Solution Manual for Digital Control System Analysis and

Design 4th Edition by Phillips ISBN 0132938316

9780132938310
Full link download:
Solution Manual:
https://testbankpack.com/p/solution-manual-for-digital-
control-system-analysis-and-design-4th-edition-by-phillips-
isbn-0132938316-9780132938310/

CHAPTER 2

2.2-1. The rectangular rules for numerical integration are illustrated in Fig. P2.2-1. The left-side rule is
depicted in Fig. P2.2-1(a), and the right-side rule is depicted in Fig. P2.2-1(b). The integral of x(t)
is approximated by the sum of the rectangular areas shown for each rule. Let y(kT ) be the
numerical integral of x(t), 0 ≤ t ≤ kT.

x(t)

x(k)
x(k + 1)

kT (k + 1)T
(a)

x(t)

x(k + 1)

kT (k + 1)T
(b)

FIGURE P2.2-1 Rectangular rules for integration: (a) left side; (b) right side.

(a) Write the difference equation relating y(k + 1) , y(k) , and x(k) for the left-side rule.

(b) Find the transfer function Y (z)/X (z) for part (a).

(c) Write the difference equation relating y(k + 1), y(k) , and x(k + 1) for the right-side rule.

(d) Find the transfer function Y (z)/X (z) for part (c).
(e) Express y(k) as a summation on x(k) for the left-side rule.
(f) Express y(k) as a summation on x(k) for the right-side rule.
 Solution:
(a) y(k +1) = y(k) + Tx(k)

Y (z)
(b) zY (z) = Y (z) + TX (z)  = T
X (z) z −1

(c) y(k +1) = y(k) + Tx(k + 1)

Y (z)
(d) zY (z) = Y (z) + TzX (z)  = Tz
X (z) z −1

(e) y(1) = y(0) + Tx(0)

y(2) = y(1) + Tx(1) = y(0) + T (x(0) + x(1))

y(3) = y(2) + Tx(2) = y(0) + T x(0) + x(1) + x(2)

k −1

 y(k) = y(0) + T  x(n)

n=0

(f) y(1) = y(0) + Tx(1)

y(2) = y(1) + Tx(2) = y(0) + T x(1) + x(2)

 y(k) = y(0) + T  x(n)

n=1

2.2-2. The trapezoidal rule (modified Euler method) for numerical integration approximates the integral
of a function x(t) by summing trapezoid areas as shown in Fig. P2.2-2. Let y(t) be the integral of
x(t) .

x(t)

x(k)

x(k + 1)

kT (k + 1)T

FIGURE P2.2-2 Trapezoidal rule for numerical integration.

(a) Write the difference equation relating y ⎡(k + 1)T ⎤ , y(kT ), x ⎡(k + 1)T ⎤ , and x(kT ) for this rule.
⎣⎢ ⎥⎦ ⎣⎢ ⎥⎦

(b) Show that the transfer function for this integrator is given by
 Y ( z) (T 2)(z + 1)
=
X ( z) z−1

Solution:
x(k) + x(k +1)
(a) y(k + 1) = y(k) + T
2
T T z +1
(b) zY (z) = Y (z) + X (z) + zX (z)  Y (z) = X (z)
2 2 z −1

2.2-3. (a) The transfer function for the right-side rectangular-rule integrator was found in Problem 2.2-1
to be Y (z)/X (z) = Tz/(z − 1) . We would suspect that the reciprocal of this transfer function should
yield an approximation to a differentiator. That is, if w(kT ) is a numerical derivative of x(t) at
t = kT ,

W (z ) z−1
=
X (z) Tz

Write the difference equation describing this differentiator.

(b) Draw a figure similar to those in Fig. P2.2-1 illustrating the approximate differentiation.

(c) Repeat part (a) for the left-side rule, where W (z)/X (z) = T / (z − 1).

(d) Repeat part (b) for the differentiator of part (c).

Solution:

(a) Tz W (z) = zX (z) − X (z)

1
w(k + 1) = x(k + 1) − x(k)
T

(b)

calculated
slope
kT (k + 1)T t

(c) TW (z) = zX (z) − X (z)

 x calculated
slope

kT (k + 1)T t

1
w(k) = x(k + 1) − x(k)
T

2.3-1. Find the z-transform of the number sequence generated by sampling the time function e(t) = t
every T seconds, beginning at t = 0 . Can you express this transform in closed form?

Tz
Solution: e(t) = t; E(z ) = 0 + Tz −1 + 2Tz−2 + =
(z − 1)2

2.3-2. (a) Write, as a series, the z-transform of the number sequence generated by sampling the time
function e(t) = −t every T seconds, beginning at t = 0 . Can you express this transform in
closed form?

(b) Evaluate the coefficients in the series of part (a) for the case that T = 0.05 s .

(c) The exponential e(t) = −bt is sampled every T = 0.2 s , yielding the z-transform

⎛ 1⎞ ⎛ 1⎞2 − ⎛ 1 ⎞3 −
E(z) = 1 + ⎜ ⎟z −1 + ⎜ ⎟ z 2 + ⎜ ⎟ z 3 +
⎟
⎝⎟
⎜2⎠ ⎜⎝ 2 ⎟⎠ ⎜⎝ 2 ⎟⎠

Evaluate b.

Solution:

(a) E(z) = 1+ −T z−1 + −2T z−2 +L

1
= 1 + (−T z −1) 1+ (−T z −1)2 + = =
z
1 − −T z−1 z − −T

z
(b) E(z) = 1+ (0.9512z−1 )1 + (0.9512z−1)2 + =
z − 0.9512

(c) −bT = −0.2b = 0.5

T =0.2
 −0.2b = ln (0.5) = −0.6931  b = −3.466

2.3-3. Find the z-transforms of the number sequences generated by sampling the following time functions
every T seconds, beginning at t = 0 . Express these transforms in closed form.

