Root Locus

Signals and Systems: 3C1

Control Systems Handout 3
Dr. David Corrigan
Electronic and Electrical Engineering
[email protected]

• Recall, the example of the PI controller car cruise control system. A con-
sequence of choosing the parameter for the integral control, Ki , is that it
affects the location of the poles and zeros of the cruise control system on
the s-plane. This different pole/zero locations for Ki = 50 and Ki = 100
are shown below

Figure 1: Pole/Zeros Diagrams in the car cruise control example for Ki = 50 and Ki = 100

• Knowing the locations of the poles and zeros is important because

1. It affects the transient response.

2. It indicates whether the system is stable or not.

1
Root Locus

• An important property of the control system is the variation of these loca-

tions as the controller gain changes. Therefore we need to quantify these
variations.

• In this handout we introduce the idea the root locus which as a graphical
means of quantifying the variations in pole locations (but not the zeros).

3C1 Signals and Systems 2 Control Systems

Root Locus 1 CLOSED LOOP SYSTEM STABILITY

1 Closed Loop System Stability

Recall that any system is stable if all the poles lie on the LHS of the s-plane.
Ensuring stability for an open loop control system, where H(s) = C(s)G(s),
is straightforward as it is sufficient merely to use a controller such that the
cascade C(s)G(s) only has poles on the LHS of the complex plane. The poles
of the open loop system will not be affected by the gain of the controller. i.e.
If C(s)G(s) is stable then KC(s)G(s) will be stable for all values of K.

However for a closed loop system this will not be the case. Considering the
following block diagram

we get a transfer function

C(s)G(s)
H(s) =
1 + C(s)G(s)
K
(s+1)(s+2)(s+3)
= K
1 + (s+1)(s+2)(s+3)
K
= (1)
s3 + 6s2 + 11s + 6 + K

Looking at the open loop transfer(H(s) = C(s)G(s)) for this system, all the
poles (s = −1, s = −2, s = −3) lie in the LHS of the s-plane and hence the
system is stable.

For the closed-loop system if the gain K = 10 the poles lie in the LHS of the
plane and so it is a stable system. However, if the gain is increased to K = 200
then there are 2 poles on the RHS of the s-plane and so the system is unstable.

3C1 Signals and Systems 3 Control Systems

Root Locus 1 CLOSED LOOP SYSTEM STABILITY

Figure 2: Pole Zero Plots for the system transfer function in Eq. (1) for K = 10 (left) and
K = 200 (right).

It is also possible to have an unstable open loop system and a stable closed
loop system. For example, consider the previous example where G(s) is now
s+1
G(s) = . (2)
(s − 1)(s + 2)(s + 3)
As G(s) has a pole at s = 1, the open loop system C(s)G(s) is clearly unstable.
However in the closed loop system, if we choose K = 10 then the poles are
located as shown in Fig. 3 and the system is stable.

Figure 3: Pole Zero Plots for the system transfer function in Eq. (2) for K = 10.

3C1 Signals and Systems 4 Control Systems

Root Locus 2 ROOT LOCUS

2 Root Locus

Revisiting the example in the last page, the system above has a transfer
function
K
H(s) = .
s3 + 6s2 + 11s + 6 + K

Starting at a value of K = 0 calculate the locations of the poles of H(s)

and repeat for increasingly large values of K. If we trace the locations of these
poles as they move across the s-plane the following plot is obtained. This plot
is called the root locus.

3C1 Signals and Systems 5 Control Systems

Root Locus 2 ROOT LOCUS

Observations

• Because we have a 3rd Order System, there are 3 separate plots on the root
locus, one for each root.

• The plot is symmetric about the Real Axis. This is because complex roots
occur in conjugate pairs.

• Each plot starts at a location equal to the location of a root of the plant
transfer function. Each plot ends at infinity.

• The asymptotes of each plot as they tend infinity, share a common centre.
Their angles are evenly spread out over 360◦ .

• Two of the plots cross the imaginary axis. This corresponds to the system
becoming unstable with increasing K.

