You are on page 1of 33

Chapter 2 Solutions

2.2-1. The characteristic polynomial is λ2 +5λ+6. The characteristic equation is λ2 +5λ+6 =


2-1
0. Also λ2 + 5λ + 6 = (λ + 2)(λ + 3). Therefore the characteristic roots are λ1 = −2
and λ2 = −3. The characteristic modes are e−2t and e−3t . Therefore

y0 (t) = c1 e−2t + c2 e−3t

and
ẏ0 (t) = −2c1 e−2t − 3c2 e−3t
Setting t = 0, and substituting initial conditions y0 (0) = 2, ẏ0 (0) = −1 in this equation
yields ¾
c1 + c2 = 2 c1 = 5
=⇒
−2c1 − 3c2 = −1 c2 = −3
Therefore y0 (t) = 5e−2t − 3e−3t

2-2 The characteristic polynomial is λ2 +4λ+4. The characteristic equation is λ2 +4λ+4 =


2.2-2.
0. Also λ2 +4λ+4 = (λ+2)2 , so that the characteristic roots are −2 and −2 (repeated
twice). The characteristic modes are e−2t and te−2t . Therefore

y0 (t) = c1 e−2t + c2 te−2t


ẏ0 (t) = −2c1 e−2t − 2c2 te−2t + c2 e−2t
and
Setting t = 0 and substituting initial conditions yields
¾
3 = c1 c1 = 3
=⇒
−4 = −2c1 + c2 c2 = 2

Therefore y0 (t) = (3 + 2t)e−2t

2.2-3. The characteristic polynomial is λ(λ + 1) = λ2 + λ. The characteristic equation is


2-3
λ(λ + 1) = 0. The characteristic roots are 0 and −1. The characteristic modes are 1
and e−t . Therefore
y0 (t) = c1 + c2 e−t
ẏ0 (t) = −c2 e−t
and
Setting t = 0, and substituting initial conditions yields
¾
1 = c1 + c2 c1 = 2
=⇒
1 = −c2 c2 = −1

Therefore y0 (t) = 2 − e−t

54

32
2.2-4.
2-4 The characteristic polynomial is λ2 + 9. The characteristic equation is λ2 + 9 = 0 or
(λ + j3)(λ − j3) = 0. The characteristic roots are ±j3. The characteristic modes are
ej3t and e−j3t . Therefore
y0 (t) = c cos(3t + θ)
and
ẏ0 (t) = −3c sin(3t + θ)
Setting t = 0, and substituting initial conditions yields
¾ ¾
0 = c cos θ c cos θ = 0 c=2
=⇒ =⇒
6 = −3c sin θ c sin θ = −2 θ = −π/2

Therefore π
y0 (t) = 2 cos(3t − ) = 2 sin 3t
2
2.2-5.
2-5 The characteristic polynomial is λ2 + 4λ + 13. The characteristic equation is λ2 + 4λ +
13 = 0 or (λ + 2 − j3)(λ + 2 + j3) = 0. The characteristic roots are −2 ± j3. The
characteristic modes are c1 e(−2+j3)t and c2 e(−2−j3)t . Therefore

y0 (t) = ce−2t cos(3t + θ)

and
ẏ0 (t) = −2ce−2t cos(3t + θ) − 3ce−2t sin(3t + θ)
Setting t = 0, and substituting initial conditions yields
¾ ¾
5 = c cos θ c cos θ = 5 c = 10
=⇒ =⇒
15.98 = −2c cos θ − 3c sin θ c sin θ = −8.66 θ = −π/3

Therefore π
y0 (t) = 10e−2t cos(3t − )
3
2.2-6. The characteristic polynomial is λ2 (λ + 1) or λ3 + λ2 . The characteristic equation is
2-6
λ2 (λ+1) = 0. The characteristic roots are 0, 0 and −1 (0 is repeated twice). Therefore

y0 (t) = c1 + c2 t + c3 e−t

and ẏ0 (t) = c2 − c3 e−t


ÿ0 (t) = c3 e−t

Setting t = 0, and substituting initial conditions yields



4 = c1 + c3  c1 = 5
3 = c2 − c3 =⇒ c2 = 2

−1 = c3 c3 = −1

Therefore y0 (t) = 5 + 2t − e−t

2.2-7. The characteristic polynomial is (λ + 1)(λ2 + 5λ + 6). The characteristic equation is


2-7
(λ + 1)(λ2 + 5λ + 6) = 0 or (λ + 1)(λ + 2)(λ + 3) = 0. The characteristic roots are −1,
−2 and −3. The characteristic modes are e−t , e−2t and e−3t . Therefore
y0 (t) = c1 e−t + c2 e−2t + c3 e−3t

and ẏ0 (t) = −c1 e−t − 2c2 e−2t − 3c3 e−3t

55

33
ÿ0 (t) = c1 e−t + 4c2 e−2t + 9c3 e−3t

Setting t = 0, and substituting initial conditions yields



2 = c1 + c2 + c3  c1 = 6
−1 = −c1 − 2c2 − 3c3 =⇒ c2 = −7

5 = c1 + 4c2 + 9c3 c3 = 3

Therefore y0 (t) = 6e−t − 7e−2t + 3e−3t

2-8 The zero-input response for a LTIC system is given as y0 (t) = 2e−t + 3. Since two
2.2-8.
modes are visible, the system must have, at least, the characteristic roots λ1 = 0 and
λ2 = −1.
(a) No, it is not possible for the system’s characteristic equation to be λ + 1 = 0
since the required mode at λ = 0 is missing.

(b) Yes, it is possible for the system’s characteristic equation to be 3(λ2 + λ) = 0
since this equation has the two required roots λ1 = 0 and λ2 = −1.
(c) Yes, it is possible for the system’s characteristic equation to be λ(λ+1)2 = 0. This
equation supports a general zero-input response of y0 (t) = c1 +c2 e−t +c3 te−t . By
letting c1 = 3, c2 = 2, and c3 = 0, the observed zero-input response is possible.
2.3-1. The characteristic equation is λ2 + 4λ + 3 = (λ + 1)(λ + 3) = 0. The characteristic
2-9
modes are e−t and e−3t . Therefore
yn (t) = c1 e−t + c2 e−3t
ẏn (t) = −c1 e−t − 3c2 e−3t

Setting t = 0, and substituting y(0) = 0, ẏ(0) = 1, we obtain


¾ 1
0 = c1 + c2 1 = 2
=⇒
1 = −c1 − 3c2 c2 = − 21

Therefore 1 −t
yn (t) = (e − e−3t )
2
h(t) = [P (D)yn (t)]u(t) = [(D + 5)yn (t)]u(t) = [ẏn (t) + 5yn (t)]u(t) = (2e−t − e−3t )u(t)

2.3-2. The characteristic equation is λ2 + 5λ + 6 = (λ + 2)(λ + 3) = 0. and


2-10

yn (t) = c1 e−2t + c2 e−3t


ẏn (t) = −2c1 e−2t − 3c2 e−3t

Setting t = 0, and substituting y(0) = 0, ẏ(0) = 1, we obtain


¾
0 = c1 + c2 c1 = 1
=⇒
1 = −2c1 − 3c2 c2 = −1

Therefore yn (t) = e−2t − e−3t


and
[P (D)yn (t)]u(t) = [ÿn (t) + 7ẏn (t) + 11yn (t)]u(t) = (e−2t + e−3t )u(t)
Hence
h(t) = bn δ(t) + [P (D)yn (t)]u(t) = δ(t) + (e−2t + e−3t )u(t)

56

34
2.3-3. The characteristic equation is λ + 1 = 0 and
2-11

yn (t) = ce−t

In this case the initial condition is ynn−1 (0) = yn (0) = 1. Setting t = 0, and using
yn (0) = 1, we obtain c = 1, and
yn (t) = e−t
P (D)yn (t) = [−ẏn (t) + yn (t)]u(t) = 2e−t u(t)

Hence h(t) = bn δ(t) + [P (D)yn (t)]u(t) = −δ(t) + 2e−t u(t)

