ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 1

Solution to HW5

AP6.3 We are given a system with open loop transfer function

(s + α)
G(s) = (1)
s3 + (1 + α)s2 + (α − 1)s + (1 − α)

and unity negative feedback. We are asked to determine (a) the range of values for α such
that the error in response to a step input is not more than 5%, (b) the range of α for
which the system is stable, and (c) a value of α that satisfies both requirements.
Solution:
The closed loop transfer function is

(s + α)
T (s) = . (2)
s3 + (1 + α)s2 + αs + 1

(a) By the final value theorem,

 
1
ess = lim s R(s)[1 − T (s)] (3)
s→0 s
= 1−α (4)

so if we choose α such that 0.95 ≤ α ≤ 1.05, we have satisfied the design criterion for
the steady state error.
(b) The Routh array is given in the Table 1.

Table 1: Routh array for AP6.3b

s3 : 1 α
s2 : 1+α 1
s1 : 1+α [1 − α(1 + α)]
−1

s0 : 1

For stability, all elements of the first column must be positive since the first one is.
The second row gives us the constraint α > −1. (Note the strict equality. We want
the entry to be positive not zero.) The third row simplifies to the constraint

α2 + α − 1 > 0. (5)
√
α2 + α − 1 = 0 are α = −1/2 ± √5/2, yielding the constraint that either
The roots of √
α < −1/2 − 5/2 ≈ −1.618 or α > −1/2 + 5/2 ≈ 0.618.
(c) α = 1, for example, satisfies both the stability and steady state error constraints. (In
fact, we’ve shown that any value that satisfies the steady state error constraint leads
to a stable system.)
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 2

AP7.1 We are given a closed loop system with unity negative feedback and forward path
transfer function
K(s + 6)
G(s) = . (6)
s(s + 4)(s2 + 4s + 8)
(a) We are asked to sketch the root loci and determine the gain K such that the damping
coefficient of the complex poles near the imaginary axis is maximized. (b) We are asked
to evaluate the roots corresponding to this value of K and predict the step response for
this value. (c) Finally we are asked to determine the actual response and compare it to
our prediction.
Solution:

(a) We’ll work our way through the list of steps in Table 7.2 of the textbook.
1. Because we have unity negative feedback, our closed loop characteristic equation

K(s + 6)
1+ =0 (7)
s(s + 4)(s2 + 4s + 8)

is already in the form 1 + KP (s) = 0 without any further manipulation. We need

only factor the quadratic in the denominator to obtain

(s2 + 4s + 8) = (s − (−2 + j2)) (s − (−2 − j2)) . (8)

We see that we have two poles (one at the origin and one at −4) and one zero
(at −6) on the real axis and two complex poles. There being 4 poles and only
one zero, there will be four loci.
2. The real axis segments that will be part of the locus are the segment from −4 to
0 and the one from −∞ to -6.
3. The loci extend to the zeros at ∞ with asymptotes centered at σA and angles φA
where
4 + 2(2) − 6 2
σA = =− (9)
4 − 1  3
2q + 1
φA = π = ±π/3, π. (10)
3

4. We determine the points at which the loci cross the imaginary axes by using the
Routh array to determine what value of K leads to instability. The Routh array
is shown in Table 2.

Table 2: Routh array for AP7.1a

s4 : 1 24 6K
s3 : 8 32 + K
s2 : (160 − K)/8 6K
s1 : (K 2 + 256K − 5120)/(K − 160)
s0 : 6K
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 3

We see that we need K < 160 from line 3 and must factor√the numerator of line
4. Factoring K 2 + 256K − 5120 = 0 yields K = −128 ± 32 11 ≈ 18.64, −274.64.
To determine whether we want K between these values or outside these values
we try a test point. If K = 0 we have (−5120)/(−160) > 0 so we want the values
between −274.64 and 18.64. From line 5 we need K positive so, summarizing,
we have the requirement that 0 < K < 18.64. If K = 0, the denominator of T (s)
does not have imaginary roots so the value we are looking for is K ≈ 18.64. (We
can verify this later by doing to root locus plot in Matlab.) When K = 18.64,
the denominator of the transfer function of the closed loop system is

s(s + 4)(s2 + 4s + 8) + K(s + 6) = s4 + 8s3 + 24s2 + 50.64s + 118.84(11)

