This action might not be possible to undo. Are you sure you want to continue?
What is Angle Modulation?
In angle modulation, information is
embedded in the angle of the carrier.
We define the angle of a modulated carrier
by the argument of...
s t ( ) = A
c
cos u t ( ) ( )
Allanki Sanyasi Rao 3
Phasor Form
In the complex plane we have
t=1
t=0
t=3
Phasor rotates with nonuniform speed
Allanki Sanyasi Rao 4
Angular Velocity
Since phase changes nonuniformly vs.
time, we can define a rate of change
This is what we know as frequency
e
i
=
du
i
(t)
dt
s t ( ) = A
c
cos 2tf
c
t + 
c
u
i
t ( )

\

.

¬
du
i
dt
= 2tf
c
Allanki Sanyasi Rao 5
Instantaneous Frequency
We are used to signals with constant
carrier frequency. There are cases where
carrier frequency itself changes with time.
We can therefor talk about instantaneous
frequency defined as
f
i
t ( ) =
1
2t
du
i
t ( )
dt
Allanki Sanyasi Rao 6
Examples of Inst. Freq.
Consider an AM signal
Here, the instantaneous frequency is the
frequency itself, which is constant
s t ( ) = 1+ km(t)  cos 2tf
c
t + 
c
u
i
t ( )

\

.

