Chapter #7 Exercise with Solution

Problem 1. Problem 7-5 (vii) and (viii).

Solution:

(vii) Let Q(z) = z 3 − 2.2z 2 + 1.55z − 0.35. Hence, n = 3 and

a3 = 1, a2 = −2.2, a1 = 1.55, a0 = −0.35.

By the Jury’s test, for the system to be stable, we must have Q(1) > 0. However, we
compute Q(1) = 0. Hence, the system is at most marginally stable. Indeed, using the
Matlab command roots([1 − 2.2 1.55 − 0.35]), we find that the system has three
poles: 0.5, 0.7, and 1, corresponding to the natural response terms (0.5)k , (0.7)k , and
1, for k = 0, 1, . . ., respectively. The last one, 1, is the one that makes the system
marginally stable.

(viii) Let Q(z) = z 3 − 1.9z 2 + 1.4z − 0.45. Hence, n = 3 and

a3 = 1, a2 = −1.9, a1 = 1.4, a0 = −0.45.

We apply the Jury’s test by constructing the array:

z0 z1 z2 z3
-0.45 1.4 -1.9 1
1 -1.9 1.4 -0.45
-0.7975 1.27 -0.545

For the system to be stable, we must have:

Q(1) = 0.05 > 0

(−1)3 Q(−1) = 4.75 > 0
|a0 | = 0.45 < a3 = 1
|b0 | = 0.7975 > |b2 | = 0.545,

all of which are satisfies. Hence, the system is stable. By Matlab, the three roots,
π
namely, closed-loop poles, are: 0.9, 0.5 ± j0.5 = √12 e±j 4 , corresponding to the natural
response terms: (0.9)k , Re[(0.5 + j0.5)k ] = (√12)k cos( kπ 4
), and Im[(0.5 + j0.5)k ] =
√1 sin( kπ ), for k = 0, 1, . . ., respectively.
( 2)k 4

Problem 2. Problem 7-9 (a)-(e).

Solution:

1
(a) The characteristic equation is
0.0415
1 + D(z)G(z)H = 1 + K = 0. (1)
z − 0.748
(b) Using the bilinear map
1 + (T /2)w 1 + 0.3w 10 + 3w
z= = = ,
1 − (T /2)w 1 − 0.3w 10 − 3w
the characteristic equation is transformed to
0.0415
1 + K 10+3w =0
10−3w
− 0.748
0.0415(10 − 3w)
⇔ 1+K =0 (2)
10 + 3w − 0.748(10 − 3w)
⇔ (5.244 − 0.1244K)w + (0.415K + 2.52) = 0. (3)
Since the above polynomial is first order, the Ruth Hurwitz criterion is trivial: the two
coefficients have to be of the same sign, i.e. either
5.244 − 0.1244K > 0 and 0.415K + 2.52 > 0,
or in the second case,
5.244 − 0.1244K < 0 and 0.0415K + 2.52 < 0.
The second case is impossible, and from the first case we have: −6.0723 < K < 42.145.
(If K ≥ 0 is assumed, then the range of K should be 0 ≤ K < 42.145 then)
(c) The characteristic equation (1) is equivalent to
z + (0.0415K − 0.748) = 0.
Let Q(z) = z + (0.0415K − 0.748). By the Jury test, for stability we need to have
Q(1) = 0.0415K + 0.252 > 0 ⇒ K > −6.0723
− Q(−1) = 1.748 − 0.0415K > 0 ⇒ K < 42.1205
|a0 | = |0.0415K − 0.748| < a1 = 1 ⇒ −6.0723 < K < 42.1205.
To sum up, the range of K for stability is −6.0723 < K < 42.145.
(d) If T = 0.06 s, then
4(1 − e−0.5T )
 
−1 2 0.1182
G(z) = (1 − z )Z = −0.5T
= .
s(s + 0.5) z−e z − 0.9704
Hence the characteristic equation becomes
0.1182
1+K 0.04 = 0 ⇔ z − 0.9704 + 0.0047K = 0,
z − 0.9704
whose root is easily seen to be 0.9704 − 0.0047K which is a real number. Thus, for
stability, we need to have −1 < 0.9704 − 0.0047K < 1, or equivalently,
−6.2979 < K < 419.234.

