Chapter 2 Continuous-Wave Modulation

2.1 Introduction

1
2.2 Amplitude Modulation
c(t )  Ac cos(2fct ) (2.1)
Ac : carrier amplitude
fc : carrier frequency
1+kam(t) S(t)
X

Accos(2fct)
The output of the modulator
s(t )  Ac1  kam(t )cos(2fct ) (2.2)
Where m(t) is the baseband signal , ka is the amplitude sensitivity.
1. kam(t )  1, for all t (2.3)
2. f c  W (2.4)
where W is the hightest freqency of m(t ) 2
 s(t )  Ac cos(2fct )  Ackam(t ) cos(2fct ) (2.2)
Recall
cos(2fct ) 
1
 ( f  fc)   ( f  fc)
2
m(t ) cos(2fct )  M ( f  fc )  M ( f  fc )
1
2
s( f )   ( f  fc )   ( f  fc )  M ( f  fc)  M ( f  fc)
Ac kaAc
(2.5)
2 2
where M ( f ) is the Fourier Transform of m(t )

1.Negative frequency component of m(t) becomes visible.

2.fc-W  M(f)  fc lower sideband
fc  M(f)  fc+W upper sideband
3.Transmission bandwidth BT=2W 3
Virtues and Limitations of Amplitude Modulation
Transmitter

Receiver

Major limitations
1.AM is wasteful of power. 4
2.AM is wasteful of bandwidth.
2.3 Linear Modulation Schemes
Linear modulation is defined by
s(t )  sI (t ) cos(2fct )  sQ (t ) sin(2fct ) (2.7)
sI (t )  In - phase component
sQ (t )  Quadrature component

Three types of linear modulation:

1.Double sideband-suppressed carrier (DSB-SC) modulation
2.Single sideband (SSB) modulation
3.Vestigial sideband (VSB) modulation

5
Notes:
1.sI(t) is solely dependent on m(t)
2.sQ(t)is a filtered version of m(t).
The spectral modification of s(t) is solely due to sQ(t).
6
Double Sideband-Suppressed Carrier (DSB-SC) Modulation
s(t )  Acm(t ) cos(2fct ) (2.8)

The Fourier transform of S(t) is

1 
s( f )  Ac  M ( f  fc )  M ( f  fc ) (2.9) 7
2
Coherent Detection (Synchronous Detection)

The product modulator output is

v (t )  Ac' cos(2fct   ) s(t )
 Ac' Ac cos(2fct ) cos(2fct   )m(t )
1 1
 AcAc' cos(4fct   )m(t )  AcAc' cos( )m(t ) (2.10)
2 filtered out 2
Let V(f) be the Fourier transform of v(t)
1
v 0(t )  AcAc' cos  m(t ) (Low pass filtered) (2.11)
2

8
Costas Receiver
I-channel and Q-channel are coupled together to
form a negative feedback system to maintain synchronization
 0
1 2 1 2 2
Ac cos  sin  m (t )  Ac m (t )sin(2 )
2

4 8
1 2 2
 Ac m (t )  (sin2  2 )
4
The phase control signal ceases with modulation.

(multiplier +
very narrow band LF)

9
Quadrature-Carrier Multiplexing (or QAM)
Two DSB-SC signals occupy the same channel
bandwidth, where pilot signal (tone ) may be
needed.
s(t )  Acm1(t ) cos(2fct )  Acm2(t ) sin(2fct )

10
Single-Sideband Modulation (SSB)
The lower sideband and upper sideband of AM signal
contain same information .
The frequency-discrimination method consists of a
product modulator (DSB-SC) and a band-pass filter.
The filter must meet the following requirements:
a.The desired sideband lies inside the passband.
b.The unwanted sideband lies inside the stopband.
c.The transition band is twice the lowest frequency of
the message.

