Solution 5.8.1.31
G
H
C
R
+
-
Figure 1: Standard Closed Loop Conguration
For the system shown above, let
GH =
K(s +2)
(s ; 1 ; j1)(s; 1+j1)
Drawtheroot locus.
Solution
First of all the closed loop transfer function is
C
R
=
G
1+GH
The root locus is merely a geometric means of nding all the possible solu-
tions to the equation
1+GH =0;;
for all positivevalues of K.The basic idea is to rewrite the equation as
GH = ;1=1
6
; 180
;;
By so doing weisolate on the left hand side a known, factored quantity
with the variable gain K merely a scaling factor. The poles and zeroes of
GH serveaslandmarks or points of reference that tell us where to look for
the closed loop poles. The rules are based upon decomposing GH into a
magnitude and angle. Let
s + a
i
= js + a
i
j
6
i
s + b
j
= js + b
j
j
6
j
1
Then
GH =
K
Q
m
i=1
(s + a
i
)
Q
m
=1
(s + b
j
)
=
K
Q
m
i=1
js + a
i
j
P
n
i=1
i
Q
n
j=1
js + b
j
j
P
n
j=1
j
=
K
Q
m
i=1
js + a
i
j
Q
n
j=1
js + b
j
j
!
0
@
m
X
i=1
i
;
n
X
j=1
j
1
A
Wenowcanwrite two equations:
K
Q
m
i=1
js + a
1
j
Q
n
j=1
js + b
j
j
!
=1;;
and
0
@
m
X
i=1
i
;
n
X
j=1
j
1
A
= ;180
We could arriveatthisdecomposition algebraically,bysimply substitut-
ing a value for s in the expression for GH and reducing it to a single complex
number. A more useful approach, if you don't haveamoderncalculator, is
to represent eachofthe terms s + a
i
and s + b
j
byavector in the s-plane.
This representation is shown in Figure 2 for the term s + b
j
, but it works
equally well for a zero.
This representation is extremely useful in thinking about almost all prob-
lems in so called `classical' control.
For the particular GH under study,we rst use the fundamental rule of
root locus:
The root locus on the real axis is locatedtotheleft of an odd count of
poles and zeros of GH.
We start at the far rightofthe real axis in the s-plane, and movetothe
left counting poles and zeros of GH whenever the countisodd, namely 1, 3,
5 :::there will be root locus in that section of the real axis. In the present
case there will be root locus to the left of the zero at s ; 2.
The second rule wecanuse is
The root locus startsatthepoles of GH and ends at the zeros of GH
As a consequence the root locus is as shown in Figure 3. Wecould nd
the break-in pointbytrialanderror, but in this case wecanuse the fact
that:
For any GH with one nite zeroandtwonite poles, the root locus o
the real axis will beaportion of a circle centeredatthezero
2
Re(s)
Im(s)
b
j
s
s + b
j
Figure 2: Vector representation of a pole or zero
As wecanseethe root locus has two`limbs,' with the limbs starting at
the twopoles of GH and meeting at the point s = ;5:16. One limb then
continuues on to the nite zero at s = ;2, while the other limbs migrates
out to the so-called `zero at ;infty.That is, every limbmigrates to a zero
of GH,it's just that some of the zeros lie innitely far from the origin of
the s-plane, at the `ends' of the asymptotes. The numberofasymptotes, of
course, is the dierence between the number of poles of GH and the number
of nite zeros of GH
3
Re(s)
Im(s)
X
X
Figure 3: Root locus
4