(a) e(t) = −at

(b) e(t ) = −(

t−T )
u (t − T )

(c) e(t ) = −(

t−5T )
u (t − 5T )

Solution:
z
(a) e(t) = −at  E(z ) = 1+ −aT z−1 + −2aT z−2 + = 2-3.
z − −aT

(b) e(t) = −(t−T)u(t − T )


E(z ) = z−1 + −T z−2 + −2T z−3 + = z−1 z (=1 −T
 −T

z− z−

(c) e(t) = −(t−5T)u(t − 5T)

 z  1
E(z ) = z−5 + −T z−6 + −2T z−7 + = z−5 =
 −T (
 z −   z4(z − −T )

2.4-1. A function e(t) is sampled, and the resultant sequence has the z-transform

z3 − 2z
E( z ) =
z4 − 0.9z2 + 0.8

Solve this problem using E(z) and the properties of the z-transform.

(a) Find the z-transform of e(t − 2T )u(t − 2T ) .

(b) Find the z-transform of e(t + 2)u(t) .

(c) Find the z-transform of e(t − T )u(t − 2T ) .

 Solution:

(z3 − 2z)z−2
(a) �[e(t − 2T )u(t − 2T )] =
z4 − 0.9z2 + 0.8

(b) e(0) = 0, e(1) = 1

�[e(t + T )u(t)] = z[E(z) − e(0) − e(1)z−1]

 z3 − 2z 1 −1.1z2 + 0.8
=
z  −  =
z4 − 0.9z2 + 0.8 z  z4 − 0.9z2 + 0.8

(c) �[e(t − T )u(t − 2T )] = e(T )z −2 + e(2T )z −3 +

= z−1[E(z) − e(0)] = z−1E(z), since e(0) = 0

z2 − z
=
z4 − 0.9z2 + 0.8

2.4-2. A function e(t) is sampled, and the resultant sequence has the z-transform

z −b
E (z ) =
z2 − cz2 + d

Find the z-transform of akT e(kT ) . Solve this problem using E(z) and the properties of the z-
transform.

Solution:

By complex translation
z−aT − b
�⎡  akT e(kT )⎤ = E(z −aT ) =
⎣⎢ ⎦⎥ z2−2aT − cz2−2aT + d

2.5-1. From Table 2-3,

⎡ ⎤ z(z − cos aT )
�⎢⎣cos akT ⎥⎦ =
z2 − 2z cos aT + 1
 (a) Find the conditions on the parameter a such that ⇥ ⎡⎢⎣cos akT ⎤⎥⎦is first order (pole-zero cancellation
occurs).

(b) Give the first-order transfer function in part (a).

(c) Find a such that ⇥ ⎡⎢cos akT ⎤⎥= ⇥ ⎡u(kT )⎤ , where u(kT ) is the unit step function.
⎣ ⎦ ⎢⎣ ⎥⎦

Solution:

(a) poles: z = z cos a  4 cos a − 4 = cos(a)  j sin(a)

2

 pole = cos a, provided sin a = 0  a = 0,  π,  2π, K ,  nπ

Then cos a = (−1)n  poles = cos a

(b) E(z) = z(z − cos a) =

z
, a =  nπ, n = 0, 1, 
(z − cos a)(z − cos a) z − cos a

(c) E(z) = z
=
z , cos a = 1, a = 0,  2π,  4π, 
z − cos a z −1

2.5-2. Find the z-transform, in closed form, of the number sequence generated by sampling the time
function e(t) every T seconds beginning at t = 0 . The function e(t) is specified by its Laplace
transform,

E (s ) =
(
2 1 − −5s ), T = 1s
s( s + 2)

Solution:

E (s) = 2 1 −1
= +
1
s(s + 2) s s + 2

e1(t) = (1− −2t )u(t)  e1(kT ) = (1− −2kT )u(kT)

 E1 (z) = (1+ z−1 + z−2 + ) − (1 − −2T z−1 + −4T z−2 + )

 1 1 =
z
− z (1 − −2 )z
=
,T=1


= − z−1 z − −2T (z − 1)(z − −2 )

1−z −1 1 − −2 z−1

−5 (1− −2 )(z5 − 1) 0.8647(z5 −1)

E(z) = E1 (z) − z E1 (z) = =
z4 (z −1)(z − −2 ) z4 (z − 1)(z − 0.1353)

2.6-1. Solve the given difference equation for x(k) using:

1, k = 0, 1
x(k) − 3x(k − 1) + 2x(k − 2) = e(k), e(k) =

 0, k  2

x(−2) = x(−1) = 0

(a) The sequential technique.

(b) The z-transform.

(c) Will the final-value theorem give the correct value of x(k) as k → ∞ ?

Solution:

(a) x(0) = e(0) = 1

x(1) = e(1) + 3x(0) = 4

x(2) = e(2) + 3x(1) − 2x(0) = 10

x(3) = 0 + 3(10) − 2(4) = 22

x(4) = 0 + 3(22) − 2(10) = 46

z +1
(b) [1 − 3z + 2z ]X (z) = E(z) = 1 + z =
−1 −2 −1

z
z2 z +1 z(z +1)  −2 3
X (z) = 
(z − 1)(z − 2) z = =z +
(z − 1)(z − 2)  z − 1 z − 2 

 x(k) = −2 + 3(2)k

(c) No, since the final value does not exist.

2.6-2. Given the difference equation

y(k + 2) − y(k + 1) + y(k) = e(k)

3 1
4 8

where y(0) = y(1) = 0, e(0) = 0 , and e(k) = 1, k = 1, 2,… .

(a) Solve for y(k) as a function of k, and give the numerical values of y(k), 0 ≤k ≤ 4.

(b) Solve the difference equation directly for y(k), 0 ≤ k ≤ 4, to verify the results of part (a).

(c) Repeat parts (a) and (b) for e(k) = 0 for all k, and y(0) = 1, y(1) = −2 .