To recover the signal at the receiver, a pilot carrier or a stable oscillator

11
is needed (Donald Duck effect ).
Vestigial Sideband Modulation (VSB)
When the message contains near DC component

The transition must satisfy

a. H ( f  f c )  H ( f  f c )  1
b.The phase response is linear :
H ( f  fc )  H ( f  fc )  1 for  W  f  W (2.13)
BT  W  fν (2.14)
12
Consider the negative frequency response:
Hf

 fc  W  fc  fv  fc  fc  fv fc  fv fc fc  fv fc  W

Here, the shift response │H(f-fc)│ is

H  f  fc 

W  fv 0 fv 2 fc  fv 2 fc 2 fc  fv 2 fc  W

13
and │H(f+fc)│ is
H  f  fc 

2 f c  W 2 f c  f v 2 f c 2 f c  f v  fv 0 fv W

14
So, we get │H(f-fc)│ +│ H(f+fc)│ is
H  f  fc 

W  fv 0 fv 2 fc  fv 2 fc 2 fc  fv

H  f  fc 

2 f c  f v 2 f c 2 f c  f v  fv 0 fv W

15
Consider –W<f<W we get:

W  fv 0
fv
W
Which is equal to

W W

So, │H(f-fc)│ + │H(f+fc)│ =1 for -W<f<W

16
 1 1
s(t )  Acm(t ) cos(2fct )  Acm' (t ) sin(2fct ) (2.15)
2 2
± corresponds to upper or lower sideband

m(t) m’(t)
HQ(f)

HQ( f )  j H ( f  fc)  H ( f  fc) for  W  f  W (2.16)

17
Television Signals (NTSC)

18
2.4 Frequency Translation

A cos(2 f t )

Up conversion
f2=f1+fl , fl=f2-f1
Down conversion
f2=f1-fl , fl=f1-f2
19
2.5 Frequency-Division Multiplexing (FDM)

20
2.6 Angle Modulation
Basic Definitions:
Better discrimination against noise and interference
(expense of bandwidth).
s(t )  Ac cosi (t ) (2.19)
The instantaneous frequency is
f i (t )  lim f Δt (t )
Δt 0

i (t  t )  i (t ) 
 lim  
Δt 0
 2t
1 d i ( t )
 (2.21)
2 dt
For an unmodulated carrier, i (t ) is
i (t )  2fct  c (2.22)
where c is constant 21
1. Phase modulation (PM)
i (t )  2f ct  k p m(t )
kp : phase sensitivity of the modulator
s(t)  Ac cos2f ct  k p m(t ) (2.23)
2. Frequency Modulation (FM)
fi (t )  f c  k f m(t ) (2.24)
t
i (t )  2π f ct  2π k f  m( )d (2.25)
0

s(t)  Ac cos 2π f ct  2π k f  m( )d 

 t
(2.26)
 0 
k f :frequency sensitivity of the modulator
t
compare (2.23) and (2.26)  k p m'(t)  2π k f  m( )d
0

22
generating FM signal generating PM signal
2.7 Frequency Modulation
FM is a nonlinear modulation process , we can not apply
Fourier transform to have spectral analysis directly.
1.Consider a single-tone modulation which produces a
narrowband FM (kf is small)
2.Next consider a single-tone and wideband FM
(kf is large)

let m(t )  Am cos(2f mt ) (2.27) (deterministic)

f i (t )  fc  k f Am cos(2f mt )
 f c  f cos(2f mt ) (2.28)
Δf  k f Am : frequency deviation
23
 t
Recall (2.25), i (t )  2  f i ( )d
0

f
 2πfct  sin(2f mt ) (2.30)
fm
f
M odulation index  (2.31)
fm
i (t )  2πfct   sin(2f mt ) (2.32)
(2.19) => s(t )  Ac cos2fct   sin(2f mt ) (2.33)
Narrowband FM ,  is smaller th an one radian.
Wideband FM ,  is larger tha n one radian.
24
 Narrowband FM
s(t )  Ac cos2f ct   sin(2f mt )
 Ac cos(2f ct ) cos sin(2f mt )  Ac sin(2f ct ) sin sin(2f mt ) (2.34)
Because  is small,
cos sin(2f mt )  1
sin sin(2f mt )   sin(2f mt )
s(t )  Ac cos(2f ct )  Ac sin(2f c t ) sin(2f mt ) (2.35)

25
The output of Fig 2.21 is
s' (t )  Ac cos(2f ct )  Ack f  m( )d sin(2f ct )

s(t) differs from ideal condition in two respects:

