4 Exponential CW Modulation

4.1 Introduction
In the previous chapter we discussed linear CW modulation process in which the message signal
modulated the amplitude of the carrier. The spectrum of the modulated signal was that of the
message spectrum but translated on to a higher frequency of the carrier. Angle modulation on the
other hand is a method where the message signal modulates the phase angle of the carrier. The two
basic types of exponential modulation are phase modulation (PM) and frequency modulation (FM).
Since the angular part of the carrier is varied in accordance with the baseband signal, the modulation
is basically a nonlinear process. The spectrum of the modulated wave is not related in a simple way
to the spectrum of the message signal. Also, the transmission bandwidth of exponential modulated
signal is usually much larger than that of the amplitude modulated signal. In spite of these, angle
modulation finds a variety of applications in commercial FM radio broadcasting, audio section of
TV transmission, mobile radio communication, microwave and satellite communications, and so on.
The advantages of angle modulation over amplitude modulation are noise reduction, improved system
fidelity and efficient use of power. However, angle modulation requires wider bandwidth and utilises
complex circuits in the transmitter and receiver sections.

4.2 Angle Modulation: Phase and Frequency

We start with a CW signal of constant amplitude and time-varying phase of the form

sc (t) = Ac cos[ωc t + θ(t)] = Ac cos φc(t), (4.1)

where Ac = peak carrier amplitude (volts)

ωc = angular frequency of carrier in rad/s,

θ(t) = instantaneous phase angle.

The quantity φ c (t) = ωc t+θ(t) is the total instantaneous phase angle in which carrier reference phase
is ωc t and instantaneous phase deviation is θ(t). If the message information is embedded in φc (t),
we have indeed accomplished angle or exponential modulation. The term exponential modulation
stems from the fact that the message signal s(t) and the modulated signal sc (t) are related through
the exponential form as
2 Principles of Electronic Communications { Analog and Digital

sc (t) = Ac cos φc (t) = Ac Re[ejφc(t) ]. (4.2)

The instantaneous frequency f(t) of sc (t) is defined as the rate of change of the instantaneous phase.
Thus, instantaneous frequency is expressed mathematically as
 
1 d 1 d
f(t) = φc(t) = {ωc t + θ(t)}
2π dt 2π dt
1 dθ
= fc + (t) Hz. (4.3)
2π dt
The second term is also referred to as instantaneous frequency deviation, whereas the angle θ(t) is
called the instantaneous phase deviation.
dθ
The term is called instantaneous frequency deviation.
dt
Note that the instantaneous frequency f(t) is a time-varying property of sc (t) if there is exponential
modulation. In the absence of modulation dθ/dt = 0 and hence the instantaneous frequency is the
same as the carrier frequency fc .
In the case of frequency modulation (FM), we modulate the instantaneous frequency with the
message signal s(t). That is,

dθ
f(t) = fc + kf s(t), or equivalently, = 2πkf s(t) (4.4)
dt
where kf is the constant of proportionality which gives the deviation sensitivity. The quantity kf has
the unit of Hz/volt or V−1 s−1 . Note that f(t) is varied around a positive value fc rather than around
zero. This makes it feasible for separation of various FM broadcast stations in the frequency domain.
The restriction on kf is that during the negative peak of s(t) the instantaneous frequency should
never become negative. We normally take the maximum frequency deviation to be much smaller than
fc so that the signal sc (t) can be treated as a bandpass signal.
Since the phase is the integration of instantaneous frequency, the phase is of the form
Z t  Z t 
φ(t) = 2π f(τ )dτ = 2π fc t + kf s(τ )dτ + φ(t0 ). (4.5)
t0 0

If t0 is taken such that φ(t0 ) = 0, then

Rt
φ(t) = 2π[fc t + kf 0
s(τ )dτ ]. (4.6)

The FM waveform is then

Rt
sFM (t) = Ac cos[ωct + 2πkf 0
s(τ )dτ ]. (4.7)

Phase modulation (PM) is another form of angle modulation in which the phase angle θ(t) is varied
linearly with the message signal s(t). That is,
 Exponential CW Modulation 3

θ(t) = kp s(t), (4.8)

where the proportionality constant kp defines the phase sensitivity of the modulator and has units of
radian/volt. The PM signal is thus described by

sPM (t) = Ac cos[ωct + kps(t)]. (4.9)

A comparison of (4.7) and (4.9) reveals that there is no basic difference between FM and PM except
that the message is direct in PM whereas it appears as integration in FM. Thus direct FM is indirect
PM and vice versa. One can relate the two types by finding the instantaneous frequency. In (4.3),
the instantaneous frequency is obtained as
1 d
f(t) = fc + kp s(t), (4.10)
2π dt
and that for the FM (stated in ( 4.4)) is

f(t) = fc + kf s(t), or ω(t) = 2πf(t) = ωct + 2πkf s(t) (4.11)

The FM signal is thus described by

sPM (t) = Ac cos[ωct + 2πkf s(t)]. (4.11a)

The above relations clearly indicate that an FM wave can be generated by first integrating
the message signal s(t) and then using the result as input to the phase modulator. Alternately, a
differentiator followed by frequency modulation results in PM wave as shown in Fig. 4.1.

Fig. 4.1 Illustrating the relations between PM and FM: (a) FM generation by phase modulator, (b) PM
generation by frequency modulator.

In the case of tone modulation (modulation signal is a single frequency sinusoidal) it is hardly possible
to distinguish between PM and FM waves. Taking s(t) = Am cos ωm t, the expressions for PM and
FM waves are given by

sPM (t) = Ac cos[ωct + kpAm cos ωm t] (4.12a)

and
2πkf Am kf Am
sFM (t) = Ac cos[ωct + sin ωm t] = Ac cos[ωct + sin ωm t]. (4.12b)
ωm fm
4 Principles of Electronic Communications { Analog and Digital

The arguments of the cosine terms in the above expressions consist of ωc t plus the sinusoidal com-
ponents. It is impossible to distinguish an FM waveform from a PM waveform without knowing the
modulating signal. On a cathode ray oscilloscope, the two waveforms look identical. A closer look
reveals that when sin ωm t in FM attains a maximum, cos ωm t in PM is zero; the maximum phase
deviation in PM occurs at a quarter of a wave early or later than the FM wave. This is illustrated in
Fig. 4.2 which shows the PM and FM waves for a sinusoidal message signal. Thus, unless one knows
the modulating signal, it is impossible to distinguish an FM from a PM waveform. The distinction
between exponential and linear modulations is now straightforward: the zero-crossing points of an
exponential modulated wave are non-periodic while these takes place at regular intervals for linear
modulation. Another difference is that the envelope of linear modulation depends on the message
signal whereas for an FM or a PM wave it is a constant.

Fig. 4.2 (a) Unmodulated sinusoidal carrier (b) modulating signal (c) FM wave (d) PM wave.

Example 4.1
Find the instantaneous frequency of the function s(t) = 5 cos[2π × 104 t + sin 2π × 100t].
 Exponential CW Modulation 5

Solution
The total instantaneous phase φ(t) is given by

φ(t) = 2π × 104 t + sin(2π × 100t).

1 dφ
Thus, f(t) = = 104 + 100 cos(2π × 100t). This is sketched in Fig. 4.3.
2π dt
In summary, in the expression sc (t) = Ac cos [ωct + θ(t)],

Instantaneous phase = ωc t + θ(t) radian.

Instantaneous phase deviation = θ(t) radian.

d
Instantaneous frequency, ω(t) = (instantaneous phase)
dt
d dθ
= (ωc t + θ(t)) = ωc + rad/sec.
dt dt
Instantaneous frequency deviation,
dθ
= rad/sec
dt
dθ/dt rad/ sec 1 dθ cycle
= = (Hz)
2π rad/cycle 2π dt sec
Deviation Sensitivity

If one can take the phase modulated signal as given by Eq. (4.9), the instantaneous phase deviation
is given by θ(t) = kp s(t). Thus, sensitivity for phase deviation kp has the unit of rad/V.
The frequency modulated signal as given by Eq. (4.11a), the instantaneous frequency deviation
dθ d ds
is given by = (2πkf s(t)) = 2πkf
dt dt dt
Thus, the sensitivity for frequency deviation kf has the unit
 
[dθ/dt] dθ/dt 1
[kf ] =   = ·   = cycle/Volt
ds 2π ds
[2π]
dt dt

4.3 Frequency Modulation

Since the FM signal defined in (4.7) is a nonlinear function of the message signal s(t), the spectrum
of sFM (t) is not related in a simple manner to that of s(t). If s(t) were known, the Fourier transform
of the FM waveform could be evaluated by using the integral property of the Fourier transform. In
finding the range of frequencies occupied by the FM signal, we perform the analysis in the time
domain for the simplest case of single-tone modulation.
Consider a message signal defined by s(t) = Am cos 2πfm t. The instantaneous frequency of the
6 Principles of Electronic Communications { Analog and Digital

FM wave is given by

f(t) = fc + kf Am cos 2πfm t

= fc + ∆f cos 2πfm t, (4.13)

where ∆f = kf Am is the maximum frequency deviation that occurs in the carrier frequency. The
total phase angle φ(t) is then obtained as (vide (4.6))
Z t Z t
φ(t) = 2π f(τ )dτ = 2π [fc + kf s(τ )]dτ
0 0

Z t
= 2πfc t + 2πkf Am cos(2πfm t)dτ
0

∆f
= 2πfc t + sin2πfm t. (4.14)
fm

Equation (4.14) can also be derived as follows.

dθ
Keeping parity with the phase modulation as in Eq. (4.8), we can write = Kf s(t) where
Z dt
rad/sec
the sensitivity for frequency modulation has the unit . Then θ(t) = kf s(t)dt and hence,
Volt
t
kf Am
Z
φ(t) = ωc t + kf s(t)dt = ωc t + sin 2πfm t
0 2πfm

∆f
= ωc t + sin 2πfm t (4.14a)
fm

Fig. 4.3 Instantaneous frequency

We define the modulation index β as

 Exponential CW Modulation 7

∆f kf Am
β= = . (4.15)
fm fm

The quantity β is a dimensionless quantity since k f has the units of volt−1 second −1
. The FM
signal is then given by

sFM (t) = Ac cos[2πfct + β sin 2πfm t]. (4.16)

Although the instantaneous frequency f(t) of (4.13) lies in the range fc ± ∆f, the spectrum of
FM signals does not lie in this range. Note that the modulation index is directly proportional to
the amplitude of the modulating signal and inversely proportional to its frequency. The index of
kpAm
modulation for PM signal is kpAm whereas that of FM is . Thus the modulation index of PM
fm
and FM are proportional to the amplitude of the modulation signal. Also index of modulation in PM
is independent of fm whereas it is inverse to fm in FM.

Example 4.2
Sketch the FM waves for the modulating signal s(t) as shown in Fig. 4.4 given that kf = 105 rad/V.s
and fc = 10 MHz.

Fig. 4.4 Angle modulated wave: (a) modulating wave, (b) FM wave. Solution
rad/sec
As kf is in , we must have
V
dθ
= kf s(t) and hence
dt

kf s(t)
f(t) = fc +
2π
8 Principles of Electronic Communications { Analog and Digital

kf
= fc +
2π
kf
For FM, f(t) = fc ± since s(t) is either +1 Volt or −1 Volt).
2π
0.1
= (10 ± ) MHz = 9.984 MHz or 10.016 MHz.
2π

Example 4.3
If the waveform cos(ωc t + k sin ωm t) is a frequency modulated one, find the modulating signal. Also
find the modulating signal if the carrier is phase modulated.

Solution
Comparing the given FM wave with that of (4.7), we see that
Rt
2πkf 0
s(τ )dτ = k sin 2πfm t

or, 2πkf s(t) = 2πfm k cos 2πfm t.

So, the modulating signal s(t) is given by

s(t) = k cos 2πfm t.

Similarly, a comparison with the PM signal of (4.9) gives

s(t) = sin 2πfm t, for kp = k.