= (12)

which has roots −4.0506 ± 1.3901i and 0.0506 ± 2.5451j so the loci cross the
imaginary axis near ±2.5j.
5. We will need a breakaway point on the segment of the real axis from −4 to
0 because these two poles must go to zeros either of the numerator or at ∞.
Because the zero is to the left of all the poles and there is a locus segment from
−∞ to −6, we will need a breakin point somewhere to the left of −6 so that one
locus can go to the zero at −6 and the other can go to −∞.
To determine the break points we put (7) over a common denominator and set
the numerator equal to zero. This yields

s4 + 8s3 + 24s2 + 32s = −(s + 6)K (13)

which we solve for

s4 + 8s3 + 24s2 + 32s
K = p(s) = − . (14)
s+6
Now we take the partial derivative of p(s) with respect to s and set it equal to
zero. The roots are candidate break points. We obtain

∂p(s) 4s3 + 24s2 + 48s + 32 s4 + 8s3 + 24s2 + 32s

− = − (15)
∂s s+6 (s + 6)2
3s4 + 40s3 + 168s2 + 288s + 192
= =0 (16)
(s + 6)2

which has roots s ≈ −7.30, −3.08, −1.48 ± j0.81 (obtained using Matlab). We
need two breakpoints and we have two possible breakpoints (the other two can-
didates not being on the real axis) so we can see from their locations that −3.08
is the breakaway point and −7.3 is the breakin point.
6. To determine the angle of departure of the loci from the complex poles, I proceed
as follows. I define the angle from the positive real axis to the segment connecting
the pole at the origin to the pole at −2 + 2j to be θ1 , the angle to the segment
connecting the complex poles to be θ2 , the angle connecting the pole at −4 to
the pole at −2 + 2j to be θ3 , and the angle to the segment connecting the zero
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 4

at −6 to the pole at −2 + 2j to be φ1 . I define the angle of departure from the

pole at −2 + 2j relative to the standard position to be θ. Then
θ + θ1 + θ2 + θ3 − φ1 = 180 mod 360 deg (17)
where θ1 + θ3 = 180 deg (by trigonometry), θ2 = 90 deg and φ1 = tan−1 (1/2) ≈
26.6 deg so θ ≈ −63.4 deg.
The resulting plot should look something like that shown in Figure 1.

Figure 1: Root locus generated by Matlab for AP7.1a

Root Locus for AP7.1a

10

2
Imaginary Axis

−2
System: untitled1
Gain: 3.05
Pole: −1.61 − 1.46i
−4 Damping: 0.742
Overshoot (%): 3.1
Frequency (rad/sec): 2.17

−6

−8

−10
−12 −10 −8 −6 −4 −2 0 2 4 6
Real Axis

From the root locus plot obtained using Matlab, we estimate the gain yielding the
maximal ζ value to be 3.07, yielding ζ = 0.742. I don’t offhand see a good way to
determine this analytically.
(b) The closed loop transfer function is
G(s)
T (s) = (18)
1 + G(s)
K(s + 6)
= (19)
s(s + 4)(s2 + 4s + 8) + K(s + 6)
K(s + 6)
= (20)
s + 8s + 24s2 + (32 + K)s + 6K
4 3
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 5

Matlab tells us that the roots of the closed loop-system corresponding to this value of
K are −3.7342, −1.6071 ± 1.4518j, and −1.0517. Matlab predicts overshoot of 3.1%
and ωn = 2.17 rad/s, which would yield a 2%-settling time of
4
Ts2% ≈ ≈ 0.59 s (21)
ζωn

(c) The actual response is shown in Figure 2. It indicates no overshoot and 2%-settling
time of 4.44 seconds. (Obtained by checking values of y.) This is not impressive
agreement and suggests that we should be rather cautious in using the estimated val-
ues of natural frequency and damping coefficient provided by Matlab. In particular,
one reason that they may not be very helpful is that they presumably represent the
values associated with the particular complex pair only. The remaining poles and
zeros will affect the behavior of the step response, and, as seen here, may have a very
significant effect.

Figure 2: Step Response generated by Matlab for AP7.1c

Step Response for AP7.1c
1

0.9

0.8

0.7

0.6
y(t)

0.5

0.4

0.3

0.2

0.1

0
0 1 2 3 4 5 6 7 8 9 10
Time (s)
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 6

AP7.4 We are given a system with unity negative feedback and forward path transfer function

(s + α)
G(s) = . (22)
s3 + (1 + α)s2 + (α − 1)s + 1 − α

We are given as a design objective that the steady state position error for a step input
should be at most 10%of the magnitude of the step input. We are asked to

(a) Sketch the root locus as a function of the parameter α.