1.The envelope contains a residual AM.
(FM has constant envelope)
2.i(t) contains odd order harmonic distortions
3 5 7
x x x
(sin x  x     )
3! 5! 7!
For narrowband FM, β≤ 0.3 radians.
26
Recall (2.35)
s (t )  Ac cos (2f c t )  Ac sin (2f c t )sin(2f mt ) (2.35)

 Ac cos (2fct )  Ac cos2 ( f c  f m )t   cos2 ( f c  f m )t  (2.36)

1
2
For AM with sinusoidal modulating wave , m(t )  cos(2fmt )
sAM (t )  Ac 1  k a m(t )cos (2f c t ) (2.2)
 Ac cos (2f c t )  k a Ac cos (2f c t ) cos(2f mt )

 Ac cos (2f c t )  Ac cos2 ( f c  f m )t   cos2 ( f c  f m )t  (2.37)

1
2

Narrow band FM

27
AM
 Wideband FM (large β)
s (t )  Ac cos2f c t   sin( 2f mt ) (2.33)
exp (jx)  cos x  j sin x
s (t )  ReAc exp( j 2f c t  j sin( 2f mt ))
 Re~
s (t ) exp( j 2f c (t )) (2.38)
where Re  denotes the real part and
~
s (t ) is the complex envelope defined by
~
s (t )  Ac exp[ j sin( 2f mt )] (2.39)

~
s (t )  c
n  
n exp( j 2 nf mt ) (2.40)
28
Complex Fourier Transform
 1 f
cn  f  s (t )exp(  j 2 nf mt )dt
2 m
m 1 f
2 m

exp  j  sin(2 f mt )  j 2 nf mt ) dt
1 f
 fm A 
2 m
c 1 f (2.41)
2 m

Let x  2 f mt (2.42)
Ac 
cn   exp  j(  sin x  nx ) dx (2.43)
2 
Define the nth order Bessel function of the first kind as
2
d y dy
（A3, x 2
2
 x  ( x 2
 n 2
) y  0)
dx dx
1 
Jn ( )   exp  j(  sin x  nx ) dx (2.44)
2  

cn  Ac J n (  )

s (t )  Ac J
n 
n (  ) exp( j 2 nf mt ) (2.45)
29
  
s(t )  Ac Re  J n (  ) exp  j 2 ( fc  nfm )t  (2.47)
  

 Ac  J n (  ) cos2 ( f c  nf m )t  (2.48)


The Fourier transform of s(t ) is


(  ) ( f  f c  nf m )   ( f  f c  nf m ) (2.49)
Ac
S( f ) 
2
J

n

30
Figure 2.23 Plots of Bessel functions of the first kind for varying order.
Properties of J n (  )
1.J n (  )  (  1)n J  n (  ), for all n (2.50)
2.If  is small
J0(  )  1

J1(  ) 
2
Jn(  )  0 n2 (2.51)

3. J n2 (  )  1
-

Observation of FM
1.An FM signal contains f c , f m , 2 f m , 3 f m , components.
2.For small  , the FM signal is effectively composed of a carrier and
a single pair of side freqencies at f c  f m  narrowband FM
3.The amplitude of carrier depends on 

1 2 Ac2
P  Ac 
2 2
 n ( )
J

2
(2.54) 31
Example 2.2

32
Transmission Bandwidth of FM signals
With a specified amount of distortion , the FM signal is
effectively limited to a finite number of significant side
frequencies.
A.Carson’s rule
1 f
BT  2f  2 f m  2f (1  ) ,  = , f   f m (2.55)
 fm

33
 f
B. BT  2nmax f m , J nmax (  )  0.01 , BT  2nmax


Universal curve for evaluating the 1 percent bandwidth of an FM wave

34
Example 2.3
In north America, the maximum value of frequency deviation f is
fixed at 75kHz for commercial FM broadcasting by radio. If we take
the modulation frequency W=15kHz, which is typically the
“maximum” audio frequency of interest in FM transmission, we find
that corresponding value of the deviation ratio is
75
D 5
15
Using Carson’s rule of Equation (2.55) , replacing by D , and
replacing fm by W , the approximate value of the transmission
bandwidth of the FM signal is obtained as
BT=2(75+15)=180kHz
On the other hand , use of the curve of Figure 2.26 gives the
transmission bandwidth of the FM signal to be
BT=3.2 f =3.2x75=240kHz
In practice , a bandwidth of 200kHz is allocated to each FM
transmission . On this basis , Carson’s rule underestimates the
transmission bandwidth by 10 percent , whereas the universal curve
of Figure 2.26 overestimates it by 20 percent.
35
Generation of FM signals