4.4 Bandwidth of FM Signal: Frequency Analysis

We will now find the range of frequencies of the FM signal. Taking the FM waveform to be

sFM (t) = Ac cos(2πfc t + β sin 2πfm t),

we will now expand this to find the individual frequency component. We convert sFM (t) to its
exponential form

sFM (t) = Re [Ac exp(j2πfc t) exp(jβ sin 2πfm t). (4.17)

The term containing the message signal is

exp(jβ sin 2πfm t) = cos(β sin 2πfm t) + j sin(β sin 2πfm t), (4.18)

the real part of which is cos(β sin 2πfm t). This is an even and periodic function with period 1/fm .
We can expand the term in an exponential Fourier series as
 Exponential CW Modulation 9

∞
ejβ sin 2πfm t = Cn ejn2πfm t ,
P
(4.19)
n=−∞

where the series coefficients are given by the integral

1/2fm
1
Z
Cn = exp(jβ sin 2πfm t)e−jn2πfm t dt. (4.20)
fm −1/2fm

The above integral cannot be evaluated in a closed form. The coefficients Cn are, of course, real
and function of n and β. It is not a function of the modulating frequency fm . The function Cn 's
are known as Bessel functions of the first kind and of order n, denoted by Jn (β).

Now, (4.17) takes the form

∞
X
j2πfc t
sFM (t) = Re [Ace Jn (β)ejn2πfm t ]
n=−∞

" ∞
#
X
j2π(fc +nfm )t
= Re Ac Jn (β)e .
n=−∞

Taking the real part,

∞
X
sFM (t) = Ac Jn (β) cos 2π(fc + nfm )t. (4.21)
n=−∞

Since for integer values of n, J−n (x) = (−1)n Jn (x), (4.21) reduces to

sFM (t) = Ac [J0 (β) cos ωc t + J1 (β){cos 2π(fc + fm )t − cos 2π(fc − fm )t}

+J2 (β){cos 2π(fc + 2fm )t + cos 2π(fc − 2fm )t}

J3 (β){cos 2π(fc + 3fm )t − cos 2π(fc − 3fm )t} + · · · ]. (4.22)

Note that the spectrum is composed of a carrier weighted by J0 (β) and infinite number of sidebands
situated symmetrically on both sides of the carrier with the frequency separations fm , 2fm , 3fm ,
· · · , etc. The magnitude spectrum is illustrated in Fig. 4.5 for positive frequencies. The amplitude
of the nth sideband is proportional to Jn (β).

The total power in sP M (t) can be calculated as follows.

Power in the carrier is A2c J02 (β)/2, power in the first sideband component is 2A2c J12 (β)/2 = A2c J12 (β)
and so on. The total power is thus

A2c 2
P = [J (β) + 2J12 (β) + 2J22 (β) + · · · ] (4.22a)
2 0
10 Principles of Electronic Communications { Analog and Digital

Fig. 4.5 Line spectra of FM of equation (4.22) for sinusoidal modulation

It appears from (4.22) that the bandwidth of FM signal is infinite. However, in practice, we
consider only a reasonable number of sidebands that contain significant part of the energy. The
quantum of energy outside this band is negligible for most applications and the resulting distortion
is tolerable. Before we proceed further, we need to discuss the behaviour of the Bessel functions of
order n with argument β.
Figure 4.6 shows Bessel functions Jn (β) plotted as a function of β for values of n = 0, 1 and 2.

Fig. 4.6 Bessel functions plotted as a function of β

We see that J0 (β) approaches unity and J1 (β), J2 (β) tend to zero for small β. At β = 0, J0 (0) = 1
and all other Jn 's are zero. This indicates that if there is no modulation, only the carrier term is
 Exponential CW Modulation 11

present, as expected. As β increases slightly, J0 (β) decreases while both J1 (β) and J2 (β) increase.
But the value of J0 (β) continues to be significant in comparison with J1 (β) and J2 (β) which are
still neglected. In fact, for a given n and β << 1, the Bessel function can be approximated by
 n
∼ 1 β
Jn (β) = n > 0. (4.23)
n! 2
In particular, for n = 0,
 2
β
J0 (β) ' 1 − . (4.24)
2
By way of an example, let us take a small value of β equal to 0.2, say. Then, J0 = 0.99, J1 =
0.1, J2 = 0.005, and higher-order Bessel functions are still smaller. The first carrier null occurs
when β = 2.4 and J0 (β) = 0. It is then reasonable to consider J0 and J1 terms that include the
carrier and a single pair of first sidebands located at fc ± fm , giving a total bandwidth of 2fm . As
β becomes larger, J2 becomes significant and must be considered. In general, with the increase of
the index β, higher-order terms must be considered. The selection rule can be stated as follows:
A sideband pair will not be considered unless it has an amplitude equal to or larger than one percent
of the unmodulated carrier amplitude.
This rule gives the approximate bandwidth of the FM as a function of fm and β. Since the power
of a periodic signal depends on the square of the amplitude, this rule can also be stated in terms of
power considerations. In approximating the bandwidth, we will consider those number of sidebands
that contain 98% or more of the total transmitted energy. This will, of course, produce some amount
of distortion in the received signal but tolerable for most purposes. In fact, for any value of β, only
those Jn 's need be considered which have significant nonzero values. Figure 4.7 shows the plot of
Jn (5) as a function of n. We can observe that for n larger than 5, the Bessel function asymptotically
approaches zero. The power contained in the terms with n = 0 through 6 is

1 2
P = J (5) + J12 (5) + J22 (5) + J32 (5) + J42 (5) + J52 (5) + J62 (5)
2 0
1
= (−0.1776)2 + (−0.3276)2 + (0.0465)2 + (0.3648)2 + (0.3912)2
2
+(0.2611)2 + (0.1310)2 = 0.496.

(The data is available in the Bessel function table.)

The total power in the FM signal is

∞
!
1 2
X 1 1
J0 + 2 Jn = [J02 + 2J12 + 2J22 + · · · ] = ,
2
2 n=1
2 2

since the value of the quantity within the square bracket is 1. Hence the power up to the six sidebands
is 99% of the total transmitted power. We therefore conclude that for given β, the number of sidebands
12 Principles of Electronic Communications { Analog and Digital

for 98% percent power transmission always occurs for n > β + 1. Hence for sinusoidal modulation,
the approximate bandwidth of FM signal is

B = 2nfm = 2(β + 1)fm . (4.25)

where n is the number of significant sidebands. This rule was proposed by John Carson and is called
Carson’s rule.

Fig. 4.7 A qualitative plot of Jn (5) as a function of n

When β is small compared to unity, the bandwidth approximates to B = 2fm . This is the case of
a narrowband FM (NBFM), where only the first sideband pair is considered. The band of frequencies
occupied by the NBFM is fc − fm to fc + fm . For the other extreme case of β >> 1, the bandwidth
is approximately B = 2βfm . This is the nominal bandwidth of a wideband FM (WBFM).
Remember that the bandwidth relation (4.25) is stated for a sinusoidal modulation of a single
frequency fm . It can be generalised to the case of a sinusoidal information signal containing a
continuum of frequencies up to a maximum of fm ; the maximum frequency deviation being ∆f =
βfm , (4.25) generalises to

B = 2(∆f + fm ) = 2fm (1 + β). (4.26)

Example 4.4
A carrier of frequency 5 kHz and amplitude 5 volts is frequency modulated by a sinusoidal modulating
wave of frequency 1 kHz and amplitude 100 volts with kf = 10 Hz/V. Write down the expression
for the FM wave. Find the approximate band of frequencies occupied by the FM.

Solution
Here the carrier wave = 5 sin(2π × 5 × 103 t) volts and message signal, s(t) = 100 sin(2π × 103 t).
Since kf = 10 Hz/V, the FM signal is,
 Exponential CW Modulation 13

Z t
sFM (t) = Ac sin[ωct + 2πkf sin(τ )dτ ].
0

kf × 100
= Ac sin[2π × 5 × 103 t − cos(2π × 103 t)]
103
The maximum frequency deviation is
∆f 103
∆f = kf Am = 10 × 100 = 103 Hz and index of modulation β = = 3 =1
fm 10
Thus, BW = 2(∆f + fm ) = 2(103 + 103 ) = 4 kHz.
1 1
The frequency range = (fc − BW ) to (fc + BW ) = 3 to 7 kHz.
2 2
Example 4.5
A carrier of amplitude 5 volts and frequency 10 MHz is frequency-modulated by a sinusoidal signal
of frequency 2 kHz. The maximum frequency deviation achieved is 75 kHz. The modulated signal
attains its positive peak at t = 0. Write down the expression for the modulated carrier. Also, find
the bandwidth of this waveform.

Solution
Since the modulated carrier attains positive peak at t = 0, so the modulation signal will be of the
form Am cos ωm t the FM signal can be put in the form

sFM (t) = Ac cos(ωc t + β sin ωm t).

Here, Ac = 5 volts, ωc = 2π × 107 Hz, ωm = 2π × 2 × 103 Hz, maximum frequency deviation is

∆f = βfm . Thus,

75 × 103 = β × 2 × 103 ⇒ β = 37.5.

Hence, sFM (t) = 5 cos(2π × 107 t + 37.5 sin 4π103 t) volts.

The approximate bandwidth is given by

BW = 2(∆f + fm )

= 2(75 × 103 + 2 × 103 ) = 154 kHz.

The case is that of a WBFM since ∆f >> fm or β >> 1.

4.5 Narrowband FM and PM Signals: Phasor Representation

The narrowband FM signal has been defined as a special case of FM for β << 1. To study the
properties of the narrowband angle-modulated signals, we start with the angle modulated signal of
the form
14 Principles of Electronic Communications { Analog and Digital

sm (t) = Ac cos[ωct + km(t)], (4.27)

where the message signal is m(t). For phase modulation, the message signal directly relates m(t)
Z t
with the constant k = k p. For frequency modulation, it is the integral of message signal s(τ )dτ
0
that relates m(t) with the constant k = 2πk f . Expanding (4.27) yields

sm (t) = Ac cos ωc t cos[km(t)] − Ac sin ωc t sin[km(t)]. (4.28)

If k is very small so that the angle km(t) still remains small, then one can write cos[km(t)] ≈ 1 and
sin[km(t)] ≈ 0. In doing so, (4.28) simplifies to

sm (t) = Ac [cos ωc t − km(t) sin ωc t]. (4.29)

This is the expression for the narrowband signals. The spectra of the signal is obtained by the Fourier
transform of (4.29). Thus,

Ac Ac k
Sm (f) = [δ(f − fc ) + δ(f + fc )] − [M (f − fc ) − M (f + fc )], (4.30)
2 2j
Z t
where M (f) is the spectrum of m(t). For frequency modulation, m(t) = s(τ )dτ and hence
0
1 1
M (f) = S(f) = S(f).
jω 2πf

The Fourier transform of the NBFM is thus

 
Ac Ac k 1 S(f − fc ) S(f + fc )
SNBFM (f) = [δ(f − fc ) + δ(f + fc )] − −
2 2j j2π f − fc f + fc
 
Ac Ac k S(f − fc ) S(f + fc )
= [δ(f − fc ) + δ(f + fc )] + − . (4.31)
2 4π f − fc f + fc

Figure 4.8 shows the magnitudes of the Fourier transform of the PM and FM waves for a given
message signal. Since we do not know the shape of the message spectra, we have given a represen-
tative plot of S(f). For PM signals, m(t) = s(t) and the spectrum looks like that of a DSB with
1
carrier, as shown in Fig. 4.8(c). For FM signals, M (f) = S(f) and the low-frequency parts
j2πf
are emphasized, as observed in Figs. 4.8(b) and (c).
We now present several phasor diagrams of angle-modulated waves. We start with a modulating
signal of the form s(t) = cos ωm t. The signal m(t) is then
Z t
Am
m(t) = s(τ )dτ = sin ωm t.
0 2πfm
 Exponential CW Modulation 15

Fig. 4.8 The magnitude spectrum of : (a) S(f) of the message signal, (b) the form S(f)/f , (c) NBPM,
(d) NBFM.