2.3-4. The characteristic equation is λ2 + 6λ + 9 = (λ + 3)2 = 0. Therefore


2-12

yn (t) = (c1 + c2 t)e−3t


ẏn (t) = [−3(c1 + c2 t) + c2 ]e−3t

Setting t = 0, and substituting yn (0) = 1, ẏn (0) = 1, we obtain


¾
0 = c1 c1 = 0
=⇒
1 = −3c1 + c2 c2 = 1

and yn (t) = te−3t


Hence
h(t) = [P (D)yn (t)]u(t) = [2ẏn (t) + 9yn (t)]u(t) = (2 + 3t)e−3t u(t)

2.4-1.
2-13
Z ∞
Ac = c(t) dt
−∞
Z ∞ ·Z ∞ ¸
= x(τ )g(t − τ ) dτ dt
−∞ −∞
Z ∞ ·Z ∞ ¸
= x(τ ) dτ g(t − τ ) dt
−∞ −∞
Z ∞
= Ax g(t − τ ) dt
−∞
= Ax Ag

This property can be readily verified from Examples 2.7 and 2.8. For Example 2.6,
we note that Z ∞
1
e−at dt =
−∞ a
Use of this result yields Ax = 1, Ah = 0.5, and Ay = 1 − 0.5 = 0.5 = Ax Ah For
example 2.8, Ax = 2, Ag = 1.5, and
Z 1 Z 2 Z 4
1 2 1
Ac = − (t + 1)2 dt + t dt + − (t2 − 2t − 8) dt
−1 6 1 3 2 6
4 14
= +1+ = 3 = Ax Ag
9 9

57

35
2.4-2.
2-14
Z ∞
x(at) ∗ g(at) = x(aτ )g[a(t − τ )] dτ
−∞
Z ∞
1
= x(w)g(at − w) dw
a −∞
1
= c(at) a≥0
a
When a < 0, the limits of integration become from ∞ to −∞, which is equivalent to
the limits from −∞ to ∞ with a negative sign. Hence, x(at) ∗ g(at) = | a1 |c(at).
2.4-3. Let x(t) ∗ g(t) = c(t). Using the time scaling property in Prob. 2.4-2
2-15 2-14 with a = −1,
we have x(−t) ∗ g(−t) = c(−t). Now, if x(t) and g(t) are both even functions of t,
then x(t) = x(−t) and g(−t) = g(t). Clearly c(t) = c(−t). Using a parallel argument,
we can show that if both functions are odd, c(t) = c(−t), indicating that c(t) is even.
But if one is odd and the other is even, c(t) = −c(−t), indicating that c(t) is odd.
2-16
2.4-4.
Z t Z t
−at −bt −aτ −b(t−τ ) −bt
e u(t) ∗ e u(t) = e e dτ = e e(b−a)τ dτ
0 0
¯t
e−bt (b−a)τ ¯¯ e−bt (b−a)t e−at − e−bt
= e ¯ = [e − 1] =
b−a 0 b−a a−b

Because both functions are causal, their convolution is zero for t < 0. Therefore
µ −at ¶
e − e−bt
e−at u(t) ∗ e−bt u(t) = u(t)
a−b
2.4-5.
2-17 (i)
Z t Z t ¯t
¯
u(t) ∗ u(t) = u(τ )u(t − τ ) dτ = dτ = τ ¯¯ = t for t ≥ 0
0 0 0
= 0 for t < 0

Therefore
u(t) ∗ u(t) = tu(t)
(ii) Because both functions are causal
Z t Z t
e−at u(t) ∗ e−at u(t) = e−aτ e−a(t−τ ) dτ = e−at dτ
0 0
= te−at t≥0

and
e−at u(t) ∗ e−at u(t) = te−at u(t)
(iii) Because both functions are causal
Z t
tu(t) ∗ u(t) = τ u(τ )u(τ − t) dτ
0

The range of integration is 0 ≤ τ ≤ t. Therefore τ > 0 and τ − t > 0 so that

58

36
u(τ ) = u(τ − t) = 1 and
Z t
t2
tu(t) ∗ u(t) = τ dτ = t≥0
0 2
and 1 2
tu(t) ∗ u(t) = t u(t)
2
2.4-6.
2-18 (i)
µZ t ¶
sin tu(t) ∗ u(t) = sin τ u(τ )u(t − τ ) dτ u(t)
0

Because τ and t − τ are both nonnegative (when 0 ≤ τ ≤ t), u(τ ) = u(t − τ ) = 1,


and µZ t ¶
sin t u(t) ∗ u(t) = sin τ dτ u(t) = (1 − cos t)u(t)
0

(ii) Similarly µZ t ¶
cos t u(t) ∗ u(t) = cos τ dτ u(t) = sin t u(t)
0

2.4-7. In this problem, we use Table 2.1 to find the desired convolution.
2-19

(a) y(t) = h(t) ∗ x(t) = e−t u(t) ∗ u(t) = (1 − e−t )u(t)


(b) y(t) = h(t) ∗ x(t) = e−t u(t) ∗ e−t u(t) = te−t u(t)
(c) y(t) = e−t u(t) ∗ e−2t u(t) = (e−t − e−2t )u(t)
(d) y(t) = sin 3tu(t) ∗ e−t u(t)
Here we use pair 12 (Table 2.1) with α = 0, β = 3, θ = −90◦ and λ = −1. This
yields · ¸
−3
φ = tan−1 = −108.4◦
−1
and
(cos 18.4◦ )e−t − cos(3t + 18.4◦ )
sin 3t u(t) ∗ e−t u(t) = √ u(t)
10
0.9486e−t − cos(3t + 18.4◦ )
= √ u(t)
10
2.4-8.
2-20 (a)

y(t) = (2e−3t − e−2t )u(t) ∗ u(t) = 2e−3t u(t) ∗ u(t) − e−2t u(t) ∗ u(t)
· ¸
2(1 − e−3t ) 1 − e−2t
= − u(t)
3 2
µ ¶
1 2 −3t 1 −2t
= − e + e u(t)
6 3 2
(b)

(2e−3t − e−2t )u(t) ∗ e−t u(t) = 2e−3t u(t) ∗ e−t u(t) − e−2t u(t) ∗ e−t u(t)
· −t ¸
2(e − e−3t ) e−t − e−2t
= − u(t)
2 1
= (e−2t − e−3t )u(t)

59

37
(c)

y(t) = (2e−3t − e−2t )u(t) ∗ e−2t u(t) = 2e−3t u(t) ∗ e−2t u(t) − e−2t u(t) ∗ e−2t u(t)
· −2t ¸
2(e − e−3t ) −2t
= − te u(t)
1
= [(2 − t)e−2t − 2e−3t ]u(t)

2.4-9.

y(t) = (1 − 2t)e−2t u(t) ∗ u(t) = e−2t u(t) ∗ u(t) − 2te−2t u(t) ∗ u(t)
·µ ¶ µ ¶¸
1 − e−2t 1 1 −2t
= − − e − te−2t u(t)
2 2 2
= te−2t u(t)

2.4-10. (a) For y(t) = 4e−2t cos 3t u(t) ∗ u(t), We use pair 12 with α = 2, β = 3, θ = 0,
λ = 0. Therefore · ¸
−1 −3
φ = tan = −56.31◦
2
and · ¸
cos(56.31◦ ) − e−2t cos(3t + 56.31◦ )
y(t) = 4 √ u(t)
4+9
4 £ ¤
= √ 0.555 − e−2t cos(3t + 56.31◦ ) u(t)
13
(b) For y(t) = 4e−2t cos 3tu(t) ∗ e−t u(t), we use pair 12 with α = 2, β = 3, θ = 0,
and λ = −1. Therefore
· ¸
−3
φ = tan−1 = −71.56◦
1

and · ¸
cos(71.56◦ )e−t − e−2t cos(3t + 71.56◦ )
y(t) = 4 √ u(t)
10
4 £ ¤
= √ 0.316e−t − e−2t cos(3t + 71.56◦ ) u(t)
10
· ¸
1
= 4 e−t − √ e−2t cos(3t + 71.56◦ ) u(t)
10

2.4-11.
2-21 (a) y(t) = e−t u(t) ∗ e−2t u(t) = (e−t − e−2t )u(t)
£ ¤
(b) e−2(t−3) u(t) = e6 e−2t u(t), and y(t) = e6 e−t u(t) ∗ e−2t u(t) = e6 (e−t − e−2t )u(t)
(c) e−2t u(t − 3) = e−6 e−2(t−3) u(t − 3). Now from the result
£ in part (a) and the
¤ shift
property of the convolution [Eq. (2.34)]: y(t) = e−6 e−(t−3) u(t) − e−2(t−3) u(t−
3)
(d) x(t) = u(t) − u(t − 1). Now y1 (t), the system response to u(t) is given by

y1 (t) = e−t u(t) ∗ u(t) = (1 − e−t )u(t)

The system response to u(t − 1) is y1 (t − 1) because of time-invariance property.