Frequency Multiplier
v(t )  a1s(t )  a 2 s 2 (t )   ans n (t ) (2.56)

s( t )  Ac cos 2 fct  2 kf  m( )d 

 t

 0 
The frequency multiplier output

s '(t )  Ac 'cos 2 nfct  2 nkf  m( )d  (2.58)

 t

 0 
36
fi '(t )  nfc  nkf m(t ) (2.59)
Varactor diode VCO FM modulator

32-1
Crosby Direct FM Transmitter

32-2
Demodulation of FM signals
The frequency discrimination consists of a slope circuit
followed by an envelope detector

Consider Fig 2.29a , the frequency response of a slope

circuit is
 BT BT BT
 j 2a( f  f c  2 ), fc 
2
 f  fc 
2
 BT BT BT
H 1( f )   j 2a( f  f c  ),  f c   f   fc  (2.60)
 2 2 2
 0, elsewhere
 33
H1( f  fc )  2H1( f ) , f  0

H 2 ( f  f c )  2H 2 ( f ) , f  0

34
Appendix 2.3 Hilbert Transform
Fourier Transform-frequency-selective
Hilbert Transform-phase-selective
(±900shift)
Let g(t)G(f)
Denote the Hilbert transform of g(t) as
1  g ( )
gˆ (t )   d (A2.31)
  t  
The inverse Hilbert transform
1  gˆ ( )
g (t )    d (A2.32)
   t 
35
 1
  j sgn( f ) (A2.33)
t
1 f 0

sgn( f )   0 f 0 (A2.34)
 1 f 0

The Fourier transform of g (t ) is
Gˆ ( f )   j sgn( f )G( f ) (A2.35)

g(t) gˆ (t )
H(f)

36
Properties of the Hilbert Transform
(time domain operation)

If g(t) is real
1.gˆ (t ) and g (t ) have the same magnitude spectrum
2.Hilbert transform of gˆ (t ) is  g (t ) (take H.F ofg(t )and
compare with A2.32)

3. g( t )ĝ(t )dt  0  g( t )  gˆ (t )
-

37
 For a band-pass system , we consider
x(t)  X( f )
X( f ) is limited within ± W Hz
W  fc
x(t )  xI (t )cos(2 f ct )  xQ (t )sin(2 f ct ) (A2.48)
The complex evelope of x(t) is
x(t )  xI (t )  j xQ (t ) (A2.49)
x(t) band pass y(t)
system, fc
H(f) 2B
f
fc
h(t )  hI (t )cos(2 f ct )  hQ (t )sin(2 f ct ) (A2.50)
38
Define the complex impluse response
~
h (t )  hI (t )  j hQ (t ) (A2.51)
The complex representation of h(t )
 ~
h(t )  Re h (t ) exp( j 2 f c t )  (A2.52)
~
hI (t ), hQ (t ) and h (t ) are low - pass functions
From (A2.52) we have ( z  v  ju , 2v  z  z*)
~ ~
2h(t )  h (t ) exp( j 2 f c t )  h * (t ) exp(  j 2 f c t ) (A2.53)
ApplyFourier transform to (A2.53)
~ ~
2 H ( f )  H ( f  f c )  H * ( f  f c ) (A2.54)
Since h(t ) is real
H * ( f )  H ( f )
~
and H ( f ) is limited to f  B with B  f c
~
 H ( f  fc )  2H ( f ) , f  0 (A2.55)
~ ~
We can obtain H ( f ) from H ( f ) , H ( f ' )  2 H ( f ' f c ) 39
 x(t) band-pass h(t) y(t)
system

y(t )  Re  y(t )exp( j 2 fct ) (A2.57)


 h( )x(t   )d (A2.58)


Define the pre-envelope of h(t ) as

 
h (t )  h(t )  j h(t ), h(t ) : Hilbert T. of h(t )
H  ( f )  H ( f )  sgn( f ) H ( f )
2 H ( f ) f  0