Equation (4.29) becomes

kAm
sm (t) = Ac [cos ωc t − sin ωm t sin ωct]
2πfm
   
1 kAm 1 kAm
= Ac [cos ωc t − cos(ωc − ωm )t + cos(ωc + ωm )t]
2 2πfm 2 2πfm
     
jωc t 1 Ac Am j(ωc −ωm )t 1 Ac kAm j(ωc +ωm )t
= Re Ac e − e + e
2 2πfm 2 2πfm
16 Principles of Electronic Communications { Analog and Digital

  
jωc t Ac kAm −jωm t Ac kAm jωm t
= Re e Ac − e + e . (4.32)
4πfm 4πfm

We assume a coordinate system that rotates anticlockwise with an angular velocity ωc . The carrier
term is fixed in this coordinate system and oriented in the horizontal direction as shown in Fig. 4.9.
c kAm jωm t
The phasor for the term −A 4πfm e rotates in the clockwise sense with angular velocity ωm
(with respect to the carrier phasor). The phasor for the term A4πfc kAm jωm t
m
e , similarly rotates in the
anticlockwise direction with angular velocity ωm with respect to ωc. The two sideband phasors result
in a sum which is always perpendicular to the carrier phasor. The magnitude ∆ of this phasor is
calculated as follows:

∆2 = a2 + b2 + 2ab cos θ
 2
Ac kAm
= [1 + 1 + 2 cos(180◦ − 2ωm t)]
4πfm
 2
Ac kAm
= × 2[1 − cos 2ωm t]
4πfm
 2
Ac kAm
= × 4 sin2 ωm t.
4πfm

Ac kAm
Thus ∆ 1 = sin ωm t. (4.33)
2πfm

kf Am
Note that (4.29) represents an FM signal, k = 2πkf , ∆1 = Ac sin ωm t = Ac β sin ωm t.
fm

Fig. 4.9 Phasor plot for NBFM signal

 Exponential CW Modulation 17

The resultant phasor R attains its maximum value when ∆1 = Ac β and minimum when ∆1 = 0.
The phasor R is ahead of the carrier phasor by an angle φ given by
∆1 kf Am
tan φ = = = β. (4.34)
Ac fm
Since β << 1, the maximum value of φ is β.
Let us now compare the phasor diagram of the AM signal with that of the FM. For a sinusoidal
message, the AM signal is

sAM (t) = Ac [1 + ks(t)] cos ωc t = Ac [cos ωc t + ks(t) cos ωct]. (4.35)

For a comparison of the time-domain representation, we rewrite the expression for NBFM:

sNBFM (t) ≈ Ac [cos ωc t − km(t) sin ωc t]. (4.36)

The first term of NBFM is cos ω ct and the second term involves sin ωc t, a quadrature component.
But in AM both the first and second terms involve cos ωc t, an in-phase relationship.
We now express (4.35) as the real part of a complex exponential with s(t) = Am cos ωm t.

sAM (t) = Ac [cos ωc t + kAm cos ωc t cos ωc t]

 
kAm
= Ac cos ωct + {cos(ωc + ωm )t + cos(ωc − ωm )t}
2
 
kAm j(ωc +ωm )t kAm j(ωc −ωm )
= Ac Re ejωc t + e + e
2 2
  
kAm jωm t kAm −jωm t
= Ac Re ejωc t 1 + e + e . (4.37)
2 2

Following the approach of NBFM, we now plot the phasor of (4.37) in Fig. 4.10. Note that the
resultant phasor R is in phase with the carrier term. The magnitude of the resultant of the two
sidebands phasors ∆ is obtained as follows:
 2
2 Ac kAm
∆ = [1 + 1 + 2 cos 2ωm t]
2
 2
kAc Am
= × 4 cos2 ωm t,
2
that is, ∆ = kAc Am cos ωm t = Ac m cos ωm t. (4.38)

The resultant phasor is given by

R = Ac + mAc cos ωct = Ac (1 + m cos ωc t). (4.39)

The resultant is in parallel to the carrier phasor with the magnitude varying between Ac (1 + m) and
Ac (1 − m).
18 Principles of Electronic Communications { Analog and Digital

Fig. 4.10 Phasor plot for an AM signal

4.6 Frequency Modulation by Non-Harmonic Modulating Waves

Consider an RF carrier of frequency ωc angle modulated by two sinusoidal modulating waves of
non-harmonic frequencies ω1 and ω2 . The FM waveform is given by the equation

v(t) = Ac cos(ωct + β1 sin ω1 t + β2 sin ω2 t)

= Ac [cos ωc t cos(β1 sin ω1 t + β2 sin ω2 t)

− sin ωc t sin(β1 sin ω1 t + β2 sin ω2 t)], (4.40)

where β1 and β2 are the depths of modulation. Since for small angle x, cos x ≈ 1 − x2 /2 and
sin x ≈ x,

v(t) = Ac [cos ωc t 1 − 12 (β1 sin ω1 t + β2 sin ω2 t)2



− sin ωct(β1 sin ω1 t + β2 sin ω2 t)], (4.41)

which on simplification leads to

β 2 + β22
 
β1 Ac
v(t) = Ac 1− 1 cos ωc t + [cos(ωc + ω1 )t − cos(ωc − ω1 )t]
4 2

β2 Ac
+ [cos(ωc + ω2 )t − cos(ωc − ω2 )t]
2
 Exponential CW Modulation 19

β12 Ac
+ [cos(ωc + 2ω1 )t + cos(ωc − 2ω1 )t]
8

β22 Ac
+ [cos(ωc + 2ω2 )t + cos(ωc − 2ω1 )t]
8
β1 β2 Ac
+ [cos(ωc + ω1 + ω2 )t + cos(ωc − ω1 − ω2 )t]
4
β1 β2 Ac
− [cos(ωc + ω1 − ω2 )t + cos(ωc − ω1 + ω2 )t]. (4.42)
4
Thus we see that in addition to the carrier frequency, the spectrum consists of sidebands separated
from the carrier at ω1 , ω2 , 2ω1 , 2ω2, ω1 + ω2 , ω1 − ω2 . If higher-order terms in the approximations
for cos x and sin x are considered, the sidebands are separated from the carrier frequency at multiples
of ω1 , ω2 , ω1 ± ω2 .
If the depths of modulations β1 and β2 in (4.40) are sufficiently small such that |β 1 sin ω1 t +
β2 sin ω2 t| << π/2, then

v(t) ∼
= Ac [cos ωc t − (β1 sin ω1 t + β2 sin ω2 t) sin ωc t]. (4.43)

4.7 Pulsed FM Signal

Frequency-modulated CW-radar utilises the linear FM pulse whose instantaneous frequency varies
linearly with time. This linear FM is also called pulsed FM. Consider an RF pulse of rectangular
envelope and duration τp as shown in Fig. 4.11. The instantaneous frequency ω(t) of the pulsed
carrier is defined as

ω(t) = ω0 + at, (4.44)

where ω0 is the unmodulated carrier frequency. The constant a determines the rate of linear frequency
change and has the units of rad/s2 . Over the duration τp , the total frequency deviation is

2ωd = aτp . (4.45)

The total instantaneous phase θ(t) of the carrier is

Rt
θ(t) = 0
ω(t)dt = ω0 t + 12 at2 . (4.46)

The linear FM signal is then

sLFM (t) = A0 cos(ω0 t + 21 at2 ) −τp /2 < t < τp /2

(4.47)
=0 |t| > τp .
20 Principles of Electronic Communications { Analog and Digital

Fig. 4.11 (a) RF pulse, (b) linear frequency variation, (c) linear FM pulse.

The spectrum of sLFM (t) is obtained by the Fourier transform of (4.47). Thus,

τp /2
1
Z
SLFM (f) = A0 cos(ω0 t + at2 )e−jωt dt
−τp /2 2

τp /2 τp /2
A0 A0
Z Z
j[(ω−ω0 )t− 12 at2] 1 2
= e dt + e−j[(ω+ω0 )t+ 2 at ] dt2 ]dt. (4.48)
2 −τp /2 2 −τp /2

The first integral gives a spike near the positive frequency ω = ω0 and the second one gives the same
near the negative frequency ω = −ω0 . To find the S(f) in the region ω > 0, the second integral is
discarded as the two peaks are well separated and hence negligible overlap occurs. The spectrum for
ω < 0 may be derived in a similar fashion or by using the property of the Fourier transform.
 2  2
1 2 1 2 1 ω − ω0 ω − ω0
Now, at − (ω − ω0 )t = a[t2 − (ω − ω0 )t] = a t − − √ .
2 2 a 2 a 2a
 Exponential CW Modulation 21

√ √
 
ω − ω0
Let πx = a t − (4.49)
a
√ √
so that πdx = adt. The first term of (4.48) becomes
x2  √
1 2 (ω − ω0 )2
 
A0 π
Z
S1 (f) = exp j πx − √ .dx
2 −x1 2 2a a
"   2 # Z x2
A0 √ ω − ω0 2
= πa exp −j √ ejπx /2 dx, (4.50)
2 2a −x1

   
aτp /2 + (ω − ω0 ) aτp /2 + (ω − ω0 )
where x1 = − √ and x2 = − √ . (4.51)
πa πa

Recall the Fourier integrals

Rx Rx
c(x) = 0
cos(πx2 /2)dx; s(x) = 0
sin(πx2 /2)dx. (4.52)

Then,
Z x2 Z 0 Z x2
2
ejπx /2
dx = [cos(πx2 /2) + j sin(πx2 /2)]dx + [cos(πx2 /2) + j sin(πx2 /2)]dx
−x1 −x1 0

= C(x1 ) + C(x2 ) + j[S(x1 ) + S(x2 ). (4.53)

Thus (4.50) reads as

r "  2 #
A0 π ω − ω0
S1 (f) = exp −j √ . {C(x1 ) + C(x2 ) + j[S(x1 + S(x2 )]}. (4.54)
2 a 2a

The magnitude spectrum is

r
A0 π 2
|S1 (f)| = [{C(x1 ) + C(x2 )} + {S(x1 ) + S(x2 )}2 ] (4.55)
2 a

and the phase spectrum is

(ω − ω0 )2
 
S(x1 ) + S(x2 )
θ(ω) = − + arctan . (4.56)
2a C(x1 ) + C(x2 )

The magnitude and phase spectra are obviously complicated functions of frequency. Define a dimen-
sionless quantity B as the product of the frequency deviation and the pulse width.
22 Principles of Electronic Communications { Analog and Digital

That is, B = 2ωd τp = aτp2 . (4.57)

r r r
aτp a2 τp2 aτp2 B
At ω = ω0 , −x1 = √ = = = = x2 .
2 πa 4πa 4π 4π

For large B(B >> 1 is the case of practical interest),

C(x1 ) = C(x2 ) ≈ 0.5 and S(x1 ) = S(x2 ) ≈ 0.5.

p
So |S1 (ω)| = A0 π/2a. Thus the magnitude spectrum approaches a rectangle shape with the
spectral width close to 2ωd , as shown in Fig. 4.12. The power spectral density of the signal is given
by

πA20
G(f) = |S1 (ω)|2 = , a constant. (4.58)
2a

Fig. 4.12 Spectrum of pulsed FM signal

4.8 Amplitude−Frequency Modulated Signal (AM−FM)

The general form of a carrier which is modulated both in amplitude and frequency is given by

vc (t) = A(t) cos[ωc t + θ(t)]. (4.59)

Here the time-varying amplitude A(t) varies in accordance with the message signal of frequency fm .
The phase θ(t) also varies with time following the message signal. For a sinusoidal message signal
of the form cos ωm t, the expression for the AM−FM signal can be written as

vc (t) = Ac (1 + m cos ωm t) cos(ωc t + β sin ωm t). (4.60)

The quantities m and β denote the depths of modulation for amplitude and frequency, respectively.
For small β, the FM part can be simplified as is done in NBFM. In doing so,

vc (t) ≈ Ac (1 + m cos ωm t)[cos ωc t − β sin ωm t sin ωct]. (4.61)

 Exponential CW Modulation 23

Example 4.6
Show that the signal vc (t) given by

vc (t) = cos(2πfc t) + 0.02 cos[2π(fc + fm )t]

represents an AM-FM signal. Find the modulation indices m and β. Write an expression for the
instantaneous frequency f(t).

Solution

vc (t) = cos(2πfc t) + 0.02[cos(2πfc t) cos(2πfm t) − sin(2πfc t) sin(2πfm t)]

= [1 + 0.02 cos(2πfm t)] cos(2πfc t) − 0.02 sin(2πfm t) sin(2πfc t).

4.9 FM Generation
There are two methods of FM generation, namely, the direct method and the indirect method. The
basic principle of FM generation in the direct method is to change the frequency of an electronic
oscillator by the modulating signal. In the indirect method, the NBFM signal is first generated,
which is then converted to an WBFM by the use of frequency multipliers. We will discuss both the
methods now.