(b) Determine the range of values for α that result in acceptable steady state error.
(c) Locate the roots for the allowable value of α that achieves the required steady-state
error and estimate the system step response.

Solution: The closed loop transfer function is

s+α
T (s) = . (23)
s3 + (1 + α)s2 + αs + 1

(a) We want to know when the denominator of the closed loop transfer function is zero.
To isolate the value of α, we rewrite

s3 + (1 + α)s2 + αs + 1 = 0 (24)

as
s3 + s2 + 1 + α(s2 + s) = 0 (25)
from which we obtain by dividing by s3 + s2 + 1

α(s2 + s)
1 + p(s) = 1 + = 0. (26)
s3 + s2 + 1
The rational function p(s) then has zeros at 0 and −1 and poles at −1.4656 and
0.2328 ± 0.7926 (using Matlab). We thus have three poles and zeros on the real axis
and the segments of the real axis that are part of the root locus will be the seg-
ments (−∞, −1.4656] and [−1, 0]. The segment between the two zeros must contain
a breakin point.
We determine the location of that breakin point by taking the partial derivative of
p(s) (technically of −p(s) the way we have defined it here) with respect to s and
setting it equal to zero. We then have

3s2 + 2s
 2
s + s2 + 1

∂p
= − (2s + 1) (27)
∂s s2 + s (s2 + s)2
s4 − 2s3 + s2 − 2s − 1
= (28)
(s2 + s)2

which has roots −1.2 ± 0.98j, 0.88, and −0.469. The only one of these on the segment
is −0.469 so this must be the breakin point.
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 7

The angle of departure of the locus from the complex poles is determined from the
equation
       
−1 0.7926 0.7926 −1 0.7926
θ − − tan + 180 + 90 − tan −1
+ tan = 180
0.2328 1 + 0.2328 1.4656
(29)
where we have used the geometry of the root locations to determine the arguments
of the inverse tangents. The result obtained is θ = 200 degrees.
The resulting root locus should look something like that shown in Figure 3.

Figure 3: Root locus generated by Matlab for AP7.4a

Root Locus
0.8

0.6

0.4

0.2
Imaginary Axis

−0.2

−0.4

−0.6

−0.8
−3 −2.5 −2 −1.5 −1 −0.5 0 0.5
Real Axis

(b) To obtain the required steady state error we apply the Final Value Theorem to obtain
 
1
ess = lim s (1 − T (s)) = 1 − α (30)
s

so we need 0.9 ≤ α ≤ 1.1.

(c) The step responses for three different values of α in this range (giving acceptable
steady state error) are shown in Figure 4.
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 8

Figure 4: Step Responses generated by Matlab for AP7.4c with α = 0.9, 1.0, and 1.1.
AP7.4 Step Responses
1.8
α=0.9
α=1
1.6 α=1.1

1.4

1.2

1
Amplitude

0.8

0.6

0.4

0.2

0
0 10 20 30 40 50 60
Time (sec)

AP7.12 We are given a system with unity negative feedback and forward path transfer function

K(s + 1)2
G(s) = . (31)
s(s + 8)(s + 20)(s2 + 3.2s + 3.56)

We are asked to (a) sketch the root locus, (b) select a value of K such that the step
response has overshoot less than 20% and settling time less than 5 seconds.
Solution:

(a) The open loop system has two zeros at −1 and poles at −8, −20, and −1.6 ± j. The
segments of the real axis contained in the root loci are thus [−1, 0], [−8, −1], and
(−∞, −20]. The asymptotes (there are three more poles than zeros so three loci must
go to infinity) will have angles ±60 and 180 degrees. The intercept of the asymptotes
with the real axis will be
(−8 + (−20) + 2(−1.6) − 2(−1))
σA = ≈ −11.07. (32)
5−2
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 9

To determine the breakin and breakout points we take the partial derivative with
respect to s of

s(s + 8)(s + 20)(s2 + 3.2s + 3.56) s5 + 31.2s4 + 253.16s3 + 611.68s2 + 596.6s

p(s) = =
(s + 1)2 (s + 1)2
(33)
obtaining after some calculation,

∂p s5 + 36.2s4 + 377.96s3 + 1371.16s2 + 1792.96s + 569.6

= . (34)
∂s (s + 1)2

This is zero when s = −21.2879, −9.3628, −3.0542, −2.0354, and −0.4597. Only the
last three of these are break point candidates since the others are not on the segment
of the real axis that coincides with the real line. The last is not a breakpoint because
the locus that starts at the pole at the origin goes directly to the zero at −1.
The root locus plot generated by Matlab is shown in Figures 5 and 6.