Solution:
 z  1
(a) E(z) = �[u(k − 1)] = z−1 =
 z − 1 z−1
 
 3 1
z2 − z+ Y (z) = E(z)
 4 
8 
1 −8 8 3
Y (z) 1 · z −1 = + + −16 + 64 3
z = 
z z−
1  1
z− z z −1 z − 1 z −1 4
2
  


 2  4
 1 k 64  1 k
 y(k ) = −8δ(0) + −16   +  
2  3  4 
7
 y(0) = 0; y(1) = 0; y(2) = 0; y(3) = 1; y(4) =
4
3 1
(b) y(k + 2) = e(k) + y(k + 1) − y(k)
4 8
3 1
y(2) = 0 + (0) − (0) = 0
4 8
3 1
y(3) = 1 + (0) − (0) = 1
4 8

3 1
y(4) = 1 + (1) − (0) = 7 4
4 8
 (

2.6-3. Given the difference equation

x ( k ) − x(k − 1) + x(k − 2) = e(k)

where e(k) = 1 for k ≥ 0.

(a) Solve for x(k) as a function of k, using the z-transform. Give the values of x(0), x(1) , and
x(2) .

(b) Verify the values x(0), x(1) , and x(2) , using the power-series method.

(c) Verify the values x(0), x(1) , and x(2) by solving the difference equationdirectly.

(d) Will the final-value property give the correct value for x(∞)?
 Solution:

z
(a) [1− z−1 + z−2 ]X (z) = E(z) =
z −1

z3 1

X (z) = , poles: z = j = 1  60

(z − 1)(z2 − z + 1) 2 2

X (z) 1 k1 k*
= + +  1 with p =  
1 60
z z −1 z − p1 z − p1*

z2 1120
k1 = =
(z −1)(z −1 − 60) (.5 + j.866 −1)(.5 + j.866 − .5 + j.866)
z=160

1120 
= = 0.5774 − 90
1120 [ j2(0.866)]
π
 aT = ln ( p ) = 0; bT = arg p =
1 1
3

A = 2 k1 = 1.155; θ = arg k1 = −90

π  π 
 x(k) = 1 + 1.155 cos k − 90 = 1 + 1.155 sin k
  3 
3  
x(0) = 1, x(1) = 2, x(2) = 2

1+ 2z−1 + 2z−2 +  x(0) = 1

(b) z3 − 2z2 + 2z − 1 z3 x(1) = 2
z − 2z + 2z − 1
3 2 x(2) = 2

2z2 − 2z + 1
2z2 − 4z + 4 − 2z−1
2z +

(c) x(k) = 1+ x(k −1) − x(k − 2)

x(0) = 1+ 0 − 0 = 1

x(1) = 1+1− 0 = 2

x(2) = 1 + 2 −1 = 2

(d) No, 3 poles for X(z) on the unit circle.

2.6-4. Given the difference equation

x(k + 2) + 3x(k + 1) + 2x(k) = e(k)

where
⎧⎪ 1, k=0
e( k ) = ⎟⎨
⎪ 0, otherwise
⎟⎩

x(0) = 1

x(1) = −1

(a) Solve for x(k) as a function of k.

(b) Evaluate x(0), x(1) , x(2) , and x(3) in part(a).

(c) Verify the results in part (b) using the power-series method.

(d) Verify the results in part (b) by solving the difference equation directly.

Solution:

(a) z2[X (z) − x(0) − x(1)z−1] + 3z[X (z) − x(0)] + 2X (z) = E(z) = 1

1 + z2 − z + 3z z2 + 2z + 1 z + 1
 X (z) = 2 = =
z − 3z + 2 z2 + 3z + 2 z + 2

 1

 X (z) = z  z +1  = z +
2

2 1
  
z(z + 2)  z z + 2 

1 1
 x(k) = (k) + (−2)k
2 2

(b) x(0) = 1, x(1) = −1, x(2) = 2, x(3) = −4

 1 − z−1 + 2z−2 − 4z−3 +
(c) z + 2 z + 1
z +2
−1
−1− 2z−1
2z− 1
2z−1 + 4z−2
−4z−2

(d) x(k + 2) = e(k) − 3x(k +1) − 2x(k)

x(2) = 1− 3(−1) − 2(1) = 2

x(3) = 0 − 3(2) − 2(−1) = −4

2.6-5. Given the difference equation

x(k + 3) − 2.2x(k + 2) + 1.57x(k + 1) − 0.36x(k) = e(k)

where e(k) = 1 for all k ≥ 0, and x(0) = x(1) = x(2) = 0 .

(a) Write a digital computer program that will calculate x(k) . Run this program solving for x(3),
x(4) , . . . , x(25) .

(b) Using the sequential technique, check the values of x(k), 0 ≤ k ≤ 5.

(c) Use the z-transform and the power-series method to verify the values x(k), 0 ≤ k ≤ 5.

Solution:

(a) x0 = 0;

x1 = 0;

x2 = 0;

for k = 0:5;

x3 = 2.2*x2 – 1.57*x1 + 0.36*x0 + 1

x0 = x1;
 x1 = x2;

x2 = x3;

end

(b) x(k + 3) = e(k) + 2.2x(k + 2) −1.57x(k +1) + 0.36x(k)

x(3) = 1+ 0 − 0 + 0 = 1

x(4) = 1+ 2.2(1) − 0 + 0 = 3.2

x(5) = 1+ 2.2(3.2) −1.57(1) = 6.47

z
(c) [z3 − 2.2z2 +1.57z − 0.36]X (z) = E(z) =
z −1

z
X (z) =
(z −1)(z − 2.2z + 1.57z − 0.36)
3 2

z−3 + 3.2z−4 + 6.47z−5 +

z4 − 3.2z3 + 3.77z2 − 1.93z + 0.36z
z − 3.2 + 3.77z−1 −
3.2 − 3.77z−1
3.2 − 10.24z−1
6.47z−1 +

 x(3) = 1
x(4) = 3.2
x(5) = 6.47

2.7-1. (a) Find e(0) , e(1) , and e(10) for

0.1
E(z ) =
z(z − 0.9)

using the inversion formula.