H  ( f )   H (0) f 0 ( A2.37)
 0 f  0
 
A2.58  y(t )   Re  h ( )Re  x (t   ) d ( A2.59) 40

Recall h  t   h (t )  jhˆ( t )
h (t )  Re  h (t ) 
x (t )  Re  x (t ) 
To prove (A2.60)
Re  h ( ) x (t   )d 
 

  

 Re  [h ( )  jhˆ( )][ x ( t   )  jxˆ ( t   )]d 

 

  
 
  h ( ) x (t   )d   hˆ( ) xˆ (t   )d
 
   t     ,   t 
 
-
h( )x(t- )d  
 
1
  
1
 u h (u ) xˆ (t   )d du,
d  d
  
 

h ( ) x (t   )d  
 
1
  t 
1
u  xˆ ( )d h (u )du
 
 

h ( ) x (t   )d  

h (u ) x (t  u )du

 2 h ( ) x (t   )d


 2 Re[h ( )] Re[ x ( t   )]d 41

(A2.58) becomes
y (t )   Reh ( )Rex (t   ) d

(A2.59)


 Re  h ( )x (t   )d 
1  
2   
1   ~
 Re  h ( ) exp( j 2fc ) ~x (t   ) exp( j 2fc (t   ))d 
2   
1 
x (t   )d 
 ~
 Re exp( j 2fct )  h ( )~
2   

42
 
(1) F{e j 2nfc t
} e j 2nfc t  j 2f c t
e dt


 e  j 2 ( f  nfc ) t
dt


  ( f  nf c )
 
dk
(2) F {e j 2nfct }   e j 2nfct e  j 2f ct dt , 令 nt  k , dt 

n
 f
 j 2 k dk

j 2f c k
e e n
,n  0

={ 



e j 2f c k
e
 j 2
f
n
k
n
dk
n
,n  0

 f
1  j 2 ( n  f c ) k
 e dk , n  0

={ n 


n 
e

 1  j 2 ( n  f c ) k
f
dk , n  0

1 f
 (  f c ), n  0

{ n n
= 1 f
n
 (  f c ), n  0
n
1 f
  (  f c )   ( f  nf c )
n n
Comparing (A2.57) and (A2.61) we have
 ~
2 y (t )   h ( ) ~
~ x (t   ) d (A2.62)

~
or 2 y (t )  h (t ) * ~
~ x (t ) (A2.63)
We can represent bandpass signals and systems
by theequivalent lowpass functions ~ x(t), ~y(t)
~
and h(t) without the factor exp( j 2fct )

43
2~y (t )  hI (t )  jh Q (t ) xI (t )  jxQ (t ) (A2.64)
 hI (t )  xI (t )  hQ (t )  xQ (t )
 jhQ (t )  xI (t )  hI (t )  xQ (t ) (A2.65)
let ~
y (t )  ~
y I (t )  j~y Q (t ) (A2.66)
2yI(t)  hI (t )  xI (t )  hQ (t )  xQ (t ) (A2.67)
2yQ(t)  hQ (t )  xI (t )  hI (t )  xQ (t ) (A2.68)

44
45
Procedure for evaluating the response
of a band-pass system

1. Replace x(t ) by ~ x (t )
x(t )  Re ~x (t ) exp( j 2fct ) 
~
2. h(t )  Re h (t ) exp( j 2fct ) 
~
3. Obtain 2 y (t )  h (t ) * ~
~ x (t )
4. y (t )  Re ~y (t ) exp( j 2fct ) 

46
To simplify the analysis
1. shift H 1( f ) to the right by f c to align to the band-pass frequency
2. set H 1( f  f c )  2 H 1( f ) , for f  0 (2.61)
Recall
 BT BT BT
 j 2πa( f  f c  ) f c   f  f c 
2 2 2

