• How do we get the periodic version of an aperiodic signal?

•
•
•
How do we get the Fourier Transform of any signal?
How do we get the Inverse Fourier Transform?
Draw Rect(t/τ) function. What’s its Fourier Transform?
Review
• Draw Triangle(t/τ) function. What’s its Fourier Transform?
• What is a sync function and what are its some major attributes?
• Write the Fourier Transform of
– unit impulse (δ(t))
– Cos(ω0 t)
– Sin(ω0 t)
– e jωct
– e -jωct
• Discuss the following properties of FT
– Symmetry/Duality
– Scaling
– Time shifting
– Frequency shifting
• What is the FT of the unit impulse train?
• What is unit impulse response?
• What is convolution and what is its utility?
• What is the differentiation and integration rule for FT?
• What is an LTI system? Why we are interested in LTI systems
• What is distortionless transmission?
• Why an All-Pass filter is not necessarily distortionless system?
• What is the nature of distortion in audio and video signal?
• Why ideal filters not realizable?
• Can you determine from the transfer function if the filter is low pass, band pass or high pass?
• What does Parseval’s Theorem say?
• What is ESD and PSD?
• If you know the energy of a signal, can you determine the energy of the modulated signal?

1
 Outline
• Baseband vs. Carrier Communication
• Double-sideband Amplitude Modulation
• Amplitude Modulation
• Bandwidth Efficient Amplitude Modulation
• Amplitude Modulation: Vestigial Sideband
• Local Carrier Synchronization
• Frequency Division Multiplexing
• Phase-Locked Loop and Applications
• NTSC Television Broadcasting System

2
 Must Reading
• Chapter No. 4: Amplitude Modulations and
Demodulations

4
 Modulation
The process by which some characteristics of a
carrier wave is varied in accordance with an
information-bearing signal.
• Two Types of Communication:
– Baseband: Communication system that does not
use modulation
– Carrier Modulation: Communication that uses
modulation
Baseband designates the band of
frequencies of the source signal. e.g.
Audio Signal (4 kHz), Video (4.3 MHz)
5
 Why Modulation?
• To
– use the range of frequencies more suited to the
medium
– allow the number of signals to be transmitted
simultaneously (Frequency Division Multiplexing)
– reduce the size of antennas in wireless links
Modulation causes a shift in the range of
frequencies in a signal
Modulation can be analog or digital
Traditional based on analog: e.g. AM/FM radios and NTSC TV
Modern based on digital: e.g. 2G, 3G 4G and upcoming 5G
cellular phone systems, HDTV, DSL
5
 Continuous-wave Modulation
• Amplitude modulation
• Frequency modulation
• Phase modulation

• AM modulation family
– Amplitude modulation (AM)
– Double sideband-suppressed carrier (DSB-SC)
– Single sideband (SSB)
– Vestigial sideband (VSB)

6
 Amplitude Modulation
A carrier wave whose amplitude is varied in
proportion to the instantaneous amplitude of a
modulating voltage

Carrier Signal: cos(2 f c t ) or cos(c t )

Modulating Message Signal: m(t ) : cos(2 f mt ) or cos(mt )

The AM Signal: s AM (t ) = m(t ) cos(2 f ct )

7
 Amplitude Modulation: Double Side
Band (DSB)
• Modulating signal, base band signal, information
signal: m(t ) M ( )
• Carrier signal:
c(t ) = cos(ct +  c )
• with Spectrum:
C ( ) = 0.5 ( − c ) +  ( + c )
Modulation is the product of the base band with the
carrier
m(t ) cos( c t )  M ( +  c ) + M ( −  c )
1
2
8
 Amplitude Modulation: Double Side
Band (DSB)
• DSB modulation simply shifts the frequency contents
of m(t) to the carrier frequency

9
 Demodulation
• To demodulate we multiply the signal by cos(c t )
1 1 
 m (t ) cos( c 
t ) cos( c t ) = m (t ) cos 2
( c t ) = m (t ) +
 2 2 cos(2 t
c )

• Therefore the FT of this signal is
1 1  1 1
m(t )  + cos(2ct )   M ( ) +  M ( + 2c ) + M ( − 2c ) 
2 2  2 4
1
• If we lowpass filter this signal we recover m(t )
2