(b) Check the value of e(0) using the initial-value property.

(c) Check the values calculated in part (a) using partial fractions.
 (d) Find e(k) for k = 0, 1, 2, 3, and 4 if ⇥ ⎡e(k)⎤ is givenby

E( z ) =
1.98z

(z 2
− 0.9z + 0.9 )(z − 0.8)(z 2
− 1.2z + 0.27 )
(e) Find a function e(t) which, when sampled at a rate of 10 Hz (T = 0.1s), results in the

transform E(z ) = 2z /(z − 0.8) .

(f) Repeat part (e) for E(z) = 2z/(z + 0.8).

(g) From parts (e) and (f), what is the effect on the inverse z-transform of changing the sign on a
real pole?

Solution:
0.1zk−1 zk−2
(a) e(k) =  z(z − 0.9)
=  z − 0.9

residues residues

0.1 0.1
k = 0 : fcn = , residue = = 0.1235
z2(z − 0.9) z=0.9
(0.9)2

d  0.1  −0.1 = −0.1235

residue = = −0.1(1) =

dz z − 0.9  z=0

z =0
(z − 0.9)2 z =0 (0.9)2

 e(0) = 0

0.1 0.1
k = 1 : e(1) = + =0
z − 0.9 z=0 z z =0.9

k = 10 : e(10) = 0.1(0.9)8

(b)(b) e(0) = lim E(z) = lim 0.1

=0
z→ z→ z(z − 0.9)

k1 k2 k3
(c)(c) E(z) = 0.1 = + +

z z2 (z − 0.9) z2 z z − 0.9

−0.1 1
k= =− ;k 0.1 1
=
1 3
0.9 9 (0.9)2 8.1
d −1
k= 0.1  = , from (a)
2 dz z −0.9 
 z=0 8.1
 −1 1 1
 e(k) = (k) − (k −1) + (0.9)k
8.1 9 8.1

1 1 1 0.9
x(0) = − + 0 + = 0; x(1) = − 0 − + =0
8.1 8.1 9 8.1

0.1
x(10) = − 0 − 0 + (0.9)10 = 0.1(0.9)8
(0.9)2

1.98z
(d) E(z) = = 1.98z−4 + ()z−5 + ()z−6 +
z +
5

e(0) = e(1) = e(2) = e(3) = 0; e(4) = 1.98

2z
(e) E(z) = = 2z
−aT = 0.8  aT = 0.2231
z − 0.8 z − −aT
0.2231

a= = 2.231,  e(t) = 2−2.231tu(t)

0.1

2z
(f) E(z) = ; −aT  jπ = −0.8  aT = 2.231
z − (−0.8)
ωs
e(t) = 2e−2.231t cos10πt where = 10π
2

(g) (e) e(k) =(0.8)k ; (f ) e(k) = (−0.8)k

 sign alternates on e(k ).

2.7-2. For the number sequence e(k),

E( z ) =
z

( z + 1)
2

(a) Apply the final-value theorem to E(z).

(b) Check your result in part (a) by finding the inverse z-transform of E(z).

(c) Repeat parts (a) and (b) with E ( z ) = z / ( z − 1) .

2

(d) Repeat parts (a) and (b) with E ( z ) = z / ( z − 0.9) .

2
 (e) Repeat parts (a) and (b) with E ( z ) = z / ( z − 1.1) .
2

Solution:

z(z − 1)
(a) e() = lim (z − 1)E(z) = =0
z→1 (z +1)2 z =1

 z 
(b) e(k) = z−1 = k(−1)k,  e()unbounded
 
(z −1)2  z
(c) (a) e() = lim (z −1) ,
unbounded

z→1 (z −1)2

(b) e(k) = k, unbounded

z
(d) (a) e() = lim (z −1) =0
z→1 (z − 0.9)2

(b) e(k) = k(0.9)k ; e() → 0

z
(e) (a) e() = lim (z −1) =0
z→1 (z −1.1)2

(b) e(k) = k(1.1)k ; e() is unbounded.

2.7-3. Find the inverse z-transform of each E(z) below by the four methods given in the text. Compare
the values of e(z) , for k = 0, 1, 2, and 3, obtained by the four methods.
0.5
(a) E(z) = 0.5z (b) E(z ) =

(z − 1)(z − 0.6 ) (z − 1)(z − 0.6)

0.5(z + 1) z(z − 0.7)
(c) E(z) = (d) E(z ) =
( z − 1)(z − 0.6 ) (z − 1)(z − 0.6)

(e) Use MATLAB to verify the partial-fraction expansions.

Solution:
 0.5z−1 + 0.8z−2 + 0.98z−3 +
(a) (i) z − 1.6z + 0.6 0.5z
2

0.5z − 0.8 + 0.3z−1

0.8 − 0.3z−1
0.8 − 1.28z−1
0.98z−1

(ii) E(z) =0.5 1.25 −1.25 1.25z 1.25z

= + ;  E(z) = −
z (z −1)(z − 0.6) z −1 z − 0.6 z −1 z − 0.6

e(k) = 1.25(1− 0.6k )u(k)

k
(iii) zk − E(z) = 0.5z1
(z − 1)(z − 0.6)
0.5(1)k 0.5(0.6)k k
e(k) = + = 1.25(1 − 0.6 )u(k)
1 − 0.6 0.6 −1