 BT BT BT
H 1( f )   j 2πa( f  f c  )  fc   f   fc  (2.60)
 2 2 2
 0 elsewhere


From (2.60) and (2.61), we get
 BT BT BT
 j 4 a( f  )   f 
H 1( f )   2 2 2
 (2.62)
0 elsewhere
47
Recall FM signal s( t )

s(t )  Ac cos 2 f ct  2 k f  m( )d 

 t

 0 
The complex envelope is

s (t )  Ac exp j 2 k f  m( )d 

 t
(2.63)
 0 
Let s 1 (t ) denote the complex envelope of the slope ckt. response output.
Recall (A2.63) 2 y(t )  h(t )  x(t ) , we have
1
S1( f )  H 1( f )S ( f ) (upper arm of Fig 2.30 in text)
2
 B B B
 j 2 a( f  T )S ( f )  T  f  T (2.64)
 2 2 2

 0 elsewhere
 d s (t ) 
 s 1(t )  a   j BT s (t ) (2.65)
 dt 
From (2.63) and (2.65) , we have
 2k f 
m(t ) exp j 2 k f  m( )d 
 t
s 1(t )  j BT aAc 1  (2.66)
  48
 BT 
0
 s1(t )  Re~
s 1(t ) exp( j 2 fct )
 2k f   t 
  BT aAc 1  m(t ) cos 2 f ct  2 k f  m( )d   (2.67)
 BT   0 2
 sin 2 fct  2 k f  m( )d 
t

 0 
s1(t ) is a hybrid-modulated signal (amplitude , frequency)
2k f
However, provided that we choose m(t )  1, for all t
BT
using an envelope detector, we have
 2k f 
s 1(t )   BT aAc 1  m(t ) (2.68)
 BT 
The bias term  BT aAc can be removed by a second frequency
discriminator with H 2 ( f ) , where H 2 ( f )  H1(  f ).

49
 Balanced Frequency Discriminator

Let the transfer function of the second branch of Fig 2.30

be (complementary slope circuit)
~ ~
H 2( f )  H 1( f ) (2.69)
~  2k f 
s 2(t )   BT aAc 1  m(t ) (2.70)
 BT 
s0 (t )  ~
s 1(t )  ~
s 2(t )
 4 k f aAc m(t ) (2.71) 50
 FM Stereo Multiplexing
Two factors which influence FM stereo standards
1.Operation within the allocated FM channels.
2.Compatible with monophonic radio receiver.

m(t )  ml (t )  mr (t )  ml (t )  mr (t )cos(4 f ct )  K cos(2 f ct ) (2.72)

51
 Stereo FM

Figure 9-40. FM stereo generation block diagram.

51-1
 Stereo FM
 In Figure 9-40, audio signals from both left and right
mircrophones are combined in an linear matrixing network
to produce an L+R signal and an L-R signal.

 Both L+R and L-R are signals in the audio band and must
be separated before modulating the carrier for transmission.
This is accomplished by translating the L-R audio signal
up in the spectrum.

 As seen in Figure 9-40, the frequency translation is

achieved by amplitude-modulating a 38-kHz subsidiary
carrier in a balanced modulator to produce DSB-SC.

51-2
 Stereo FM Transmitter

Stereo FM transmitter using frequency-division multiplexing.

51-3
 Stereo FM Transmitter

Stereo FM transmitter: (a) block diagram; (b) resulting spectrum.

SAC: Subsidiary Communication Authorization 51-4
 Stereo FM
 The stereo receiver will need a frequency-coherent 38-kHz
reference signal to demodulate the DSB-SC.

 To simplify the receiver, a frequency- and phase-coherent

signal is derived from the subcarrier oscillator by frequency
division (÷2) to produce a pilot.

 The 19-kHz pilot fits nicely between the L+R and DSB-SC L-
R signals in the baseband frequency spectrum.

51-5
 Stereo FM
 As indicated by its relative amplitude in the baseband
composite signal, the pilot is made small enough so that
its FM deviation of the carrier is only about 10% of the
total 75-kHz maximum deviation.

 After the FM stereo signal is received and demodulated to

baseband, the 19-kHz pilot is used to phase-lock an
oscillator, which provides the 38-kHz subcarrier for
demodulation of the L-R signal.

 A simple example using equal frequency but unequal

amplitude audio toned in the L and R microphones is used
to illustrate the formation of the composite stereo (without
pilot) in Figure 9-41.