Therefore the response y(t) to x(t) = u(t) − u(t − 1) is given by

y(t) = y1 (t) − y1 (t − 1) = (1 − e−t )u(t) − [1 − e−(t−1) ]u(t − 1)

60

38
(d)
Z 3+t
c(t) = e−τ dτ = e−t (1 − e−3 ) = 0.95e−t t≥0
t
Z 3+t
= e−τ dτ = 1 − e−(3+t) = 1 − 0.0498e−t 0 ≥ t ≥ −3
0
= 0 t ≤ −3
(e)
R −1+t 1 π
c(t) = −∞
dτ = tan−1 (t − 1) +
τ 2 +1 2 t≤1
R0 ¯0
1 −1 ¯
c(t) = −∞ 2
τ +1 dτ = tan τ ¯ = π2 t≥1
−∞

(f)
Rt
c(t) = 0
e−τ dτ = 1 − e−t 0≤t≤3
Rt
c(t) = t−3
e−τ dτ = e−(t−3) − e−t t≥3
c(t) = 0 t≤0

(g) This problem is more conveniently solved by inverting x1 (t) rather than x2 (t)
R t+1 1
c(t) = t
(τ − t) dτ = 2 t≥0
R t+1
c(t) = 0
(τ − t) dτ = 21 (1 − t2 ) 0 ≥ t ≥ −1
c(t) = 0 for t≥0

(h) x1 (t) = et , x2 (t) = e−2t , x1 (τ ) = eτ , x2 (t − τ ) = e−2(t−τ ) .


Z 0 Z 0
1 −2t
c(t) = eτ e−2(t−τ ) dτ = e−2t e3τ dτ = [e − et−3 ] 0≤t≤1
−1+t −1+t 3
Z t Z t
1 t
c(t) = eτ e−2(t−τ ) dτ = e−2t e3τ dτ = [e − et−3 ] 0 ≥ t ≥ −1
−1+t −1+t 3
Z t Z t
τ −2(t−τ ) −2t 1 t
c(t) = e e dτ = e e3τ dτ = [e − e−2(t+3) ] − 1 ≥ t ≥ −2
−2 −2 3
c(t) = 0 t ≤ −2

2.4-19.
2-29

ẋ(t) = δ(t) − δ(t − 2)

By inspection, we find Z µ ¶
t
t−1
W (τ )dτ = ∆
0 2
Therefore µ ¶
t−1
x(t) ∗ W (t) = [δ(t) − δ(t − 2)] ∗ ∆
2
µ ¶ µ ¶
t−1 t−3
= ∆ −∆
2 2

64

42
2-31 y(t) = x(t) ∗ h(t). For t < −1, y(t) = 0;
2.4-21.
Rt ¯t
For −1 ≤ t < 0, y(t) = −1 (τ + 1)dτ = τ 2 /2 + τ ¯τ =−1 = t2 /2 + t + 1/2.
R0
For t ≥ 0, y(t) = −1 (τ + 1)dτ = 1/2.
Thus, 
 0 t < −1
y(t) = t2 /2 + t + 1/2 −1 ≤ t < 0 .

1/2 t≥0

2.4-22.
2-32 RUsing the graph of the system response, h(t) = (−t/2 + 1)(u(t) − u(t − 2)). y(1) =

h
−∞ total
(τ )x(1 − τ )dτ . Since x(t) is causal, the upper limit of the integral is one.
Furthermore, since h(t) is causal, the total response htotal (t) = h(t) ∗ h(t) is also
causal, which makes the lower limit of the integral zero. Over [0, 1], x(t) = u(t) = 1.
R1
Thus, y(1) = 0 htotal (τ )dτ . To compute y(1), it is only necessary to know htotal (t)
up to t = 1.
Rt Rt
Over (0 ≤ t < 2), htotal (t) = 0 (−τ /2 + 1)(−(t − τ )/2 + 1)dτ = 0 (−τ /2 + 1)(τ /2 +
Rt¡ ¢ t3 3
1 − t/2)dτ = 0 −τ 2 /4 + τ (1 − 1 + t/2)/2 + (1 − t/2) dτ = − 12 + t8 + (1 − t/2)t =
t3 t2
24 − 2 + t.
Thus,
Z 1 ¯1
3 2 τ4 τ3 τ 2 ¯¯ 1 1 1 11
y(1) = (τ /24 − τ /2 + τ )dτ = − + ¯ = − + = = 0.34375.
0 96 6 2 τ =0 96 6 2 32

2.4-23.
2-33 (a) Using KVL, x(t) = vL (t) + y(t). Also, iC (t) = C dy diL diC
dt and vL (t) = L dt = L dt =
2
LC ddt2y . Combining yields

d2 y 1 1
+ y(t) = x(t).
dt2 LC LC

(b) The characteristic equation is


1
λ2 + = 0.
LC
The characteristic roots are
±
λ1,2 = √ .
LC
(c) The form of the ¯zero-input response is y0 (t) = c1 eλ1 t + c2 eλ2 t . Using λ1 = −λ2 ,
¯
ic (0) = 0 = C dy
dt ¯ = C(c1 λ1 + c2 λ2 ) = C(c1 λ1 − c2 λ1 ) = Cλ1 (c1 − c2 ). Thus,
t=0
c1 = c2 . Also, vc (0) = 1 = y(0) = c1 + c2 . Combining yields √2c1 = 2c2 =√ 1 or
c1 = c2 = 0.5. The zero-input response is thus y0 (t) = 0.5(et/ LC + e−t/ LC ).
Using Euler’s identity yields

y0 (t) = cos(t/ LC).

(d) MATLAB is used to plot y0 (t) for a short time after t ≥ 0.


>> t = linspace(0,6*pi,201); y_0 = cos(t); plot(t,y_0,’k’);
>> axis tight; xlabel(’t(LC)^{-1/2}’); ylabel(’y_0(t)’);

66

44
1

0.8

0.6

0.4

0.2

y0(t)
0

−0.2

−0.4

−0.6

−0.8

−1
0 2 4 6 8 10 12 14 16 18
t(LC)−1/2

Figure S2.4-23d:
S2-33d Plot of y0 (t)

Since y0 (t) is a non-decaying sinusoid, it continues forever; the initial conditions


never die out.
(e) Since L = C = 1, λ1,2 = ±. Let ỹ0 (t) = c̃1 et +c̃2 e−t . Using ỹ0 (0) = 0 = c̃1 +c̃2 ,
(1)
we know c̃1 = −c̃2 . Combining with ỹ0 (0) = 1 = c̃1 − c̃2 , we know 2c̃1 = 1 or
t −t
c̃1 = −0.5. Thus, c̃2 = 0.5 and ỹ0 (t) = e −e 2 = sin(t). From this, the system
1
response is determined to be h(t) = LC sin(t)u(t) = sin(t)u(t).
Rt
Next,
³ the zero-state response
´ is computed³ as x(t) ³ τ e−(t−τ ) dτ =
∗ h(t) = 0 sin´´
R t R t
e−t 0 Imag (eτ eτ ) dτ u(t) = Imag e−t 0 eτ (1+) dτ u(t) =
µ µ ¯ ¶¶ ³ ³ ´´
τ (1+) ¯ t t(1+)
Imag e−t e 1+ ¯ u(t) = Imag e−t e 1+−1 u(t) =
τ =0
³ ³ t −t ´´
Imag e 1+−e
u(t) = (Imag (0.5et − 0.5et − 0.5(1 − )e−t )) u(t) =
(0.5 sin(t) − 0.5 cos(t) + 0.5e−t ) u(t).
Summing the zero-state response and the zero-input response calculated
in 2.4-23c yields the total response, y(t) = x(t) ∗ h(t) + y0 (t) =
(0.5 sin(t) − 0.5 cos(t) + 0.5e−t + cos(t)) u(t). Thus,
¡ ¢
y(t) = 0.5 sin(t) + 0.5 cos(t) + 0.5e−t u(t).