0.5z
(iv) E1(z) =  e1 (k) = 0.5(0.6)k
z − 0.6

1
E (z) =  e2 (0) = 0; e2 (k) = 1, k  1
2
z −1

e(0) = e1(0)e2 (0) = (0.5)(0) = 0

e(1) = e1(0)e2 (1) + e1(1)e2 (0) = (0.5)(1) + (0.3)(0) = 0.5

e(2) = e1(0)e2 (2) + e1(1)e2 (1) + e1(2)e2 (0)

= 0.5 1 + 0.3 1 + 0.18  0 = 0.8

e(3) = 0.51+ 0.31+ 0.181+ 0.108 0 = 0.98

(b) e(0) = 0

e(k) = 1.25 − 2.083(0.6)k , k  1

E(z) = 0.5z−2 + 0.8z−3 + 0.98z−4 + 1.088z−5 +

(c) e(0) = 0; e(k) = 2.5 − 3.33(0.6)k , k  1

E(z) = 0.5z−1 + 1.30z−2 + 1.78z−3 + 2.068z−4 + 2.2408z−5 +

(d) e(k) = 0.75 + 0.25(0.6)k

 E(z) = 1+ 0.9z−1 + 0.84z−2 + 0.804z−3 +

(e) num=[0 0 0.5];

den=[1 –1.6 0.6];

[r, p, k] = residue (num, den)

2.8-1. Given in Fig. P2.8-1 are two digital-filter structures, or realizations, for second-order filters.
e(k)

y(k)

+ + y(k)

(a)

b2

b1

+
(k) + f(k) + y(k)
T T b0
- +
-

a1

a0

(b)

FIGURE P2.8-1 Digital-filter structures: (a) 3D; (b) 1D.

 (a) Write the difference equation for the 3D structure of Fig. P2.8-1(a), expressing y(k) as a
function of y(k − i) and e(k − i).

(b) Derive the filter transfer function Y (z)/E(z) for the 3D structure by taking the z-transform of
the equation in part (a).

(c) Write the difference equation for the 1D structure of Fig. P2.8-1(b). Two equations are
required, with one for f (k) and one for y(k) .

(d) Derive the filter transfer function Y (z)/E(z) for the 1D structure by taking the z-transform of
the equations in part (c) and eliminating F (z) .

(e) From parts (b) and (d), relate the coefficients αi , βi to ai , bi such that the two filters realize the
same transfer function.

(f) Write a computer-program segment that realizes the 3D structure. This program should be of
the form used in Example 2.10.

(g) Write a MATLAB-program segment that realizes the 1D structure. This program should be of
the form used in Example 2.10.

Solution:

(a) y(k) = 2e(k) + 1e(k −1) + 0e(k − 2) − 1y(k −1) − 0 y(k − 2)

(b) 1 + 1z−1 + 0 z−2  Y (z) = 2 + 1z−1 + 0 z−2  E(z)

Y (z) 2 z2 + 1z + 0
=
E(z) z2 + 1z + 0

(c) f (k) = e(k) − a1 f (k −1) − a0 f (k − 2)

y(k) = b2 f (k) + b1 f (k −1) + b0 f (k − 2)

E(z)
(d) F(z) = E(z) − (a1z + a0 z )F (z)  F(z) =
−1 −2

1 + a1z−1 + a0 z−2

b2 z2 + b1z + b0
Y (z) = (b + b z−1 + b z−2 ) F (z) = E(z)
2 1 0 z2 + a z + a
1 0

(e) i = ai and βi = bi , i = 1, 2
(f)

ykminus2 = 0;

ykminus1 = 0;

ekminus2 = 0;

ekminus1 = 0;

ek = 1;

for k = 0:5

yk=b2*ek+b1*ekminus1+b0*ekminus2–a1*ykminus1-a0*ykminus2;

[k, ek, yk]

ekminus2 = ekminus1;

ekminus1 = ek;

ykminus2 = ykminus1;

ykminus1 = yk;

end

(g)

fkminus2 = 0;

fkminus1 = 0;

ek = 1;

for k = 0:5

fk=ek-a1*fkminus1–a0*fkminus2;

yk = b2*fk+b1*fkminus1+b0*fkminus2;

[k, ek, yk]

fkminus2 = fkminus1;

fkminus1 = fk;

end
2.8-2. Shown in Fig. P2.8-2 is the second-order digital-filter structure 1X.

e(k) + f1(k) y(k)

+
3

+
+ f2(k)
g4

FIGURE P2.8-2 Digital-filter structure 1X.

This structure realizes the filter transfer function

D(z) = b + A + A 
2
z− p z−p

where p and p  (conjugate of p) are complex. The relationships between the filter coefficients
and the coefficients in Fig. P2.8-2 are given by

g1 = Re( p) g3 = −2 Im(A)

g2 = Im( p) g4 = 2 Re( A)

(a) To realize this filter, difference equations are required for f1(k), f2 (k), and y ( k ) . Write these
equations.

(b) Find the filter transfer function Y (z)/E(z) by taking the z-transform of the equations of part (a)
and eliminating F1(z) and F2 (z) .

(c) Verify the results in part (b) using Mason’s gain formula.

(d) Write a MATLAB-program segment that realizes the 1X structure. This program should be of
the form of that is used in Example 2.10.
 Solution:

(a) f1(k) = g1 f1(k −1) − g2 f2 (k −1) + g3e(k)

f2 (k) = g1 f2 (k −1) + g1 f1(k −1) + g4e(k)

y(k) = b2e(k) + f2 (k −1)

(b) (1) F1(z) = g1z−1F1(z) − g2 z−1F2 (z) + g3E(z)

(2) F2 (z) = g1z−1F2 (z) + g2 z−1F1(z) + g4 E(z)

(3) Y(z) = b0E(z) + z−1F2(z)

(1) (z − g1)F1(z) + g2 F2 (z) = g3 zE(z)

(2) − g2 F1(z) + (z − g1)F2 (z) = g4 zE(z)

z − g1 g3zE(z)
g4 zE(z) = (g4 z − g1g4 z + g2 g3 z) E(z)
2
 F2 (s) = −g2
z − g1 g2 (z − g1 )2 + g 22
−g2 z − g1