¬
du
i
dt
= 2tf
c
Allanki Sanyasi Rao 7
Impressing a message on
the angle of carrier
There are two ways to form a an angle
modulated signal.
– Embed it in the phase of the carrier
Phase Modulation(PM)
– Embed it in the frequency of the carrier
Frequency Modulation(FM)
Allanki Sanyasi Rao 8
Phase Modulation(PM)
In PM, carrier angle changes linearly with
the message
Where
– 2πf
c
=angle of unmodulated carrier
– k
p
=phase sensitivity in radians/volt
s t ( ) = A
c
cos u
i
t ( ) ( ) = A
c
cos 2tf
c
t + k
p
m t ( )
( )
Allanki Sanyasi Rao 9
Frequency Modulation
In FM, it is the instantaneous frequency
that varies linearly with message
amplitude, i.e.
f
i
(t)=f
c
+k
f
m(t)
Allanki Sanyasi Rao 10
FM Signal
We saw that I.F. is the derivative of the
phase
Therefore,
f
i
t ( ) =
1
2t
du
i
t ( )
dt
u
i
t ( ) = 2tf
c
t + 2tk
f
m t ( )
0
t
}
s t ( ) = A
c
cos 2tf
c
t + 2tk
f
m(t)dt
0
t
}
¸
(
¸
(
Allanki Sanyasi Rao 11
FM for Tone Signals
Consider a sinusoidal message
The instantaneous frequency
corresponding to its FM version is
m(t) = A
m
cos 2tf
m
t ( )
f
i
t ( ) = f
c
+ k
f
m(t)
= f
c
resting frequency
+ k
f
A
m
cos 2tf
m
t ( )
Allanki Sanyasi Rao 12
Illustrating FM
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
1
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
FM
message
Inst.frequency
Moves with the
Message amplitude
Allanki Sanyasi Rao 13
Frequency Deviation
Inst. frequency has upper and lower
bounds given by
f
i
t ( ) = f
c
+ Af cos 2tf
m
t
( )
where
Af = frequency deviation= k
f
A
m
then
f
i
max
= f
c
+ Af
f
i
min
= f
c
÷ Af
Allanki Sanyasi Rao 14
FM Modulation index
The equivalent of AM modulation index is
 which is also called deviation ratio. It
quantifies how much carrier frequency
swings relative to message bandwidth
 =
Af
W
baseband
or
Af
f
m
tone
Allanki Sanyasi Rao 15
Example:carrier swing
A 100 MHz FM carrier is modulated by an
audio tone causing 20 KHz frequency
deviation. Determine the carrier siwng
and highest and lowest carrier frequencies
Af = 20KHz
frequency swing = 2Af = 40KHz
frequency range:
f
high
=100MHz + 20KHz = 100.02MHz
f
low
=100MHz ÷ 20KHz = 99.98MHz
Allanki Sanyasi Rao 16
Example: deviation ratio
What is the modulation index (or deviation
ratio) of an FM signal with carrier swing of
150 KHz when the modulating signal is 15
KHz?
Af =
150
2
= 75KHz
 =
Af
f
m
=
75
15
= 5
Allanki Sanyasi Rao 17
Myth of FM
Deriving FM bandwidth is a lot more
involved than AM
FM was initially thought to be a bandwidth
efficient communication because it was
thought that FM bandwidth is simply 2Af
By keeping frequency deviation low, we
can use arbitrary small bandwidth
Allanki Sanyasi Rao 18
FM bandwidth
Deriving FM bandwidth is a lot more
involved than AM and it can barely be
derived for sinusoidal message
There is a graphical way to illustrate FM
bandwidth
Allanki Sanyasi Rao 19
Piecewise approximation of
baseband
Look at the following representation
1/2W
Baseband bandwidth
=W
Allanki Sanyasi Rao 20
Corresponding FM signal
FM version of the above is an RF pulse for
each square pulse.
The frequency of the kth RF pulse at t=t
k
is
given by the height of the pulse. i.e.
f
i
= f
c
+ k
f
m t
k
( )
Allanki Sanyasi Rao 21
Range of frequencies?
We have a bunch of RF pulses each at a
different frequency.
Inst.freq corresponding to square pulses
lie in the following range
f
i
max
= f
c
+ k
f
m
max
f
i
min
= f
c
+ k
f
m
min
m
min
m
max
Allanki Sanyasi Rao 22
A look at the spectrum
We will have a series of RF pulses each at
a different frequency. The collective
spectrum is a bunch of sincs
f
highest
lowest
4W
Allanki Sanyasi Rao 23
So what is the bandwidth?
Measure the width from the first upper
zero crossing of the highest term to the
first lower zero crossing of the lowest
term
f
highest
lowest
Allanki Sanyasi Rao 24
Closer look
The highest sinc is located at f
c
+k
f
m
p
Each sinc is 1/2W wide. Therefore, their
zero crossing point is always 2W above
the center of the sinc.
f
2W
Allanki Sanyasi Rao 25
Range of frequenices
Above range lies
<f
c
k
f
m
p
2W,f
c
+k
f
m
p
+2W>
f
highest
lowest
Allanki Sanyasi Rao 26
FM bandwidth
The range just defined is one expression
for FM bandwidth. There are many more!
B
FM
=4W+2k
f
m
p
Using =∆f/W with ∆f=k
f
m
p
B
FM
=2(+2)W
Allanki Sanyasi Rao 27
Carson’s Rule
A popular expression for FM bandwidth is
Carson’s rule. It is a bit smaller than what
we just saw
B
FM
=2(+1)W
Allanki Sanyasi Rao 28
Commercial FM
Commercial FM broadcasting uses the
following parameters
– Baseband;15KHz
– Deviation ratio:5
– Peak freq. Deviation=75KHz
B
FM
=2(+1)W=2x6x15=180KHz
Allanki Sanyasi Rao 29
Wideband vs. narrowband
FM
NBFM is defined by the condition
– ∆f<<W B
FM
=2W
– This is just like AM. No advantage here
WBFM is defined by the condition
– ∆f>>W B
FM
=2 ∆f
– This is what we have for a true FM signal
Allanki Sanyasi Rao 30
Boundary between narrowband and
wideband FM
This distinction is controlled by 
– If >1 > WBFM
– If <1>NBFM
Needless to say there is no point for going
with NBFM because the signal looks and
sounds more like AM
Allanki Sanyasi Rao 31
Commercial FM spectrum
The FM landscape looks like this
FM station B FM station A
FM station C
25KHz guardband
150 KHz
200 KHz
carrier
Allanki Sanyasi Rao 32
FM stereo:multiplexing
First, two channels are created; (left+right)
and (leftright)
Left+right is useable by monaural
receivers

Left channel
Right channel
+
+
+
mono
Allanki Sanyasi Rao 33
Subcarrier modulation
The mono signal is left alone but the
difference channel is amplitude modulated
with a 38 KHz carrier
Left channel
Right channel
+
+
+
mono
DSBSC
f
sc
=38 kHz
+
fsc=
38KHz
freq
divider
Composite baseband

Allanki Sanyasi Rao 34
Stereo signal
Composite baseband signal is then
frequency modulated
Left channel
Right channel
+
+
+
mono
DSBSC
f
sc
=38 kHz
+
fsc=
38KHz
freq
divider
Composite baseband
FM
transmitter

Allanki Sanyasi Rao 35
Stereo spectrum
Baseband spectrum holds all the
information. It consists of composite
baseband, pilot tone and DSBSC
spectrum
38 KHz 19 KHz
15 KHz
Left+right
DSBSC
Allanki Sanyasi Rao 36
Stereo receiver
First, FM is stripped, i.e. demodulated
Second, composite baseband is lowpass
filtered to recover the left+right and in
parallel amplitude demodulated to recover
the leftright signal
38 KHz 19 KHz
15 KHz
Left+right
DSBSC
Allanki Sanyasi Rao 37
Receiver diagram
FM
receiver
lowpass
filter(15K)
bandpass
at 38KHz
X lowepass
VCO
Divide 2
X lowpass
+
+

+
+
+
Left+right
left
right
PLL
coherent detector
38 KHz 19 KHz
15 KHz
Allanki Sanyasi Rao 38
Subsidiary communication
authorization(SCA)
It is possible to transmit “special
programming” ,e.g. commercialfree
music for banks, department stores etc.
embedded in the regular FM programming
Such programming is frequency
multiplexed on the FM signal with a 67
KHz carrier and ±7.5 KHz deviation
Allanki Sanyasi Rao 39
SCA spectrum
38 KHz 19 KHz
15 KHz
Left+right
DSBSC
59.5 67 74.5 f(KHz)
SCA signal
Allanki Sanyasi Rao 40
FM receiver
FM receiver is similar to the superhet
layout
RF
mixer
LO
limiter
Discrimi
nator
deemphasis
AF power
amp
IF
Allanki Sanyasi Rao 41
Frequency demodulation
Remember that message in an FM signal
is in the instantaneous frequency or
equivalently derivative of carrier angle
s t ( ) = A
c
cos 2tf
c
t + 2tk
f
m(t)dt
0
t
}
¸
(
¸
(
' s t ( ) = A
c
2tf
c
+ 2tk
f
m t ( )
 
sin 2tf
c
t + 2tk
f
m(t )dt
÷·
t
}

\

.