2.4-24.
2-34 (a) MATLAB is used to sketch h1 (t) and h2 (t).
>> h1 = inline(’(1-t).*((t>=0)-(t>=1))’);
>> h2 = inline(’t.*((t>=-2)-(t>=2))’);
>> t = linspace(-2.5,2.5,501);
>> subplot(211),plot(t,h1(t),’k’);
>> axis([-2.5 2.5 -2.5 2.5]); xlabel(’t’); ylabel(’h_1(t)’);
>> subplot(212),plot(t,h2(t),’k’);
>> axis([-2.5 2.5 -2.5 2.5]); xlabel(’t’); ylabel(’h_2(t)’);
(b) For a parallel connection, hp (t) = h1 (t) + h2 (t). MATLAB is used to plot hp (t).
>> h1 = inline(’(1-t).*((t>=0)-(t>=1))’);
>> h2 = inline(’t.*((t>=-2)-(t>=2))’);
>> t = linspace(-2.5,2.5,501); hp = h1(t)+h2(t);

67

45
2

h1(t)
0

−1

−2

−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5


t

h2(t)
0

−1

−2

−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5


t

Figure S2.4-24a:
S2-34a: Plots of h1 (t) and h2 (t).

>> plot(t,hp,’k’);
>> axis([-2.5 2.5 -2.5 2.5]); xlabel(’t’); ylabel(’h_p(t)’);

2.5

1.5

0.5
hp(t)

−0.5

−1

−1.5

−2

−2.5
−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5
t

Figure S2.4-24b:
S2-34b: Plot of hp (t).

(c) For a series connection, hs (t) = h1 (t) ∗ h2 (t).


For (t < −2), hs (t) = 0.
R t+2 R t+2 ¡ ¢
For (−2 ≤ t < −1), hs (t) = 0 (1 − τ )(t − τ )dτ = 0 t − τ (t + 1) + τ 2 dτ =
¯t+2
tτ − (t + 1)τ 2 /2 + τ 3 /3¯τ =0 = t(t + 2) − (t + 1)(t + 2)2 /2 + (t + 2)3 /3 = −t3 /6 +
t2 /2 + 2t + 2/3.
R1 R1¡ ¢
For (−1 ≤ t < 2), hs (t) = 0 (1 − τ )(t − τ )dτ = 0 t − τ (t + 1) + τ 2 dτ =
¯1
tτ − (t + 1)τ 2 /2 + τ 3 /3¯τ =0 = t − (t + 1)/2 + 1/3 = t/2 − 1/6.
R1
For (2 ≤ t < 3), hs (t) = t−2
(1 − τ )(t − τ )dτ =
R1 ¡ 2
¢ 2 3 ¯1
¯
t−2
t − τ (t + 1) + τ dτ = tτ − (t + 1)τ /2 + τ /3 τ =t−2 = t/2 −
¡ 2 3
¢
1/6
¡ 3 − 2t(t − 2) − (t + 1)(t ¢ − 2) /2 + (t − 2) /3 = t/2 − 1/6 −
−t /6 + t /2 + 2t − 14/3 = t3 /6 − t2 /2 − 3t/2 + 9/2.
For (t > 3), hs (t) = 0.

68

46
Combining all pieces yields


 −t3 /6 + t2 /2 + 2t + 2/3 −2 ≤ t < −1

t/2 − 1/6 −1 ≤ t < 2
hs (t) = 3 2 .

 t /6 − t /2 − 3t/2 + 9/2 2≤t<3

0 otherwise

MATLAB is used to plot hs (t).


>> t = linspace(-2.5,3.5,501);
>> hs = (-t.^3/6+t.^2/2+2*t+2/3).*((t>=-2)&(t<-1));
>> hs = hs+(t/2-1/6).*((t>=-1)&(t<2));
>> hs = hs+(t.^3/6-t.^2/2-3*t/2+9/2).*((t>=2)&(t<3));
>> plot(t,hs,’k’); xlabel(’t’); ylabel(’h_s(t)’);

0.8

0.6

0.4

0.2
hs(t)

−0.2

−0.4

−0.6

−0.8
−3 −2 −1 0 1 2 3 4
t

Figure S2.4-24c:
S2-34c: Plot of hs (t).

1 1
2.4-25.
2-35 (a) Using KVL, x(t) = RC ẏ(t) + y(t) or ẏ(t) + RC y(t) = RC x(t). The characteristic
−1
root is λ = RC .
The zero-input response has form y0 (t) = c1 e−t/(RC) . Using the IC, y0 (0) = 2 =
c1 . Thus, y0 (t) = 2e−t/(RC) .
The zero-state response is x(t) ∗ h(t), where h(t) = b0 δ(t) + [P (D)ỹ0 (t)]u(t).
For this first-order system, ỹ0 (t) = c̃1 e−t/(RC) and ỹ0 (0) = 1 = c̃1 . Using
1
ỹ0 (t) = e−t/(RC) , b0 = 0, and P (D) = RC , the impulse
³R response is´ h(t) =
1 −t/(RC) t 1 −τ /(RC)
e u(t). Thus, the zero-state response is e dτ u(t) =
³RC ¯ t
´ ¡ ¢
0 RC

−e−τ /(RC) ¯τ =0 u(t) = 1 − e−t/(RC) u(t).


For t ≥ 0, the total response is the sum of the zero-input response and the zero
state response, ³ ´
y(t) = 1 + e−t/(RC) u(t).

2-35a, we know the zero-input response is y0 (t) = y0 (0)e−t/(RC) . Since


(b) From 2.4-25a,
the system is time-invariant, the unit step response from 2.4-25a
2-35a is shifted by
one
¡ to provide the
¢ response to x(t) = u(t − 1). Thus, the zero-state response is
1 − e−(t−1)/(RC) u(t − 1). Summing the two parts together and evaluating at

69

47
t = 2 yields y(2) = 1/2 = y0 (0)e−2/(RC) + (1 − e−1/(RC) ). Solving for y0 (0) yields

y0 (0) = e1/(RC) − 0.5e2/(RC) .

2.4-26.
2-36 Notice, x(2t) is a compressed version of x(t). The convolution y(t) = x(t) ∗ x(2t) has
several distinct regions.
For t < 0 and t ≥ 3/2, y(t) = 0.
Rt ¯t
For 0 ≤ t < 1/2, y(t) = 0 2τ (t − τ )dτ = tτ 2 − 2τ 3 /3¯τ =0 = t3 /3.
R 1/2 ¯1/2
For 1/2 ≤ t < 1, y(t) = 0 2τ (t − τ )dτ = tτ 2 − 2τ 3 /3¯τ =0 = t/4 − 1/12.
R 1/2 ¯1/2
For 1 ≤ t < 3/2, y(t) = t−1 2τ (t − τ )dτ = tτ 2 − 2τ 3 /3¯τ =t−1 = t/4 − 1/12 − (t3 −
2t2 + t − 2t3 /3 + 2t2 − 2t + 2/3) = −t3 /3 + 5t/4 − 3/4.
Thus, 

 t3 /3 0 ≤ t < 1/2

t/4 − 1/12 1/2 ≤ t < 1
y(t) = .

 −t3 /3 + 5t/4 − 3/4 1 ≤ t < 3/2

0 otherwise

MATLAB is used to plot y(t).

>> t = linspace(-1/2,2,251);
>> y = (t.^3/3).*((t>=0)&(t<1/2));
>> y = y+(t/4-1/12).*((t>=1/2)&(t<1));
>> y = y+(-t.^3/3+5*t/4-3/4).*((t>=1)&(t<3/2));
>> plot(t,y,’k’); xlabel(’t’); ylabel(’y(t)’);

0.2

0.18

0.16

0.14

0.12
y(t)

0.1

0.08

0.06

0.04

0.02

0
−0.5 0 0.5 1 1.5 2
t

Figure S2.4-26:
S2-36 Plot of y(t) = x(t) ∗ x(2t).