4.9.1 Indirect method

An angle modulated signal is written as Ac cos[ωct + km(t)], where m(t) relates the modulating
Rt
signal. For frequency modulation, m(t) = 0 s(τ )dτ , where s(t) is the actual modulating signal and
k = 2πkf . If the modulation is narrowband, the waveform is approximated as

sNBFM (t) = Ac [cos ωc t − 2πkf m(t) sin ωc t]

Z t
= Ac [cos ωc t − 2πkf s(τ )dτ · sin ωc t].
0

The equation suggests the possible method of NBFM generator, whose block diagram representation
is shown in Fig. 4.13. If we desire phase modulation rather than frequency, we simply remove the
integration and replace the input 2πk f m(t) by kp s(t).
The signals thus generated have strictly a narrow bandwidth since the maximum frequency devi-
ation ∆f is small compared to highest modulating frequency. The NBFM signal can be converted
to an WBFM signal by the use of a frequency multiplier that increases the frequency deviation. Let
us briefly discuss the operational principle of a frequency multiplier. Frequency multiplication is
achieved by sending a periodic signal through a nonlinear device that generates harmonics, followed
by bandpass filtering as shown in Fig. 4.14.
The input to the nonlinear device is a periodic signal, not necessarily sinusoidal. The output
waveform has a similar fundamental frequency but is rich in higher-order harmonics. To attain an
n-fold increase in frequency, the nonlinear device must offer nth order nonlinearity. The bandpass
24 Principles of Electronic Communications { Analog and Digital

filter tuned to the nth harmonic selects the signal of frequency nf. We may multiply the frequency by
any arbitrary integer n, at least in principle. But, in practice, the amplitudes decrease with increasing
harmonic number. Normally, a multiplication factor of 2 to 5 is taken in a single stage. If it requires
a 12-fold frequency multiplication, one can use a device offering 12th order nonlinearity. The same
can be achieved by cascading two second-order devices and one third-order device.

Fig. 4.13 NBFM generator

Fig. 4.14 A frequency multiplier

Fig. 4.15 Block diagram of the modulator for WBFM

Returning to the NBFM, let the signal be applied at the input of the frequency multiplier. Since
the instantaneous frequency of the NBFM signal is f(t) = fc + kf s(t), the resulting signal has the
frequency nf(t) = nfc + nkf s(t), where n is the multiplier. Both the carrier and the frequency
deviation are multiplied by the same integer. Note that the modulation index is proportional to the
frequency deviation for a given modulating frequency and hence this scheme increases the modulation
 Exponential CW Modulation 25

index by factor n and the waveform is converted to a wideband signal. In practice, the higher carrier
frequency is down-converted by heterodyning. The complete scheme for generating wideband FM is
shown in Fig. 4.15.

Example 4.7
Show that the frequency multiplier, although a nonlinear device, does not distort the signal.

Solution
Suppose that the nonlinearity is of second order so that the multiplier is described by the following
input{output relation:

y(t) = a0 + a1 x(t) + a2 x2 (t).

When x(t) is an FM signal of the form x(t) = A cos(ωc t + km(t)), the output is

y(t) = a0 + a1 A cos[ωct + km(t)] + a2 A2 cos2 [ωct + km(t)]

a2 A2 a2 A2
 
= a0 + + a1 A cos[ωct + km(t)] + cos[2ωct + 2km(t)].
2 2

We observe that the output contains the dc term which can be filtered out by ac coupling. The second
term is the original FM weighted by a1 . The last term is also an FM signal of carrier frequency
2ωc with frequency deviation twice that of the original FM wave. The message signal m(t) remains
unaltered in both the terms. The result can be extended to the case where the multiplier shows nth
order nonlinearity.

Example of Armstrong FM system

In the Armstrong method, a relatively low frequency carrier is phase-shifted by 90◦ and fed to a
balanced modulator (multiplier) (see Fig. 4.13), where it is mixed with the input modulating signal
(fm ). The output is a narrowband FM with low index of modulation and hence low frequency
deviation. Since small frequency deviations are possible in the basic Armstrong systems, extensive
frequency multiplication and mixing are used to increase the deviation to the desired value (Fig. 4.16).
Suppose we want to design an Armstrong FM modulator to generate an FM signal of carrier
frequency 96 MHz and frequency deviation ∆f = 75 kHz using the indirect method. We first
generate a NBFM signal with a carrier frequency fc1 = 200 kHz, as shown in Fig. 4.16. The
frequency chosen is such that the stable crystal oscillator and the product modulator required for
NBFM generation can easily be constructed (see Fig. 4.13 for NBFM generator). We select the
frequency deviation ∆f = 25 Hz at this stage based on the following consideration.
For commercial FM broadcast, the baseband spectrum ranges from 50 Hz to 15 kHz. The FM
signal, cos(ωc t + φ(t)), turns into a narrowband FM provided the depth of modulation is low enough
for |φ(t)| << π/2 = 1.57. In the worst case we can take a value of φ as large as 0.5. For tone
modulation, φ(t) = β sin(2πfm t), and so β can attain a maximum value of 0.5. Since β = ∆f fm ,
∆f = βfm = 0.5 × 50 Hz= 25 Hz.
Since we require ∆f = 75 kHz at the final output, the multiplication factor is 75000/25 = 3000.
26 Principles of Electronic Communications { Analog and Digital

But, instead of multiplying directly, we multiply the frequency of NBFM in two stages: by 64 before
heterodyning and by 48 after heterodyning. These values are selected based on the property that
multiplication can be achieved using factors of 2 and 3, i.e., multiplication by 64 can be realised using

Fig. 4.16 Indirect method of FM generation for Armstrong modulator

a multiplier strip consisting of six doublers in cascade (64 = 26 ). The first multiplication results in a
new carrier frequency fc2 = 64 ×200 kHz = 12.8 MHz and a frequency deviation ∆f2 = 64 ×25 Hz
= 1.6 kHz. The frequency translation is achieved in a mixer with a carrier of frequency 10.8 MHz.
The whole spectrum now shifts around an intermediate frequency of fc3 = 12.8 − 10.8 = 2 MHz.
The frequency deviation is of course ∆f3 = 1.6 kHz, as it is not affected by the frequency-shifting.
The FM signal is obtained by multiplying this signal a second time by 48. This gives the carrier
frequency fc = 2 MHz ×48 = 96 MHz and a frequency deviation equal to 75 kHz. The crystal
oscillator of carrier frequency 10.8 MHz is generated from the crystal oscillator of 200 kHz to avoid
any phase mismatch.
Example 4.8
In the above FM modulator, if the oscillator at the mixer stage suffers a frequency drift of δfLO ,
what will be the carrier frequency of the FM signal ?
Solution
The frequency of the oscillator is
fL = 10.8 MHz ±δfLO Hz.
The output signal of the mixer has the carrier frequency
fc3 = 12.8MHz −(10.8 MHz ±δfLO Hz) = 2MHz ∓δfLO .
Hence the FM carrier is at the frequency
fc = 48 × fc3 = 96 MHz ±48 × δfLO .
To suppress the effect of this drift, the mixing signal is not taken from a separate oscillator but from
the 200 kHz crystal oscillator through the use of frequency multipliers.
 Exponential CW Modulation 27

Example 4.9
Assume that the 10.8 MHz mixing signal is derived from the 200 kHz oscillator. If the 200 kHz
oscillator drifts by 0.2 Hz ( 1 part per millions), find the drifts in the 10.8 MHz signal and the
resulting FM signal.

Solution
We need a frequency multiplier of 10.8MHz 3
200kHz = 54 = 2 × 3 . (This is obtained by the cascade of a
frequency doubler and three triplers.) The carrier frequency of the NBFM is fc1 = 200 kHz ±0.2
Hz. After the first multiplication by 64,

fc2 = 12.8 MHz ± 12.8 Hz.

For a drift of 0.2 Hz in the 200 kHz signal, the frequency of the mixing signal is

fLO = 10.8 MHz ± 54 × 0.2 Hz = 10.8 MHz ±10.8 Hz.

For the intermediate frequency after the mixer stage,

fc3 = (12.8 MHz ± 12.8 Hz) − (10.8 MHz ± 10.8 Hz)

= 2 MHz ± 2 Hz.

Hence the resulting FM signal drifts by 48 × (2) = ±96 Hz.

4.9.2 Direct method of FM generation

The direct way of FM generation is based on varying the output frequency of an oscillator in proportion
to the baseband signal. The heart of this modulator is the oscillator. There are a wide variety of
oscillators that accomplish this task, e.g., the Hartley or the Colpitt oscillator consisting of a tank
circuit that produces sinusoidal waveform of frequency ωc determined by the values of L and C (
1
since ωc = √LC ). The variation of the frequency of oscillation can be accomplished by varying any
of the reactive parameters (L or C) of the tank circuit. If, for example, the capacitor variation is
controlled by the baseband signal, an FM signal results. One way to accomplish this is with the use
of semiconductor diodes that act as capacitors whose values vary with reverse bias voltage. Such
capacitors are commonly called the varicaps.
Let the capacitance variation be of the form Cm = αs(t), where s(t) is the baseband signal. The
instantaneous frequency of the oscillator is then

1 1 h α i−1/2
ω(t) = p = √ 1 + s(t)
L[C + αs(t)] LC C
h α i α 
≈ ωc 1 − s(t) , since  s(t) << 1.
 
2C c

= ωc − ks(t) (4.62)
28 Principles of Electronic Communications { Analog and Digital

αωc
where k = + . Thus FM signal is achieved. The instantaneous frequency depends on the instan-
2C
taneous value of the modulating signal. The maximum capacitance deviation is ∆C = αAm , where
Am is the signal amplitude. Thus,
 
∆C αAm 2kAm 2 × 2π∆f ∆f
= = = =2 .
C C ωc 2πfc fc

Since ∆f << fc , the fractional change in capacitance may be small compared to C.

There are other devices namely PIN diodes and FETs that can be used successfully to generate FM
signals. Figure 4.17 shows a FET reactance modulator whose effective capacitance at the terminals
(1,1’) can be controlled by the modulating signal. If this capacitance variation be included in the tank
circuit of an oscillator, we have accomplished the task of generating FM.

Fig. 4.17 FET reactance modulator

With the applied voltage v(t), the bias current ib (t) is given by

v(t)
ib (t) = , (4.63)
R − jXC

where Xc = 1/ωC.

The voltage at the gate of the FET is given by

R
vg (t) = ib (t)R = v(t). (4.64)
R − jXC

The drain current i d (t) is then

Rgm
id (t) = gm vg (t) = v(t), (4.65)
R − jXC
 Exponential CW Modulation 29

where gm is the FET’s transconductance. The impedance Z(ω) at the terminals (1,1’) between the
drain and the ground is thus
 
v(t) R − jXc 1 Xc
Z(ω) = = = 1−j . (4.66)
id (t) gm R gm R

By the appropriate choice of C and the frequency of operation, we assume Xc >> R and so

Xc 1 j
Z(ω) ≈ −j = −j =− , (4.67)
gm R ωRCgm ωCm

where Cm = gm RC is the effective capacitance. Any modulating signal applied between the gate
and the source of Fig. 4.17 produces a periodic variation of vgs , the gate-to-source voltage, thereby
making a proportional change in gm . Thus the effective capacitance Cm can be varied with bias
voltage.
An oscillator whose frequency of oscillation can be varied by the voltage of the modulating signal
is called a voltage-controlled oscillator (VCO). The direct method of FM generation using VCO is
attractive due to the fact that it produces sufficient frequency deviation and requires no frequency
multiplication. The disadvantage of this method is that it has poor frequency stability. Of course, a
crystal-oscillator may be used for carrier generation.
VCOs are now available in IC form. VCO-566 is one of the family that can be used for generating
carrier frequencies less than 1 MHz.

Example 4.10
In the direct method of FM generation, the controlled capacitance Cm is related to the modulating
signal vm (t) as
100
Cm = √ pF.
1 + 2vm
It is given that the capacitance C0 of the resonant circuit is 200 pF. The inductance L is adjusted for
resonance to occur at 5 MHz for Vm = 4 volts. If the modulating signal

vm (t) = 4 + 0.05 sin(2π × 103 t) volts,

find the expression for the angle-modulated signal.

Solution
The carrier frequency is given by
1
fc = p .
2π L(C0 + Cm )

100
Here, C0 + Cm = 200 + p pF
1 + 2{4 + 0.05 sin(2π × 103 t)}
30 Principles of Electronic Communications { Analog and Digital

 
700 1 3
' 1− sin(2π × 10 t) pF.
3 1260
r  
3 1
Thus, (C0 + Cm )−1/2
= 1+ sin(2π × 103 t) × 106 (F −1/2).
700 2520

100
For vm = 5 volts, fc = 5 MHz and Cm = pF,
3
1 106
5 × 106 = q = q H. Thus L = 4.342 µH.
2π L 200 + 100 700


3
× 10 −12 2π 3
L

The instantaneous carrier frequency is given by

1 1
f(t) = √
2π L (C0 + Cm )1/2
r r  
1 700 3 1
= × 10π × 1+ sin(2π × 10 t) × 106 Hz
3
2π 3 700 2520
 
1
=5 1+ sin(2π × 103 t) MHz.
2520

The angle modulated signal is v(t) = Ac cos[5 × 106 t + 1984 sin(2π × 103 )t] volts.