Figure 5: Root Locus Plot generated by Matlab for AP7.12

AP7.12
40

30

20

10
Imaginary Axis

−10

−20

−30

−40
−60 −50 −40 −30 −20 −10 0 10 20
Real Axis
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 10

Figure 6: Closeup of Root Locus Plot generated by Matlab for AP7.12

Closeup for AP7.12

15

10

5
Imaginary Axis

−5

−10

−15

−15 −10 −5 0 5 10 15
Real Axis

(b) The closed-loop transfer function is

K(s + 1)2
T (s) = (35)
s(s + 8)(s +20)(s2 + 3.2s + 3.56) + K(s + 1)2
K(s + 1)2
= . (36)
s5 + 31.2s4 + 253.16s3 + (611.68 + K)s2 + (569.6 + 2K)s + K

We may start by selecting a location (on the locus that goes from one of the complex
poles to one of the zeros on the real axis) where the overshoot is just less than
20%. I found one such with gain approximately 900. I tested this gain and obtained
a satisfactory step response as shown in Figure 7. I verified that the final value
y(5) = 0.9801 so the settling time criterion as well as the overshoot criterion is
satisfied. Had the criteria not been satisfied, I would have iterated until I found an
acceptable gain. (Note that there is probably a wide range of gains that meet the
design criteria. The gain given by the solution manual is much larger than the one I
obtained. Their gain does produce overshoot, unlike my choice.)
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 11

Figure 7: Step Response for K = 900 in AP7.12

Step Response of AP7.12 with K=900
1

0.9

0.8

0.7

0.6
y(t)

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

DP6.2 We are given a closed loop system with forward path transfer function
10
G(s) = (37)
s2 (s + 10)

and feedback path transfer function H(s) = Ks + 1. We are asked to determine (a)
acceptable values of K to keep the system stable, (b), the value of K for which one of the
two closed loop poles is s = −5, and (c) the value of the other two poles in this case. We
are then asked to (d) generate the (closed-loop) system step response for the K determined
in part (b).
Solution: The closed loop transfer function is
10
T (s) = . (38)
s3 + 10s2 + 10Ks + 10
(a) The corresponding Routh array is given in Table 3. From the third row we determine
that we need K > 0.1 for stability.
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 12

(b) By long division we determine that

s3 + 10s2 + 10Ks + 10 = (s + 5)(s2 + 5s + (10K − 25)) (39)

with remainder 10 − 50K + 125. Requiring the remainder to be zero yields K =

135/50 = 2.7.
(c) The quadratic factor being s2 + 5s + (10K − 25),
√ substituting for K ands setting the
expression equal to zero yields roots −5/2 ± 17/2.
(d) The step response is then obtained with the Matlab command
y = step(tf([10],[1 10 27 10]));
and is shown in Figure 8.

MP7.5 We are given a system having open loop transfer function

k(s + 1)
G(s) = (40)
s2
and unity negative feedback. We are asked to find the value of K for which the damping
factor of the closed loop poles is ζ = 0.707.
Solution:
Of course we could use Matlab to plot the root locus and use the cursor to identify the
gain K corresponding to the point at which the real and imaginary parts of the poles have
the same magnitude, since this is the condition that corresponds to the value ζ = 0.707.
In this case, it seems silly to accept such an approximate answer when we can calculate
the exact answer easily enough. For the given system, the closed loop transfer function is

k(s + 1)
T (s) = (41)
s2 + k(s + 1)

so ωn2 = k and 2ζωn = k. 2 2

√ Solving for ζ yields k = (2ζ) = (2(0.707)) = 1.9994. In fact, if
k = 2 we obtain ζ = 1/ 2 ≈ 0.707, which is probably what we wanted anyway.

Table 3: Routh array for DP6.2a

s3 : 1 10K
s2 : 10 10
s: 10 (10 − 100K)
−1

s0 : 10
ECE382/ME482 Spring 2005 Homework 5 Solution April 16, 2005 13

Figure 8: Plot of step response for the automatically guided vehicle of DP6.2.
Step Response for DP6.2 with K=2.7
1

0.9

0.8

0.7

0.6
y(t)

0.5

0.4

0.3

0.2

0.1

0
0 2 4 6 8 10 12 14 16 18 20
Time (s)