EE320A

Tutorial 8
Date: 29th Sept. 2023

C Ci (t)
L

Figure 1: Equivalent circuit of the frequency determining network of a VCO.

1. The equivalent circuit of the frequency determining network of a VCO

is shown in Figure 1. Frequency modulation is produced by applying
the modulating signal Vm sin(2πfm t) plus a bias Vb to a varactor diode
connected across the parallel combination of a 200-µH inductor (L) and a
100-pF capacitor (C). The capacitance in the varactor diode is related to
the voltage V (t) applied across its terminals by:
p
Ci (t) = 100/ V (t) pF.
The instantaneous frequency of oscillation is given by:
1 1
fi (t) = p .
2π L(C + Ci (t))
The unmodulated frequency of oscillation is 1 MHz. The VCO output is
applied to a frequency multiplier to produce an FM signal with carrier
frequency of 64 MHz and a modulation index of 5.
(a) Determine the magnitude of the bias voltage Vb .
(b) Find the amplitude Vm of the modulating wave, given that fm = 10
kHz.
Assume that Vb ≫ Vm .
2. Let
s(t) = a(t) cos(2πfc t + φ(t))
where a(t) > 0, be applied to a hard limiter whose output z(t) is defined
by:
z(t) = sgn[s(t)]

+1 for s(t) > 0
=
−1 for s(t) < 0.
K Vasudevan Faculty of EE IIT Kanpur ([email protected]) 2

(a) Show that z(t) can be expressed in the form of a Fourier series as
follows:

4 X (−1)n
∞
z(t) = cos[2πfc t(2n + 1) + (2n + 1)φ(t)].
π n=0 2n + 1

(b) Compute the output when z(t) is applied to an ideal bandpass filter
with center frequency fc and bandwidth BT , where BT is the trans-
mission bandwidth of s(t) in the absence of amplitude modulation.
Assume that fc ≫ BT .
3. Show that the magnitude of the correlation coefficient |ρ| is always less
than or equal to unity.

Y = g(X)
5

−2 −1 0 1 2

Figure 2: The transformation Y = g(X).

4. Let a random variable X have a uniform pdf over −2 ≤ X ≤ 2. Compute

the pdf of Y = g(X) as shown in Figure 2, where

2X 2

for |X| ≤ 1
g(X) =
3|X| − 1 for 1 ≤ |X| ≤ 2.

2
5. A Gaussian distributed random variable X of zero-mean and variance σX
is transformed by a rectifier with input-output relation given below:

X if X ≥ 0
Y =
0 if X < 0

Compute the pdf of Y .

 EE320A
Solutions for Tutorial 8
Date: 29th Sept. 2023

1. Solution: The instantaneous frequency of oscillation is given by:

1 1
fi (t) = . (1)
2π L(C + Ci (t))
p

The unmodulated frequency of oscillation is given as 1 MHz. Thus

1 1
fc =
2π L(C + C0 )
p

⇒ C + C0 = 126.65 pF
⇒ C0 = 26.651 pF
p
⇒ 100/ Vb = 26.651
⇒ Vb = 14.078 volts. (2)

Since the final carrier frequency is 64 MHz, the frequency multiplication

factor is 64. Thus, the modulation index at the VCO output is
5
β= = 0.078. (3)
64
Thus, the FM signal at the VCO output is narrowband. The instantaneous
frequency of oscillation at the VCO output is:
1 1
fi (t) = . (4)
2π L(C + 100(Vb + Vm sin(2πfm t))−0.5 )
p

Since Vb ≫ Vm the instantaneous frequency can be approximated as:

1 1
fi (t) ≈ . (5)
2π L(C + 100V −0.5 (1 − (V /2V ) sin(2πf t)))
q
b m b m

Hence the instantaneous frequency becomes:

1 1
fi (t) =
2π L(C + 100/Vb − (50Vm /Vb1.5 ) sin(2πfm t)))
p
0.5

1 1
= p
2π L(C + 100/Vb0.5
1
p
1 − (50Vm /Vb )(1/(100 + CVb0.5 )) sin(2πfm t)))
 
25Vm
≈ fc 1 + sin(2πfm t) . (6)
Vb (100 + CVb0.5 )
K Vasudevan Faculty of EE IIT Kanpur ([email protected]) 4

The modulation index of the narrowband FM signal is given by:

25Vm fc
β= = 0.078. (7)
Vb fm (100 + CVb0.5 )

Substituting fc = 1 MHz, fm = 10 kHz and Vb = 14.078 volts, we get:

Vm = 0.2087 volts. (8)

2. Solution: Since a(t) > 0, z(t) can be written as:

z(t) = sgn[cos(2πfc t + φ(t))]. (9)

Let

α(t) = 2πfc t + φ(t). (10)

Then z(t) can be rewritten as:

z(α(t)) = sgn[cos(α(t))]. (11)