4.10 FM Demodulators
In FM signals, the information resides in the instantaneous frequency. Recovery of the information
signal can be done in two ways. In one, the demodulators employ discriminators (a frequency-selective
network that yields an output proportional to the instantaneous frequency). A diode envelope detector
following the discriminator detects the message signal. The frequency discriminator converts FM
signals to AM signals and then demodulates the AM envelope using peak detector. The second
method employs a phase-locked loop (PLL) that demodulates the message signal by matching the
output of a local oscillator (VCO in this case) to the modulated carrier. We will discuss here both
types of demodulators.

4.10.1 Detection based on discriminator

Consider the FM signal

sFM (t) = Ac cos[ωct + 2πkf m(t)]

Rt
where m(t) = 0
s(τ )dτ , s(t) being the message signal. The derivative of sFM (t) is given by
 Exponential CW Modulation 31

 Z t 
dsFM (t)
= −Ac [ωc + 2πkf m(t)] sin ωc t + 2πkf s(τ )dτ . (4.68)
dt 0

The signal defined by (4.68) is an AM−FM waveform with amplitude Ac [ωc + 2πkf s(t)]. Since the
amplitude varies proportionately to the message signal, an envelope detector can be used to recover
s(t). Figure 4.18 shows such a scheme. For a phase modulated input, the envelope detector yields a
signal that is the derivative of s(t). In such a case an integrator must follow the detector.
A differentiator in the frequency domain is a network having transfer function of the form H(f) =
j2πf. Its output is related to its input as S out (f) = j2πfSin (f), i.e., the magnitude of the output
is linearly proportional to the frequency of the input. We require such a linear transfer characteristic
around the carrier frequency. A bandpass filter can be used if the operation is confined to a limited
range of frequencies, as illustrated in Fig. 4.19.

Fig. 4.18 FM demodulation by differentiation followed by diode detection

Fig. 4.19 Bandpass filter showing linear portion

The range of linear operation is required to be extended so as to accommodate sufficient number

of sidebands. This is accomplished in a balanced demodulator as shown in Fig. 4.20. As we take
32 Principles of Electronic Communications { Analog and Digital

the difference between two demodulator outputs, the overall transfer function of the bandpass filter
discriminator is as shown in Fig. 4.21.
The bandpass filter consists of a tank circuit as shown in Fig. 4.22. For an FM input the discrim-
inator produces an output voltage proportional to the input frequency. The magnitude of the output
voltage attains a maximum for an input frequency equal to the resonant frequency (of either f1 or
f2 ). The transfer characteristic of Fig. 4.21 can be realised in a circuit of Fig. 4.23. The upper half
of the secondary winding and C2 is tuned to f2 . The lower half and C1 form the BPF centred at f1 .
The precisely linear transfer characteristic of the filter's tail is a theoretical assumption. In practical
filters the magnitude of the transfer function at the tail can be written as

|H(f)| = a0 + a1 (f − fc ) + a2 (f − fc )2 + · · · . (4.69)

Fig. 4.20 Balanced FM demodulator

Fig. 4.21 Bandpass filter discriminator

 Exponential CW Modulation 33

Fig. 4.22 Possible frequency selective network (a) frequency-to-voltage converter

Fig. 4.23 The circuit realising the transfer function of Fig. 4.21

Assuming a second-order nonlinearity, the transfer function of BPF2 is

|H2(f)| = a0 + a1 (f − fc ) + a2 (f − fc )2 (4.70)

and that of BPF1 is

|H1(f)| = a0 + a1 (f − fc ) + a2 (f − fc )2 . (4.71)

The difference signal at the output has the amplitude

V0 = Vi [|H2(f)| − |H1(f)|]

= Vi × 2a1 (f − fc ). (4.72)

So far, we have considered that the amplitude of the FM carrier remains constant. But in an FM
system, channel noise, fading, etc. cause the amplitude Ac to vary. This leads to an additional
term dAc/dt in (4.69). Such amplitude variations can be removed by passing the signal through a
hard-limiter followed by a bandpass filter as shown in Fig. 4.24. The limiting of the peaks of the
34 Principles of Electronic Communications { Analog and Digital

envelope does not create any detrimental effect on message recovery as the message is impressed on
the carrier in the form of frequency variation.

Fig. 4.24 (a) Hard-limiter followed by BPF to suppress the amplitude variation; (b) the input and output
of the hard-limiter.

Example 4.11
Explain how a hard-limiter and a bandpass filter can suppress the amplitude variation in the received
FM signal.

Solution
Let the received input be vi (t) = Ai (t) cos(φc t), where φc (t) = ωc t + 2πkf (t) (see Fig. 4.25 (a)).
The output of the hard-limiter is V or −V depending on whether v i (t) is positive or negative. That
is, irrespective of the magnitude of vi (t), the hard-limiter generates either +V or −V . Hence a
square wave results, the duration of which is determined by the zero-crossings of the input signal.
Since the input signal is angle-modulated, the output is also angle-modulated.
The input−output relation of the hard-limiter can be written as

v0 (t) = V when vi (t) > 0

= −V when vi (t) < 0,

Fig. 4.25 Illustrates the action of hard-limiter

 Exponential CW Modulation 35

or equivalently,

 +V when cos φc (t) ≥ 0
v0 [φc(t)] =
−V when cos φc (t) < 0.


A plot of v0 [φ] versus φ is shown in Fig. 4.25 (b). Note that v0 (φ) is a periodic function with period
2π. Fourier series expansion of v0 (φ) gives
4V 1 1 1
v0 (φ) = (cos φ − cos 3φ + cos 5φ − cos 7φ + · · · ).
π 3 5 7
Since φ = ωct + 2πkf s(t),
 
4V 1
v0 (t) = cos [ωc t + 2πkf s(t)] − cos[3ωct + 6πkf s(t)] + · · · .
π 3
Each term of the above equation represents an FM wave. We see that the hard-limiter output contains
the original FM wave in addition to the FM waves of carrier frequencies 3ωc , 5ωc, etc. The bandpass
filter tuned to ω c passes only the original FM wave with constant amplitude.

4.10.2 Detection using phase-locked loops

The phase-locked loop (PLL) is a feedback system that can be used for FM demodulation, especially
under high noise conditions when a conventional FM discriminator fails to yield satisfactory recovery
of baseband signal. The PLL comprises two basic components namely a voltage-controlled oscillator
(VCO) and a phase comparator (Fig. 4.26). A PLL FM demodulator does not require any tuned
circuit. It automatically compensates for changes in the carrier frequency. Once frequency lock
occurs, the VCO tracks the frequency changes in the input signal. When the PLL input is a deviated
FM (frequency different from the natural frequency of VCO), the output of the phase comparator
is proportional to the frequency deviation. This is the correction voltage to VCO. We have indeed
demodulated the FM signal.

Loop Loop

Fig. 4.26 Block diagram of the PLL FM demodulator

The simplest method of achieving phase comparison is using a product multiplier followed by a
lowpass filter shown in Fig. 4.27. For the two input signals v1 (t) = A sin[ωt + θ1 (t)] and v2 (t) =
B cos[ωt + θ2 (t)], the multiplier output is
36 Principles of Electronic Communications { Analog and Digital

x1 (t) = AB sin[ωt + θ1 (t)] cos[ωt + θ2 (t)]

AB
= {sin[θ1 (t) − θ2 (t)] + sin[2ωt + θ1 (t) + θ2 (t)]}. (4.73)
2
The output of the LPF is
AB
v0 (t) = sin(θ1 − θ2 ) (4.74)
2
provided that θ1 (t) and θ2 (t) have bandwidths less than 2ω. Since the instantaneous frequency of
the VCO is given by

ωVCO(t) = ωc + kv0 (t), (4.75)

the loop is locked when θ1 = θ2 and VCO oscillates at its free-running frequency ωc.

Fig. 4.27 A simple phase comparator

Example 4.12
Find and sketch the output of the phase comparator if the inputs to the phase comparator are square
waveforms as shown in Fig. 4.28 (a).

Solution
Let td be the time difference. Since the LPF gives the average value of the product signals v1 (t), v2 (t),
the final output is calculated as follows.

T /2 td T /2
2 2 2
Z Z Z
Time average = v1 (t)v2 (t)dt = (−AB)dt + (AB)dt
T 0 T 0 T td

   
2 T 4td
= (AB) −td + − td = AB 1 − .
T 2 T

Thus the comparator output amplitude varies linearly with the time difference (or equivalently, the
phase difference) as shown in Fig. 4.28(b).
 Exponential CW Modulation 37

v2(t)

Fig. 4.28 Illustrates phase comparison and average output.

A phase comparator based FM demodulator is drawn in Fig. 4.29. Returning to the discussions
on VCO, suppose the output of the VCO is a sinusoidal voltage of amplitude B and its free-running
frequency is ωc. The instantaneous frequency of the VCO is given by (4.75), where the constant
k defines the measure of the sensitivity of the VCO. If one defines the VCO sensitivity as the rate
dωVCO
of change of the instantaneous frequency with respect to the controlling voltage, then k = .
dv0 (t)
Thus the VCO output is given by
Z t
e0 (t) = B cos[ωc t + φ(t)], where φ(t) = 2πk s0 (τ )dτ . (4.76)
0
Rt
Let the FM signal be v1 (t) = A sin[ωct + θ(t)], where θ(t) = 2πkf m(t) and m(t) = 0
s(τ )dτ and
s(t) is the message signal. The multiplier output is the error signal ed (t) given by
38 Principles of Electronic Communications { Analog and Digital

ed (t) = v1 (t)e0 (t)

 Z tv Z t 
AB
= sin 2π kf s0 (τ )dτ − k s0 (τ )d(τ ) + high-frequency terms.
2 0 0

1
= kd AB sin[θ(t) − φ(t)] (4.77)
2
where kd is the constant associated with the multiplier.

Fig. 4.29 A first-order phase-locked loop

The output of the loop filter is then

 Z t Z t 
AB
s0 (t) = sin 2π kf s0 (τ )dτ − k s0 (τ )dτ .
2 0 0

1
= µkd AB sin[θ(t) − φ(t)] (4.78)
2
The phase difference β(t) is given by
Z t Z t
β(t) = θ(t) − φ(t) = 2πkf s(τ )dτ − 2πk s0 (τ )dτ. (4.79)
0 0

To perform transient analysis, we differentiate (4.79) to get

dβ
= 2πkf s(t) − 2πks0 (t). (4.79)
dt
For a linear phase detector,
 Z t Z t 
1
s0 (t) = µkd AB2π kf s(τ )(τ )dτ − k v0 (τ )dτ , (4.80)
2 0 0
 Exponential CW Modulation 39

which may be derived from (4.78) for small angles as for small phase error, θ(t) − φ(t) is small and
sin[θ(t) − φ(t)] ≈ θ(t) − φ(t). Differentiating (4.80), we have
ds0 1 dβ
= πµkd AB[kf s(t) − ks0 (t)] = µkd AB . (4.81)
dt 2 dt
dβ
This represents the relation between the frequency offset and the output voltage s0 (t). The
dt
ds0
equilibrium steady-state is obtained by setting = 0. Then,
dt
kf s(t) − ks0 (t) = 0

k
or, s(t) = s0 (t). (4.82)
kf

We have thus accomplished FM demodulation. We rewrite (4.81) as

dv0
+ πABkv0 (t) = πABkf v(t)
dt
which is of the form
dv0 v0 kf
+ = v(t), (4.83)
dt τ kτ
1
where τ = πABk is the time constant of the transient response.

The transfer function H(s) of the PLL is obtained by taking the Laplace transform of (4.83) and
then the finding the ratio of the output signal to the input signal. Thus,
1 kf
sV0 (s) + V0 (s) = V (s)
τ kτ
V0 (s) kf /(kτ )
or, H(s) ≡ = . (4.84)
V (s) (s + 1/τ )

The function H(s) has a single pole at s = − τ1 and that is why the PLL is referred to as a first-order
PLL. The PLL output is thus a distorted version of v(t) (as if the signal has passed through a lowpass
simple RC network of time constant τ ). To reduce distortion, the cutoff frequency of the PLL is
made much higher than the highest modulating frequency.

The unit impulse response is obtained by taking the inverse Laplace transform of Eq. (4.84).
In doing so,
 
kf 1/τ kf 1 −t/τ
h(t) = ·L−1
= · ·e u(t) (4.84a)
k s + 1/τ k τ

kf
Where = π µ kd AB.kf = kt is the loop gain of the PLL. As loop gain tends to infinity,
kτ
40 Principles of Electronic Communications { Analog and Digital

we have Lt h(t) → δ(t), Loop gain → ∞ which is synonymous with the condition φ(t) ≈ θ(t) for
large loop gain.

Example 4.13
Assume an input to the FM modulator is a unit-step function u(t). Find the demodulated output using
a first-order PLL.

Solution
The FM signal in this case

v1 (t) = A sin [ωct + θ(t)],

Rt
where θ(t) = 2πkf m(t) = 2πkf s(t)d(t), and s(t) = u(t).
0

How, for linear phase detector we have the relation:

ds0 1 d
= µ kd AB [2πkf s(t) − 2πks0 (t)].
dt 2 dt
d s0 s0 kf
+ = s(t)
dt τ kτ
1 kf
or, s S0 (s) + S0 (s) = .S(f)
τ kτ
1
Since s(t) = u(t), we have, S(s) = .
s
 
1 kf 1
So, s + S0 (s) = ·
τ kτ s

kf 1 1
or, S0 (s) = ·
kτ s s + 1/τ
 
kf 1 1
= −
k s s + 1/τ

The demodulated output is given by

kf  
s0 (t) = 1 − e−t/τ u(t) (4.84b)
k
 Exponential CW Modulation 41

low τ
s(t) = u(t)

kf k

0.63 kf k s0(t)

O τ t

Fig. 4.29(a) Demodulated output for a step input signal.

and is shown in Fig. 4.22(a). The PLL output is thus a distorted version of the input signal. If the
line constant of the loop (τ ) is made small (that is, large loop gain), the demodulated output becomes
closer to the information signal u(t).

4.11 Stereophonic FM Broadcast: Transmission and Reception

The FM broadcast band extends from 88 MHz to 108 MHz (assigned by the FCC) with an adjacent
channel separation of 200 kHz. The instantaneous maximum frequency deviation is limited by the
FCC to 75 kHz. In monophonic commercial FM broadcasts, a single audio baseband signal is
transmitted and the receiver reproduces the original sound via the loudspeaker. But in stereophonic
FM broadcasts, two audio baseband signals are transmitted, and at the receiver the signals are applied
to two independent loudspeakers. The microphones at the studio are physically separated as well.
The speakers are also positioned some distance apart. This arrangement helps the listener to get a feel
of the `natural' sound; this is known as the stereophonic effect. For AM broadcasting, the maximum
frequency of the information signal is set at 5 kHz in contrast to 15 kHz set for FM broadcasting.
Inclusion of higher spectral components accounts for the improved tonal quality of FM reception
(closer to the baseband) compared to AM. If one uses narrowband FM (commonly used in two-way
wireless communications), the system requires a bandwidth of 2fm = 30 kHz although the bandwidth
allocated to each FM station is much larger, i.e., 200 kHz. This avoids interference from the adjacent
stations.
The monophonic FM broadcast is well established and there are millions of monophonic FM radio
receivers. To facilitate their continued usability, the FCC has framed a rule that the stereophonic FM
transmission should be compatible with the original monophone systems so that the old radio sets
would continue to receive the monophonic version of the stereophonic transmission.

4.11.1 FM stereo
In commercial monaural FM broadcast systems, a single 50 Hz to 15 kHz audio channel makes
up the entire information spectrum. This modulates the high frequency carrier and is transmitted
through an FM communication channel with a bandwidth of 200 kHz. The receiver reproduces the
42 Principles of Electronic Communications { Analog and Digital

information signal which appears to have originated from a point source. No spatial separation can
be reproduced so that the sound looks like that from an extended source. Stereo FM transmission
became commercially available in 1961, which made spatial separation of information signals possible.
During transmission, the message signal is spatially divided into two 50 Hz audio channels, namely a
left audio channel L and a right audio channel R. These two channels are transmitted simultaneously
and independently. Two speakers at the receiver reproduce the information signal.
As shown in the Fig. 4.30, the two channels L and R are passed through a matrix which produces
two audio channels, L + R and L − R. The L + R audio channel, 50 Hz to 15 kHz, modulates the
carrier for monaural transmission. The difference signal (L − R) modulates a subcarrier of 38 kHz
and produces the L − R stereo channel occupying the frequency band from 23 kHz−53 kHz. These
two channels, L + R monaural and L − R stereo, are combined in a combining network and fed to
the FM transmitter. In a stereo receiver, the L − R channel is added to L + R channel to produce
the L channel; the difference produces the R-channel.
FM stereo basically multiplexes two independent audio channels. Consider the system shown in
Fig. 4.31 that multiplexes two signals s1 (t) and s2 (t), each bandlimited to 15 kHz.
As shown in the figure, we amplitude modulate a carrier at 38 kHz with s2 (t) and then add s1 (t)
to it. The combined signal is then s1 (t) + s2 (t) cos 4πfc t. Note that the individual spectrum of
s1 (t) and s2 (t) are still preserved. The signal s2 (t) cos 4πfc t occupies the frequency band 2fc ± fm ,
that is, from 23 to 53 kHz. If we frequency-modulate another carrier with this combined signal, the
resulting NBFM wave uses only 106 kHz of the assigned 200 kHz.
Figure 4.32 depicts the FM stereo receiver for recovering the baseband signals s1 (t) and s2 (t).
The signal s1 (t) is easily recovered by a lowpass filter with cutoff frequency fm = 15kHz. Although
the spectrum of s2 (t) is located at 2fc = 38 kHz, we cannot recover s2 (t) by a bandpass filter; it
requires a synchronous demodulator. The reason for this is that for envelope detection to yield s2 (t),
the carrier frequency of the envelope must be higher than the frequency of the envelope.

Fig. 4.30 FM stereo transmitter

 Exponential CW Modulation 43

Fig. 4.31

Fig. 4.32 FM stereo and mono receiver

We do not need to generate the carrier at 38 kHz in synchronous detection. Figure 4.30 shows
that a carrier at 19 kHz is also being transmitted. This is separated at the receiver using a bandpass
filter centred at 19 kHz with a bandwidth no more than 4 kHz. The carrier at 19 kHz is doubled and
used to down-convert the spectrum so that an LPF can reconstruct the signal s2 (t).
In the above discussions, s1 (t) represents the L + R channel, compatible with mono-receiver, and
s2 (t) represents the L − R channel. These two are combined in a stereo decoder to produce L and
R independently. These are then applied to the respective speakers.

.
44 Principles of Electronic Communications { Analog and Digital

4.12 Power Relations in FM waves

An FM signal contains infinite number of sidebands. It is important to know the distribution of
power among the sidebands. The effective bandwidth of the FM signal is calculated on the basis
of the significant power content in the sidebands. In amplitude modulation, it has been derived that
the total power of the modulated wave is more than the carrier power and the modulating signal
contributes power. In contrast, the total power in an angle-modulated signal is the same as the power
of the unmodulated carrier. So the distribution of power among the sidebands comes from the carrier
power. The average power of the modulated carrier remain the same, being independent of the depth
of modulation.
For the modulated signal of (4.1), the total power is

Pt = sc (t)/R

A2c
= cos2 [ωc t + θ(t)]
R

A2c
= {1 + cos[2ωct + 2θ(t)]}. (4.85)
2R

Here R is the resistance in which power is dissipated. The modulating signal is impressed upon θ(t).
The second term of (4.85) consists of an infinite number of sinusoidal side frequencies. The average
value of the second term is thus zero. So, the total power in the modulated wave is

A2c
Pav = , (4.86)
2R
which is the average power of the unmodulated carrier. Again, the modulated carrier power is the
sum of all sideband powers. Following (4.22), the total power is then

J02 (β)
Pt = + J12 (β) + J22 (β) + . . . + Jn2 (β) + . . ., (4.87)
2
where the first term gives the modulated carrier power.

4.13 Frequency-Division Multiplexing (FDM)

Frequency division multiplexing is the process of combining transmissions from several commu-
nication channels and transmitting them on one transmission facility (such as metallic or optical
fibre cable, radio-frequency channel etc.). FDM is used extensively for combining many narrowband
sources to form a wideband channel. Because of the bandwidth and power requirement considerations,
single-sideband modulation is the most common application in the FDM systems.
The principle of combining several input messages in the FDM method is illustrated in Fig. 4.33,
where N input channels x 1 (t), x2 (t), . . . , xN (t) modulate the subcarriers fc1 , fc2 , . . . , fcN in SSB
 Exponential CW Modulation 45

scheme. The individual channels are first band limited by the low-pass filters. The modulator signals
are then summed in a linear manner to produce the baseband signal xb (t). The band pass filter used
after SSB modulation is to allow only one of the sidebands.

X1(f)
x1(t) (0 − w) SSB
LPF BPF

−W1 0 W1 f ∼ fc1

X2(f) xc(t)

x2(t) (0 − w) SSB xb(t)

LPF BPF Linear Carrier
summer Mod
−W2 0 W2 f
∼ fc2

∼ fc
XN(f)

xN(t) (0 − w) SSB
LPF BPF
−WN 0 WN f
∼ fcN

Fig. 4.33 SSB FDM system block diagram

The subcarrier frequencies fc1 , fc2 , . . . are properly chosen so that the composite wide spectra
include non-overlapping spectra of individual channels, as in Fig.4.34. The baseband signal xb (t)
thus formed can be transmitted directly or it can be used to modulate another carrier of frequency fc ,
as shown in the figure to form xc (t). If the subcarrier frequencies are not separated enough, the tail of
one message spectrum may couple with other adjacent message spectra. To avoid that, the modulated
spectra are spaced out in frequency by guard bands. If we select the upper SSB modulation, the
guard band frequency is calculated as follows: (vide Fig. 4.31)

guard band
xb(f)

0 fc1 fc1 + w fc2 fcN f

fc2 + w

Fig. 4.34 Baseband spectra of FDM

46 Principles of Electronic Communications { Analog and Digital

U-SSB spectra of Message 1 range from fc1 to fc1 + W .

U-SSB spectra of Message 2 range from fc2 to fc2 + W .
So, guard band = fc2 − (fc1 + W ).
The baseband bandwidth of the composite wideband channel is thus the sum of the message bandwidth
(N.W) plus the guard bandwidth.
The demodulation of FDM consists of three successive steps, namely, recovery of baseband sig-
nal xb (t), separation of subcarriers and finally recovery of the individual messages, as depicted in
Fig. 4.35.

xc1(t)
BPF Detector x1(t)

Central at fc1

xc(t) Carrier xb(t)

BPF Detector x2(t)
demid

Central at fc2

BPF Detector xN(t)

Central at fcN

Fig. 4.35 FDM Demodulation

A familiar example of FDM system are the commercial AM and broadcast, TV broadcast, telephone
and data communication system. We will illustrate here only two such examples, namely, the com-
mercial AM broadcast and the AT & T type L4 carrier system.

Commercial AM broadcast
This occupies the frequency spectrum from 535 kHz to 1605 kHz, a total of 1070 kHz bandwidth and
accepts the information signal of bandwidth ranging from 0 to 5 kHz (the signal may be voice grade
or numerical signals). The subcarrier frequencies of adjacent channels are separated by 10 kHz. Each
channel amplitude modulates a subcarrier and produces a 10 kHz signal around that subcarriers, as
in Fig.(4.36). Since a total AM bandwidth of 1070 kHz is available, these will be 1070/10 = 107
slots in the frequency domain.
 Exponential CW Modulation 47

xi(f)

0
5 kHz fc1 − 5 kHz fc1 fc1 + 5 kHz f

10 kHz

Fig. 4.36 Channel bandwidth.

The AT & T type L4 carrier system

The system has become the standards for long-distance communication in the United States. This
includes 3600 voice channels each of 4 kHz bandwidth, multiplexed together for transmission. The
voice channels are combined to form groups; groups are combined to form super-groups etc. The
digital hierarchy that we have used for T1 carrier system multiplexes the signals on TDM basic and
is an example of short-distance communication. A similar kind of hierarchy is also adopted here
in L4 system but on FDM basis. The voice band channel is actually band limited to approximately
a 300Hz to 3,000Hz frequency band, but for practical design consideration, we consider a 4 kHz
channel bandwidth.
The first level of FDM hierarchy is the formation of basic groups. The group multiplexes 12 input
voice channels and stacks them on subcarriers 64, 68, . . . , 108 kHz using lower SSB modulation.
The subcarrier frequencies for the channel bands are synthesised from a common oscillator and are
of multiples of 4 kHz. These are determined by the formula
fcin = 112 − 4nkHz, n = 1, 2, . . . , 12 (4.88)
for n = 12, fc = 1124×12 = 112 − 48 = 64 kHz and we use the lower SSB. Thus, for channel
no.12, the frequency band occupied is from 60 to 64 kHz. Channel no.1 has the subcarriers frequency
fc = 112 − 4 = 108 kHz and the spectrum occupies 104 to 108 kHz. In effect, the resulting group
spectrum is 48 kHz wide and ranges from 60 kHz to 108 kHz, as shown in Fig. (4.37).