• All parts of the real axis to the left of an odd number of poles are part of
one of the plots.

• 2 of the plots break off from the real axis when they meet. Before they
break off the system has 3 real poles, after they break off 2 roots are
complex and one is real.

• A root locus plot also exists for negative values of K. The root locus plot
for 0 ≥ K > −∞ shown below. In 3c1 we focus almost exclusively on the
root locus for positive K.

3C1 Signals and Systems 6 Control Systems

Root Locus 2 ROOT LOCUS

2.1 The Root Locus Concept

We will now look at the root locus in general and look in more detail at some
of the observations we made from our example.

Consider a closed loop system above with a simple feedback loop that uses
simple proportional controller. It has a transfer function
KG(s) n(s)
H(s) = =
1 + KG(s) d(s)
The poles occur at the roots of d(s). For system transfer function above these
roots occur where
1 + KG(s) = 0. (3)
This is referred to as the characteristic equation of the system. Therefore,
to determine the poles it is necessary that

|KG(s)| = 1 (4)
∠G(s) = 180◦ ± k360◦ for k ∈ Z. (5)

3C1 Signals and Systems 7 Control Systems

Root Locus 2 ROOT LOCUS

These two equations are referred to as the Magnitude and Angle criteria
respectively.

Therefore, the root locus is the path of the roots of the characteristic equation
with respect to K as K is varied between 0 and infinity.

3C1 Signals and Systems 8 Control Systems

Root Locus 2 ROOT LOCUS

Example

We will show that by manipulating the denominator polynomial it is possible to

generate a root locus plot for the variation of other transfer function parameters.

For the system above the characteristic equation of the root locus due to
variations in K can be written directly from Eq. 3 as
1
1+K = 0 for a fixed τ .
s(s + τ )

However we can also rewrite the characteristic equation if we wished to see

the effect of variations of τ on the roots of the control system
1
1+K =0
s(s + τ s)
⇒ s2 + τ s + K = 0
s
⇒1+τ 2 = 0 for a fixed K.
s +K

Comment In general, it is often possible to generate a characteristic equation

of the form in Eq. 3 for another choice of parameter. It can be also be used for
more complicated closed loop system architectures. Although it is not necessary
to generate the equation in this form to plot a root locus, it allows the keypoints
of the root locus to be easily identified.

3C1 Signals and Systems 9 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

3 Root Locus Procedure

We will now show the procedure for rapid sketching of the root locus of a transfer
function H(s) as values of the tuning parameter K varies from 0 ≤ K < ∞.
Before we start we note that as the complex poles of H(s) will always occur in
conjugate pairs the root locus will be symmetric about horizontal real axis.

Step 1: Prepare the sketch

Firstly, identify the parameter to be varied (K) and generate the appropriate
characteristic equation of the form

1 + KP (s) = 0 (6)

Secondly, we consider the positions of the roots of the characteristic equation

for the start (K = 0) and end (K = ∞) points for the root locus. To do this,
we factorise P (s) and write the polynomial in terms of its poles and zeros as
follows QM
(s − zi )
1 + K QNi=1 =0
j=1 (s − pj )
where zi and pj are the zeros and poles of P (s) respectively.

When K = 0, we can find the roots of the characteristic equation by re-

writing it as
N
Y M
Y
(s − pj ) + K (s − zi ) = 0,
j=1 i=1
N
Y
⇒ (s − pj ) = 0.
j=1

Hence when K = 0, the roots of the transfer function H(s) are given by the
the poles of P (s).

3C1 Signals and Systems 10 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

When K = ∞, we can find the roots of the characteristic equation by re

writing it as
N M
1 Y Y
(s − pj ) + (s − zi ) = 0,
K j=1 i=1
M
Y
⇒ (s − zi ) = 0.
i=1

Hence when K = ∞, the roots of the transfer function H(s) are given by the
the zeros of P (s).

Observations

• The number of separate plots on the root locus sketch will be the same as
the number of poles of P (s).

• Each plot will start at a pole of P (s) for K = 0 and end at a zero of P (s)
for K = ∞.