Y (z) g4 z + g2 g3 − g1g4
 = b2+
E(z) (z − g )2 + g 2
1 2

Re( A) + j Im(A) Re( A) − jIm( A)

also, D(z) = b2 + z − Re( p) − j Im( p) + z − Re( p) + jIm( p)

1
(g4 − jg3 ) 1 (g + jg )3
= b2 + 2
+ 2 4

z − g1 − jg2 z − g1 + jg2

g4 z − g1g4 + g2 g3
= b2 + (z − g )2 + g 2
1 2

g2 g3 z−2 + g4 (1 − g1z−1)
(c) D(z) = b0 +
1 − g1 z−1 − g1z−1 + g12 z−2 + g2 2 z−2

g4 z + g2 g3 − g1g4
= b0 + z2 − 2g z + g 2 + g 2
1 1 2

(d) f1kminus1 = 0;

f2kminus1 = 0;
 ek = 1;

for k = 0:5

yk = b0*ek+f2kminus1;

[k, ek, yk]

f1k = g1*f1kminus1 – g2*f2kminus1 + g3*ek;

f2k = g1*f2kminus1 + g2*f1kminus1 + g3*ek;

f1kminus1 = f1k;

f2kminus1 = f2k;

end

2.8-3. Given the second-order digital-filter transfer function

D(z) = 2z − 2.4z + 0.72

2

z2 − 1.4z + 0.98

(a) Find the coefficients of the 3D structure of Fig. P2.8-1 such that D(z) is realized.

(b) Find the coefficients of the ID structure of Fig. P2.8-1 such that D(z) is realized.

(c) Find the coefficients of the IX structure of Fig. P2.8-2 such that D(z) is realized.

The coefficients are identified in Problem 2.8-2.

(d) Use MATLAB to verify the partial-fraction expansions in part (c).

(e) Verify the results in part (c) using Mason’s gain formula.

Solution:

(a) 2 = 2, 1 = −2.4, 0 = 0.72, 1 = −1.4, 0 = 0.98

(b) b2 = 2, b1 = −2.4, b0 = 0.72, a1 = −1.4, a0 = 0.98

( )
1

1.4  1.42 − 4(0.98)

2

(c) poles: z = = 0.7  j0.7 = 0.99   45

2
 A A*
D(z) = 2 + +
z − 0.7 − j0.7 z − 0.7 + j0.7

 j1.96 − (1.68 + j1.68) + 0.72

A = 2z − 2.4z + 0.72
2
=
z − 0.7 + j0.7
z =0.9945
j1.4

= 0.2 + j0.6857

 g1 = 0.7 g3 = 1.371

g2 = 0.7 g4 = 0.4

(d) num =[2 -2.4 .72];

den =[1 -1.4 0.98];

[r,p,k,]=residue(num, den)

(e)  = 1− (0.7z−1 + 0.7z−1 + 0.4z−2 ) + 0.49z−2

= 1−1.4z−1 + 0.98z−2
1
D(z) = 2 + [1.371 (0.7)z−2 + 0.4z−1(1 + 0.7z−1)]

0.4z − 1.24 2z 2 − 2.4z + 0.72
=
2+ =
z2 − 1.4z + 0.98 z2 − 1.4z + 0.98

2.9-1. Find two different state-variable formulations that model the system whose difference equation is
given by:

(a) y(k + 2) + 6 y(k + 1) + 5y(k) = 2e(k)

(b) y(k + 2) + 6 y(k + 1) + 5y(k) = e(k + 1) + 2e(k)

(c) y(k + 2) + 6 y(k + 1) + 5y(k) = 3e(k + 2) + e(k + 1) + 2e(k)

Solution:

(a) Y (z) =2

U (z) z2 + 6z + 5

(1) control canonical:

 u(k) + x2(k) x1(k) y(k)

0 1 0
x(k + 1) = x(k) + u(k)
−5
−6 1
  

y(k) = 2 0x(k)

(2) observer canonical:

u(k)

Z O O
x2(k) x1(k) y(k)

−6 1 0
x(k + 1) = x(k) + u(k)
−  
   5

y(k) = 1 0x(k)

z+2 x(k + 1) = same as (a)

(b) Y (z) = (1) control canonical:
U (z) z2 + 6z + 5 y(k) = [2 1]x(k)

(2) observer canonical:

−6 1 1
x(k +1)= x(k) + u(k)
−  
   5

y(k) = 1 0x(k)

Y (z) 3z2 + z + 2 x(k + 1) = same as (a)

(c) = (1) control canonical:
U (z) z2 + 6z + 5 y(k) = [−13 −17]x(k) + 3u(k)
 (2) observer canonical:

−6 1 1
x(k + 1) = x(k) + u(k)
−  
   5

y(k) = 1 0x(k) + 3u(k)

2.9-2. Write the state equations for the observer canonical form of a system, shown in Fig. 2-10, which
has the transfer function given in (2-51) and (2-61)

b zn−1 + bz+b
G(z) =
n−1 1 0

n−1 1 0

Solution:

0 
an 1
1 0  
bn−1
 −   b 
x(k + 1) =  an−2 0 x(k ) + n−2 u(k )
0 1 
  
   
 0 0 0 0   
b
 0 

y(k ) =  1 0 0 □ 0x(k )

2.10-1.Find a state-variable formulation for the system described by the coupled second-order difference
equations giv en. The system output is y(k) , and e1(k) and e2 (k) are the system inputs. Hint:
Draw a simulation diagramfirst.

x(k + 2)+ v(k +1)= 4e1 (k) + e2 (k)

v(k + 2) − v(k) + x(k ) = 2e1 ( k )

y ( k ) = v(k + 2)− x(k + 1) + e1 (k)

