Control System I

Chapter 5
The Root-locus Design Method

Optimism is the faith that leads to achievement.

Nothing can be done without hope and confidence.
- Helen Keller

當一扇快樂的門關起時，另一扇門打開。但我們經常一直注視關
上的門，以致沒看見開啟的門。– 海倫‧凱勒
When one door of happiness closes, another opens, but
often we look so long at the closed door that we do not see
the one that has been opened for us. – Helen Keller

Chap 5, No. 1
Control System I
A Perspective of the Root-Locus Design Method
In Chapter 3
Rise time t r : the time it takes the system to reach the vicinity of its new set
point.
Settling time t s : the time it takes the system transients to decay.
Overshoot MP: the maximum amount the system overshoots its final value
divided by its final value (often expressed as a percentage).
Pole locations in terms of  , n , and  .
In Chapter 4
We study how the feedback (P, PI, PD,
PID) can influence stability and
dynamic response by changing the
system’s pole locations.
In this chapter, we study root-locus method.
This technique plots the locus of the closed-loop pole locations in the s-
plane as a parameter varies; this produces a root locus. Root locus is the
most used to study the effect of open-loop gain variations.
With Matlab, the rules of Root Locus are no longer needed for detail
plotting, but for a control designer it is essential to understand how
dynamic controllers will influence a locus in the design process.
Chap 5, No. 2
Control System I
5.1 The Root Locus of Feedback System
The closed-loop transfer function
Y s Dc  s  G  s 
  s  ,
R s 1  Dc  s  G  s  H  s  Dc  s  G s
The characteristic equation:
1  Dc  s  G  s  H  s   0, (*)
H s
Rewrite (*) as
bs
1  KL  s   0, where L  s   , K is the parameter to be adjusted
as
K varies from 0 to  , and then use the resulting plot to select the best
value of K. The plot of the result can be used to determine Dc  s  as the
compensation in the loop for some required dynamics.
The graph of ALL possible root of equation 1  KL  s   0, relative the
parameter K is called the root-locus, and the set of rules to construct
this graph is called the root-locus method of Evans.
L(s) is a rational function, b(s) is monic polynomial of degree m, while
a(s) is monic polynomial of degree n, where n  m.

Chap 5, No. 3
Control System I
5.1 The Root Locus of Feedback System
Thus b  s   s m  b1s m 1   bm   s  z1  s  z2  ...  s  z m    m
i 1  s  z i 

a  s   s n  a1s n 1   bn   s  p1  s  p2  ...  s  pn    ni1  s  pi 

The roots of b(s) = 0 are the zeros of L(s), while the roots of a(s) = 0 are
the poles of L(s). n  m
bs
1  KL  s   1  K  0,  a  s   Kb  s   0,
as
The roots of a  s   Kb  s   0
a  s   Kb  s    s  r1  s  r2  ...  s  rn   0
The equivalent and useful way to express the above equation.
bs 
1  KL  s   0, 1  K  0, 
as  The roots-locus form of a

a  s   Kb  s   0, L  s    
1 characteristic equation.
K
The root-locus method can be thought of as a method for inferring
dynamic properties of the closed-loop system as K changes.

Chap 5, No. 4
Control System I
Example 5.1 Root Locus of a Motor Position Control
m  s  Y s A
  G s  Let Dc  s   H  s   1, and c = 1
Va  s  U s s s  c
Solution: 1  DcGH  1  KL  0
1
Ls   0, K  A,  a  s   s  s  1 , b  s   1
s  s  1
a  s   Kb  s   s 2  s  K  0. Find the roots directly using the
1 1  4K quadratic formula
r1 , r2   
2 2
k 1

k 0 k  1/ 4
k 0 k 0

Chap 5, No. 5
Control System I
Example 5.1 Root Locus of a Motor Position Control
Summary for Example 5.1:
1
1. When K = 0, it is open loop. a  s   Kb  s   s 2  s  K  0. G  s  
pole: 0, -1; zero: none s  s  1
2. Two roots with two branches
3. When 0  K  1 4 , the roots move toward breakaway
to each other, and meet at s = -1/2 (K = point
1/4) with repeated roots., called a
breakaway point.
4. If K > 1/4 , the roots move to infinity with
equal real parts.
5. In some points along this locus
correspond to satisfactory transient
response, we can complete the design by
choice of the corresponding value of K.