• In general, system could have more poles than zeros. In this case, there
will be a number of plots that tend to infinity, with the number being
determined by the difference between the number of poles and zeros (i.e.
. N − M ). Such a system is said to contain N − M zeros at infinity.

3C1 Signals and Systems 11 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Case Study

Consider the closed loop system with a simple proportional controller as follows

The transfer function for this system is

Ks
H(s) = .
s2 + (2 + K)s + 2

To get the root locus as K varies we obtain the characteristic equation from
the denominator of H(s) as follows

s2 + (2 + K)s + 2 = 0
⇒s2 + 2s + 2 + Ks = 0
s
⇒1 + K 2 = 0.
s + 2s + 2

We then find the zeros and poles of P (s) which mark the start and end points
of each branch. For
s
P (s) =
s2
+ 2s + 2
we have one zero at s = 0 and two poles at −1 ± j. These points are marked
on the root locus plot

3C1 Signals and Systems 12 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

3C1 Signals and Systems 13 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Step 2: Determine the Parts of the Real Axis that are on the Root
Locus

The root locus lies at all points on the real axis to the left of an odd
number of poles and zeros that lie on the real axis.

This arises because of the angle criterion (Eq. 5) and the symmetry of the
root locus.

For example if we had a characteristic equation of

s+2
1+K =0
s(s + 4)
in which P (s) has a zero at s = −2 and poles at s = 0 and s = −4, then the
parts of the root locus marked on the real axis are marked by the solid line in
the plot below

3C1 Signals and Systems 14 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Case Study

Revisiting our example from Step 1, we see that we have only one zero on the
real axis (at s = 0) and so all points on the left of s = 0 must lie on the root
locus. We add that to the root locus plot.

3C1 Signals and Systems 15 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Step 3: Determine the Asymptotes for the loci tending to zeros at

infinity

We know that if we have a characteristic equation P (s) that has more poles N
than zeros M then N − M of the root locus bracnches tend to zeros at infinity.
These asymptotes intercept the real axis at a point, σA , given by
PN PM
j=1 p j − i=1 zi
σA = .
N −M
In other words P P
(finite poles) − (finite zeros)
σA =
N −M
The angles of the asymptotes φk are given by
2k − 1
φA = π for k = 1, 2, . . . , N − M
N −M
In other words the orientations of the assymptotes are evenly spread out over
360◦ .

For example, a 4th Order System with a characteristic equation

s+1
1+K = 0,
s(s + 2)(s + 3)(s + 5)
with 1 zero and 4 poles, 3 branches of the root locus will tend to zeros at infinity.
The centre of the 3 asymptotes is
(0 − 2 − 3 − 5) − (−1) −9
σA = = =3
4−1 3
and have an angles at
2k − 1
φA = π for k = 1, 2, 3
3
Therefore the angles of the asymptotes are π/3, π and 5π/3 = −π/3. The
asymptotes are marked for this system are shown in Fig. 4 in the left plot and
are superimposed on the full root locus for this system shown on the right.

3C1 Signals and Systems 16 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Figure 4: Left: Asymptotes shown for the above example. Right: Asymptotes superimposed
on the completed root locus.

Case Study

In the case study example we have N − M = 2 − 1 = 1.

Therefore, there is only one asymptote that has an angle

2−1
φA = π = π = 180◦
1
and “intercepts” the real axis at
((−1 + j) + (−1 − j) − (0)
σA = =2
1

As the asymptote is parallel to the real axis it does not intersect with it,
the value of σA is meaningless. However, it does lie on the real axis and so is
consistent with the rule.

A casual glance at the partially completed root locus shows that step 3
merely confirms the result of step 2. All points along the real axis to the left
of s = 0 are on the root locus.

3C1 Signals and Systems 17 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Step 4: Determining the Breakaway Points

From the examples we have seen so far, we can see that often there is a point
on the real axis at which the root locus either breaks off or onto the real axis.