Solution:
 e1(k) x1(k)
x2(k) x(k)
y(k)

e2(k) x3(k)
x4(k) n(k)

0 1 0 0 0 0 

0 0 0 −1 4 1 
e(k)
 x(k) + 
x(k + 1)= 
0 0 0 1 0 0
   
 −1 0 1 0 2 0 

y(k) = x4 (k +1) − x2 (k) + e1(k) = −x1(k) + x3 (k) − x2 (k) + e1(k)

 y(k) = −1 −1 1 0x(k) + 1 0e(k)

2.10-2.Consider the system described by

⎡ 0 1 ⎤ ⎡1 ⎤
x (k + 1 )= ⎢ ⎥ x( k ) + ⎢
⎢1 ⎥ u(k)
⎥
⎢03⎥
⎣ ⎦ ⎣ ⎦
y(k) = ⎡ −2 1 ⎤ x(k)
⎣⎢ ⎥⎦

(a) Find the transfer function Y (z)/U (z).

(b) Using any similarity transformation, find a different state model for this system.

(c) Find the transfer function of the system from the transformed state equations.

(d) Verify that A given and Aw derived in part (b) satisfy the first three properties of similarity
transformations. The fourth property was verified in part (c).

(a)
2.10-3.Consider the system of Problem 2.10-2. A similarity transformation on these equations yields

⎢⎡ d1 0⎥⎤
w( k + 1) =
⎢ ⎥ ( k ) + B wu(k)
w

0 d2 ⎥
⎣⎢ ⎦

y(k) = Cwx(k)

(a) Find d1 and d2 .

(b) Find a similarity transformation that results in the Aw matrix given. Note that this matrix is
diagonal.

(c) Find Bw and Cw .

(d) Find the transfer functions of both sets of state equations to verify the results of this problem.

2.10-4.Repeat Problem 2.10-2 for the system described by

⎡ 1 0 ⎤ ⎡ ⎤
x (k + 1 ) = ⎢ ⎥ x ( k) + ⎢ 2 ⎥ u ( k )
⎣⎢ 0 0.5 ⎥⎦ ⎣⎢ 1 ⎥⎦

y(k) = ⎡⎢ 1 2 ⎤⎥ x(k)
⎣ ⎦

(a) Find the transfer function Y (z)/U (z).

(b) Using any similarity transformation, find a different state model for this system.

(c) Find the transfer function of the system from the transformed state equations.

(d) Verify that A given and Aw derived in part (b) satisfy the first three properties of similarity
transformations. The fourth property was verified in part (c).

Solution:

(a)
2.11-1.Consider a system with the transfer function
Y (z )
G(z) = = 2

U (z ) z(z − 1)

(a) Find three different state-variable models of this system.

(b) Verify the transfer function of each state model in part (a), using (2-84).

Solution:
2 2z−2
(a) G(z) = G1 (z)G2 (z) = =
z2 − z 1 − z−1

(1)

x1(k) x2 x1 y(k)

x1(k)

x1(k) y(k)

x2(k)

change Z to 2
(3)(3)

u(k) x2 + x1 y(k)

change Z to 2

(b) (1)
 2.11-2.Consider a system described by the coupled difference equation

y(k + 2) − v(k) = 0

v(k + 1)+ y(k + 1)= u(k)

where u(k) is the system input.

(a) Find a state-variable formulation for this system. Consider the outputs to be y(k + 1) and v(k) .
Hint: Draw a simulation diagram first.

(b) Repeat part (a) with y(k) and v(k) as the outputs.

(e) Verify the transfer function V (z)/U (z) in part (d)

(c) Repeat part (a)
by taking the z-transform of the given system
with the single
output v(k) . difference equations and eliminating Y(z).

(d) Use (2-84) to (f) Verify the transfer function V (z)/U (z) in
calculate the part (d) by using Mason’s gain formula on
system the simulation diagram of part (a).
transfer
y(k + 2) y(k + 1) y(k)
x2(k) x1(k)

u(k) + u(k + 1) u(k1)

x3(k)

function with
v(k) as the Solution:
system
output, as in (a)
part (c); that
is, find V
(z)/U (z) .

1
(b) (e) z2Y (z) − V (z) = 0  Y (z) = V (z)
z2

1
zV (z) + zY (z) = zV (z) + V (z) = U (z)
z

V (z) Y0(z) 1 z
 = = 1=
U (z) U (z) z+ z +1
2
z
 (c) From (a):

1 y0(z)
u(z) 1

make u and y capital letters

Y0 (z) z−1 z
 = = 2
U (z) 1 + z −2
z +1

2.11-3.Given the system described by the state equations

⎡ 1 0 0 ⎤
x(k + 1) = + ⎡ 1 ⎤
⎢ 1 1 0 ⎥ x(k)⎥ ⎢ 0 ⎥ u⎢ (k)⎥

⎢ ⎥ ⎢ ⎥
⎢ 0 1 0 ⎥ ⎢ 0 ⎥
⎣ ⎦ ⎣ ⎦

y(k) = ⎡⎢ 0 0 1 ⎤⎥ x(k)
⎣ ⎦
 (a) Calculate the transfer function Y (z)/U (z) , using (2-84).

(b) Draw a simulation diagram for this system, from the state equations given.

(c) Use Mason’s gain formula and the simulation diagram to verify the transfer function found in
part (a).

