You are on page 1of 27

Design Via Root Locus

Improving Transient Response


%OS should be same, but response should be faster. We have pole at A on the root locus, but we want response like at B.

Compensator is used to improve the transient response of a system.

Compensator presents additional poles and zeros to the system.

CEN455: Dr. Ghulam Muhammad

Improving Steady-State Error


Compensator is also used to improve steady-state error.

If gain is higher, steady-state error is smaller, but %OS is larger.


Pure integration for improving steady-state error. Ideal compensator Pure differentiation for improving transient response.

Method of implementing compensator:


1. 2. 3. Proportional control systems: feed the error forward to the plant. Integral control systems: feed the integral of the error to the plant. Derivative control systems: feed the derivative of the error to the plant.

CEN455: Dr. Ghulam Muhammad

PI (Proportional + Integral) Controller


Steady-state error improvement: Place an open-loop pole at the origin. (Because this increases system type by one.) For step input: Type 0 finite error Type 1 zero error

Introducing a pole at origin changes the root locus! (Not desired)


CEN455: Dr. Ghulam Muhammad 3

PI Controller (contd.)
Adding a zero close to the pole at origin solves the problem. Why?

ZC = PC
Length (A to PC) 1 Length (A to ZC)
System type has been increased. Point A is still on root locus.

Ideal Integral Compensator: Pole at origin Zero at close to origin.


CEN455: Dr. Ghulam Muhammad 4

Ideal Integral Controller: Example

Operating at damping ratio = 0.174 Dominant poles: locate at the intersection of root locus and damping ratio line. Find gain K for those poles. K = 164.6 Find the third pole to the left of -10 with K = 164.6 Kp = 8.23. Steady-state error: 1 / (1+Kp) = 0.108.
CEN455: Dr. Ghulam Muhammad 5

Ideal Integral Controller: Example-(contd.)


After integral compensation:

Comparison in system response:

CEN455: Dr. Ghulam Muhammad

PI Controller Implementation

K2 K1 s + K1 K2 GC ( s ) = K1 + = s s
The value of zero can be adjusted by varying K2 / K1.

CEN455: Dr. Ghulam Muhammad

Lag Compensation
Ideal integral compensation: pole is in the origin, requires active network (costly). Real (passive) integral compensation: pole is close to origin (not in the origin), cheaper. Type is not increased. What about steady-state error? Static error constants:

K vo =

Kz1 z 2 .. p1 p2 ..

K vN =

( Kz1 z 2 ..)( zc ) ( p1 p2 ..)( pc )

CEN455: Dr. Ghulam Muhammad

Lag Compensation (contd.)


Effect on transient response: Angular contributions of zc and pc cancel out each other. Effect on gain, K: Lengths (P to zc) and (P to pc) are similar. (no problem) (no problem)

Effect on steady-state error:

K vN = K vo

zc > K vo pc

Though error is not zero, but error constant is increased by zc / pc.

Before lag compensation

After lag compensation


9

CEN455: Dr. Ghulam Muhammad

Lag Compensator Design


Refer to system in slide 5. Compensate the system to improve the steady-state error by a factor of 10. Before the error was 0.108, so the new error is 0.0108.

e( ) =

1 = 0.0108 K p = 91.59 1+ K p

Before it was 8.23

Hence, improvement,

zc K pN 91.59 = = = 11.13 8.23 pc K pO

For example, if pc = 0.01, zc = 11.13pc = 0.111

Lag compensated system.


CEN455: Dr. Ghulam Muhammad 10

Lag Compensator Design (contd.1)

Less error!

Root locus. Comparison


CEN455: Dr. Ghulam Muhammad 11

Lag Compensator Design (contd.2)


Different lag compensators
Faster steadystate response

Poles closer to origin has slower response! Skill Assessment Exercise 9.1

CEN455: Dr. Ghulam Muhammad

12

Improving Transient Response


PD Controller
To speed up the system: add a single zero to the path.

Same damping ratio line: same %OS Larger negative value for real part: shorter settling time

Larger imaginary part: Smaller peak time

CEN455: Dr. Ghulam Muhammad

13

Ideal Derivative Compensator

CEN455: Dr. Ghulam Muhammad

14

Ideal Derivative Compensator Design


Design for 16%OS, with a threefold reduction in settling time.
Ts = 4 = 4 = 3.320 1.205

Now settling time will be 1.107

4 = 3.613 Ts

d = 3.613 tan(180 120.26) = 6.193


CEN455: Dr. Ghulam Muhammad 15

Ideal Derivative Compensator Design (contd.)


zero angles - pole angles = (2k+1)180 -275.6 Zero angle = 275.6-180= 95.6
6.193 = tan(180 95.6) = 3.006 3.613

CEN455: Dr. Ghulam Muhammad

16

PD Controller

Ideal derivative compensator has two disadvantages: (1) needs active circuit, (2) differentiation is a noisy process.

