b) PID-controlled
filter.
A–8–10. In some cases it is desirable to provide an input filter as shown in Figure 8–61(a). Notice that
the
input filter is outside the loop. Therefore, it does not affect the stability of the closed-
loop portion of the system.An advantage of having the input filter is that the zeros of the closed-loop
transfer function can be modified (canceled or replaced by other zeros) so that the closed-
Show that the configuration in Figure 8–61(a) can be modified to that shown in Figure 8–61(b),
(8–15)
Thus
or
(8–16)
= Gf(s)
Gc(s)Gp(s)
1 + Gc(s)Gp(s)
C(s)
1 + Gc(s)Gp(s)
C(s)
1 + Gc(s)Gp(s)
C(s) = Gp(s)U(s)
C(s)
R(s) = Gf(s)
Gc(s)Gp(s)
1 + Gc(s)Gp(s)
Gf(s)
(a)
(b)
Gc(s)
R(s) C(s)
G Gp(s) f(s)
Gc(s)
Gd(s)
Gp(s)
U(s)
+–
+–+
+ Figure 8–61
diagram.
Openmirrors.com
which is the same as Equation (8–15). Hence, we have shown that the systems shown in Figures
It is noted that the system shown in Figure 8–61(b) has a feedforward controller In
such a case, does not affect the stability of the closed-loop portion of the system.
A–8–11. A closed-loop system has the characteristic that the closed-loop transfer function is nearly
equal
to the inverse of the feedback transfer function whenever the open-loop gain is much greater
than unity.
The open-loop characteristic may be modified by adding an internal feedback loop with a
characteristic equal to the inverse of the desired open-loop characteristic. Suppose that a
Determine the transfer function H(s) of the element in the internal feedback loop so that the inner
Solution. Figure 8–62(a) shows the original system. Figure 8–62(b) shows the addition
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: