Generation and Demodulation of

Frequency Modulated Signals

Gaurav S. Kasbekar
Dept. of Electrical Engineering
IIT Bombay
 Introduction
• Consider a carrier wave:
1) 𝑐 𝑡 = 𝐴𝑐 cos(2𝜋𝑓𝑐 𝑡),
where 𝐴𝑐 > 0 is “carrier amplitude” and
𝑓𝑐 > 0 is “carrier frequency”
• Let 𝑚(𝑡) be message signal, which is a baseband
signal
• Frequency modulated (FM) wave given by:
𝑡
2) 𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
where 𝑘𝑓 > 0 called “frequency sensitivity”
• We will study circuits for generating FM signals and
demodulating them
 Narrow-band FM (NBFM) Generation
• Recall: FM signal given by:
𝑡
1) 𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
𝑡
• Let 𝑎 𝑡 = ‫׬‬0 𝑚 𝜏 𝑑𝜏
• Recall: when 𝑘𝑓 𝑎 𝑡 ≪ 1 for all 𝑡, the signal 𝑠(𝑡) called
narrow-band FM signal
• NBFM signal can be
approximated by:
2) 𝑠 𝑡 ≈ 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 −
𝐴𝑐 2𝜋𝑘𝑓 𝑎(𝑡) sin 2𝜋𝑓𝑐 𝑡
• Signal in 2) can be
generated using:
 a DSB-SC modulator as
Ref: B.P. Lathi, Z. Ding, “Modern Digital and Analog
shown in fig. Communication Systems”, 4th ed.
 Distortion in NBFM Signal
• Recall: NBFM signal can be approximated by:
1) 𝑠 𝑡 ≈ 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 − 𝐴𝑐 2𝜋𝑘𝑓 𝑎(𝑡) sin 2𝜋𝑓𝑐 𝑡
• Due to above approximation, there is some distortion in NBFM
signal generated using above scheme
• In terms of envelope and phase, relation in 1) can be rewritten as:
 𝑠 𝑡 ≈ 𝑎 𝑡 cos 2𝜋𝑓𝑐 𝑡 + 𝜙(𝑡)
 where:
o 𝑎 𝑡 = 𝐴𝑐 1 + 4𝜋 2 𝑘𝑓2 (𝑎 𝑡 )2
o 𝜙 𝑡 = tan−1 2𝜋𝑘𝑓 𝑎(𝑡)
• Ideally, envelope 𝑎 𝑡 should be a constant
 since it is not constant, “amplitude distortion” is said to occur
• Ideally, phase should be 𝜙 𝑡 = 2𝜋𝑘𝑓 𝑎(𝑡)
 since 𝜙 𝑡 = tan−1 2𝜋𝑘𝑓 𝑎(𝑡) , “frequency distortion” said to occur
• However, note that since 𝑘𝑓 𝑎 𝑡 ≪ 1, above distortions are
small
 Frequency Translation
• We will discuss generation of a wide-band FM (WBFM) signal
• One of the components needed for this is a “frequency
translator”
also known as “frequency converter” or “frequency mixer”
• Suppose we have a modulated signal 𝑠1 (𝑡), whose spectrum
is centred on carrier frequency 𝑓1 :
𝑠1 𝑡 = 𝑎 𝑡 cos 2𝜋𝑓1 𝑡 + 𝜙(𝑡)
• Using 𝑠1 𝑡 , we want to generate a modulated signal 𝑠2 (𝑡),
whose spectrum is centred on carrier frequency 𝑓2 :
𝑠2 𝑡 = 𝑎 𝑡 cos 2𝜋𝑓2 𝑡 + 𝜙(𝑡)
• How can we generate 𝑠2 𝑡 from 𝑠1 𝑡 ?
by multiplying 𝑠1 𝑡 with 2cos 2𝜋𝑓𝑙 𝑡 and band-pass filtering the
product
where 𝑓𝑙 = 𝑓2 − 𝑓1 if 𝑓2 > 𝑓1 and 𝑓𝑙 = 𝑓1 − 𝑓2 if 𝑓1 > 𝑓2
 Frequency Multiplication
• Frequency multiplier is another of the components used in
generation of WBFM signal
• Consider signal of form:
1) 𝑠 𝑡 = 𝐴cos 2𝜋𝑓𝑐 𝑡 + 𝜙(𝑡)
• Frequency multiplier can be used to generate, from 𝑠 𝑡 ,
signal of form:
2) 𝑠 ′ 𝑡 = 𝑘cos 2𝜋𝑛𝑓𝑐 𝑡 + 𝑛𝜙(𝑡) ,
where 𝑛 is a positive integer and 𝑘 is a constant
• How can we generate signal in 2) with 𝑛 = 2 from that in 1)?
generate (𝑠 𝑡 )2 from 𝑠 𝑡 and band-pass filter it
• Generalization: how can we generate signal in 2) with
arbitrary 𝑛 from that in 1)?
generate (𝑠 𝑡 )𝑛 from 𝑠 𝑡 and band-pass filter it
 Indirect Method of Armstrong
• Recall: FM signal given by:
𝑡
𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
𝑡
• Let 𝑎 𝑡 = ‫׬‬0 𝑚 𝜏 𝑑𝜏
• Recall: when 𝑘𝑓 𝑎 𝑡 ≪ 1 for all 𝑡, the signal 𝑠(𝑡) called
NBFM signal; otherwise called WBFM signal
• In indirect method of Armstrong:
first NBFM signal is generated (possibly using DSB-SC modulator
method we discussed)
then NBFM signal converted to WBFM signal
• How can NBFM signal be converted to WBFM signal of
desired center frequency and frequency deviation?
 by using frequency multipliers and frequency converters
• Want to generate WBFM signal with: Example
 carrier frequency 𝑓𝑐 = 91.2 MHz and
 frequency deviation ∆𝑓 = 76.8 kHz
• We initially generate NBFM signal with ∆𝑓 = 25 Hz and then
convert it to required WBFM signal as shown in fig.
• Note that we need to multiply 25 Hz by 3072 to get 76.8 kHz
• However, if carrier frequency of 200 kHz multiplied by 3072,
then we get 614.4 MHz
• Hence, we need a frequency converter stage

Ref: B.P. Lathi, Z. Ding, “Modern Digital

and Analog Communication Systems”, 4th
ed.
 Example
• Want to design an Armstrong indirect FM modulator to
generate FM signal with:
carrier frequency 97.3 MHz and ∆𝑓 = 10.24 kHz
• An NBFM generator is available with:
𝑓𝑐1 = 20 kHz and ∆𝑓1 = 5 Hz
• Only frequency doublers can be used as multipliers
• Also, a local oscillator (LO) with adjustable frequency
between 400 and 500 kHz is available for frequency mixing
• Solution:
NBFM output is input to frequency multiplier with factor 16
Frequency multiplier output is input to mixer with frequency
440.15 kHz followed by band-pass filter that retains band around
760.15 kHz
Band-pass filter output is input to frequency multiplier with
factor 128
 Direct Generation of Wideband FM Signals
• In a voltage-controlled oscillator (VCO), frequency of output
sinusoid controlled by input voltage
• FM signal can be generated using a VCO with input being:
 message signal 𝑚(𝑡)
• Instantaneous frequency:
 𝑓𝑖 𝑡 = 𝑓𝑐 + 𝑘𝑓 𝑚(𝑡)
• VCO can be implemented:
 using an operational amplifier and comparator
• VCO can also be implemented:
 by varying the capacitance value according to 𝑚(𝑡) in an oscillator
containing inductor and capacitor
 e.g., Hartley oscillator, Colpitts oscillator
• Implementation of capacitor with variable capacitance:
 reverse-biased diode acts as capacitor whose capacitance varies with
bias voltage
 variable capacitor also known as “varicap”, “varactor” or “voltacap”
Direct Generation of Wideband FM Signals (contd.)
• In Hartley or Colpitts oscillator, frequency of oscillation given by:
1
 𝑓0 =
2𝜋 𝐿𝐶
• Suppose capacitance 𝐶 varied using 𝑚(𝑡):
 𝐶 = 𝐶0 − 𝑘𝑚(𝑡)
• Then 𝑓0 :
1

2𝜋 𝐿(𝐶0 −𝑘𝑚 𝑡 )
• 𝑓0 ≈:
1 𝑘𝑚 𝑡 𝑘𝑚 𝑡
 1+ when ≪1
2𝜋 𝐿𝐶0 2𝐶0 𝐶0
• So 𝑓0 = 𝑓𝑐 + 𝑘𝑓 𝑚(𝑡),
1
 where 𝑓𝑐 =
2𝜋 𝐿𝐶0
𝑘𝑓𝑐
 𝑘𝑓 =
2𝐶0
• Max. capacitance deviation:
 ∆𝐶 = 𝑘𝑚𝑝 , where −𝑚𝑝 ≤ 𝑚(𝑡) ≤ 𝑚𝑝
∆𝐶
• :
𝐶0
2∆𝑓
 ,
𝑓𝑐
 where ∆𝑓 = 𝑘𝑓 𝑚𝑝 (frequency deviation)
∆𝑓
• In practice, is small even in WBFM; hence distortion due to above approximation is small
𝑓𝑐
• So direct FM generation can be used to produce sufficient frequency deviation and does not
require much frequency multiplication
Demodulation of FM Signals
 Demodulation Using Differentiator
• Recall: FM signal given by:
𝑡
𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• If a differentiator circuit is available, how can we
demodulate FM signal?
by inputting 𝑠 𝑡 to differentiator and inputting its output to
envelope detector
• Output of differentiator:
𝑡
 𝐴𝑐 2𝜋𝑓𝑐 + 2𝜋𝑘𝑓 𝑚(𝑡) sin 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏 − 𝜋
• Since 𝑘𝑓 𝑚𝑝 < 𝑓𝑐 , envelope detector can be used to
recover 𝑚(𝑡) from output of differentiator
 Demodulation Using Slope Circuit
• Demodulation can be performed using “frequency discriminator”:
 consists of “slope circuit” followed by envelope detector
• Suppose we want to demodulate FM signal 𝑠(𝑡) of carrier frequency 𝑓𝑐 and
transmission bandwidth 𝐵𝑇
• Ideal slope circuit has following transfer function:
𝐵𝑇 𝐵𝑇 𝐵𝑇
𝑗2𝜋𝑎 𝑓 − 𝑓𝑐 + , 𝑓𝑐 − ≤ 𝑓 ≤ 𝑓𝑐 + ,
2 2 2
 𝐻1 𝑓 = 𝑗2𝜋𝑎 𝑓 + 𝑓 − 𝐵𝑇
, −𝑓𝑐 −
𝐵𝑇
≤𝑓≤
𝐵
−𝑓𝑐 + 2𝑇 ,
𝑐 2 2
0, else.
 where 𝑎 > 0 is a constant
• Want to find output, say 𝑠1 (𝑡), of slope circuit
when input is FM signal:
𝑡
 𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• Convenient to use:
 complex envelopes
• Complex transfer function of 𝐻1 (𝑓):
𝐵𝑇 𝐵𝑇 𝐵𝑇
෩1 𝑓 = ቐ 𝑗4𝜋𝑎 𝑓 + , − ≤𝑓≤ ,
𝐻 2 2 2
0, else.
Ref: “Communication
• Complex envelope of FM signal 𝑠 𝑡 : Systems” by S. Haykin
𝑡 and M. Moher, 5th ed
• Recall: Demodulation Using Slope
 Complex transfer function of 𝐻1 (𝑓):

෩ 𝑗4𝜋𝑎 𝑓 + , −
𝐵𝑇
≤ 𝑓 ≤ ,
Circuit
𝐵𝑇 (contd.) 𝐵𝑇
o 𝐻1 𝑓 = ቐ 2 2 2
0, else.
 Complex envelope of FM signal 𝑠 𝑡 :
𝑡
1) 𝑠ǁ 𝑡 = 𝐴𝑐 exp 𝑗2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• Fourier transform of complex envelope of output of slope circuit:
𝐵𝑇 𝐵𝑇 𝐵𝑇
𝑗2𝜋𝑎 𝑓 + ሚ
𝑆(𝑓), − ≤𝑓≤ ,
 𝑆ሚ1 𝑓 = ቐ 2 2 2
0, else.
• Complex envelope of output of slope circuit:
ǁ
𝑑𝑠(𝑡)
2) 𝑠1ǁ 𝑡 = 𝑎 𝑑𝑡
+ 𝑗𝜋𝐵𝑇 𝑠(𝑡)
ǁ
• By 1) and 2):
2𝑘𝑓 𝑡
 𝑠1ǁ 𝑡 = 𝑗𝜋𝐵𝑇 𝑎𝐴𝑐 1 + 𝑚(𝑡) exp 𝑗2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
𝐵𝑇
• Output of slope circuit:
2𝑘𝑓 𝑡 𝜋
 𝑠1 𝑡 = 𝜋𝐵𝑇 𝑎𝐴𝑐 1 + 𝐵𝑇
𝑚(𝑡) cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏 + 2
• How can we obtain the message signal 𝑚(𝑡) from 𝑠1 𝑡 ?
2𝑘𝑓
 using an envelope detector, since 𝑚(𝑡) < 1 for all 𝑡
𝐵𝑇
• Output of envelope detector:
2𝑘𝑓
 𝜋𝐵𝑇 𝑎𝐴𝑐 1 + 𝑚(𝑡)
𝐵𝑇
 Demodulation Using Slope Circuit
(contd.)
• Recall: output of envelope detector:
2𝑘𝑓
𝜋𝐵𝑇 𝑎𝐴𝑐 1 + 𝑚(𝑡)
𝐵𝑇

• DC bias term can be removed by:

inputting output of envelope detector to a
high-pass filter
 FM Stereo Multiplexing
• In stereo audio, audio signal split into two signals,
𝑚𝑙 (𝑡) and 𝑚𝑟 (𝑡):
sent to left headphone and right headphone respectively
• E.g.: 𝑚𝑙 (𝑡) and 𝑚𝑟 (𝑡) may correspond to a vocalist and
an accompanist
• FM stereo multiplexing enables transmission of the two
signals 𝑚𝑙 (𝑡) and 𝑚𝑟 (𝑡) using the same FM carrier
• FM stereo multiplexing must satisfy following
requirements:
1) The transmission must operate within allocated FM
channel
2) Has to be compatible with monophonic radio receivers
• Frequency deviation of FM signal must be set taking
requirement 1) into account
 FM Stereo Multiplexing (contd.)
• Let 𝑓𝑐 be a 19 kHz subcarrier
Note: this subcarrier different from the carrier used for FM
modulation
• FM stereo transmitter shown in Fig.
• Generated signal:
𝑚 𝑡 = 𝑚𝑙 𝑡 + 𝑚𝑟 𝑡 + ሾ𝑚𝑙 𝑡 −

• The signal 𝑚 𝑡 then frequency

modulates the main carrier to
produce transmitted signal
• The term 𝐾cos 2𝜋𝑓𝑐 𝑡 included
to:
 provide a reference for coherent
detection of difference signal
𝑚𝑙 𝑡 − 𝑚𝑟 𝑡 at receiver Ref: “Communication
Systems” by S. Haykin
 called “pilot signal” and M. Moher, 5th ed
 FM Stereo
• Recall: modulating signal: Multiplexing (contd.)
𝑚 𝑡 = 𝑚𝑙 𝑡 + 𝑚𝑟 𝑡 + ሾ𝑚𝑙 𝑡 −

• How does a
stereo receiver
process 𝑚 𝑡 ?
as shown in Fig.

Ref: “Communication
Systems” by S. Haykin
and M. Moher, 5th ed
 Zero-Crossing Detectors
• Want to demodulate FM signal:
𝑡
𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• First, FM signal is input to a hard limiter, i.e., circuit
with following input-output characteristic:
1, if 𝑣𝑖 𝑡 > 0,
𝑣𝑜 𝑡 = ቊ
−1, if 𝑣𝑖 𝑡 < 0.
• Output of hard limiter is:
square wave with time-varying pulse widths
• How should square wave be processed to achieve
FM demodulation?
• Square wave is input to a digital counter:
rate at which counter increments is directly proportional
to instantaneous frequency of 𝑠(𝑡)
 Phase-Locked Loop (PLL)
• System that can be used for FM demodulation
• Consists of following components connected together in form of feedback loop (see fig.):
 Multiplier
 Loop filter
 Voltage-Controlled Oscillator (VCO)
• Input to PLL is FM signal:
𝑡
 𝑠 𝑡 = 𝐴𝑐 sin 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• Initially, VCO adjusted such that when control voltage 𝑣(𝑡) is 0:
1) Frequency of VCO is exactly carrier freq. 𝑓𝑐 of unmodulated carrier wave
2) VCO output has a 90° phase-shift w.r.t. unmodulated carrier wave
• VCO output:
𝑡
 𝑟 𝑡 = 𝐴𝑣 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑣 ‫׬‬0 𝑣 𝜏 𝑑𝜏
 where 𝑘𝑣 is frequency sensitivity of VCO
• Objective: To recover message signal at VCO output, i.e., to ensure that:
 𝑣 𝑡 = 𝑐𝑚(𝑡), where 𝑐 is a constant

Ref: “Communication
Systems” by S. Haykin
and M. Moher, 5th ed
• Recall:
 𝑠 𝑡 = 𝐴𝑐 sin 2𝜋𝑓𝑐 𝑡 + 𝜙1 (𝑡) , Non-Linear Model of PLL
 𝑟 𝑡 = 𝐴𝑣 cos 2𝜋𝑓𝑐 𝑡 + 𝜙2 (𝑡) ,
𝑡 𝑡
 where 𝜙1 𝑡 = 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏 and 𝜙2 𝑡 = 2𝜋𝑘𝑣 ‫׬‬0 𝑣 𝜏 𝑑𝜏
• Multiplier output is low-pass filtered and 𝑒(𝑡) is output of LPF
• So 𝑒(𝑡):
 𝑘𝑚 𝐴𝑐 𝐴𝑣 sin 𝜙𝑒 (𝑡)
 where 𝜙𝑒 𝑡 = 𝜙1 𝑡 − 𝜙2 (𝑡) and gain of multiplier is assumed to be 2𝑘𝑚
• If loop filter is a linear system with impulse response ℎ(𝑡), then 𝑣(𝑡):
∞
 ‫׬‬−∞ 𝑒 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
• So we get the following integro-differential equation:
𝑑𝜙𝑒 𝑡 𝑑𝜙 𝑡
1 ∞
1) 𝑑𝑡
= 𝑑𝑡 − 2𝜋𝐾0 ‫׬‬−∞ sin 𝜙𝑒 (𝜏) ℎ 𝑡 − 𝜏 𝑑𝜏
 where 𝐾0 = 𝑘𝑣 𝑘𝑚 𝐴𝑐 𝐴𝑣
• Equation in 1) difficult to analyse due to non-linear term sin 𝜙𝑒 (𝜏)
• So next, we linearize the model to simplify the analysis

Ref: “Communication
Systems” by S. Haykin
and M. Moher, 5th ed
 Linear Model of PLL
• Recall:
𝑑𝜙𝑒 𝑡 𝑑𝜙1 𝑡 ∞
1) 𝑑𝑡
=
𝑑𝑡
− 2𝜋𝐾0 ‫׬‬−∞ sin 𝜙𝑒 (𝜏) ℎ 𝑡 − 𝜏 𝑑𝜏
• When 𝜙𝑒 𝑡 = 0, the PLL said to be in phase-lock
• When |𝜙𝑒 𝑡 | ≪ 1 for all 𝑡, PLL said to be in near phase-
lock
then we use the approx. sin 𝜙𝑒 (𝜏) ≈ 𝜙𝑒 (𝜏)
• Equation 1) reduces to:
𝑑𝜙𝑒 𝑡 ∞ 𝑑𝜙1 𝑡
2) 𝑑𝑡
+ 2𝜋𝐾0 ‫׬‬−∞ 𝜙𝑒 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 =
𝑑𝑡
• How can we analyze 2)?
take Fourier transforms on both sides

Ref: “Communication
Systems” by S. Haykin
and M. Moher, 5th ed
• Recall: Linear Model of PLL (contd.)
𝑑𝜙𝑒 𝑡 ∞ 𝑑𝜙1 𝑡
1) 𝑑𝑡
+ 2𝜋𝐾0 ‫׬‬−∞ 𝜙𝑒 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 = 𝑑𝑡
• Taking Fourier transforms, we get:
1
 Φ𝑒 𝑓 = 1+𝐿 Φ1 (𝑓),
𝑓
𝐻(𝑓)
 where 𝐿 𝑓 = 𝐾0 𝑗𝑓
𝐾0
• Recall: 𝑒(𝑡) ≈ 𝜙 (𝑡)
𝑘𝑣 𝑒
• 𝑉(𝑓):
𝑗𝑓𝐿(𝑓)
 Φ (𝑓)
𝑘𝑣 (1+𝐿 𝑓 ) 1
• How should 𝐿(𝑓) be chosen in order to ensure that 𝑣(𝑡) is approx. proportional
to 𝑚(𝑡)?
 We choose 𝐿(𝑓) such that 𝐿 𝑓 ≫ 1 in the baseband
• 𝑣(𝑡) ≈:
1 𝑑𝜙1
 2𝜋𝑘𝑣 𝑑𝑡
• So 𝑣(𝑡) ≈:
𝑘𝑓
 𝑚(𝑡)
𝑘𝑣
• Thus, frequency demodulation achieved
Ref: “Communication
Systems” by S. Haykin
and M. Moher, 5th ed
 Effect of Sending FM Signal over Non-linear
• Consider FM signal:
Channel
 𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 𝜙(𝑡)
𝑡
 where 𝜙 𝑡 = 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• Sent over a channel with following input-output characteristic:
 𝑣𝑜 𝑡 = 𝑎1 𝑣𝑖 𝑡 + 𝑎2 𝑣𝑖2 𝑡 + 𝑎3 𝑣𝑖3 𝑡
 where 𝑎1 , 𝑎2 and 𝑎3 are constants
1
• Recall the trigonometric identity: cos 3 𝜃 = 4 3cos(𝜃) + cos(3𝜃)
• Output of channel:
1 3 1 1
 𝑣𝑜 𝑡 = 𝑎2 𝐴2𝑐 + 𝑎1 𝐴𝑐 + 𝑎3 𝐴3𝑐 cos 2𝜋𝑓𝑐 𝑡 + 𝜙(𝑡) + 𝑎2 𝐴2𝑐 cos 4𝜋𝑓𝑐 𝑡 + 2𝜙(𝑡) + 𝑎3 𝐴3𝑐 cosሾ6𝜋𝑓𝑐 𝑡 +
2 4 2 4
 Effect of Sending AM Signal over Non-linear
Channel
• Consider AM signal:
 𝑠 𝑡 = 𝐴𝑐 1 + 𝑘𝑎 𝑚(𝑡) cos 2𝜋𝑓𝑐 𝑡
• Sent over a channel with following input-output characteristic:
 𝑣𝑜 𝑡 = 𝑎1 𝑣𝑖 𝑡 + 𝑎2 𝑣𝑖2 𝑡 + 𝑎3 𝑣𝑖3 𝑡
 where 𝑎1 , 𝑎2 and 𝑎3 are constants
• Output of channel:
1 3
 𝑣𝑜 𝑡 = 𝑎2 𝐴2𝑐 1 + 𝑘𝑎 𝑚(𝑡) 2
+ ቀ𝑎1 𝐴𝑐 1 + 𝑘𝑎 𝑚(𝑡) + 𝑎3 𝐴3𝑐 ሾ1 +
2 4
 Amplitude Variations in FM Signal
• Recall: FM signal:
𝑡
 𝑠 𝑡 = 𝐴𝑐 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚
𝜏 𝑑𝜏
• When FM signal sent over a channel, several factors (e.g., channel noise
and fading) can cause amplitude to vary
• So received signal is of form:
𝑡
 𝑟 𝑡 = 𝐴 𝑡 cos 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏
• If 𝑟 𝑡 directly input to FM demodulator:
 output would be distorted version of 𝑚(𝑡)
• E.g.:
 Recall that FM signal can be demodulated by inputting it to differentiator and
inputting output of differentiator to an envelope detector
 Output of this differentiator when 𝐴 𝑡 = 𝐴𝑐 :
𝑡
o 𝐴𝑐 2𝜋𝑓𝑐 + 2𝜋𝑘𝑓 𝑚(𝑡) sin 2𝜋𝑓𝑐 𝑡 + 2𝜋𝑘𝑓 ‫׬‬0 𝑚 𝜏 𝑑𝜏 − 𝜋
 When 𝐴 𝑡 is not a constant:
𝑑𝐴(𝑡)
o There is an additional term containing in output of differentiator
𝑑𝑡
o Even if this term is neglected, output of envelope detector is 𝐴(𝑡) 2𝜋𝑓𝑐 + 2𝜋𝑘𝑓 𝑚 𝑡 ;
hence, output is distorted version of 𝑚(𝑡)
• Hence, technique required to suppress variation in amplitude of received
signal 𝑟 𝑡 before inputting it to FM demodulator
 Amplitude Variations in FM Signal (contd.)
• Recall: received signal is of form 𝑟 𝑡 = 𝐴 𝑡 cos ቂ2𝜋𝑓𝑐 𝑡 +