1 Group
LPF 108
Multi-
12 11 1
plexer
60 64 68 104 108 f
2 (in kHz)
LPF 104
BW = −18 kHz

12
LPF 64

Fig. 4.37 Group multiplexing

48 Principles of Electronic Communications { Analog and Digital

The next higher level in the super-group in which five such groups are lower SSB modulated on the
subcarriers that differ by 48 kHz. The group carrier frequencies are derived from the formula:
fc = 372 + 48n kHz, n = 1, 2, . . . , 5. (4.89)
The first group is modulated on the subcarrier frequency
fc = 372 + 48 × 1 = 420 kHz and the 5t h group (n =5) on
fc = 372 + 48 × 5 = 372 + 240 = 612 kHz. As we consider the lower SSB modulation, the range
of the super-group spectrum can be calculated as follows:
Group 1: fout = 420 kHz − (60 kHz to 108 kHz) = 312 kHz to 360 kHz
Group 2: fout = 468 kHz − (60 kHz to 108 kHz) = 360 kHz to 408 kHz
Group 3: fout = 516 kHz − (60 kHz to 108 kHz) = 408 kHz to 456 kHz
Group 4: fout = 564 kHz − (60 kHz to 108 kHz) = 456 kHz to 504 kHz
Group 5: fout = 612 kHz − (60 kHz to 108 kHz) = 504 kHz to 552 kHz
The super-group spectrum thus occupies the frequency range from 312 kHz to 552 kHz, as shown in
Fig. (4.38). An interesting point in the figure is that the voice channel number 1 (104 − 108 kHz)
appears is an upper SSB signal (312 − 316 kHz). We also note that the subcarrier for Group 1 (420
kHz) falls in the Group 3 spectrum, subcarrier for Group 2 (468 kHz) falls in the Group 4 spectrum.

(60 − 108 kHz)

Group 1 Super group
420
multtiplexs
Group 2 sub-carrier BW = 552 − 312 = 240 kHz
468

Group 5
612

Group 1 Group 2 Group 5

}
}
}

1
312 360 408 504 552 f
(in kHz)

Fig. 4.38 Super-group multiplexing in FDM The third higher level in FDM hierarchy is the master-

group which is formed by combining 10 super-group spectra. In the bandwidth of each super-group
spectrum is 240 kHz. In between two super-groups, a guard band of 8 kHz is included. Table 1
lists all the carrier frequencies of the super-groups forming the master-group. Note that all the carrier
 Exponential CW Modulation 49

frequencies are 248 kHz apart from the preceding frequencies except the carrier frequency of the
super-group 7 which differs by 56 kHz. We now calculate the spectra of master-group as follows:

Table 4.1 Super-group carrier frequencies

Super-group Carrier frequency in kHz
1 1116
2 1364
3 1612
4 1860
5 2108
6 2356
7 2652
8 2900
9 3148
10 3396

Super-group 1: fout = 1116 kHz − (312 kHz to 552 kHz) = 564 kHz to 804 kHz

Super-group 10: fout = 3396 kHz − (312 kHz to 552 kHz) = 2844 kHz to 3084 kHz

The spectra of master-group occupy the frequency band from 564 kHz to 3084 kHz, a bandwidth of
2.52 MHz.

In a similar fashion, 6 master-groups can be combined to form a bigger group, called a jumbo group.
The final baseband spectra occupy the frequency range from 0.5 MHz to 17.5 MHz including the
guard bands. Table 2 summarizes the AT & T FDM hierarchy.

Table 4.2 AT & T FDM hierarchy: L4 carrier spectrum.

Frequency range Bandwidth No of voice band channel
Group 60−108 kHz 48 kHz 12
Super-group 312−552 kHz 240 kHz 60
Master-group 564−3084 kHz 2.52 MHz 600
Jumbo group 0.5−17.5 MHz 17 MHz 3600

4.14 Non-linear distortion and Interference in FM signals

Interference is the contamination of true signal by another similar signal. Severe interference may
lead to unsuccessful recovery of true message signal since FM signal is of constant amplitude and as
such, less affected by system nonlinearities. Assume a generalised non-linear distortion dictated by
the input r(t) and output s(t) relation

s(t) = a0 + a1 r(t) + a2 r 2 (t) + · · · + an r n (t) (4.90)

Clearly, a0 adds to the dc value while a1 r(t) is the signal amplification by gain a1 ; the remaining
terms are non-linear distortion. For an angle-modulation, the signal is
50 Principles of Electronic Communications { Analog and Digital

sc (t) = Ac cos(ωc t+?(t)) = Ac cos φc (t) (4.91)

(vide Eq. (4.1)). The output of nonlinear system is given by

s (t) = a0 + a1 Ac cos (ωc t + θ (t)) + a2 A2c cos2 (ωc t + θ (t)) + · · · + an Anc cosn (ωc t + θ (t))

= c0 + c1 cos(ωc t + θ(t)) + c2 cos(2ωct + 2θ(t)) + · · · + cn cos(nωc t + nθ(t)) (4.92)

Since ωc is large, each harmonic in the above equation s(t) is separable in the frequency domain. So,
any band-pass filter centred at ωc and with bandwidth of FM (or PM) signal, can recover the desired
component cos(ωc t + θ(t)). Thus angle-modulated signals are immune to non-linear distortion.

In discussing interference, it may be caused due to multipath propagation. Since FM is of constant

amplitude, any amplitude change due to fading can be adjusted by automatic gain control. Of course,
if the received signal contains two or more similar unwanted signals in the frequency band, it requires
a little thought.

Let the receiver be turned to a particular carrier frequency ωc. The received signal r(t) consists of
unmodulated carrier plus other interfering carrier of slightly different frequency. Thus,

r(t) = Ac cos ωct + Ai cos((ωc + ωi )t + φi ) (4.93)

where Ai is the amplitude of the interfering signal of frequency (ωc +ωi ) and φi is the relative phase.
We may write the above equation as

r(t) = Ac cos ωct + Ai cos ωct cos(ωi t + φi) − Ai sin ωc t sin(ωi t + φi)

= [Ac + Ai cos(ωi t + φi)] cos ωct − Ai sin(ωi t + φi)] sin ωc t

Let Ac + Ai cos(ωi t + φi) = A(t) cos θ(t)

Ai sin(ωi t + φi) = A(t) sin θ(t)

so that

A2 (t) = A2c + A2i + 2Ac Ai cos(ωi t + ϕi )

p 
or A (t) = Ac · 1 + ρ2 + 2ρ cos (ωi t + φi ) 
ρ sin(ωi t + φi ) (4.94)
and θ (t) = tan−1
1 + ρ cos(ωi t + φi )


where ρ = Ai /Ac , a measure of interfering strength. Thus

r(t) = A(t) cos(ωc t + θ(t)) (4.95)

in envelope-and-phase form. This shows that the interfering signal produces both amplitude and
phase modulation. As a special case of low-level interference, ρ = Ai /Ac  1, we have amplitude

A (t) ≈ Ac (1 + ρcos(ωi t + ϕi ))
 Exponential CW Modulation 51

and phase

θ (t) ≈ ρ sin(ωi t + ϕi ))

Thus,

r (t) = [Ac (1 + ρcos(wi t + φi ))] · cos(ωc t + ρ sin(ωi t + ϕi )) (4.96)

The above equation resembles Equation (4.60). The time-varying amplitude is with AM mod-
ulation, with modulation index ? and the carrier undergoing phase or frequency modulation. On the
other hand, if ρ  1, then

A (t) ≈ Ac · [ρ2 + 2ρcos(ωi t + φi )]1/2

≈ Ac ρ [1ρ2 + ρ−1 cos(ωi t + ϕi )] = Ai [1 + ρ−1 cos(ωi t + ϕi )]

θ (t) ≈ ωi t + φi

r(t) = Ai (1 + ρ−1 cos(ωi t + ϕi )) · cos(ωc t + ωi t + ϕi ) (4.97)

Thus, the envelope looks like AM and phase of the carrier is shifted by ωi in addition to φ i .

Solved Numericals

1. In an FM system, the message signal is given by s(t) = 2 cos(2π500) volts and the frequency
deviation is ∆f = 5 kHz. If the carrier is given by vc (t) = 5 cos(2π × 106 t) volts, determine
the following
(a) Index of modulation, and
(b) The FM Waveform
Repeat the problem if the amplitude of the modulating signs is increased to 3 volts.
Solution
By Eq. (4.16), we know the FM signal as
sFM (t) = Ac cos [2π fc t + β sin(2πfm t)]
∆f
where β modulation index β = and ∆f = kf Am where kf is a constant for FM.
fm
Here, Am = 2 volts, fm = 500 Hz, fc = 106 Hz, Ac = 5 volts; ∆f = 5 kHz
∆f 5kHZ
(a) β = = = 10
fm 500HZ
(b) sF M (t) = 5 cos(2π · 106 t + 10 sin 2π · 500t)volts
52 Principles of Electronic Communications { Analog and Digital

∆f 5 × 103 Hz
Since kf = , we have kf = = 2.5 × 103 V−1 S−1
Am 2volts
Further, if amplitude A m of the modulating signal is increased to 3 volts, ∆f = kf Am =
2.5 × 103 × 3 = 7.5 kHz and so
7.5 × 103
= 15 and sF M (t) = 5 cos 2π106 t + 15 sin(2π · 500t) Volts.


β=
500

2. In an FM system, the message signal is given by s(t) = 2 cos(2π500t) volts and the frequency
deviation is kp Am = 5 kHz. If the carrier is given by vc (t) = 5 cos(2π × 106 t) volts, answer
the following.

(a) Index of modulation, and

(b) The PM waveform

Repeat the problem if the amplitude of the modulating signals is increased to 3 volts.

Solution

By Eq. (4.9), we know the PM signal as

sP M (t) = Ac cos(2πfc t + kpAm cos ωm t)

where modulation index k is (kp Am ), and kp is a constant for PM.

Here Ac = 5 volts, Am = 2 volts, fc = 106 Hz

(a) The frequency deviation can be obtained from Eq. (4.10).

1 d 
∆f = kp [s(t)]max
2π dt
1 d 
= kp (Am cos 2πfm t)max
2π dt
1
= kp · (−Am.2πfm ) = kp fm .Am (magnitude only)
2π
The constant kp is obtained as

∆f 5 × 103 Hz
kp = = = 5V−1
fm Am 500 × 2Volt Hz
and modulation index kp Am = 5 × 2 = 10

(b) The PM waveform is given as

sPM (t) = 5 cos(2π · 106 t + 10 cos 2π500t) volts when the amplitude of the modulating signal
is increased to 3 volts, then kpAm = 5 × 3 = 15 = modulation index.
 Exponential CW Modulation 53

The waveform is then

sPM (t) = 5 cos(2π · 106 t + 15 cos 2π500t) volts

3. An angle modulated signal is described by

sF M (t) = 5 cos(2π × 106 t + 2 sin 2π500t + 3 sin 2π1000t) volts.

Find

(a) Power dissipation

(b) Frequency and phase deviation

(c) Modulation index

(d) Bandwidth of the signal

Solution:

(a) Since the carrier amplitude is AC = 5 volts, normalised power dissipation will be

AC 2 52
p= = = 12.5 watt.
2 2
(b) To find frequency deviation, the instantaneous frequency is given by
1 
2π · 106 + 4π × 500cos(2π500t) + 6π × 1000cos(2π1000t)

f(t) =
2π
 6 
= 10 + 1000cos(1000πt) + 3000cos(2000πt)

So, deviation from carrier frequency is

1000cos(1000πt) + 3000cos(2000πt)

Since these two sinusoids will add in phase at some time, the maximum frequency deviation
would be

∆f = 1000 + 3000 = 4000 Hz

The phase deviation can be calculated as follows

sF M (t) = 5cos(ωc t + θ(t))

where the phase term θ(t) = 2 sin(2π500t) + 3 sin(2π1000t)

Maximum value of this phase term is 2 + 3 = 5 radian and this is the phase deviation ∆θ.