2.4-27.
2-37 Notice, vR (t) = vL1 (t) = vL2 (t) = v(t).
(a) KCL at the top node gives x(t) = y(t)+iL1 (t)+iR . Since v(t) = L2 ẏ(t), we know
iR (t) = v(t)/R = LR2 ẏ(t). Thus, x(t) = y(t) + LR2 ẏ(t) + iL1 (t). Differentiating
(1) (1)
this expression yields ẋ(t) = ẏ(t) + LR2 ÿ(t) + iL1 (t). However, iL1 (t) = v(t)/L1 =

70

48
L2 L2 L2
L1 ẏ(t). Thus, ẋ(t) = ẏ(t) + R ÿ(t) + L1 ẏ(t) or
µ ¶
R R R
ÿ(t) + + ẏ(t) = ẋ(t).
L1 L2 L2
³ ´
(b) The characteristic equation is λ2 + LR1 + R
L2 λ = 0 which yields characteristic
³ ´
roots of λ1 = 0 and λ2 = − LR1 + LR2 .
(c) The zero-input response has form y0 (t) = c1 + c2 eλ2 t . Each inductor has an
initial current of one amp each. Thus, y0 (0) = 1 = c1 + c2 . The initial resistor
current is iR (0) = −iL1 (0) − iL2 (0) = −2 and the initial resistor voltage is
2L1
v(0) = iR (0)R = −2R. Thus, ẏ0 (0) = − 2R
L2 = λ2 c2 . Solving yields c2 = L1 +L2
and c1 = 1 − c2 = L 2 −L1
L1 +L2 . Thus,

L2 − L1 2L1
y0 (t) = + e−tR/L1 −tR/L2 .
L1 + L2 L1 + L2
2-38 Since the system step response is s(t) = e−t u(t)−e−2t u(t), the system impulse response
2.4-28.
d
is h(t) = dt s(t) = −e−t u(t) + δ(t) + 2e−2t u(t) − δ(t) = (2e−2t − e−t )u(t). The input

x(t) = δ(t − π) − cos( 3)u(t) is just a sum of a shifted delta function and a scaled
step function. Since the system is LTI, the output is quickly computed using just h(t)
and s(t). That is,
√ √
y(t) = h(t−π)−cos( 3)s(t) = (2e−2(t−π) −e−(t−π) )u(t−π)−cos( 3)(e−t −e−2t )u(t).

2.4-29.
2-39 Since x(t) is (T = 2)-periodic, the convolution y(t) = x(t) ∗ h(t) is also (T = 2)-
periodic. Thus, it is sufficient to evaluate y(t) over any interval of length two.
Rt R 3/2 ¯
2 ¯t
¯
2 ¯3/2
For 0 ≤ t < 1/2, y(t) = 0 τ dτ + t+1 τ dτ = τ2 ¯ + τ2 ¯ = t2 /2 + 9/8 − (t2 /2 +
τ =0 τ =t+1
t + 1/2) = −t + 5/8.
Rt
For 1/2 ≤ t < 1, y(t) = τ dτ = t2 /2.
0
Rt ¯
2 ¯t
For 1 ≤ t < 3/2, y(t) = t−1 τ dτ = τ2 ¯ = t2 /2 − (t2 /2 − t + 1/2) = t − 1/2.
τ =t−1
R 3/2 ¯
2 ¯3/2
For 3/2 ≤ t < 2, y(t) = t−1 τ dτ = τ2 ¯ = 9/8−(t2 /2−t+1/2) = −t2 /2+t+5/8.
τ =t−1
Combining, 

 −t + 5/8 0 ≤ t < 1/2


 t2 /2 1/2 ≤ t < 1
y(t) = t − 1/2 1 ≤ t < 3/2 .



 −t2 /2 + t + 5/8 3/2 ≤ t < 2

y(t + 2) ∀t

MATLAB is used to plot y(t) over (−3 ≤ t ≤ 3). This interval includes three periods
of the (T = 2)-periodic function y(t).

>> t = linspace(-3,3,601); tm = mod(t,2);


>> y = (-tm+5/8).*((tm>=0)&(tm<1/2));
>> y = y+(tm.^2/2).*((tm>=1/2)&(tm<1));
>> y = y+(tm-1/2).*((tm>=1)&(tm<3/2));
>> y = y+(-tm.^2/2+tm+5/8).*((tm>=3/2)&(tm<2));
>> plot(t,y,’k’); xlabel(’t’); ylabel(’y(t)’);

71

49
1

0.9

0.8

0.7

0.6

y(t)
0.5

0.4

0.3

0.2

0.1
−3 −2 −1 0 1 2 3
t

Figure S2.4-29:
S2-39: Plot of y(t) = x(t) ∗ h(t) over (−3 ≤ t ≤ 3).

2.4-30.
2-40 (a) Using KVL, x(t) = i(t)R+vC1 (t)+y(t) = RC2 ẏ(t)+vC1 (t)+y(t). Differentiating
yields ẋ(t) = RC2 ÿ(t) + v̇C1 (t) + ẏ(t) = RC2 ÿ(t) + C11 i(t) + ẏ(t) = RC2 ÿ(t) +
C2
C1 ẏ(t) + ẏ(t). Thus,
µ ¶
1 1 1
ÿ(t) + + ẏ(t) = ẋ(t).
RC1 RC2 RC2

(b) Since R = 1, C1 = 1, and C2 = 2, the differential equation becomes ÿ(t) +


3/2ẏ(t) = 1/2ẋ(t).
The characteristic equation is λ2 + 3/2λ = 0, and the characteristic roots are
λ1 = 0 and λ2 = −3/2. Thus, the form of the zero-input response is y0 (t) =
c1 + c2 e−3t/2 . Using the first IC, y(0) = 1 = c1 + c2 . The initial voltage across
the resistor is vR (0) = −3 which yields iR (0) = −3/R = −3. Also, iR (0) = −3 =
iC2 (0) = C2 ẏ(0) = 2ẏ(0). Thus, ẏ(0) = −3/2 = −3c2 /2. Solving yields c2 = 1
and c1 = 0. Thus,
y0 (t) = e−3t/2 .
The zero-state response is x(t) ∗ h(t), where h(t) = b0 δ(t) + [P (D)ỹ0 (t)]u(t). For
(1)
this second-order system, ỹ0 (t) = c̃1 + c̃2 e−3t/2 , ỹ0 (0) = 0 = c̃1 + c̃2 and ỹ0 (0) =
1 = −3c̃2 /2. Thus, c̃2 = −2/3 and c̃1 = 2/3. Using b0 = 0, and P (D) = 0.5D,
the impulse response is h(t) = 0.5D(ỹ0 (t))u(t) = 0.5D(2/3 − 2/3e−3t/2 )u(t) =
0.5(−2/3(−3/2)e³−3t/2 )u(t) = 0.5e−3t/2 u(t). Using ´ x(t) =³4te
−3t/2
u(t), the
´ zero-
Rt −3τ /2 −3(t−τ )/2 −3t/2 t
R
state response is 0 (4τ e )(0.5e )dτ u(t) = 2e 0
τ dτ u(t) =
¡ −3t/2 2 ¢
2e t /2 u(t). Thus,

x(t) ∗ h(t) = t2 e−3t/2 u(t).

Since the input is driving a natural mode, resonance is expected; thus, the t2
term seems sensible.
For (t ≥ 0), the total response is the sum of the zero-input response and the
zero-state response.
³ ´
y(t) = y0 (t) + x(t) ∗ h(t) = e−3t/2 + t2 e−3t/2 u(t).