10
Modulation & Demodulation

11
Example

12
 Frequency Mixer or converter
• We wanted to change the modulated signal from wc to wI
• The product x(t) is x (t ) = 2m(t ) cos  t cos  t
c mix

x(t ) = m(t )cos( c −  mix )t + cos( c +  mix )t 

• Down conversion if we select  mix =  c −  I
x(t ) = m(t )cos( c − ( c −  I ) )t + cos( c +  c −  I )t 
x(t ) = m(t )cos( I ) )t + cos(2 c −  I )t 

• Up conversion if we select mix = c + I

x(t ) = m(t )cos(c − (c + I ) )t + cos(c + c + I )t 
x(t ) = m(t )cos( I ) )t + cos(2 c +  I )t 
13
Frequency Mixer (cont)

14
 AMPLITUDE MODULATION
(Transmitted Carrier DSB-TC)
• In this case we send the carrier with the signal
 AM (t ) = A cos ct + m(t ) cos ct
Carrier Sidebands

• We can think as the modulating signal to be

 
 AM (t ) =  
A + m(t )
 
 cos ct
mod ulating 
 signal with a DC 

• The spectrum of this signal is

1 1
 AM ( ) =  M ( + c ) + M ( − c )  + A  ( + c ) +  ( − c ) 
2 2
DSB − SC spectrum Carrier spectrum

15
16
17
 AM (cont)

• A is large enough that A + m(t )  0. The demodulation can be

achieved by a simple envelope detector

• Let’s consider the peak value of m(t ) to be m p. Then the

condition for envelope detection of AM signal is

A + m(t )  0
• Which is equivalent to

A  mp

18
 Modulation Index

• We define the modulation index as

mp
=
A
• Therefore we can see that if we want to maintain the condition
A  mp
• We have

0   1

19
Example 4.4

20
Example 4.4 (cont)

21
 Percentage Modulation

Under modulated (<100%) 100% modulated Over Modulated (>100%)

Envelope Detector Envelope Detector

Can be used Gives Distorted signal

22
 Sideband and Carrier Power

• The advantage of envelope detection in AM has its price

 AM (t ) = A cos  c t + m(t ) cos  c t



 
carrier sidebands

• The carrier Power is (see example 2.2)

A2
Pc =
2
• The sidebands Power is half of modulating signal (baseband)

1 ~~~~~~~
Ps = m 2 (t )
2
23
 Sideband and Carrier Power (cont)

• The sideband power is the useful power and the Carrier Power
is the power wasted

• We define the Power Efficiency as

~~~~~~~
1 2 ~~~~~~~
m (t ) 2
UsefulPowe r Ps m (t )
= = = 2 = *100%
TotalPower Pc + Ps A2 1~~~~~~~ ~~~~~~~

+ m (t )
2 A 2
+ m 2
(t )
2 2

24
 Sideband and Carrier Power (cont)

For the special case of tone modulation

m(t ) = A cos  m t
then its power is
~~~~~~~
m 2
(t ) =
( A)2
2
then

1 (A) (A)
2 2

 2
 = 2 2 2 2 *100% 2 *100% = *100%
A 1 (A) (A) 2+
2 2
+ A +
2

2 2 2 2
The max value when  =(100%
1 modulation) is  = 33%
25
 Demodulation of AM Signals

• We do not need a local generated carrier in this case

• If we have undermodulation then we can use

1. Rectifier detection
2. Envelope detection

26
Rectifier detector

27
 Rectifier detector (cont)

• If the AM wave is applied to diode and resistor circuit. The

negative part of the AM is supressed. This is like saying that
we have half wave rectified the AM

Mathematically

1 2  
v R ' = A + m(t )cos  c t  +  cos  c t − cos 3 c t + cos 5 c t.....
1 1
2   3 5 

vR ' =
1
A + m(t ) + other terms


28
 Rectifier detector (cont)

• If we pass this voltage thru a LPF we get

DC Component: 0 frequency Baseband, modulating signal

v filtered =
1
A + m(t )

• If we use a capacitor, we block the DC and we obtain

vout =
1
m(t )


29
 Time required to charge
a capacitor through
Envelope detector resistor by 63.2%
Time constant of an Time period of
RC circuit the carrier

Bandwidth of the
baseband
modulating signal

Required
condition for
envelope detection
30
 Quadrature Amplitude Modulation
(QAM)
• DSB signals occupy twice the bandwidth required for the base
band.

• This disadvantage can be overcome by transmitting two DSB

signals using carriers of the same frequency but in phase
quadrature

 QAM = m1 (t ) cosc t + m2 (t ) sin c t

• The message signals m1(t) & m2(t) are in-phase & quadrature-
phase components of φQAM(t)

31
Modulation and Demodulation of QAM

32
 QAM (cont)
 QAM = m1 (t ) cosc t + m2 (t ) sin c t
• We can obtain both the signals by using two local carriers in
phase quadrature
x1 (t ) = 2 QAM cosc t = 2m1 (t ) cosc t + m2 (t ) sin c t cosc t
x1 (t ) = m1 (t ) + m1 (t ) cos 2 c t + m2 (t ) sin 2 c t

• Similarly the output of the lower branch can be shown as

• A slight error in phase leads to distortion and mixing of signals

x1 (t ) = 2 QAM cosc t = 2m1 (t ) cosc t + m2 (t ) sin c t cos(c t +  )
x1 (t ) = m1 (t ) cos  − m2 (t ) sin 
33
QAM Simulation

34
 Amplitude Modulation: Single Sideband
To improve the
Remember: Both side bands (SSB) bandwidth efficiency
(USB and LSB) contain complete
information of the message

• The idea is to transmit either the USB or LSB

35
36
 SSB (cont)

• Let m+(t) and m-(t) be the complex conjugates of m(t)

m+ (t ) = m(t ) + jm h (t )
1
2
m− (t ) = m(t ) − jm h (t )
1
2

• Where mh(t) is unknown

37
 SSB (cont)

• To determine mh(t) we know that

M + ( ) = M ( )u ( ) 1
U( ) = 1 + sgn( )
2
1
= M ( ) 1 + sgn( ) 
2
1 1
= M ( ) + M ( ) sgn( )
2 2
• Comparing above equation with the one on the previous slide
m+ (t ) =
1
m(t ) + jmh (t )
jmh (t )  M ( ) sgn( ) 2

38
 Multiplication in jmh (t )  M ( ) sgn ( )

• Hence frequency domain  jM h ( ) = M ( ) sgn ( )

M ( ) sgn ( )
M h ( ) = − jM ( ) sgn( )  M h ( ) =
j
 M h ( ) = − jM ( ) sgn ( )

• Applying duality prop. to pair 12 of table 3.1 yields

2
sgn ( t ) 
j

2
Convolution in
ByDuality : f ( t )  F ( ) thenF ( t ) = 2 f ( − )
2
   = 2 sgn ( − )     = −2 sgn ( )
 jt   jt 

time domain
1
 2  1
   = − sgn ( )     = − j sgn ( )
 j 2 t   t 

 − j sgn( )
t
M h ( )  mh (t ) = m(t ) 1  t

• Which gives 
∞

m( ) 1 𝑥 𝑡 ∗ 𝑦 𝑡 = න 𝑥 𝜏 𝑦 𝑡 − 𝜏 𝑑𝜏
mh (t ) =  d
−∞
Replace:
By definition  − t −  𝑥 𝑡 →𝑚 𝑡
1
y 𝑡 →
• This is the Hilbert transform of m(t) 𝜏→𝛼
𝜋𝑡

39
 Hilbert Transform

• We can Hilbert transform m(t) if we pass it through a filter

with
H ( ) = − j sgn( )
e jx = Cos ( x ) + jSin ( x ) ;
e − jx = Cos ( x ) − jSin ( x ) ;
e +e − jx − j = 1e − j 2
0
=
jx
Cos ( x ) = ;
j 2
 j = 1e 0
2
e − e − jx
jx
Sin ( x ) =
2
• It follows that |H(ω)|=1 and that

 h ( ) = −  2 for   0 and  2 for   0

40
Hilbert Transform (cont)

41
42
 e jx = Cos ( x ) + jSin ( x ) ; 1
m+ (t ) =  m(t ) + jmh (t )
e − jx = Cos ( x ) − jSin ( x ) ;
e +e
jx − jx
Mathematics 2
m− (t ) = m(t ) − jm h (t )
1
Cos ( x ) = ; 2
2
e jx − e − jx
Sin ( x ) =
2
USB (t ) = m+ (t )e j t + m− (t )e − j t
c c

1 1
USB (t ) =  m(t ) + jmh (t ) e jct +  m(t ) − jmh (t )  e − jct
2 2
1 1
USB (t ) = m(t ) e jct + e − jct  +  jmh (t )e jct − jmh (t )e − jct 
2 2
• From previous figure jmh (t ) USB spectrum can be jmhexpressed
(t ) as
USB (t ) = m(t )Cos (ct ) +
USB ( ) = M + (2 −Cos (c)c +
   ) + jSin (c ) −
M − ( + c2) Cos (c ) − jSin (c )
 
USB (t ) = m(t )Cos (ct ) + jmh (t ) ( jSin (c ) )
• The inverse transform yields
jc t − jc t
USB (t )= m(t()Cos
USB t ) = m +c m− (t )e
(  c t ) −
+ (t h)e
m (t ) Sin (  )
• Substituting m+(t) and m-(t) from slide 38 in the above Eq.
USB (t ) = m(t ) cos ct − mh (t ) sin ct
43
 Mathematics (cont)

• Similarly we can show

LSB (t ) = m(t ) cos ct + mh (t ) sin ct

• General SSB signal can expressed as

SSB (t ) = m(t ) cos ct mh (t ) sin ct

44
SSB Simulation

45
 Generation of SSB Signals

• Two methods are used for the generation of SSB signals

– Selective Filtering Method

– Phase Shift Method

46
 Selective Filtering Method

• Selective Filtering using filters with sharp cutoff

characteristics. Sharp cutoff filters are difficult to design

• The audio signal spectrum has no dc component, therefore ,

the spectrum of the modulated audio signal has a null around
the carrier frequency

• This means a less than perfect filter can do a reasonably good

job of filtering the DSB to produce SSB signals

47
Filtering (cont)

48
Filtering (cont)

49
Phase Shift Method

50
 Demodulation
• SSB signals can be coherently demodulated in the same way
as DSB-SC

SSB (t ) = m(t ) cos ct mh (t ) sin ct

1
Cos 2 ( u ) = 1 + Cos ( 2u ) 
2

51
 Amplitude Modulation: Vestigial
Sideband (VSB)
• The generation of SSB signals is rather difficult in practice

• To produce SSB signal from DSB signal ideal filters should be

used to split the spectrum in the middle so that the bandwidth
of bandpass signal is reduced by one half

• The selective filtering method demands dc null in the

modulating signal

• The generation of DSB signals is simple, but DSB signals

require twice the signal bandwidth of SSB

52
 Vestigial Sideband (cont)

• Vestigial sideband (VSB) modulation was designed to provide

a compromise between DSB and SSB

• In VSB instead of rejecting one sideband completely, we do a

gradual cutoff of one sideband

• It can be detected with a synchronous detector in conjunction

with an appropriate filter at the receiver output

• If a carrier is sent along the transmission, the VSB can be

recovered by an envelope or a rectifier detector

53
 Vestigial Sideband (cont)

• Generation of VSB is done by multiplying m(t) by 2cos(ωct)

and applying this signal to a filter Hi(ω)

VSB ( t ) =  2m ( t ) cos c ( t )  hi ( t ) 
54
 Vestigial Sideband (cont)

• Because the VSB is not a SSB, the bandwidth is 25 to 33%

larger only, but it also makes the band-pass filter easier to
realize

55
 Vestigial Sideband (cont)

• The VSB signal spectrum is given by

VSB ( ) =  M ( + c ) + M ( − c )  H i ( )

• Where Hi(ω) is VSB shaping filter, which allows the

transmission of one sideband and suppresses the other
sideband gradually

56
 Vestigial Sideband (cont)

• We can recover the message by using synchronous

demodulation
• Multiply the incoming VSB signal by 2Cos(ωct)

57
 Vestigial Sideband (cont)

• The product e(t) is given by

VSB ( t ) =  2m ( t ) cos c ( t )  hi ( t ) 
e ( t ) = 2VSB ( t ) cos c t
=  2m ( t ) cos c ( t )  hi ( t )  2 cos c ( t )
= 4m ( t ) cos 2 (c t )  2hi ( t ) cos (ct )
1
Cos 2 ( u ) = 1 + Cos ( 2u ) 
2

=  2m ( t ) + 2m ( t ) cos ( 2ct )   2hi ( t ) cos (ct )

=  2m ( t )  2hi ( t ) cos c ( t )  +  2m ( t ) cos 2c ( t )  2hi ( t ) cos c ( t ) 

58
 e ( t ) =  2m ( t )  2hi ( t ) cos c ( t )  +  2m ( t ) cos 2c ( t )  2hi ( t ) cos c ( t ) 
• The Fourier Transform of e(t)

E ( ) = M ( )  H i ( + c ) + H i ( − c )  +
 M ( + 2c ) + M ( − 2c )   H i ( + c ) + H i ( − c ) 
High Frequency Terms

• Low pass filter will block high frequency component and we

will be left with
E ( )
M ( ) =
 H i ( + c ) + H i ( − c ) 

M ( ) = E ( ) H o ( )

• Where H o ( ) =
1
,   2 B
 H i ( + wc ) + H i ( − c ) 
59
Spectrum of Hi(w) and Ho(w)

60
Example:

• Hi(f):

• Ho(f)=?
1
Ho ( f ) = , f B
 H i ( f + f c ) + H i ( f − f c ) 

61
 Linearity of Amplitude Modulation

• In all types of AM discussed before, linearity is satisfied

• k1 m1 (t ) + k 2 m2 (t ) produces the modulated signal k11 (t ) + k 2 2 (t )

• The modulation system following the superposition theorem of

spectra is known as linear modulation system

– The theorem states that the sideband spectrum of a multiple tone AM

signal is equal to the sum of the sideband spectrum of the individual
tone modulation.

62
 Carrier Acquisition

• In SC transmissions, we have to generate a carrier with the

same frequency and phase that the carrier at the transmitter.

• Consider the case of DSB-SC where a received signal is

 DSB − SC (t ) = m(t ) cos c t

• and the local carrier is

2 cos( c +  )t +  
• therefore, we have errors in frequency and phase

63
 Carrier Acquisition (cont)

• The product of the received signal and the local carrier is

e(t ) =  m(t ) cos ct    2 cos ( (c +  ) t +  ) 
1
Cos ( u ) Cos ( v ) = Cos ( u − v ) + Cos ( u + v ) 
2


= m(t ) cos (  ) t +   + cos ( ( 2c +  ) t +  ) 
• After the LPF we have

eo (t ) = m(t )cos( )t +  

64
 Carrier Acquisition (cont)

• Let’s consider two cases. First  = 0

• In this case eo (t ) = m(t )cos 

• The output is proportional to m(t) because the factor is a

constant

• The output is maximum when δ=0 and minimum (zero) when

δ=±π/2 Thus, this kind of phenomenon only attenuates the
output without adding distortion

• Unfortunately delta is not constant. This may occur for

example because of variations in the propagation path.
65
 Carrier Acquisition (cont)

• Now consider the second case  = 0,   0

• In this case eo (t ) = m(t ) cos (  ) t

• The output is distorted as well, the output is

m(t) multiplied by a low frequency oscillation.

• This beating is catastrophic even for a small

frequency
66
 Carrier Acquisition (cont)

• To ensure identical carrier frequencies at the emitter and

receiver we can use crystal oscillators

In telecommunications, a pilot signal is a signal, usually a single

frequency, transmitted over a communications system for
supervisory, control, equalization, continuity, synchronization, or
reference (e.g. phase reference) purposes. [WiKi]

• Other method is to send a carrier or pilot at a reduced level

along with the sidebands. Then is filtered at the receiver with a
very narrow filter

67
 Frequency Division Multiplexing

• FDM
• When useful bandwidth of medium
exceeds required bandwidth of channel
• Each signal is modulated to a different
carrier frequency
• Carrier frequencies separated so
signals do not overlap (guard bands)
• e.g. broadcast radio
• Channel allocated even if no data

68
FDM System
• Example:

69
 Analog Carrier Systems
• AT&T (USA)
• Hierarchy of FDM schemes
• Group
– 12 voice channels (4kHz each) = 48kHz
– Range 60kHz to 108kHz
• Supergroup
– 60 channel
– FDM of 5 group signals on carriers between 420kHz and 612 kHz
• Mastergroup
– 10 supergroups
• Jumbo group
– 6 supergroups
70
 Phase Locked Loop (PLL)

• The PLL can be used to track the phase and frequency of the
carrier component of an incoming signal.

• It is then useful for synchronous demodulation of AM signals

with suppressed carrier or with a pilot

• PLL has three basic components:

– A VCO or Voltage Controlled Oscillator
– A multiplier, serving as a phase detector or a phase comparator
– a loop filter H(s)

71
Phase Locked Loop (cont)

72
 Phase Locked Loop (cont)

• PLL works just like feedback system, the signal fed back tends
to follow the input signal to minimize the error. The quantity to
compare is the phase in this case

• The VCO oscillates linearly with the input voltage

 (t ) =  c + ceo (t )
• Where “c” is a constant and “ωc” is the free running frequency
of the VCO. This is the one when the input signal is zero

73
 PLL Operation

• Let the input to the PLL be

A sin ( c t +  i )
• Let the output of VCO be
B cos( c t +  o )
• The multiplier output x(t) will be
x(t ) = AB sin(ct + i ) cos (ct +  o )
1
Sin ( u ) Cos ( v ) =  Sin ( u + v ) + Sin ( u − v ) 
2

=
AB
sin( i −  o ) + sin( 2c t +  i +  o )
2

74
 PLL Operation (cont)

• The filter is a low pass narrow filter therefore the

error signal is
eo (t ) =
AB
sin( i −  o )
2
AB
= sin( e )
2

• Where θe is the phase error, which is linear for small

error

75
 PLL Operation (cont)

• We have two cases: Input frequency changes or phase changes

• If input frequency is increased, the input changes to

A sin (( c + k )t +  i ) (
= A sin ct + ˆi )
• Where ˆi = kt +  i
• Thus the increase in frequency causes 𝜽෡ 𝒊 to increase thereby
increasing θe which in turn increases input voltage to the VCO

76
 PLL Operation (cont)

• The VCO increases the frequency because the input voltage is

increased to match the increase in the input frequency

• If the input frequency decreases the same reasoning applies

• The PLL tracks the input sinusoid. The two signals are said to
be phase coherent or in phase lock

77
 PLL Operation (cont)

• A PLL tracks the incoming frequency only over a finite range

of frequency shift. This range is called the hold-in or locks
range

• The frequency range over which the input will cause the loop
to lock is called the Pull-in or Capture range

78
 Carrier Acquisition in DSB-SC

• Signal Squaring Method

• Costas Loop

79
 Signal Squaring Method

• This method is explained in the following block diagram

x(t ) = m(t ) cos wct  = m (t ) + m 2 (t ) cos 2wct

2 1 2 1
2 2

• The squarer output will be

1 2 1 2
x ( t ) =  m ( t ) cos ct  = m ( t ) + m ( t ) cos 2ct
2

2 2
• Now m2(t) is a non negative signal and therefore has non zero
average value in contrast to m(t)

80
 Signal Squaring Method (cont)

• Let the average value, which is the dc component of m2(t)/2,

be k then

m (t ) = k +  (t )
1 2
2
()
• Where  t is a zero mean baseband signal minus its dc
component
1 2 1 2
x ( t ) = m ( t ) + m ( t ) cos 2ct
2 2
1 2
= m ( t ) + k cos 2ct +  ( t ) cos 2ct
2

81
Costas Loop

82
Superhetrodyne AM Receiver

83
 Why IF

• At very high frequencies, signal processing circuitry

performs poorly

• It is difficult to build amplifiers, filters, and detectors

that can be tuned to different frequencies

• It is also used to improve frequency selectivity

84
 Example

Solution:
Image frequency = 2 * 455 kHz = 910 kHz apart
Station heard if tuned to 1500 – 910 = 590 kHz
Reason:
Oscillator frequency = fLO = 590 + 455 = 1045 kHz
This is multiplied by incoming signal of fc = 1500 kHz, yielding sum and
difference frequencies i.e. 1500 ± 1045 kHz = 2545 kHz which is
suppressed, and 455 kHz which is received.

85
 Exercise

Solution:
LO frequency range = fc + fLO = 1+8 = 9 and 30+8 = 38 MHz
Another station at 10 + 2x8 = 26 MHz will mix with fLO=18 MHz to
produce signals centered at 8 and 44 MHz.
44 MHz is suppressed by IF filter but 8 MHz passes through, causing
interference.

86