CEN455: Dr. Ghulam Muhammad

17

Lead Compensator Design


30%OS, reduce settling time by a factor of 2.
4 Ts 1.007 = = 1.986 2 2

For compensator pole,


d = n = 4 / Ts = 2.014 d = 2.014 tan(110.98) = 5.252

Assume compensator zero at -5. zero angles - pole angles = (2k+1)180 Yields, angle of pole = -7.31

5.252 = tan 7.31 pc = 42.96 pc 2.014

CEN455: Dr. Ghulam Muhammad

18

PID Compensator
Improving Steady-State Error and Transient Response
2 K1 K2 K3 s + s+ 2 K3 K3 K1 s + K 2 + K 3 s K2 Gc ( s ) = K1 + + K3s = = s s s
Two zeros, and one pole at origin Ideal Integral Compensator: One pole at origin, and one zero Ideal Derivative Compensator: One zero

CEN455: Dr. Ghulam Muhammad

19

PID Compensator Design -1


Problem: Design a PID controller so that the system can operate with a peak time that is two-thirds that of the uncompensated system at 20% OS and with zero steady-state error for a step input.

Solution: Step 1: Evaluate the performance of the uncompensated system. Draw 20%OS line ( =0.456) <left-half plane>.

= 180 cos 1 (0.456) = 117.13

Find dominant poles at root locus. (angles adding up to 180) -5.415j10.57, and -8.169 with K = 121.5 Peak time = 0.297.

Tp =

d
20

CEN455: Dr. Ghulam Muhammad

PID Compensator Design -2


Step 2: Design the PD controller to meet the transient response.

d =

Tp

(2 / 3)(0.297)

= 15.87

d
tan 117.13

= 8.13

zc = 198.37 180 = 18.37

15.87 = tan 18.37 zc = 55.92 zc 8.13

Thus , PD controller is: GPD(s) = (s + 55.92)


CEN455: Dr. Ghulam Muhammad

Uncompensated
21

PID Compensator Design -3


Steps 3 and 4: Check step responses and redesign if needed.

Step 5: Design the PI controller. Let,


GPI ( s ) = s + 0.5 s

PD compensated

CEN455: Dr. Ghulam Muhammad

22

PID Compensator Design -4


Step 6: Find gains K1, K2, K3.
GPID ( s ) = K ( s + 55.92)( s + 0.5) 4.6( s + 55.92)( s + 0.5) = s s 4.6( s 2 + 56.42 s + 27.96) = s

Matching this equation with the equation in slide 19, we get:

K1 = 259.5, K 2 = 128.6, K 3 = 4.6


Steps 7 and 8: Summarize the findings and redesign if needed.

PID compensated

CEN455: Dr. Ghulam Muhammad

23

Physical Realization of PID Controller


Z1(s) C1 R2 Z2(s) C2

R1 Transfer function for unity gain,

1 R2 C1 R1C2 + + + R C s 2 1 R C s 2 1

179

R2 C1 + = 56.42, R1 C2

R2C1 = 1,

1 = 27.96 R1C2

358

Four unknown variables, and three equations. So, arbitrary select C2 = 0.1 F Then we get, R1 = 357.65 k, R2 = 179 k , C1 = 5.6 F.
CEN455: Dr. Ghulam Muhammad 24

Lag-Lead Compensator Design -1


Problem: Design a Lag-Lead compensator so that the system can operate at 20% OS and twofold reduction in settling time. Also tenfold improvement in steady-state error for a ramp input.

Solution:

Ts

Must be increased by a factor of 2.

n = 2(1.794) = 3.588

d = n tan 117.13 = 7.003


25

CEN455: Dr. Ghulam Muhammad

Lag-Lead Compensator Design -2


Let, lead compensator zero is at -6. Find and pc. (see slide 21)

So, we need improvement of 10 / 2.122 = 4.713

192.1 s ( s + 6)( s + 10) 1977 GLC ( s ) = s ( s + 10)( s + 29.1) G (s) =

Kv = 3.201 Kv(LC) = 6.794

Ratio = 2.122

CEN455: Dr. Ghulam Muhammad

26

Lag-Lead Compensator Design -3


Choose arbitrary lag compensator pole at -0.01. So, compensator zero is at -0.04713.

Glag ( s ) =

( s + 0.04713) ( s + 0.01) K ( s + 0.04713) GLLC ( s ) = s ( s + 10)( s + 29.1)( s + 0.01)

CEN455: Dr. Ghulam Muhammad

27

You might also like