2
 (e) It is easy to see that the analog system without the sampler is stable for all K. Com-
paring the results in (b) and (d), we conclude that, by adding the sampler and the data
hold, the stability of the system might be compromised. Furthermore, the less frequent
the sampling, the easier for the system to become unstable (i.e., stability range of K
becomes narrower).

Problem 3. Problem 7-11 (a)-(d).

Solution:

(a) In this case,

0.02268z + 0.02052
G(z) = K , H(z) = 0.4.
(z − 1)(z − 0.7408)

The closed-loop transfer function can be verified as

Θ(z) D(z)G(z) G(z)

= = .
R(z) 1 + D(z)G(z)H(z) 1 + G(z)H(z)

Hence the characteristic equation is 1 + G(z)H(z) = 0, which is equivalent to

z 2 + (0.009072K − 1.7408)z + (0.008208K + 0.7408) = 0.

(b) Using the bilinear transformation z = 1+(T /2)w

1−(T /2)w
40+w
= 40−w , the characteristic equation
becomes
 2
40 + w 40 + w
+ (0.009072K − 1.7408) + (0.008208 + 0.7408) = 0
40 − w 40 − w
⇔ (40 + w)2 + (0.009072K − 1.7408)(1600 − w 2 ) + (0.008208K + 0.7408)(40 − w)2 = 0
⇔ (3.4816 − 0.000864K)w 2 + (20.736 − 0.65664K)w + 27.65K = 0.

We next find the Routh array:

w2: 3.4816 − 0.000864K 27.65K

w1: 20.736 − 0.65664K
w0: 27.65K

For the closed-loop system to be stable, the first column of the above array cannot
change sign. Thus, we must have

3.4816 − 0.000864K > 0, 20.736 − 0.65664K > 0, 27.65K > 0,

which is equivalent to

0 < K < 31.58.

3
 (c) Instead of the Routh-Hurwitz criterion, we now apply the Jury test for the polynomial
z 2 + (0.009072K − 1.7408)z + (0.008208K + 0.7408). Here, Q(z) = a2 z 2 + a1 z + a0 ,
with
a2 = 1, a1 = 0.009072K − 1.7408, a0 = 0.008208K + 0.7408.
In order for the system to be stable, the Jury test requires that

Q(1) = a2 + a1 + a0 = 0.0173K > 0 ⇒ K > 0

Q(−1) = a2 − a1 + a0 = −0.000864K + 3.4816 > 0 ⇒ K < 4029.6
|a0 | = |0.008208K + 0.7408| < a2 = 1 ⇒ K < 31.58.

Hence, the range of K for stability is 0 < K < 31.58, which is the same as in (b).

(d) In (b), plugging K = 31.58 into the equation (3.4816 − 0.000864K)w 2 + (20.736 −
0.65664K)w + 27.65K = 0 and solving for w, we have

w = ±j15.90.

Similarly, plugging K = 31.58 into the equation z 2 +(0.009072K−1.7408)z+(0.008208K+

0.7408) = 0 and solving for z, we get

z = 0.7272 ± j6865 = 1∠ ± 0.7568 rad.

Problem 4. In the system above, D(z) = K for some adjustable gain K ≥ 0, and

1 − e−sT
 
z
G(z) = Z Gp (s) = .
s (z + 2)(z 2 + 0.25)

Using your favorite method, determine the range of K ≥ 0 such that the closed-loop system
is stable.
Solution: We use the Jury’s test. The characteristic equation is equivalent to

Q(z) = (z + 2)(z 2 + 0.25) + Kz = z 3 + 2z 2 + (K + 0.25)z + 0.5 = 0.

We apply the Jury’s test by constructing the array:

z0 z1 z2 z3
0.5 K + 0.25 2 1
1 2 K + 0.25 0.5
-0.75 0.5K-1.875 0.75-K

4
For the system to be stable, we must have:

Q(1) = K + 3.75 > 0 ⇒ K > −3.75

(−1)3 Q(−1) = K − 1.25 > 0 ⇒ K > 1.25
|a0 | = 0.5 < a3 = 1
|b0 | = 0.75 > |b2 | = |0.75 − K| ⇒ 0 < K < 1.5.

To sum up, for the closed-loop system to be stable, we must have 1.25 < K < 1.5.