Note that z(α(t)) is periodic with respect to α(t), that is:

cos(α(t)) z(α(t))
0.5

-0.5

-1
-6 -4 -2 0 2 4 6
α(t)

Figure 3: z(α(t)) is periodic with respect to α(t).

z(α(t) + 2nπ) = z(α(t)). (12)

K Vasudevan Faculty of EE IIT Kanpur ([email protected]) 5

This is illustrated in Figure 3. Hence z(t) can be written in the form of a

Fourier series with respect to α(t) as follows:
∞
X
z(α(t)) = 2 an cos(nα(t)) (13)
n=1

where
Z π
1
an = z(α(t)) cos(nα(t)) dα(t)
2π α(t)=−π
1 π
Z
= z(α(t)) cos(nα(t)) dα(t). (14)
π α(t)=0

For convenience denote α(t) = x. Thus

1 π
Z
an = z(x) cos(nx) dx
π x=0
1 π/2 1 π
Z Z
= cos(nx) dx − cos(nx) dx
π x=0 π x=π/2
(
0 for n = 2m
= 2(−1)m (15)
π(2m+1) for n = 2m + 1.

Thus
∞
X 4(−1)m
z(x) = cos((2m + 1)x)
m=0
π(2m + 1)
4(−1)m
∞
X
⇒ z(α(t)) = cos((2m + 1)α(t))
m=0
π(2m + 1)
∞
X 4(−1)m
= cos((2m + 1)2πfc t + (2m + 1)φ(t)).
m=0
π(2m + 1)
(16)
Observe that the mth harmonic has a carrier frequency at (2m + 1)fc and
bandwidth (2m+1)BT , where BT is the bandwidth of s(t) with amplitude
modulation removed, that is, a(t) = Ac . If the fundamental component
at m = 0 is to be extracted from z(α(t)) we require
BT BT
fc + < (2m + 1)fc − (2m + 1)
 2 2
BT 1
⇒ 1+ < fc for m > 0. (17)
2 m
The left-hand-side of the second equation in (17) is maximum for m = 1.
Thus in the worst case we require
fc > BT . (18)
K Vasudevan Faculty of EE IIT Kanpur ([email protected]) 6

Now if z(α(t)) is passed through an ideal bandpass filter with center fre-
quency fc and bandwidth BT , the output is (assuming (18) is satisfied):
4
y(t) = cos(2πfc t + φ(t)). (19)
π
which has no amplitude modulation.
3. Solution: For any real number a, we have:
h i
2
E (a(X − mX ) − (Y − mY )) ≥ 0
⇒ a2 σX
2
− 2aE [(X − mX )(Y − mY )] + σY2 ≥ 0. (20)
Since the above quadratic equation in a is nonnegative, its discriminant is
nonpositive. Hence
4E 2 [(X − mX )(Y − mY )] − 4σX
2 2
σY ≤ 0
|E [(X − mX )(Y − mY )]|
⇒ ≤ 1. (21)
σX σY
Hence proved.

Y = g(X)
5

−2 −1 0 1 2

Figure 4: The transformation Y = g(X).

4. Solution: We first note that the mapping g(x) is many-to-one. This is

illustrated in Figure 4. Let y and x denote particular values taken by the
RVs Y and X respectively. Therefore we have
[fX (x) + fX (−x)] |dx| = fY (y) |dy|. (22)
Note that

1/4 for −2 < x < 2
fX (x) = (23)
0 elsewhere.
Substituting (23) in (22) we have:
1 dx
= fY (y). (24)
2 dy
Next we observe that:
 √
dx 1/(2 2y) for 0 < y < 2
= (25)
dy 1/3 for 2 < y < 5
K Vasudevan Faculty of EE IIT Kanpur ([email protected]) 7

Substituting (25) in (24) we get:

 √
1/(4 2y) for 0 < y < 2
fY (y) = (26)
1/6 for 2 < y < 5.

5. Solution: We start with the basic equation when a random variable X is

transformed to a random variable Y through the relation:

Y = g(X). (27)

We note that the mapping between Y and X is one-to-one and an inverse

mapping exists:

X = g −1 (Y ). (28)

Thus the following relation holds:

fX (x)
fY (y) = . (29)
|dy/dx| X=g−1 (Y )

For the given problem


dy 1 for X > 0
= (30)
dx 0 for X < 0.

Since X < 0 maps to Y = 0, we must have

P (X < 0) = P (Y = 0). (31)

Thus the pdf of Y must have a delta function at y = 0. Let

 
y2
(
kδ(y) + √ 1 2 exp − 2σ 2 for Y ≥ 0
fY (y) = 2πσX X (32)
0 for Y < 0

where k is a constant such that the pdf of Y integrates to unity. It is easy

to see that
Z ∞
fY (y) dy = k + 1/2 = 1
y=−∞
⇒k = 1/2 (33)

where we have used the fact that

Z ∞
δ(y) dy = 1. (34)
y=−∞