Chap 5, No. 6
Control System I
Example 5.2 Root Locus with Respect to Plant Open-
Loop Pole and Zero
Solution:
s  2c s  2c
G s  , 1 G  s   1 .
s s  c s s  c
The corresponding closed-loop characteristic equation is
s 2  s  c  s  2   0. or s 2   c  1 s  2c  0.
s2 c = 0.172
Ls  ,  a  s   s  s  1 ,
s  s  1 c = 5.828 s = -0.586
s = -3.41
bs  s  2
The root locus form of the characteristic x x
equation is
Break-in point
c  s  2
1  0.
s  s  1
c1 c 2  6c  1
 r1 , r2   
2 2
The root locus method must be suitable for much higher-order systems for
which explicit solutions are difficulty to obtain. The general method will be
presented in guidelines that can be followed in a step-by-step manner to
sketch any root locus.
Chap 5, No. 7
Control System I
5.2 Guidelines for Determining a Root Locus
The purpose of the root locus is to show in a graphic form the general
trend of the roots of the closed-loop system as the parameter is varied.
There are two reasons for learning how to generate a root locus by hand:
(1) Use the manual method for design of fairly simple systems.
(2) Verify and understand computer-generated loci, such as Matlab.
Definition I: The root locus is the set of values of s for which 1 + KL(s) = 0 is
satisfied as the real parameter K varies from 0 to  . Typically 1 + KL(s) = 0
is the characteristic equation of the system, and in this case the root on the
locus are the closed-loop poles of that system.
1
From L  s    , L(s) will be negative if K is positive, and the phase
K of L(s) is 180 .
Definition II: The root locus of L(s) is the set of point in the s-plane where
the phase of L(s) is 180 . To test whether a point in the s-plane is on the
locus, we define the angle to the test point from a zero as  i and the angle
from a pole as i , then Definition II is expected as those points in the s-
plane where,
L  s    i   i  180  360   1 , for an int eger .
To solve a high-order polynomial by hand is very difficult, computing
the phase of transfer function is relatively easy.
Chap 5, No. 8
Control System I
5.2 Guidelines for Determining a Root Locus
Example:
s1 s0  1  2 j
Ls  x
s  s  5   s  2   4 
2
 
Test point: s0  1  2 j
x x
L  s    i   i  180  360   1 .
L  s0    1  1  2  3  4
x
 90  116.6  0  76  26.6
 129.2 .
L  s0   180  s0 not on the root locus.
However, measuring phase at every point in the s-plane is hardly practical.
Need some general guideline for determining where the root locus is.
Assume Matlab and its equivalent is always available to make an
accurate plot of a promising locus.

Chap 5, No. 9
Control System I Rules for Determining a Positive (180 ) Root Locus
Meanwhile, an example is used to express the idea, which is:
1 Zero: none
L(s ) 
s  s  4   16 
2
poles : 0,  4  4 j ,  4  4 j
 
RULE 1: Mark an x on the s-plane for each pole and an o for each zero. The
n branch of the locus start at each pole of L(s) and end on the zeros of L(s).
Im(s )
RULE 2: The loci on the real axis to the left of
an odd number of pole plus zero. x 4
Im(s ) 1  2 ,  1   2 Re(s )
Taking a test point 𝒔𝟎 on x
1 4
x 2
the real axis, the angles
of the two complex poles x 4
s0 cancel each other, as
Re(s ) would the angles from (K = 0)
x
complex conjugate zeros.
1
x 2 Any where in the root
locus L  s   180
Chap 5, No. 10
Control System I
Rules for Determining a Positive (180 ) Root Locus
RULE 2: The loci on the real axis to the left of an odd number of pole
plus zero.
The result is that the angle of 𝑳(𝒔𝟎 ) for 𝒔𝟎 on the real axis is given by the
angles from poles and zeros on the real axis.
s0 s0
0 180

Notice that it must be true that L(s )  180  360   1

⇒ the test points must be to the left of an odd number of real axis poles
plus zeros In the demo example:
x
5 4 3 2 1

x
6 5 4 3 2 1
Root locus on the red lines.