51-6
 Stereo FM

Figure 9-41. Development of composite stereo signal. The 38 kHz alternately

multiplies L-R signal by +1 and –1 to produce the DSB-SC in the balanced AM
modulator (part d). The adder output (shown in e without piot) will be filtered to
51-7
reduce higher harmonics before FM modulation.
 Stereo FM

Spectrum of stereo FM signal.

SCA: Subsidiary communication authorization
(commercial-free program)
51-8
 51-9

Reference : G. M. Miller “Modern Electronic Communication” 5th Edition, Prentice Hall

 2.8 Nonlinear Effects in FM Systems
1.Strong nonlinearity, e.g., square-law modulators ,
hard limiter, frequency multipliers.
2.Weak nonlinearity, e.g., imperfections
Nonlinear input-output relation

v0 (t )  a1vi (t )  a2vi (t )  a3vi (t )

2 3
(2.73)

vi(t) v0(t)
Nonlinear
Channel (device)

52
For FM signal
vi (t )  Ac cos2 f c t   (t )
t
 (t )  2 k f  m( )d
0

v0 (t )  a1 Ac cos 2 f c t   (t )  a2 Ac cos 2 f c t   (t )

  2 2

 a3 Ac3 cos 3 2 f c t   (t ) (2.74)

 a2 Ac  (a1 Ac  a3 Ac3 ) cos2 f c t   (t )

1 2 3
2 4
 a2 Ac2 cos4 f c t  2 (t )
1
2
 a3 Ac3 cos6 f c t  3 (t )
1
(2.75)
4
53
 Carson' s rule , BT  2f  2 f m  2f  2W
W 2f W W 4f W

fc 2fc
In order to seperate the desired FM signal from the second
harmonic , we have
2 f c  (2f  W )  f c  f  W
f c  3f  2W (2.76)
The output of the band-pass filter is
3
v0 '(t )  ( a1 Ac  a3 Ac 3 )cos  2 f ct   (t ) ( no effect to m(t ))
4
An FM system is extremely sensitive to phase nonlinearities.
Common type of source : AM-to -PM conversion.
54
 2.9 Super Heterodyne Receiver
(Carrier-frequency tuning , filtering , amplification , and demodulation)

AM radio receiver

fIF=fLO-fRF (2.78)

A FM system may use a limiter to remove amplitude variations.

55
Commercial FM Broadcast、
Allocations and Sidebands

56
 2.10 Noise in CW modulation System
1.Channel model: additive white Gaussian noise (AWGN)
2.Receiver model: a band-pass filer followed by an ideal demodulator

The PSD of w(t) is denoted by N0

2.

57
The filtered noise in narrowband noise representation :
n(t )  nI (t ) cos(2 f c t )  nQ (t ) sin( 2 f c t ) (2.79)
The filtered signal for demodulation is
x(t )  s (t )  n(t ) (2.80)
The channel signal - to - noise ratio
average power of s (t )
(SNR )C 
average power of n(t )
The output signal - to - noise ratio
average power of the demodulated signal
(SNR )O 
average power of noise at the output
(SNR)O
Figure of merit  (2.81)
(SNR)C

58
2.11 Noise in Linear Receiver Using Coherent Detection
The DSB-SC system

s(t )  CAc cos(2 f ct )m(t ) m(t )  S M ( f )

W
P S M ( f )df (2.83)
W

C 2 Ac 2 P
(SNR)C,DSB  2
WN 0
C 2 Ac 2 P
 (baseband) (2.84)
2WN 0
C:system dependent scaling factor 59
x(t )  s (t )  n(t )
 CAc cos(2 f c t )m(t )  nI (t ) cos(2 f c t )  nQ (t ) sin( 2 f c t ) (2.85)
v(t )  x(t )cos (2 f c t )
1 1
 CAc m(t )  nI (t )
2 2
 CAc m(t )  nI (t )cos(4 f c t )  nQ (t ) sin( 4 f c t )
1 1
2 2
high frequency components
1 1
Low - pass filter  y (t )  CAc m(t )  nI (t ) (2.86)
2 2
(2.86) indicates :
1. m(t ) and nI (t ) are additive at the receiver output.
2. nQ (t ) is completely rejected by thecoherent detector.
60
 1
The average output signal ( CAc m(t )) power  C Ac
2 2 P