72

50
2.4-31.
2-41 Since h(t) is only provided for over (0 ≤ t < 0.5), it is not possible to determine with
certainty whether or not the system is causal or stable. However, when looking at
h(t) the waveform appears to have a DC offset. This apparent DC offset can be very
troubling if h(t) is truly an impulse response function. If a DC offset is present, the
system is neither causal nor stable. Imagine, a non-causal, unstable heart! Something
is probably wrong.
One simple explanation is that a blood-filled heart always has some ventricular pres-
sure. Unless removed, this relaxed-state pressure would likely appear as a DC offset
to any measurements. It would likely be most appropriate to subtract this offset when
trying to measure the impulse response function.
Another problem is that the impulse response function is most appropriate in the
study of linear, time-invariant systems. It is quite unlikely that the heart is either
linear or time-invariant. Even if the impulse response could be reliably measured at a
particular time, it might not provide much useful information.
R∞ R∞
2.4-32.
2-42 (a) x(t) ∗ x(−t) = −∞ x(τ )x(−(t − τ ))dτ = −∞ x(τ )x(τ − t)dτ = rxx (t).
(b) Since rxx (t) is an even function, we only need to compute rxx (t) for either t ≥ 0
or t ≤ 0. In either case, the autocorrelation function is computed by convolving
the original signal with its reflection.
For t < −2, rxx (t) = 0.
R t+2 ¯t+2
For −2 ≤ t < −1, rxx (t) = 0 τ dτ = 0.5τ 2 ¯τ =0 = t2 /2 + 2t + 2.
µ ¯t+1 ¶
R t+1 R1 R t+2 τ3 τ2 ¯
For −1 ≤ t < 0, rxx (t) = 0 τ (τ −t)dτ + t+1 τ dτ + 1 dτ = 3 − t 2 ¯ +
τ =0
¯ 1
τ2 ¯ t+2 3 2 3 2 2

2 ¯ + τ |τ =1 = t +3t 3+3t+1 − t +2t


2
+t
+ 21 − t +2t+1
2 + (t + 2 − 1) = −t3 /6 −
τ =t+1
t2 /2 + t/2 + 4/3.
Combining and using rxx t = rxx −t yields


 t2 /2 + 2t + 2 −2 ≤ t < −1


 −t /6 − t2 /2 + t/2 + 4/3 −1 ≤ t < 0
3

rxx (t) = t3 /6 − t2 /2 − t/2 + 4/3 0≤t<1



 2

 t /2 − 2t + 2 1≤t<2

0 otherwise

MATLAB is used to plot the result.


>> t = linspace(-2.5,2.5,501);
>> rxx = (t.^2/2+2*t+2).*((t>=-2)&(t<-1));
>> rxx = rxx+(-t.^3/6-t.^2/2+t/2+4/3).*((t>=-1)&(t<0));
>> rxx = rxx+(t.^3/6-t.^2/2-t/2+4/3).*((t>=0)&(t<1));
>> rxx = rxx+(t.^2/2-2*t+2).*((t>=1)&(t<2));
>> plot(t,rxx,’k’); xlabel(’t’); ylabel(’r_{xx}(t)’);

x(t)−0
2.4-33.
2-43 (a) KCL at the negative terminal of the op-amp yields R + C ẏ(t) = 0. Thus,

1
ẏ(t) = − x(t).
RC

(b) The zero-state response is y(t) = x(t)∗h(t), where h(t) = b0 δ(t)+[P (D)ỹ0 (t)]u(t).
This is a first order system with λ = 0, thus ỹ0 (t) = c̃1 eλt = c̃1 . Since ỹ0 (0) =
1 1
1 = c̃1 , b0 = 0, and P (D) = − RC , the impulse response is h(t) = − RC u(t).

73

51
1.4

1.2

0.8

rxx(t)
0.6

0.4

0.2

0
−2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5
t

Figure S2.4-32b:
S2-42b: Plot of rxx (t).

Thus, µZ ¶
t
1 t
y(t) = − dτ u(t) = − u(t).
0 RC RC
Notice, |y(t)| ramps toward infinity as time increases. Intuitively, this makes
sense; a DC input to an integrator should output an unbounded ramp function.
2-44 The system response to u(t) is g(t) and the response to step u(t − τ ) is g(t − τ ). The
2.4-34.
input x(t) is made up of step components. The step component at τ has a height 4f
which can be expressed as
4f
4f = 4τ = ẋ(τ )4τ

The step component at n4τ has a height ẋ(n4τ )4τ and it can be expressed as
[ẋ(n4τ )4τ ]u(t − n4τ ). Its response 4y(t) is

4y(t) = [ẋ(n4τ )4τ ]g(t − n4τ )

The total response due to all components is


X∞
y(t) = lim ẋ(n4τ )g(t − n4τ )4τ
4τ →0
n=−∞
Z ∞
= ẋ(τ )g(t − τ ) dτ = ẋ(τ ) ∗ g(τ )
−∞

2-45 Consider the input x(t) = ejωo t . Letting s = jωo in Eq. (2.47), the system response is
2.4-35.
found as
y(t) = H(jωo )ejωo t
Using Eq. (2.40), the system response to input x̂(t) = cos ωo t = Re[ejωo t ] is ŷ(t),
where

ŷ(t) = Re[H(jωo )ejωo t ]


n o
6 H(jωo )]
= Re |H(jωo )|ej[ωo t+
= |H(jωo )| cos[ωo t + 6 H(jωo )]

74

52
(c) For this system, the characteristic root is only affected by C and Rf . Using 10%
resistors, the resistor Rf is generally expected to lie in the range (0.9Rf , 1.1Rf ).
Using 25% capacitors, the capacitor C is generally expected to lie in the range
−1
(.25C, 1.25C). Since λ = CR f
, the characteristic root is expected to lie in the
range (λ/[(0.9)(0.75)], λ/[(1.1)(1.25)]). Thus,

The characteristic root is expected within the interval (1.48λ, 0.73λ).

2-48
2.4-38. Identify the output of the first op-amp as v(t).

(a) KCL at the negative terminal of the first op-amp yields x(t)R1 + C1 v̇(t) = 0 or
1
R1 C 1 x(t) = − v̇(t). KCL at the negative terminal of the second op-amp yields
v(t) y(t) R2
R2 + R3 +C2 ẏ(t) = 0 or v(t) = − R 3
y(t)−R2 C2 ẏ(t). Substituting this expression
R2
for v(t) into the first expression yields R11C1 x(t) = R 3
ẏ(t) + R2 C2 ÿ(t). Thus,

1 1
ÿ(t) + ẏ(t) = x(t).
R3 C2 R1 R2 C1 C2
1
The characteristic equation is λ2 + R3 C2 λ = 0 and the characteristic roots are

1
λ1 = 0 and λ2 = − .
R 3 C2
Substituting C1 = C2 = 10µF, R1 = R2 = 100kΩ, and R3 = 50kΩ yields

ÿ(t) + 2ẏ(t) = x(t), λ1 = 0, and λ2 = −2.

Since one root lies on the ω-axis, the circuit is not BIBO stable. In particular, a
DC input results in an unbounded output.
(b) The zero-input response has form y0 (t) = c1 + c2 e−2t . Each op-amp has an initial
output of one volt. Thus, y0 (0) = 1 = c1 + c2 . KCL at the negative terminal of
the second op-amp yields R12 + R13 + C2 ẏ0 (0) = 0 or ẏ0 (0) = − R21C2 − R31C2 =
−1−2 = −3. Thus, ẏ0 (0) = −3 = −2c2 . Thus, c2 = 3/2 and c1 = 1−3/2 = −1/2
and
y0 (t) = −1/2 + 3/2e−2t .
(c) The zero-state response is y(t) = x(t)∗h(t), where h(t) = b0 δ(t)+[P (D)ỹ0 (t)]u(t).
This is a second order system with λ1 = 0 and λ2 = −2, so ỹ0 (t) = c̃1 + c̃2 e−2t .
(1)
Solving ỹ0 (0) = 0 = c̃1 + c̃2 and ỹ0 (t) = ¡1 = −2c̃2 yields¢ c̃2 = −1/2 and
c̃1 = 1/2. Since b0 = 0 and P (D) = 1, h(t) = 1/2 − e−2t /2 u(t).
³R ´
t −2τ
Next, y(t) = x(t) ∗ h(t) = 0
(1/2 − e /2)dτ u(t) =
³ ¯ ´ ¡ ¢
t
τ /2 + e−2τ /4¯τ =0 u(t) = t/2 + e−2t /4 − 1/4 u(t).
¡ ¢
y(t) = t/2 + e−2t /4 − 1/4 u(t).

As expected, the DC nature of the unit step input results in an unbounded


output.
(d) For this system, λ1 = 0 is not affected by the components and λ2 is only affected
by C2 and R3 . Using 10% resistors, the resistor R3 is generally expected to lie
in the range (0.9R3 , 1.1R3 ). Using 25% capacitors, the capacitor C2 is generally

76

54
expected to lie in the range (.25C2 , 1.25C2 ). Since λ2 = − R31C2 , the characteristic
root is expected to lie in the range (λ2 /[(0.9)(0.75)], λ2 /[(1.1)(1.25)]). Thus,

λ1 is unaffected and λ2 is expected to lie within (−2.9630, −1.4545).

