MTRN3210 Feedback Control Systems

Problem Set 5 Solutions

T1 2025

1. (a) Root locus analysis is a graphical method for examining how the
poles of the closed-loop system change with variation of a certain
system parameter (K in this problem). Thus if a point s0 lies
on the root locus, it means that s0 is a root of the closed-loop
b(s)
system which means 1 + KG(s0 ) = 0. So if G(s) = a(s) then
a(s0 ) + Kb(s0 ) = 0.

(b) If we find the phase of both sides of the characteristic equation of

the closed-loop system we have ∠K + ∠G(s0 ) = ∠ − 1. Then if
K > 0 we have ∠G(s0 ) = 180o .

(c) We first write the characteristic equation in the form of K =

−1
G(s0 )
and then find the magnitude of both sides: K = |G(s1 0 )| .
This equation is called the magnitude relationship (or magnitude
condition, magnitude criterion) and is used to find the value of K
when the closed-loop system has a pole at s0 .

(d) If K < 0, then ∠K = 180o . So the phase equation will change to

180o + ∠G(s0 ) = 180o and thus ∠G(s0 ) = 0o . In this case the gain
equation is |K| = |G(s1 0 )| .

1
2. (a) Poles: p1 = p2 = 0, p3 = −10. Zeros: z = 3.
(b) From −10 to 3.
(c) We first find the departure angles from p1 = p2 = 0:
X X
2ϕdep = ψi − ϕi − 180 − 360(l − 1)
= 180 − 0 − 180 − 360(l − 1)
= 360(l − 1)

So the departure angle from these two poles are 0o and 180o . The
departure angle from p3 = −10 is 0o and the arrival angle at z = 3
is 180o .
(d) The number of asymptotes: n − m = 2.
o o (l−1)
Asymptotes angles: ϕl = 180 +360
n−m
, l = 1, 2. Thus ϕ1 = 90o ,
ϕ2 = 270o . P P
i−
Asymptotes center: σ = pn−m zi
= −6.5.
(e) Break-away point:

−1 s2 (s + 10) dK s(2s2 + s − 60)

K= =− ⇒ =− =0
G(s) s−3 ds (s − 3)2

Thus

 s=0 ⇒ K=0
s = 5.2329 ⇒ K = −186.8 (not acceptable as K < 0)
s = −5.7329 ⇒ K = 16.06


Figure 1 shows the root locus plot generated using numerical soft-
ware.

2
 Root Locus
25

20

15
Imaginary Axis (seconds−1)

System: untitled1
10 Gain: 16.1
Pole: −5.73
5 Damping: 1
Overshoot (%): 0
Frequency (rad/s): 5.73
0

−5

−10

−15

−20

−25
−12 −10 −8 −6 −4 −2 0 2 4
Real Axis (seconds−1)

Figure 1: Root locus

3. (a) Open loop poles: 0, 0, −3, −5, −6

Open loop zeros: −1, −8 ± 6j.
(b) (−6, −5) ∪ (−3, −1).
(c) The arrival angle at −8 + 6j:
X X
ψ= ϕi − ψi + 180
−1 6 −1 6 −1 6 −1 6


o o o o




= 2 180 − tan + 180 − tan + 180 − tan + 180 − tan
8 5 3 2
−1 6
 
o o o


− 90 + 180 − tan + 180
7
≈ 231o

Similarly, the arrival angle at −8 − 6j is 129o . We could also

find this angle using the fact that root locus has mirror symmetry
about the real axis.

3
 Departure angle from p = 0:
X X
2ϕ = ψi − ϕi − 180o − 360o (l − 1)
 6
6  o 
= 0 + 360o − tan−1 + tan−1 − 0 + 0o + 0o − 180o
8 8
o
− 360 (l − 1)

Thus the departure angle from the poles at the origin are 90o and
270o (= −90o ).
Other arrival and departure angles are trivial.
(d) The number of asymptotes: n − m = 2.
o o (l−1)
Asymptotes angles: ϕl = 180 +360
n−m
, l = 1, 2. Thus ϕ1 = 90o ,
ϕ2 = 270o . P P
i−
Asymptotes center: σ = pn−m zi
= (−6−5−3)−(−1−8−6j−8+6j)
2
=
1.5.
(e) Break-away point (not easy to calculate by hand for this problem):

 s=0 ⇒ K=0
dK
=0 ⇒ s = −3.92 ⇒ K = −0.21 (Unacceptable)
ds
s = −5.57 ⇒ K = 0.1


Note that when we solve dK/ds = 0, we are only interested in the

real values for s as we want to find the break-in and break-away
points on the real axis.
Imaginary axis crossover points: we set s = jω and put it into
the characteristic equation and solve it, which is not easy to do
by hand: ω ≈ ±2.69, K ≈ 3.7.
(f)