(c) The modulation index is

∆f 4000
β= = =4
fh 1000
54 Principles of Electronic Communications { Analog and Digital

(we have taken the highest modulating frequency)

(d) By Carson's formula (Eq. 4.26), signal bandwidth is B = 2fm(1+β) = 2×1000(1+4) =

10 kHz.

4. With reference to the FET reactance modulator of Fig.4.17, the effective capacitance of the
FET is n-times larger than the gate-to-source impedance. Find the expression for the effective
capacitance Cm . If the FET transconductance varies from 300 µs to 800 µs over the p linear
portion of the gm ugs curve, find the minimum and maximum effective capacitance of the
modulator. Assume that resistance R is one-tenth of the capacitive reactance and the resonant
frequency is 100 MHz

Solution:

By Eq. (4.67), we may write

1
Xc = = nR
Wc
1 1
Or, = nR, thus C = .
2πf0 C 2πF0 nR
Since Cm = gm RC, we have controlled capacitance
gm
Cm = .
2πf0 n
So, minimum effective capacitance
300 × 10−6 3
Cm,min = 6 = × 10−12 = 0.0477 pF.
2π × 100 × 10 × 10 20π
Maximum effective capacitance
800 × 10−6
Cm,max = = 0.1273 pF.
2π × 100 × 106 × 10

5. If the FET reactance modulator is to offer a maximum frequency deviation of 75 kHz for the
carrier frequency of f0 = 100 MHz, find the fixed capacitance C and the inductance L of the
tank circuit across which the FET is connected. Assume data of Example 4.

Solution:

At the minimum effective capacitance Cm,min , the frequency of oscillation is

1
fmax = p
2π L(C + Cm,min
and that at the maximum effective capacitance Cm,max
 Exponential CW Modulation 55

1
fmin = p
2π L(C + Cm,max )
Thus,
f 2 max C + Cm1 max
2
=
f min C + Cm1 min
Or,
 
(fmax − fmin )(fmax + fmin ) Cm,max − Cm,min
=
f 2min C + Cm,min
Or,
2(∆f) · 2f0 Cm,max − Cm,min
2 ≈
fmin C + Cm,min
(since fmax = f0 + ∆f, fmin = f0 − ∆f and fmin ≈ f0 , assumed)

Or,
Cm,max − Cm,min · f0
C + Cm,min =
4∆f
Or,
Cm,max − Cm,min · f0
C= − Cm,max
4∆f
Substituting the values,
(0.1273 − 0.0477) × 10−12 × 100 × 106
C= − 0.0477 × 10−12
4 × 75 × 103
= 26.533pF − 0.0477pF = 26.48pF
1 1
Also, f = √ ≈ p
2π LC 2π L(C + Cav )
(we assume the average value of the controlled capacitor Cm over the range, which is
Cm,max + Cm,max
= 0.0875 pF)
2
1
= √ , as C  Cav
2π LC
Thus,
1 1
L= = = 9.56 µH.
4π2f02 C 2 6 2
4π × (100 × 10 ) × 26.48 × 10−12
56 Principles of Electronic Communications { Analog and Digital

Exercises

4.1 An FM signal carrier amplitude of 2.5 volts has an instantaneous frequency ω(t) given by
f(t) = 108 + 104 cos(2π × 103 t). Find the modulation index. Write the expression for the
modulated waveform.
4.2 An angle modulated signal is described by s(t) = 10 cos(2π × 106 t + 0.01 sin 2π × 103 t).

(a) What is the instantaneous frequency f(t) of s(t)?

(b) Estimate the bandwidth of s(t).
(c) Find the modulating signal if s(t) is an FM signal.
(d) Repeat (c) if s(t) is a phase modulated signal.
4.3 An angle modulated signal is described by v(t) = 50 cos(2π × 108 t + 20 cos 2π × 103 t) volts.
(a) Find the modulating signal if the phase deviation constant kp = 100 rad/volt.
(b) Find the modulating voltage if the frequency deviation constant kf = 106 rad/volt second.
4.4 A single-tone FM signal has the frequency deviation ∆f = 75 kHz. Find the frequencies
of the modulating signal at which the carrier will be absent, given that J0 (β) = 0 for β =
2.405, 5.520, 8.654.
∆f
Hint: From (4.22) we find that the carrier frequency is absent for J 0 (β) = 0. Here β = fm
75kHz 75kHz
or, fm = ∆f
β ; f1 = 2.405 = ; f2 = = , etc.
5.520
4.5 In FM, the carrier is sinusoidal-modulated by a modulating signal of frequency 3 kHz. What
is the bandwidth occupied by the modulated wave if the frequency deviation is 5 kHz? The
amplitude of the modulating signal is increased by three times and its frequency is also lowered
by the same factor. What will be the new bandwidth?
4.6 The NBFM signal is given by (vide (4.36)) s(t) = Ac [cos ωct − km(t) sin ωc t]. Find an
expression for the instantaneous frequency.
4.7 A single tone modulating signal is applied to a frequency modulator and a phase modulator
separately. The two output spectra are found identical. Identify the spectra with the following
tests: (a) amplitude of the modulating signal is increased/decreased. (b) frequency of the
modulating signal is increased/decreased.
4.8 Discuss the scheme for detecting φ FM signal in an envelope detector. (Hint: FM-to-AM
conversion by time derivative)
4.9 Let sP m (t) = Ac cos [ωc t + kp s(t)] be the phase-modulated signal. If s(t) has a jump discon-
tinuity, then the phase term k p s(t) must be restricted to the range (−π, π); otherwise, reception
will be ambiguous. Discuss this effect by taking s(t) to be a digital signal.
4.10 The narrow-band FM (NBFM) signal is defined by Eq. (4.29).
sNBF M (t) = Ac [cosωc t − k s(t)sinωc t] .
Explain the generation scheme for the same. Estimate the bandwidth of the NBFM signal.
 Exponential CW Modulation 57

Review questions
1. (a) What is pre-emphasis? Why is it used? Draw a typical pre-emphasis circuit and explain
why de-emphasis must also be used .
(b) A 20 MHz carrier is modulated by 1 kHz audio sine wave. If the carrier voltage is 5 V
and the maximum deviation is 12.5 kHz, write the equations of this modulated wave for
FM and PM. If the modulating frequency is now changed to 3 kHz, the other remaining
constant, write the new equations for FM and PM.
(c) The block diagram shown below is a typical narrowband FM system where sin ωc t is the
carrier and m(t) = β sin ωm t is the modulating signal. What will be the expression and
type of the output signal? (KU 2000)

2. (a) Explain why modulation is needed for communications purpose. (b) Derive the mathe-
matical expression for a frequency and phase modulated waves. (c) What is the basic
difference between frequency and phase modulation? (d) A 500 Hz modulating voltage
fed into a PM generator produces a frequency deviation of 2.25 kHz. What is the modu-
lation index? If the amplitude of the modulating voltage is kept constant but its frequency
raised to 60 kHz, what is the new deviation? (KU 2001)

3. (a) Derive an expression for an FM signal.

(b) Define modulation index for signal-tone FM signal. Under what conditions do FM signals
become narrowband FM?
(c) A 25 MHz carrier is modulated by a 400 Hz audio sine wave. If the carrier voltage is
4V and the maximum deviation is 10 kHz, write an expression for the modulated wave
for (i) FM and (ii) PM. (KU 2001)

4. (a) What are the advantages of the frequency modulation over the amplitude modulation?
(b) How are the frequency modulation and phase modulation related?
(c) What is the advantage of wideband frequency-modulated signal over the narrowband
frequency-modulated signal?
(d) Obtain the peak frequency deviation (∆f) and modulation index for an FM modulator with
a deviation sensitivity k = 5 kHz/V and a modulating signal vm (t) = 2 cos(2π × 2000t).
(WBUT 2003)
58 Principles of Electronic Communications { Analog and Digital

5. (a) Explain a method of FM generation. (b) Discuss exchange of bandwidth and SNR for a
FM system.

6. (a) A tone-modulated signal is given by xc (t) = A cos(ωc t + β sin ωm t). When β << 1, we
get narrowband (NB) angle modulation.
(i) Find the spectrum of this NB angle-modulated signal.
(ii) Compare the result with that of a tone-modulated AM signal.
(iii) Discuss the similarities and differences by drawing their phasor representations.
(b) Explain the differences between frequency modulation and phase modulation. Begin with
the definitions of each type and the meaning of modulation index in each case. (WBUT
2004)

7. (a) With the help of a block diagram, describe the indirect (Armstrong) method of generating
FM signal.
(b) In an Armstrong-type FM generator, the crystal oscillator frequency is 200 kHz. The
maximum phase deviation is limited to 0.2 to avoid distortion. Let the modulating fre-
quency fm range from 50 Hz to 15 kHz. The carrier frequency at the output is 108 MHz
and the maximum frequency deviation is 75 kHz. Select the multiplier and the mixer
oscillator frequencies.
(WBUT 2004)

8. Obtain an expression for the spectrum of an angle-modulated signal. What is the difference
between a standard AM signal and narrowband FM signal? An FM signal has sinusoidal
modulation with frequency fm = 10 kHz and modulation index β = 3.0. Find the transmission
bandwidth of the FM signal. What is the percentage of the total FM signal? What is the
percentage of the total FM signal power that lies within the Carson’s rule bandwidth?
(RPE 2004)

9. (a) Write the advantages and disadvantages of amplitude modulation over frequency modu-
lation.
(b) A carrier of frequency 50 kHz and amplitude 5 volts is frequency modulated by a sinu-
soidal modulating wave of frequency 1 kHz and amplitude 10 volts with kf = 10 Hz/V.
Write the expression for the FM wave. Find the approximate band of frequencies occupied
by FM.
(c) Find the bandwidth of narrow-band FM. Also draw the frequency spectra for the same.
(MUST − Spring 2010)

10. What may be the possible reasons for FM reception to have higher received signal power than
a comparable AM reception?

Multiple-choice questions
1. When the transmission bandwidth is doubled in FM, then the SNR is
(a) doubled (b) improved four-fold (c) decreased by four-times (d) unaffected
 Exponential CW Modulation 59

2. An FM signal with a frequency deviation δ is passed through a mixer. The output of the mixer
will have the frequency deviation

(a) 5δ (b) δ/5 (c) δ (d) indeterminate

3. In FM broadcasting, the maximum frequency deviation is usually

(a) 10 kHz (b) 75 kHz (c) 200 kHz (d) 1.0 MHz

4. Varactor diode can be used for the generation of

(a) FM (b) AM (c) PM (d) both AM and FM

5. Pre-emphasis in FM systems is used for

(a) amplifying high frequency components of the modulating signal
(b) amplifying low frequency components of the modulating signal
(c) compressing the modulating signal
(d) suppressing noise components

6. If the index of modulation of an FM signal is increased, the power in carrier component

(a) increases (b) decreases (c) remains the same (d) nothing can be said

7. Two carrier components of frequencies 40 MHz and 80 MHz are frequency-modulated by a

modulating frequency of 4 kHz in such a way that the bandwidths of the FM signals are the
same. The peak frequency deviation in the two cases are

(a) 1:1 (b) 1:2 (c) 1:4 (d) 2:1

8. An FM signal with index of modulation β is passed through a frequency-doubler network. The

modulation index of the output signal is

(a) β (b) 2β (c) 3β (d) 4β.

9. In FM, the modulation index β is related to the modulating frequency fm as

2 2
(a) β ∝ fm (b) β ∝ (1/fm ) (c) β ∝ fm (d) β ∝ (1/fm )

10. In FM broadcast, the maximum modulation frequency is restricted to

(a) 5 kHz (b) 10 kHz (c) 15 kHz (d) 20 kHz

11. A 2.5 V, 500 Hz signal frequency modulates a carrier to cause a frequency deviation of 5 kHz.
The modulation index is
(a) 5 (b) 10 (c) 25 (d) 50

12 If the depth of modulation is doubled in FM, the output power

√ √
(a) increases by a factor of 2 (b) increases by a factor of 3
(c) doubles (d) remains at unmodulated value
60 Principles of Electronic Communications { Analog and Digital

13. One drawback of FM over AM is that

(a) noise is very high for high modulation frequencies
(b) larger bandwidth is required
(c) higher modulating power is required
(d) higher output power is required
14. Practical bandwidth of a very wideband FM approximately equals
(a) fm (b) 2fm (c) fd (d) 2fd .
15. The power distribution in FM sidebands is given by the Bessel function components Jn (β).
The number of significant sidebands (' 98% or more power content) always occurs for
(a) n = β (b) n > β (c) n > β + 1 (d) none