Considering the root locus of a system with

1
P (s) =
(s + 2)(s + 4)
shown below, we can see that there is a breakaway point on the real axis at
a value of s = σ = −3 and that the angle of the branches as they leave the
imaginary axis is perpendicular to the real axis. It turns out that K = 1 at the
breakaway point.

This corresponds from the system transitioning from being overdamped K <
1 to being underdamped K > 1 and so marks a transition in the transient
response of the system from non-oscillatory to oscillatory behaviour.

3C1 Signals and Systems 18 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

It is possible to determine the breakaway point as follows.

Firstly, identify the portions of the real axis where a breakaway point must
exist. Assuming we have already marked the segments of the real axis that are
on the root locus, we need to find the segments that are bookended by either
two poles or two zeros (either finite zeros or zeros at infinity).

For example, lets look again at the root locus shown in Figure 4

There are 3 portions of the root locus on the real axis yet only one portion
(the middle one) contains a breakaway point. This is because that segment of
the root locus is terminated on either side by poles. The other two have a pole
at one end and a zero at the other and hence do not contain breakaway points.

3C1 Signals and Systems 19 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

To estimate the values of s at the breakaway points, the characteristic equa-

tion
1 + KP (s) = 0 (7)
is rewritten in terms of K as
1
K=− = K(s).
P (s)

To find the breakaway points we find the values of s corresponding to the

maxima in K(s). i.e. where
dK
= K 0 (s) = 0.
ds

As the last step we check the roots of K 0 (s) that lie on the real axis segments
of the locus. The roots that lie in these intervals are the breakaway points.

3C1 Signals and Systems 20 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Case Study

Revisiting our example, after stage 3 we have a partially completed root locus
shown below.

There is one segment of the real axis on the root locus. As it is bookended
by a zero at s = 0 and a zero at infinity (s = −∞) it must contain a breakaway
point.

Since
s
P (s) =
s2 + 2s + 2
we have breakaway points where
 2 
d s + 2s + 2
K 0 (s) = − = 0.
ds s
s(2s + 2) − (s2 + 2s + 2)
=−
s2
2
s −2
=− 2 =0
s
√ √
The roots K 0 (s) are ± 2. As the root at 2 does not fall on the root locus
√
segment, the breakaway point occurs at s = − 2.

3C1 Signals and Systems 21 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

The value of K at the breakaway point is

1
K(s) = − √ .
P ( 2)
√ 2 √
2 +2 2+2
=− √
√2
4+2 2
=− √ ≈ 4.83
2

Therefore, we can update the root locus by marking the breakaway point.
As the root locus will leave the breakaway points at an angle perpendicular to
the real axis. We add this tangent to the root locus sketch.

3C1 Signals and Systems 22 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Step 5: Find the angles of departure/arrival for complex poles/zeros

In our case study, we can see that there are two complex poles (at −1 ± j) from
which branches of the root locus must start. This step wishes to determine the
angle at which the root locus leaves such poles.

When there are complex poles or zeros of P (s), the root locus branches will
either depart or arrive at an angle θ where, for a complex pole at s = p or zero
at s = z, θ is given by

θ = ∠P ∗ (p) − π for s = p
or θ = π − ∠P ∗ (z) for s = z

where
M
Q
(p − zi )
∗ i=1
P (p) = N
Q
(p − pj )
j=1
p6=pj
M
Q
(z − zi )
i=1
z6=zi
P ∗ (z) = N
Q
(z − pj )
j=1

In other words, P ∗ (p) is (s − p) × P (s) evaluated at s = p and P ∗ (z) is (s −

z)−1 × P (s) evaluated at s = z

3C1 Signals and Systems 23 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Therefore, exploiting the rules of complex numbers, we can rewrite ∠P ∗ (p)

and ∠P ∗ (z) as
M
X N
X
∗
∠P (p) = ∠(p − zi ) − ∠(p − pj )
i=1 j=1
p6=pj
M
X N
X
∗
∠P (z) = ∠(z − zi ) − ∠(z − pj )
i=1 j=1
z6=zi

3C1 Signals and Systems 24 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Case Study

Here we have two complex poles at s = −1 ± j.