2-49
2.4-39.
2.4-40. (a) Yes, the system is causal since h(t) = 0 for (t < 0).
(b) To compute the zero-state response y1 (t), the convolution of two rectangular
pulses is required: a pulse of amplitude  and width two and a pulse of amplitude
one and a width of one. The convolution involves several regions.
For t < 0, y1 (t) = 0.
Rt
For 0 ≤ t < 1, y1 (t) = 0 dt = t.
Rt
For 1 ≤ t < 2, y1 (t) = t−1 dt = (t − (t − 1)) = .
R2
For 2 ≤ t < 3, y1 (t) = t−1 dt =  (2 − (t − 1)) =  (3 − t).
For t ≥ 0, y1 (t) = 0.
Thus, 

 t 0≤t<1

 1≤t<2
y1 (t) = .

  (3 − t) 2 ≤t<3

0 otherwise
(c) To compute y2 (t), first note that x2 (t) = 2x1 (t − 1) − x1 (t − 2). Using the system
properties of linearity and time-invariance, the output y2 (t) is given by

y2 (t) = 2y1 (t − 1) − y1 (t − 2).

2.5-1.
2-50

λ2 + 7λ + 12 = (λ + 3)(λ + 4)

The natural response is


yn (t) = K1 e−3t + K2 e−4t
P (0) 1
(a) For x(t) = u(t) = e0t u(t), yφ (t) = H(0) = Q(0) = 6
1
y(t) = K1 e−3t + K2 e−4t +
6
ẏ(t) = −3K1 e−3t − 4K2 e−4t

Setting t = 0 and substituting initial conditions, we obtain


¾
0 = K1 + K2 + 61 K1 = − 32
=⇒
1 = −3K1 − 4K2 K2 = 12

and 2 1 1
y(t) = − e−3t + e−4t + t≥0
3 2 6
P (−1) 1
(b) x(t) = e−t u(t), yφ (t) = H(−1) = Q(−1) = 6

1
y(t) = K1 e−3t + K2 e−4t + e−t
6
1
ẏ(t) = −3K1 e −3t
− 4K2 e−4t
− e−t
6

77

55
Setting t = 0, and substituting initial conditions yields
¾
0 = K1 + K2 + 61 K1 = − 21
1 =⇒
1 = −3K1 − 4K2 − 6 K2 = − 23

and 1 −3t 2 −4t 1 −t


y(t) = e − e + e t≥0
2 3 6
(c) x(t) = e−2t u(t), yφ (t) = H(−2) = 0
y(t) = K1 e−3t + K2 e−4t
ẏ(t) = −3K1 e−3t − 4K2 e−4t

Setting t = 0, and substituting initial conditions yields


¾
0 = K1 + K2 K1 = 1
=⇒
1 = −3K1 − 4K2 K2 = −1

and
y(t) = e−3t − e−4t t≥0
2.5-2.
2-51 λ2 + 6λ + 25 = (λ + 3 − j4)(λ + 3 + j4) characteristic roots are −3 ± j4

yn (t) = Ke−3t cos(4t + θ)


3
For x(t) = u(t), yφ (t) = H(0) = 25 so that

3
y(t) = Ke−3t cos(4t + θ) +
25
ẏ(t) = −3Ke−3t cos(4t + θ) − 4Ke−3t cos(4t + θ)

Setting t = 0, and substituting initial conditions yields


¾ ¾
3
0 = K cos θ + 25 K cos θ = −325 K = 0.427
=⇒ =⇒
2 = −3K cos θ − 4K sin θ K sin θ = −41
100 θ = −106.3

and
3
y(t) = 0.427e−3t cos(4t − 106.3◦ ) + t≥0
25
2-52 Characteristic polynomial is λ2 + 4λ + 4 = (λ + 2)2 . The roots are −2 repeated twice.
2.5-3.

yn (t) = (K1 + K2 t)e−2t

(a) For x(t) = e−3t u(t), yφ (t) = H(−3) = −2e−3t


y(t) = (K1 + K2 t)e−2t − 2e−3t
ẏ(t) = −2(K1 + K2 t)e−2t + K2 e−2t + 6e−3t

Setting t = 0, and substituting initial conditions yields


9
¾ 17
4 = K1 − 2 K1 = 4
=⇒ 15
5 = −2K1 + K2 + 6 K2 = 2

and
17 15 −2t
y(t) = ( + t)e − 2e−3t t≥0
4 2

78

56
(b) x(t) = e−t u(t), yφ (t) = H(−1)e−t = 0
y(t) = (K1 + K2 t)e−2t
ẏ(t) = −2(K1 + K2 t)e−2t + K2 e−2t

Setting t = 0, and substituting initial conditions yields


9
¾ 9
4 = K1 K1 = 4
=⇒ 19
5 = −2K1 + K2 K2 = 2

and 9 19
y(t) = ( + t)e−2t t≥0
4 2
2.5-4.
2-53 Because (λ2 + 2λ) = λ(λ + 2), the characteristic roots are 0 and −2.

yn (t) = K1 + K2 e−2t

In this case x(t) = u(t). The input itself is a characteristic mode. Therefore

yφ (t) = βt

But yφ (t) satisfied the system equation

(D2 + 2D)yφ (t) = (D + 1)y(t) = ÿφ (t) + 2ẏφ (t) = ẋ(t) + x(t)

Substituting x(t) = u(t) and yφ (t) = βt, we obtain

1
0 + 2β = 0 + 1 =⇒ β=
2
Therefore yφ (t) = 21 t.
1
y(t) = K1 + K2 e−2t + t
2
1
ẏ(t) = −2K2 e−2t +
2
Setting t = 0, and substituting initial conditions yields
¾
2 = K1 + K2 K1 = 49
1 =⇒
1 = −2K2 + 2 K2 = − 41

and
9 1 −2t 1
y(t) = − e + t t≥0
4 4 2
2-54 The natural response
2.5-5. yn (t) is found in Prob. 2.5-1:
2-50:

yn (t) = K1 e−3t + K2 e−4t

The input x(t) = e−3t is a characteristic mode. Therefore

yφ (t) = βte−3t

Also yφ (t) satisfies the system equation:

(D2 + 7D + 12)yφ (t) = (D + 2)x(t)

79

57
or ÿφ (t) + 7ẏφ (t) + 12yφ (t) = ẋ(t) + 2x(t)
Substituting x(t) = e−3t and yφ (t) = βte−3t in this equation yields

(9βt − 6β)e−3t + 7(−3βt + β)e−3t + 12βte−3t = −3e−3t + 2e−3t

or βe−3t = −e−3t =⇒ β = −1
Therefore
y(t) = K1 e−3t + K2 e−4t − te−3t
ẏ(t) = −3K1 e−3t − 4K2 e−4t + 3te−3t − e−3t

Setting t = 0, and substituting initial conditions yields


¾
0 = K1 + K2 K1 = 2
=⇒
1 = −3K1 − 4K2 − 1 K2 = −2

and
y(t) = 2e−3t − 2e−4t − te−3t t≥0
−3t −4t
= (2 − t)e − 2e t≥0

2-55 (a)
2.6-1. λ2 + 8λ + 12 = (λ + 2)(λ + 6)
Both roots are in LHP. The system is BIBO stable and also asymptotically stable.
(b) λ(λ2 + 3λ + 2) = λ(λ + 1)(λ + 2)
Roots are 0, −1, −2. One root on imaginary axis and none in RHP. The system
is BIBO ustable and marginally stable.
√ √
(c) λ2 (λ2 + 2) = λ2 (λ + j 2)(λ − j 2)√
Roots are 0 (repeated twice) and ±j 2. Multiple roots on imaginary axis. The
system is BIBO unstable and asymptotically unstable.
(d) (λ + 1)(λ2 − 6λ + 5) = (λ + 1)(λ − 1)(λ − 5)
Roots are −1, 1 and 5. Two roots in RHP. The system is BIBO unstable and
asymptotically unstable.
2.6-2.
2-56 (a) (λ + 1)(λ2 + 2λ + 5)2 = (λ + 1)(λ + 1 − j2)2 (λ + 1 + j2)2
Roots −1, −1 ± j2 (repeated twice) are all in LHP. The system is BIBO stable
and asymptotically stable.
(b) (λ + 1)(λ2 + 9) = (λ + 1)(λ + j3)(λ − j3)
Roots are −1, ±j3. Two (simple) roots on imaginary axis, none in RHP. The
system is BIBO unstable and marginally stable.
(c) (λ + 1)(λ2 + 9)2 = (λ + 1)(λ + j3)2 (λ − j3)2
Roots are −1 and ±j3 repeated twice. Multiple roots on imaginary axis. The
system is BIBO unstable and asymptotically unstable.
(d) (λ2 + 1)(λ2 + 4)(λ2 + 9) = (λ + j1)(λ − j1)(λ + j2)(λ − j2)(λ + j3)(λ − j3)
The roots are ±j1, ±j2 and ±j3. All roots are simple and on imaginary axis.
None in RHP. The system is BIBO unstable and marginally stable.
2-57 (a) Because u(t) = e0t u(t), the characteristic root is 0.
2.6-3.
(b) The root lies on the imaginary axis, and the system is marginally stable.
R∞
(c) 0
h(t) dt = ∞
The system is BIBO unstable.