Do envelope detection on s’(t)
Allanki Sanyasi Rao 42
Receiver components:RF
amplifier
AM may skip RF amp but FM requires it
FM receivers are called upon to work with
weak signals (~1µV or less as compared to
30 µV for AM)
An RF section is needed to bring up the
signal to at least 10 to 20 µV before mixing
Allanki Sanyasi Rao 43
Limiter
A limiter is a circuit whose output is
constant for all input amplitudes above a
threshold
Limiter’s function in an FM receiver is to
remove unwanted amplitude variations of
the FM signal
Limiter
Allanki Sanyasi Rao 44
Limiting and sensitivity
A limiter needs about 1V of signal, called
quieting or threshold voltage, to begin
limiting
When enough signal arrives at the
receiver to start limiting action, the set
quiets, i.e. background noise disappears
Sensitivity is the min. RF signal to
produce a specified level of quieting,
normally
Allanki Sanyasi Rao 45
Sensitivity example
An FM receiver provides a voltage gain of
200,000(106dB) prior to its limiter. The
limiter’s quieting voltage is 200 mV. What
is the receiver’s sensitivity?
What we are really asking is the required
signal at RF’s input to produce 200 mV at
the output
200 mV/200,000= 1µV>sensitivity
Allanki Sanyasi Rao 46
Discriminator
The heart of FM is this relationship
What we need is a device that linearly
follows inst. frequency
f
i
(t)=f
c
+k
f
m(t)
Disc.output
f
Deviation limits
+75 KHz
75 KHz
f
carrier
f
carrier
is at the IF frequency
Of 10.7 MHz
Allanki Sanyasi Rao 47
Examples of discriminators
Slope detector  simple LC tank circuit
operated at its most linear response curve
This setup turns an FM signal
into an AM
f
c
f
o
output
f
Allanki Sanyasi Rao 48
PhaseLocked Loop
PLL’s are increasingly used as FM
demodulators and appear at IF output
Phase
comparator
Lowpass
filter
VCO
fin Error signal
f
vco
VCO input
Control signal:constant
When f
in
=f
vco
Output proportional to
Difference between f
in
and f
vco
Allanki Sanyasi Rao 49
PLL states
Freerunning
– If the input and VCO frequency are too far apart,
PLL freeruns
Capture
– Once VCO closes in on the input frequency, PLL
is said to be in the tracking or capture mode
Locked or tracking
– Can stay locked over a wider range than was
necessary for capture
Allanki Sanyasi Rao 50
PLL example
VCO freeruns at 10 MHZ. VCO does not
change frequency until the input is within
50 KHZ.
In the tracking mode, VCO follows the
input to ±200 KHz of 10 MHz before losing
lock. What is the lock and capture range?
– Capture range= 2x50KHz=100 KHz
– Lock range=2x200 KHz=400 KHz
Allanki Sanyasi Rao 51
Advantages of PLL
If there is a carrier center frequency or LO
frequency drift, conventional detectors
will be untuned
PLL, on the other hand, can correct itself.
PLL’s need no tuned circuits
f
c
f
o
output
f
If f
c
drifts detector has no way of
correcting itself
Slope detector
Allanki Sanyasi Rao 52
Zero crossing detector
Hard
limiter
Zero
Crossing
detector
Multi
vibrator
Averaging
circuot
FM Output
FM input
Hard limiter
ZC detector
multiV
more frequent
ZC’s means
higher inst freq
in turn means
Larger message
amplitudes
Averaging circuit
NOISE IN ANALOG
MODULATION
AMPLITUDE MODULATION
Allanki Sanyasi Rao 54
Receiver Model
The objective here is to establish a
relationship between input and and output
SNR of an AM receiver
BPF detector
Noise n(t)
Modulated signal s(t)l
output
filter
f
c
f
c
B
T
=2W
Allanki Sanyasi Rao 55
Establishing a reference
SNR
Define “channel” SNR measured at
receiver input
(SNR)
c
=avg. power of modulated signal/
avg. noise power in the message bandwidth
Allanki Sanyasi Rao 56
Noise in DSBSC Receiver
Tuner plus coherent detection
BPF LPF
DSBSC
n(t)
Cos(2πfct)
x(t) v(t)
s(t)
s t ( ) = A
c
m(t)cos 2tf
c
t
( )
< s
2
t ( ) >= avg. power = A
c
2
< m
2
(t) > / 2 = A
c
2
P/ 2
P = avg. message power
Allanki Sanyasi Rao 57
Receiver input SNR
Also defined as channel SNR:
(SNR)
c
=
A
c
2
P/ 2
WN
o
noise power in the message bandwidth
=
A
c
2
P
2WN
o
W
W
No/2 Flat noise spectrum:white noise
Noise power=hatched area
Allanki Sanyasi Rao 58
Output SNR
Carrying signal and noise through the rest
of the receiver, it can be shown that
output SNR comes out to be equal to the
input. Hence
Therefore, any reduction in input SNR is
linearly reflected in the output
SNR ( )
o
SNR ( )
c
=1
Allanki Sanyasi Rao 59
(SNR)
o
for DSBAM
Following a similar approach,
Best case is achieved for 100%
modulation index which, for tone
modulation, is only 1/3
SNR ( )
o
SNR ( )
c
=
k
2
P
1 + k
2
P
<1
k : AM modulation index
P: avg. message power
Allanki Sanyasi Rao 60
DSBAM and DSBSC noise
performance
An AM system using envelope detection
needs 3 times as much power to achieve
the same output SNR as a suppressed
carrier AM with coherent detection
This is a result similar to power efficiency
of the two schemes
Allanki Sanyasi Rao 61
Threshold effectAM
In DSBAM (not DSBSC) there is a
phenomenon called threshold effect
This means that there is a massive drop in
output SNR if input SNR drops below a
threshold
For DSBAM with envelope detection, this
threshold is about 6.6 dB
NOISE IN ANALOG
MODULATION
FREQUENCY MODULATION
Allanki Sanyasi Rao 63
Receiver model
Noisy FM signal at BPF’s output is
BFP Limiter
FM
detector
LPF
(W)
n(t)
FM
s(t)
x t ( ) = s t ( ) + n(t) =
A
c
cos 2tf
c
t +  t ( ) ( )+ r(t)cos 2tf
c
t +¢ t ( ) ( )
noise
where
 t ( ) = m(t)dt
}
Allanki Sanyasi Rao 64
Phasor model
We can see the effect of noise graphically
reference
(t)
¢(t)
The angle FM detector will extract
A(t)
Allanki Sanyasi Rao 65
Small noise
For small noise, it can be approximated
that the noise inflicted phase error is
A=[r⁄Ac]Sin(¢÷)
So the angle available to the FM detector
is +A
FM Detector computes the derivative of
this angle. It will then follow that...
Allanki Sanyasi Rao 66
FM SNR for tone modulation
Skipping further detail, we can show that
for tone modulation, we have the following
ratio
SNR rises as power of 2 of bandwidth; e.g.
doubling deviation ratio quadruples the
SNR
SNR ( )
o
SNR ( )
c
=
3
2

2
BandwidthSNR exchange
Allanki Sanyasi Rao 67
Comparison with AM
In DSBSC the ratio was 1 regardless.
For commercial FM, =5. Therefore,
(SNR)
o
/(SNR)
c
=(1.5)x25=37.5
Compare this with just 1 for AM
Allanki Sanyasi Rao 68
Capture effect in FM
An FM receiver locks on to the stronger of
two received signals of the same
frequency and suppresses the weaker one
Capture ratio is the necessary
difference(in dB) between the two signals
for capture effect to go into action
Typical number for capture ratio is 1 dB
Allanki Sanyasi Rao 69
Normalized transmission
bandwidth
With all these bandwidths numbers, it is
good to have a normalized quantity.
Define
normalized bandwidth=B
n
=B
T
/W
Where W is the baseband bandwidth
Allanki Sanyasi Rao 70
Examples of B
n
For AM:
B
n
=B
T
/W=2W/W=2
For FM
B
n
=B
T
/W~2 to 3
For =5 in commercial FM, this is a very
large expenditure in bandwidth which is
rewarded in increased SNR
Allanki Sanyasi Rao 71
Noise/bandwidth summary
AMenvelope detection
SNR ( )
o
=
µ
2
2 + µ
2
SNR ( )
c
B
n
= 2
Allanki Sanyasi Rao 72
Noise/bandwidth summary
DSBSC/coherent detection
(SNR)
o
=(SNR)
c
B
n
=2
SSB
(SNR)
o
=(SNR)
c
B
n
=1
Allanki Sanyasi Rao 73
Noise/bandwidth summary
FMtone modulation and =5
(SNR)
o
=1.5 
2
(SNR)
c
=37.5 (SNR)
c
B
n
~16 for =5
Allanki Sanyasi Rao 74
Preemphasis and
deemphasis
High pitched sounds are generally of
lower amplitude than bass. In FM lower
amplitudes means lower frequency
deviation hence lower SNR.
Preemphasis is a technique where high
frequency components are amplified
before modulation
Deemphasis network returns the
baseband to its original form
Allanki Sanyasi Rao 75
Pre/deemphasis response
Flat up to ~500Hz, rises from 50015000 Hz
500 Hz 2120 Hz 15KHz
17dB
17dB
+3dB
3dB
preemphasis
deemphasis
Deemphasis circuit
Is between the detector
And the audio amplifier
Allanki Sanyasi Rao 76
Suggested homework
3.41
5.3
5.7