Therefore, for the pole at s = −1 + j, the root locus departs the pole at an
angle θ where

∠P ∗ (−1 + j) − θ = π


⇒ ∠(s + 1 − j)P (s) −θ =π
s=−1+j

s 
⇒∠  −θ =π
s − (−1 − j) s=−1+j
⇒ ∠−1 + j − ∠−1 + j − (−1 − j) − θ = π
3π π
⇒ −0− −θ =π
4 2
3π
⇒θ=−
4

We could repeat this calculation to estimate θ for s = −1 − j. However,

exploiting the symmetry of the root locus, it is quicker to notice that the value
3π
of θ−1−j = −θ−1+j = 4 .

Therefore we can update the root locus plot by adding tangents correspond-
ing to the angles of departure for the complex poles.

3C1 Signals and Systems 25 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

Step 6: Complete the Sketch

This can be done by hand from all the information gained from the previous
steps. You could make it simpler by finding the poles of the characteristic
equation for more values of K and placing them on the locus.

The root locus for our case study is shown below.

3C1 Signals and Systems 26 Control Systems

Root Locus 3 ROOT LOCUS PROCEDURE

3.1 Summary

The process of generating the root locus for 0 ≤ K < ∞ can be summarised as
follows

Step Rule

Estimate the transfer function and 1 + KP (s) = 0

write the characteristic equation of that
the tuning parameter, K, appears as a
multiplier.

Mark the start and end points of the Find the poles (start points) and zeros
loci. (end points) of P (s).

Determine the number of separate loci It is the number of poles of P (s).

Locate the segments of the real axis that Locus lies to the left of an odd number
are root loci of zeros or poles.

Find the asymptotes of the loci tending The intersection

P
point
P
of the asymp-
pj − zi
to zeros at infinity totes is σA = N −M . The angles are
2k−1
φA = N −M π

Determine the breakaway points. Set K(s) = − P 1(s) and find K 0 (s) = 0

Determine the angles of ar- For poles the angle of departure is

rival/departure at complex zeros/poles ∠P ∗ (pk ) − π. For zeros the angle of ar-
of P (s) rival is π − ∠P ∗ (zk )

Complete the Sketch

3C1 Signals and Systems 27 Control Systems

Root Locus 4 ROOT LOCUS FOR NEGATIVE K

4 Root Locus for Negative K

A similar set of rules exists for plotting the root locus for 0 ≥ K > −∞. The
summary for plotting the negative K root is shown below with differences from
the positive K root locus highlighted in bold.

Step Rule

Estimate the transfer function and 1 + KP (s) = 0

write the characteristic equation of that
the tuning parameter, K, appears as a
multiplier.

Mark the start and end points of the Find the poles (start points) and zeros
loci. (end points) of P (s).

Determine the number of separate loci It is the number of poles of P (s).

Locate the segments of the real Locus lies to the left of an even
axis that are root loci number of zeros or poles.

Find the asymptotes of the loci The intersectionP point

P
of the asymp-
pj − zi
tending to zeros at infinity totes is σA = N −M . The angles
are φA = 2 N2k−1
−M π

Determine the breakaway points Set K(s) = − P 1(s) and find K 0 (s) = 0

Determine the angles of ar- For poles the angle of departure is

rival/departure at complex ze- ∠P ∗ (pk ). For zeros the angle of ar-
ros/poles of P (s) rival is ∠P ∗ (zk )

Complete the Sketch

3C1 Signals and Systems 28 Control Systems

Root Locus 5 CONCLUSION

Case Study

We will now plot the negative K root locus for a system with
Ks
H(s) = 2 .
s + (2 + K)s + 2
Hence,
s
P (s) = 2 .
s + 2s + 2
The root locus is plotted below.

Notice that for a value of K < −2 the roots move into the RHS of the s-plane
and so the system is unstable.

5 Conclusion

In this handout we showed how to determine the location of the poles of a

control system through a root locus plot. These plots are important because
they indicate whether a system is stable or not and can also be used to estimate
the transient response.

3C1 Signals and Systems 29 Control Systems