ECE 486, Spring 2018 Sample Midterm #1 Monday, Feb 19th

1. Problem 1: Consider the linear system given by the following differential equation

y (4) + 3y (3) + 2ÿ + 3ẏ + 2y = u̇ − u,

where u is the input and y the output.

Is this system stable? Show your work to justify your claim. Note: y (4) is the fourth
derivative of y.
Solution: Apply Laplace transform to the given differential equation with zero initial
conditions,
(s4 + 3s3 + 2s2 + 3s + 2)Y (s) = (s − 1)U (s).
The system transfer function is therefore

Y (s) s−1
= 4 .
U (s) s + 3s3 + 2s2 + 3s + 2

All the coefficients of the characteristic polynomial s4 + 3s3 + 2s2 + 3s + 2 are positive,
necessary condition for stability is met. We need to further check stability with Routh
test. (So, if necessary condition is not met, there is no need to test Routh-Hurwitz.)
Form Routh array

s4 : 1 2 2
s3 : 3 3 0
s2 : 1 2
s1 : −3 0
s0 : 2

Not all the entries in the first column are positive (−3 < 0). Therefore the system is
not stable.
Untitled Diagram
1
2. Problem 2: Consider the unity negative feedback loop with plant G(s) = s2 +6s+8 and
proportional controller C(s) = K. Assume all initial conditions are zero here.

(a) What is the transfer function from r to y?

Y K
Solution: (s) = 2 .
R s + 6s + (K + 8)

3
 (b) For what range of values of K is this closed loop system stable?
Solution: To the characteristic polynomial s2 + 6s + (K + 8), we apply the
Routh-Hurwitz criterion for second order transfer function, i.e., it is necessary
and sufficient that all coefficients are positive. We need K > −8.
(c) Suppose we want the closed loop system to be underdamped with peak time
tp < 2π seconds. Sketch the possible pole locations in the complex plane. Mark
indicative regions or points if they are helpful.
Solution: Peak time tp = ωπd ≤ 2π implies ωd ≥ 21 . But ωd is the damped
natural frequency when system is underdamped. Sketch the closed loop poles on
the complex plane
s1,2 = −σ ± jωd
with negative real part and imaginary part with absolute value ≥ 0.5.
(d) For what range of values of K is the closed loop system underdamped with peak
time tp < 2π seconds?
Solution: From (c), the closed loop poles are
√
s1,2 = −3 ± 1 − K.

When the system is underdamped, 1 − K < 0, i.e., K > 1. The closed loop poles
further become √
s1,2 = −3 ± j K − 1.
By (c), the absolute value of the imaginary part ≥ 0.5. So
√
K − 1 ≥ 0.5.
We get K ≥ 54 .

3.Untitled
ProblemDiagram 3: Consider the unity negative feedback loop with reference r, output y,
1
tracking error e, input disturbance d and plant P (s) = s2 +2s . Assume all initial condi-
tions are zero here.

+
+ +

(a) Suppose we want to design a controller C(s) so that the steady state tracking error
due to square input r(t) = t2 , t ≥ 0 is at most 0.5 (d = 0, no disturbance). With
only two possible choices, P and PID controllers, which controller would you use?
And why?
Solution: To track quadratic reference input t2 with error up to a nonzero con-
stant, we need the system type to be 2. The plant P (s) itself already has 1 pole
at origin, we need one more from the controller. Therefore we choose PID control.

4
 2
(b) Suppose we choose the PID controller C(s) = KD s +K s
P s+KI
. For square input
2
r(t) = t , t ≥ 0 without disturbance d = 0, derive the expression for the steady
state error in terms of gains KP , KD and/or KI . What conditions (if any) must
the gains satisfy in order to achieve a steady state error of less than 0.5?
Solution: The transfer function from reference to error is
E 1
(s) =
R 1 + P (s)C(s)
1
= 1 KD s2 +KP s+KI
1 + s2 +2s s
s3 + 2s2
= .
s3 + (KD + 2)s2 + KP s + KI
Apply Routh-Hurwitz criterion for third order transfer function, all the coefficients
of the characteristic polynomial shall be positive
KD + 2 > 0, KP > 0, KI > 0,

and (KD + 2)KP > KI .

After we guarantee stability, we further apply the Final Value Theorem
ess = e(∞)
= lim sE(s)
s→0
 
E
= lim s (s) R(s)
s→0 R
s3 + 2s2
= lim s · 3 2
· L{t2 }
s→0 s + (KD + 2)s + KP s + KI
s3 + 2s2 2
= lim s · 3 2
· 3
s→0 s + (KD + 2)s + KP s + KI s
4
= .
KI
Steady state error ess < 0.5 implies KI > 8.
In conclusion, KP > 0, KD > −2, KI > 8 and (KD + 2)KP > KI .
(c) With the controller C(s) as in (b), suppose we want the closed loop poles to be
at s = −1, −2, −a, where a is some positive real number. For r(t) = t2 , t ≥ 0 and
d = 0, what is the minimum (or lower limit if minimum cannot be achieved) value
of a in order to achieve a steady state error of less than 0.5?
Solution: By (b), we try to associate a with KI . Note KI is the constant term of
the characteristic polynomial in (b). If the closed loop poles are s = −1, −2, −a,
the resulting characteristic polynomial is
(s + 1)(s + 2)(s + a)
= s3 + · · · + 2a.
[= s3 + (a + 3)s2 + (3a + 2)s + 2a] (This is not necessary)

5
 Matching the characteristic polynomial from (b), we have KI = 2a. Further from
(b), KI > 8. Therefore a > 4.
(d) With the controller C(s) as in (b), compute the transfer function from d to y
(r = 0). Leave the answer in terms of KP , KD and/or KI . (Do not substitute any
numerical values.)
Solution: Reshape the loop in the figure if necessary, the forward gain is P (s) =
1
s2 +2s
and the loop gain is P (s)C(s) with negative feedback. Apply our formula
forward gain
, then
1 + loop gain

Y P (s)
(s) =
D 1 + P (s)C(s)
1
s2 +2s
= 1 KD s2 +KP s+KI
1+ s2 +2s s
s
= .
s3 + (KD + 2)s2 + KP s + KI

(e) Assuming the closed loop system is stable, what is the steady state response due
to a unit step disturbance input, i.e., d(t) = 1, t ≥ 0 (r = 0)? Is this control
system good at rejecting step disturbances?
Solution: By the transfer function from disturbance to output in (d), the system
type is 1. For a 0-th degree polynomial disturbance input d(t) = 1, the disturbance
response is 0.
Alternatively, apply FVT to step disturbance response based on the transfer func-
tion in (d).
The control system is good at rejecting step disturbance inputs.