Root Locus: Figure 1: Pole/Zeros Diagrams in The Car Cruise Control Example For K 50 and K 100
Root Locus: Figure 1: Pole/Zeros Diagrams in The Car Cruise Control Example For K 50 and K 100
Root Locus: Figure 1: Pole/Zeros Diagrams in The Car Cruise Control Example For K 50 and K 100
• 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
1
Root Locus
• 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).
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
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.
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.
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
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.
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)
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.
Example
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 )
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.
Firstly, identify the parameter to be varied (K) and generate the appropriate
characteristic equation of the form
1 + KP (s) = 0 (6)
Hence when K = 0, the roots of the transfer function H(s) are given by the
the poles of P (s).
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.
Case Study
Consider the closed loop system with a simple proportional controller as follows
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
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.
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.
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◦ .
Figure 4: Left: Asymptotes shown for the above example. Right: Asymptotes superimposed
on the completed root locus.
Case Study
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.
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.
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.
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.
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.
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.
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.
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
Case Study
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
Therefore we can update the root locus plot by adding tangents correspond-
ing to the angles of departure for the complex poles.
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.
3.1 Summary
The process of generating the root locus for 0 ≤ K < ∞ can be summarised as
follows
Step Rule
Mark the start and end points of the Find the poles (start points) and zeros
loci. (end points) 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.
Determine the breakaway points. Set K(s) = − P 1(s) and find K 0 (s) = 0
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
Mark the start and end points of the Find the poles (start points) and zeros
loci. (end points) 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.
Determine the breakaway points Set K(s) = − P 1(s) and find K 0 (s) = 0
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