4
 Root Locus

4
Imaginary Axis (seconds−1)

System: untitled1
Gain: 0.102
Pole: −5.58 − 0.0151i
2 Damping: 1
Overshoot (%): 0
Frequency (rad/s): 5.58
0

−2

System: untitled1
−4 Gain: 3.7
Pole: −0.00348 − 2.69i
Damping: 0.00129
Overshoot (%): 99.6
−6 Frequency (rad/s): 2.69

−10 −8 −6 −4 −2 0 2
Real Axis (seconds−1)

Figure 2: Root locus

4. The transfer function is:

K
Y (s) K+s Ks
= 1 K
= 2 . (1)
U (s) 1 + s K+s s + Ks + K

The characteristic equation is

s2 + Ks + K = 0 ⇒ s2 + K(s + 1) = 0
s+1
⇒ 1 + K 2 = 0.
s
So we can assume we have an open-loop system G(s) = s+1 s2
and want
to find the closed-loop poles location for all positive values of K. The
root locus of this system is the same as the root locus of (1).

• Open loop zeros: s = −1

Open loop poles: s1 = s2 = 0.
• Number of branches: n = 2

5
 • n = 2 ⇒ 2 branches start at open-loop poles and n−m = 2−1 = 1
branch goes to infinity.
• Real locus: (−∞, −1].
• Break-in and break-away point(s):

s2 dK 2s(s + 1) − s2
K=− ⇒ =− =0
s+1 ds (s + 1)2
s2 + 2s
⇒ − =0
(s + 1)2

s=0 ⇒ K=0
⇒
s = −2 ⇒ K = 4

Figure 3: Root locus

5. (a) 1 LHP zero and 3 LHP poles:

Although, all poles and zeros are in the LHP, it does not neces-
sary mean that all values of gain, K, will keep the system stable.

6
 If the asymptotes are in the RHP (which means there exist jω-
crossings), then the system to be unstable with a high gain.
For example, consider a system with a zero at s = −5 and poles
at s = −2, s = −1 ± j. Although, it has 3 LHP poles and 1 LHP
zero, the root locus plot of this system shows that there are jω-
crossings and the system becomes unstable with high gain. See
Figure 4.

Figure 4: Root locus

(b) 3 LHP zeros, 1 RHP pole, 2 LHP poles:

Root locus branches start at the poles and end at the zeros. If the
number of poles is greater than the number of zeros, then some
branch(es) will go to infinity.
In this case, there is one unstable pole (RHP pole). Since there
are equal numbers of poles and zeros, all branches will end at the
zeros and because all the zeros are in the LHP, a large enough K
can stabilize the closed-loop system.
(c) 4 LHP zeros, 1 RHP zero, 5 LHP poles:

7
Similarly to the previous problem, all branches will start at the
poles and end at the zeros. This case also has equal number of
both. However, one of the zeros is in the RHP.
Therefore, a large value of K can cause the closed-loop system to
be unstable.

8
6. From the root-locus, we see the open-loop system G(s) has two poles
at s = −2 and s = 3. So the transfer function of G(s) is
α
G(s) = ,
(s + 2)(s − 3)

where α is a positive constant.

(a) We cannot cancel a RHP pole by a RHP zero. Note that if we

do this for a real system for which the location of the pole is
not accurately known, then we always have a branch in the RHP,
which means the system is unstable for all K > 0.

Figure 5: Root locus for controller (a)

(b) It is OK if we cancel a stable pole with a stable zero. In this

case, we are cancelling a pole of G(s) as s = −2 which a zero of
Gc (s). We see in Figure 5 that the closed-loop system is stable for
K > 15.

9
 Figure 6: Root locus for controller (b)

(c) The system is also stable in this case for K > 24.

Figure 7: Root locus for controller (c)

10
7. (a)
s−2
G(s) =
s(s + 1)

Figure 8: Root locus for system (a)

(b)
2−s
G(s) =
s(s + 1)
This system can be written as

(−1)(s − 2)
G(s) = .
s(s + 1)
s−2
So we can assume we want to plot the root-locus of G(s) = s(s+1)
for K < 0.

11
Figure 9: Root locus for system (b)

12