L(s ) 
1 x
s  s  4   16 
2
  Chap 5, No. 11
Control System I
Rules for Determining a Positive (180 ) Root Locus
RULE 3: For large s and K, n – m branches of the loci are asymptotic to
lines at angle  radiating out from the center s = on the real axis, where
180  360   1
  ,  1, 2, ..., n  m
nm
 pi   zi z
 . (will be explained later; pi : pole, i :zero )
nm
This can occur in two apparently different ways.
bs
 L  s   ,  L  s   0 as b  s   0. 
b( s )
0
as 1 K
a(s )
Thus for large K, the root locus goes to the zeros.  a(s )  Kb(s )  0
b( s ) s m  b1s m 1   bm  b( s )  
a(s )
 1 K  0 or 1  K n 0
a(s ) s  a1s n 1   an K

If n > m, that is if L(s) represents a physical system and K is the open-

loop gain, then
1
1  KL  s   0  L  s   ,  L  s   0 as K  
K

Chap 5, No. 12
Control System I Rules for Determining a Positive (180 ) Root Locus
Explanation:
In fact, for very large s, the poles and zeros appear to be in a cluster,
m zeros would essentially cancel the effects of m of the poles, and the
other (n - m) poles would appear to be at the same place.
 1 (s  z1 ) (s  zm ) The locus of (1)
1 K nm
 0 (1) is asymptotic
(s  ) ( s  p1 ) ( s  p m )
to the locus of
x K   1
  1 K nm
 0 (2) (2) for large K
 s   (s  ) and s.
x x x
It must satisfy the phase criterion
x 1
 180  360   1 ,
s   
nm

s0  Re j    s   
nm
  1 ,
⇒  180  360
R is fixed but large   n  m    180  360   1 ,
   n  m   180  360   1 ,  1, 2, ..., n  m
x
 pi   zi 180  360   1
 . ⇒   ,  1, 2, ..., n  m
nm nm
Chap 5, No. 13
Control System I
Rules for Determining a Positive (180 ) Root Locus
1
For the example, L(s )  ,
s (s  4)  16 
2
n=3,m=0
180  360   1
 
nm
 60 , 180 , 300 60 .   180 60

In order to sketch the asymptotic lines, 60 or 300
we need also know the value of 𝜶 .
𝜶 should stay on the real axis for
symmetrical poles and zeros
Find 𝜶
Suppose we have a monic polynomial with coefficients 𝒂𝒊 and roots 𝒑𝒊 , then
a  s   s n  a1s n 1   an  (s  p1 )(s  p2 ) (s  pn )
n
 an  ( p1)( p2 ) ( pn )   pi
i 1
n
n 1
s : a1   p1  p2  p3  pn  a1    pi
i 1

Chap 5, No. 14
Control System I
Rules for Determining a Positive (180 ) Root Locus
Similarly, b  s   s m  b1s m 1   bm  (s  z1)(s  z2 ) (s  zm )
m
 b1    zi
i 1
bs
For 1  KL  s   1  K  0,  a  s   Kb  s   0
as
The closed-loop characteristic equation (with roots 𝒓𝒊 )
s n  a1s n 1   an  K (s m  b1s m 1   bm )  0
a  s   Kb  s    s  r1  s  r2  ...  s  rn   0
If m < n – 1; that is, L(s) has at least two more poles than zeros, then
n n n
s n 1
: a1    ri    ri    pi (The common open-loop and
i 1 i 1 i 1 closed-loop sum is  a1 )
For large K, m of the root 𝒓𝒊 approach the zeros 𝒛𝒊 and (n - m) of the
roots approach the branches of the asymptotic system
1
whose poles add up to (n  m )
( s   )n  m
Chap 5, No. 15
Control System I Rules for Determining a Positive ( 180 ) Root Locus
More detail
bs 1 (s  z1) ( s  zm ) z1  z2   zm  
 p1  p2   pn  m 1  
a  s  (s  )n  m (s  p1 ) (s  pm )

System whose poles add up to 𝒏 − 𝒎 𝜶.

n m
  ri  (n  m )   zi   pi
i 1 i 1

Open-loop poles Open-loop zeros

 
 pi   zi

 a1  b1 as k  
nm nm

Conclusion: The sum of the roots equals the sum of those roots that
go to infinity plus the sum of those roots that go to the zero.

Chap 5, No. 16