80

58
(d) The integral of δ(t) is u(t). The system response to δ(t) is u(t). Clearly, the
system is an ideal integrator.
2-58
2.6-4. Assume that a system exists that violates Eq. (2.64) and yet produces a bounded
output for every bounded input. The response at t = t1 is
Z ∞
y(t1 ) = h(τ )x(t1 − τ ) dτ
0

Consider a bounded input x(t) such that at some instant t1


½
1 if h(τ ) > 0
x(t1 − τ ) =
−1 if h(τ ) < 0

In this case
h(τ )x(t1 − τ ) = |h(τ )|
and Z ∞
y(t1 ) = |h(τ )| dτ = ∞
0
This violates the assumption.
2.6-5.
2-59 (a) For this convolution, there are several regions. For (t < −2) and (t ≥ 4), y(t) = 0.
R t+2
For (−2 ≤ t < 0), y(t) = 0 τ dτ = (t + 2)2 /2 = t2 /2 + 2t + 2.
R2
For (−2 ≤ t < 0), y(t) = 0 τ dτ = 2.
R2
For (−2 ≤ t < 0), y(t) = t−2 τ dτ = 22 /2 − (t − 2)2 /2 = 2 − t2 /2 − 2t − 2 =
−t2 /2 + 2t.
Combining yields
 2

 t /2 + 2t + 2 −2 ≤ t < 0

2 0≤t<2
y(t) = 2 .

 −t /2 + 2t 2 ≤t<4

0 otherwise

MATLAB is used to plot the result.


>>t = linspace(-3,5,401);
>>y = (t.^2/2+2*t+2).*((t>=-2)&(t<0));
>>y = y+2.*((t>=0)&(t<2));
>>y = y+(-t.^2/2+2*t).*((t>=2)&(t<4));
>>plot(t,y,’k’); axis([-3 5 -.5 2.5]);
>>xlabel(’t’); ylabel(’y(t)’);
R
(b) Yes, the system is stable since h(t) = 4 < ∞.
No, the system is not causal since h(t) 6= 0 for all t < 0.
2-60 R∞
2.6-6. Yes, the system is stable since h(t) is absolutely integrable. That is, −∞ h(t)dt =
R1
0
1dt = 1 < ∞.
Yes, the system is causal since h(t) = 0 for t < 0.s
2-61
2.6-7. Expanding

X
h(t) = (0.5)i δ(t − i)
i=0

81

59
2.5

1.5

y(t)
1

0.5

−0.5
−3 −2 −1 0 1 2 3 4 5
t

Figure S2.6-5a:
S2-59a: Plot of y(t) = x(t) ∗ h(t).

yields
h(t) = (δ(t) + 0.5δ(t − 1) + 0.25δ(t − 2) + 0.125δ(t − 3) + · · ·)
(a) Yes, the system is causal since h(t) = 0 for t < 0.
(b) Yes,R theP
system is stable since the
Pimpulse response
R∞ is absolutely
P∞ integrable.1−0
That
∞ ∞ ∞
is, −∞ i=0 (0.5)i δ(t − i)dt = i=0 (0.5)i −∞ δ(t − i) = i=0 (0.5)i = 1−0.5 =
2 < ∞.
2.7-1. (a) The time-constant (rise-time) of the system is Th = 10−5 . The rate of pulse
2-62
communication < T1h = 105 pulses/sec. The channel cannot transmit million
pulses/second.
(b) The bandwidth of the channel is
1
B= = 105 Hz
Th
The channel can transmit audio signal of bandwidth 15 kHz readily.
2-63
2.7-2.
1 1
Th = = 4 = 10−4 = 0.1 ms
B 10
The received pulse width = (0.5 + 0.1) = 0.6 ms. Each pulse takes up 0.6 ms interval.
The maximum pulse rate (to avoid interference between successive pulses) is
1
' 1667 pulses/sec
0.6 × 10−3

2-64
2.7-3. Using Eqs. (2.67) and (2.68)
(a)
1
Tr = Th = − = 10−4
λ

(b) The bandwidth fc = 1/Th = 1/Tr = 104 .


(c) The pulse transmission rate is fc = 104 pulses/sec.

82

60
2.7-4. (a) For a causal system with finite duration h(t), the rise time is exactly equal to the
2-65
time when the signal is last non-zero. That is,

Tr = 4 seconds.

(b) The impulse response function h(t) is consistent with a channel that has the
following three characteristics: 1) a channel with delay from input to output (for
example, signal propagation delay), 2) a channel with low-pass character (pulse
dispersion that results in a δ(t) input spreading into a square pulse), and 3) a
channel with two signal paths (for example, a primary signal path and an echo
path).
For systems with predominantly low-pass character, digital information can be
transmitted without significant interference at a rate of Fc = T1r = 1/4. However,
this estimate is too conservative for the present system. Notice that h(t) = 0 for
0 ≤ t < 1, corresponding to a transmission delay in the primary signal path.
The remaining portion of h(t) has a width of three, so it is therefore practical
to transmit at rates of Fc = 1/3. By clever interleaving of data, it is possible
to transmit at rates of Fc = 1/2. Consider transmitting the binary sequence
{b0 , b1 , b2 , b3 , . . .} using a (t = 1)-spaced delta train weighted by the pulse se-
quence {b0 , b1 , 0, 0, b2 , b3 , 0, 0, . . .}. The output is the series of non-overlapping
unit-duration pulses given by {b0 , b1 , b0 , b1 , b2 , b3 , b2 , b3 , . . .}. The effective trans-
mission rate is 0.5 bits per unit time.
(c) The resulting convolution y(t) = x(t) ∗ h(t) has many regions.
For t < 1, y(t) = 0.
Rt
For 1 ≤ t < 2, y(t) = 1 (−1)dt = 1 − t.
R2
For 2 ≤ t < 3, y(t) = 1 (−1)dt = −1.
R2 Rt
For 3 ≤ t < 4, y(t) = t−2 (−1)dt + 3 (−1)dt = (t − 2) − 2 + 3 − t = −1.
R4
For 4 ≤ t < 5, y(t) = 3 (−1)dt = −1.
R4
For 5 ≤ t < 6, y(t) = t−2 (−1)dt = (t − 2) − 4 = t − 6.
For 6 ≤ t, y(t) = 0.
Thus, 

 1−t 1≤t<2

−1 2 ≤ t < 5
y(t) = .

 t −6 5≤t<6

0 otherwise
MATLAB is used to plot the result.
>> t = [0:.01:10]; y = zeros(size(t));
>> y = y + (1-t).*((t>=1)&(t<2));
>> y = y + (-1).*((t>=2)&(t<5));
>> y = y + (t-6).*((t>=5)&(t<6));
>> plot(t,y,’k’); axis([0 10 -1.2 .2]);
>> xlabel(’t’); ylabel(’y(t)’);

83

61
0.2

−0.2

−0.4
y(t)

−0.6

−0.8

−1

0 1 2 3 4 5 6 7 8 9 10
t

Figure S2.7-4c:
S2-65c: Plot of y(t) = x(t) ∗ h(t).

84

62

You might also like