Solution:

z − 1 0 0
(a) zI − A =  −1 z − 1 0 ;  = z − 2z + z = z(z − 1)
3 2 2

 
 0 −1 z 

 1 
z(z −1) z 1  z−1 0 0

  1 1 
Cof (zI − A) =  0 z(z − 1) z − 1 , (zI − A)−1 =  0
 (z −1)
2 z −1 
 (z −1)2 
 0 0  
1 1 1

 z(z − 1)
2 z(z − 1) z 

1 

G(z) = C[zI − A] B = 0 0 1[zI − A] 0
−1  −1

 0
1 
 1 1 1 1 1
= 0  = =

 z(z − 1)
2 z(z − 1) z  z(z − 1)2 z3 − 2z2 + z
 0

(b)

e(k) + x1(k)

x2(k)

x3(k) y(k)

(c)  = 1 − z−1 − z−1 + z−2 = 1 − 2z−1 + z−2

 z−3 1
 G(z) = =
 z3 − 2z2 + z

2.11-4.Section 2.9 gives some standard forms for state equations (simulation diagrams for the control
canonical and observer canonical forms). The MATLAB statement

⎡⎢ A,B,C,D⎤⎥ = tf 2ss(num,den)
⎣ ⎦

generates a standard set of state equations for the transfer function whose numerator
coefficients are given in the vector num and denominator coefficients in the vector den.

(a) Use the MATLAB statement given to generate a set of state equations for the transfer function

G(z) = 3z + 4
z2 + 5z + 6

(b) Draw a simulation diagram for the state equations in part (a).

(c) Determine if the simulation diagram in part (b) is one of the standard forms in Section 2.9.

Solution:

(a) n = [0 3 4];

d = [1 5 6];

[A,B,C,D] = tf2ss(n, d)
−5 −6 1
x(k + 1) = x(k) + u(k)
  
1 0
  0

y(k) = 3 4x(k)

(b)

u(k) + x1(b) x2(b) y(k)

 (c) Yes, it is the control canonical form with the states renumbered.

2.12-1.Consider the system described in Problem 2.10-2.

⎡ 0 1 ⎤ ⎡1 ⎤
x (k + 1 )= ⎢ ⎥ x( k ) + ⎢
⎢1 ⎥ u(k)
⎥
⎢03⎥
⎣ ⎦ ⎣ ⎦
y(k) = −2 1 x(k)
⎡ ⎤
⎣⎢ ⎥⎦

(a) Find the transfer function of this system.

(b) Let u(k) = 1, k ≥ 0 (a unit step function) and x(0) = 0 . Use the transfer function of part (a) to
find the system response.

(c) Find the state transition matrix Φ(k) for this system.

(d) Use (2-90) to verify the step response calculated in part (b). This calculation results in the
response expressed as a summation. Then check the values y(0), y(1) , and y(2) .

(e) Verify the results of part (d) by the iterative solution of the state equations.

Solution:

z −1 
(a) zI − A = 0 z ; = zI − A = z(z − 3) = 

− 3
 
Y (z) 1  z − 3 1 1
= C[zI − A]−1B = −2 1

U (z)  0 z  1
 

=
1
− 2z + 6 z −21= −z + 4
1
  z(z − 3)

(b) Y (z) = (−z + 4)z

z(z − 3)(z −1)

Y (z) −z + 4 4
3 −3 1

= = + 2 + 6

z z(z −1)(z − 3) z z −1 z−3

2.12-2.The system described by the equations
⎡ 1 0 ⎤ ⎡ ⎤
x (k + 1 ) = ⎢ ⎥ x ( k) + ⎢ 2 ⎥ u ( k )
⎣⎢ 0 0.5 ⎥⎦ ⎣⎢ 1 ⎥⎦
 y(k) = ⎡⎢⎣1 2⎤⎥⎦ x(k)

is excited by the initial conditions x(0) = ⎡⎢⎣ −1 2 with u( k ) = 0 for all k.

T
⎤
⎥
⎦

(a) Use (2-89) to solve for x(k), k ≥ 0.

(b) Find the output y(z) .

(c) Show that Φ(k) in (a) satisfies the property Φ(0) = I.

(d) Show that the solution in part (a) satisfies the given initial conditions.

(e) Use an iterative solution of the state equations to show that the values y(k) , for k = 0, 1, 2, and
3, in part (b) are correct.

(f) Verify the results in part (e) using MATLAB.

(a)
Solution:
2.12-3.The system described by the equations

is excited by the initial conditions x(0) = ⎡⎢⎣ −1 2 ⎤ with u(k) = 0 for all k.
T
(a) Use (2-89) to solve for x(k), k ≥ 0.

(b) Find the output y(k) .

(c) Show that Φ(k) in part (a) satisfies the property Φ(0) = I .

(d) Show that the solution in part (a) satisfies the given initial conditions.

(e) Use an iterative solution of the state equations to show that the values y(k) , for k = 0, 1, 2, and
3, in part (b) are correct.
 −15 + 14 −1
(d) x(k) = =
k =0 9 − 7  2 
   

y(0) = −3 y(2) = 1.56

(e) From (b),
y(1) = 0.6 y(3) = 1.536

−1
y(0) = Cx(0) =1 −1 = −3
 2
 
0.9
 1.1 1 −1 0.9 y(1) = 1 −1 = 0.6
;
x(1) = = 0.3 0.3
−0.3 0  2 
      
 1.29 
 1.1 1 0.9  1.29  y(2) = 1 −1  = 1.56
x(2) = = ;
−0.3 0 0.3 −  −0.27
    0.27   
 1.149 
 1.1 1  1.29   1.149  y(3) =1 −1 = 1.536
x(3) = = ;
−0.3 0 −0.27 −  −0.389
    0.387   

MATLAB:

A = [1.1 1;–0.3 0]; B = [1; 1]; C = [1 –1];

x=[–1; 2];

u = 0;

for k = 0:3

x1 = A*x + B*u;

y = C*x;

[k,y]

x = x1;

end

2.12-4.Let Φ(k) be the state transition matrix for the equations

x(k + 1) = Ax(k)

Show that Φ(k) satisfies the difference equation

Φ(k + 1) = AΦ(k)
 Solution:

x(k +1) = Ax(k); x(k) = Φ(k)x(0)

 Φ(k +1)x(0) = AΦ(k)x(0)

Since this is true for any x(0),  Φ(k +1) = AΦ(k)