Unit 7: Part 2: Refining The Root Locus: Engineering 5821: Control Systems I
Uploaded by
Nikhil PanikkarUnit 7: Part 2: Refining The Root Locus: Engineering 5821: Control Systems I
Uploaded by
Nikhil PanikkarRefining the Sketch
Unit 7: Part 2: Refining the Root Locus
Engineering 5821:
Control Systems I
Faculty of Engineering & Applied Science
Memorial University of Newfoundland
March 10, 2010
ENGI 5821 Unit 7: Root Locus Techniques
Refining the Sketch
The rules described thus far only allow us to draw a rough sketch
of the RL. We will now discuss several refinements that allow
important aspects of the curve to be specified.
Real-Axis Breakaway and Break-in Points: These are points at
which the RL leave the real axis (breakaway) or return to it. e.g.
It can be shown that at breakaway and break-in points the RL
forms a 180o /n angle with the real-axis, where n is the number of
branches involved. Often we have just two branches so a 90o angle
is formed.
As the two-closed loop poles approach the breakaway point the
gain, K , is increasing. At the breakaway point −σ1 the gain
reaches its maximum value on the real-axis between -2 and -1.
Similarly, at the break-in point σ2 the gain is at its minimum value
on the real-axis between 3 and 5.
Where we know the RL exists on the real-axis, it must be the case
−1
that K = G (σ)H(σ) . We can plot K in these regions.
We can determine the breakaway and break-in points by searching
for maximum (breakaway) and minimum (break-in) values of K .
We can perform this search using a computer, or find the
maximum/minimum points analytically by differentiating the
equation above w.r.t. σ and then solving for the values of σ which
make the slope zero.
e.g. Find the breakaway and break-in points for the following
open-loop system:
K (s − 3)(s − 5) K (s 2 − 8s + 15)
KG (s)H(s) = = =
(s + 1)(s + 2) s 2 + 3s + 12
Since we are interested in the real-axis we can replace s with σ.
Equating KG (σ)H(σ) = −1 we solve for K ,
−(σ 2 + 3σ + 2)
K=
σ 2 − 8σ + 15
Differentiate w.r.t σ, equate to zero, and solve for σ.
dK 11σ 2 − 26σ − 61
= =0
dσ (σ 2 − 8σ + 15)2
σ = −1.45 and 3.82. -1.45 is the breakaway point, 3.82 is the
break-in point.
Crossings of the jω-Axis: This is a most important refinement
since it defines the boundary between stability and instability. e.g.
Again, we can find crossings of the jω axis by searching points
along the axis and evaluating where these points lie on the RL.
A more analytical method relies on the Routh-Hurwitz criterion...
e.g. The closed-loop TF for the unity feedback system with
K (s+3)
forward gain G (s) = s(s+1)(s+2)(s+4) is as follows,
K (s + 3)
T (s) =
s 4 + 7s 3 + 14s 2 + (8 + K )s + 3K
Form the Routh table so that we can determine the value(s) of K
such that poles of T (s) lie on the jω axis.
s4 1 14 3K
s3 7 8+K
s2 90 − K 21K
−K 2 −65K +720
s1 90−K
s0 21K
s4 1 14 3K
s3 7 8+K
s2 90 − K 21K
−K 2 −65K +720
s1 90−K
s0 21K
What we are really interested in is the value of K where marginal
stability is achieved. This can only occur if this 4th -order
polynomial is divisible by an even polynomial (only EP’s have roots
symmetric about the origin). Recall that this occurs when we have
a ROZ in the Routh table. For K ≥ 0 this can only occur when:
−K 2 − 65K + 720 = 0 =⇒ K = 9.65
In order to determine the appropriate value of ω we form the EP
and find its roots for K = 9.65:
(90 − K )s 2 + 21K = 80.35s 2 + 202.7 =⇒ s = ±j1.59
Angles of Departure and Arrival: Where the RL begins at
complex poles, or ends at complex zeros, we want to provide the
angle of departure or arrival from these poles or zeros. Assume we
are on the RL, just after departing from a complex pole. We
consider a point q which is at a small distance from the pole just
departed.
The angle criterion must be satisfied. Thus we can solve for the
angle of departure,
θ1 = θ2 + θ3 − θ4 − θ5 + θ6 − (2k + 1)180o
We can apply the same process to obtain the angle of arrival to a
zero,
θ2 = θ1 − θ3 + θ4 + θ5 − θ6 + (2k + 1)180o
However, how can we determine the other angles (θ1 , θ3 , θ4 , θ5 ,
θ6 ) to q when we don’t actually know the location of q? Given that
q is a short distance from a pole or zero, we just determine these
angles from the other poles or zeros to that closest pole or zero.
e.g. Find the angle of departure from the complex poles of the
following unity feedback system:
Consider a point q close to the upper pole.
θ3 − θ1 − θ2 − θ4 = (2k + 1)180o , from which we can obtain
θ1 = 108.4o .
Plotting and Calibrating the RL: We may wish to find certain
points of interest on the RL and their associated gains. For
second-order systems or systems that are well-approximated as
second-order we wish to identify points with a particular value of ζ
or Ts .
e.g. Assume we wish to find the point that crosses the ζ = 0.45
line for the following system.
ζ = 0.45 corresponds to a line at θ = cos−1 0.45 = 63.2o to the
negative real-axis. The %OS for this value of ζ is 20.5%.
Given the line corresponding to ζ = 0.45 we search along it to
determine where it crosses the RL. The condition here for begin on
the root locus is that the following sum is an odd multiple of 180o :
θ2 − θ1 − θ3 − θ 4 − θ5
This criterion is met at a radius of 0.747 from the origin. We can
calculate the corresponding value of K ,
|A||C ||D||E |
K= = 1.71
|B|
Note that this system is not second-order. Recall the conditions for
a second-order approximation:
A higher-order pole far to the left of the dominant
second-order poles can be ignored.
Zeros that are far from the dominant poles can be ignored.
Zeros that are not far from the dominant poles may be
cancelled by very nearby poles.