2 4
Let BT  2W
1 1 2 1
The average noise ( nI (t )) power  ( ) 2WN 0  WN 0
2 2 2
2 2 P
C Ac 2 2
C Ac P
(SNR ) O,DSBSC  4  (2.87)
W N0 2WN 0
2
(SNR)O
1 (2.88)
(SNR)C DSB-SC
1. Coherent SSB has the same figure of merit of DSB - SC
2. No trade - off between performance and bandwidth.
Serious problem! 61
 2.12 Noise in AM Receivers Using Envelope Detection
s(t )  Ac 1  ka m(t )cos(2 f ct ) (2.89)
 Ac cos(2 f ct )  Ac ka m(t ) cos(2 f ct )

Ac (1  k a P)
2 2
(SNR )C, AM  (2.90)
2WN 0
At theoutput of the filter :
x(t )  s(t )  n(t )
 Ac  Ac k a m(t )  nI (t )cos(2 f c t )  nQ (t ) sin( 2 f c t ) (2.91)

62
y (t )  envelope of x(t )

 A  A k m(t )  n (t )
c c a I
2 2

 nQ (t )
1
2
(2.92)

Assume Ac  Ac k a m(t )  nI (t )  nQ (t )
y (t )  Ac  Ac k a m(t )  nI (t )
2
A
1. c  WN 0 (carrier power > noise power)
2
2. k a  1
2 2
A k P
(SNR)O, AM  c a (2.94)
2WN 0
2
(SNR)O ka P
 (2.95)
(SNR)C AM 1  k a 2 P 63
Supplements
Define the pre-demodulation SNR as
The average power of the modulated signal
SNR pre-de = The average noise power at the input of the demodulator

modulated Band pass

signal demodulator m(t)
s(t) filter
n(t) (SNR)pre-de (SNR)o
The Bandwidth of the bandpass filter is B T then the average noise power at the input of
the demodulator is N o BT S (f) N
N0
2
f
BT BT
Ac (1  k a p) Ac (1  k a p)
2 2 2 2
AM 2
For an AM system SNR  
pre  de N o BT 2 N o BT
Ac (1  k a p)
2 2
AM
If B T =2W SNR 
pre  de 4 N oW
Supplements
For a DSB-SC system, DSB  SC C 2 Ac P
2
2 2
SNR  2  C Ac p
pre  de N o BT 4 N oW
為與課本一致加C 2
For an FM system 2 2
FM Ac 2 Ac
SNR  
pre  de N o BT 2 N o BT
If using Carson’s rule, we have
B T =2Δf+2fm>> fm =w
For the purpose of comparing different CW modulation systems, we define
The average power of the modulated signal
(SNR)c= The average power of channel noise in the message band

Message signal with LP filter

the same power as output
modulated wave with bandwidth w
noise
n(t)
The equivalent baseband transmission model.
Supplements
More precisely, we may express the DSB-SC
as m(t) S‘(t)

cos(2πfc t+θ)
θ is uniformly distributed over ﹝0, 2π﹞
S'(t)=Ac m(t) cos(2πfc t+θ)
At the receiver we may write
S(t)=C Ac m(t) cos(2πfc t+θ)

Ps  E S 2 (t )  R s (0)

 

S x ( f ) df

 E (CAc m(t ) cos( 2f c t   )) 2 

 C 2 Ac E cos 2 ( 2f c t   ) E m 2 (t )
2
  
 C 2 Ac Rm (0) 2  C 2 Ac P
2 2
2
w
R m ( 0)  P  w
S m ( f ) df

The average noise power in –w<f<w w N0

Pn  
w 2
df  N oW
Supplements
The average power of S(t)
SNRc=
The average power of channel noise in the message band
The average power of the modulated signal
=
The average power of channel noise in the message band
2
Ps C 2 Ac P
= =
Pn 2 N oW
For convenience we write the modulated signal
as S (t )  CAc m(t ) cos(2f c t ) θ不出現

Since cos(2f c t   ) is ergodic and we take cos(2f c t ) as a sample function

Ps  C 2 Ac2 Rm (0) [time average of [cos (2f c t ) ]]

2

 C 2 Ac2 P 2

C 2 Ac2 P 2 C 2 Ac2 P
SNRc  
N 0W 2 N 0W
64
Threshold Effect

noise power > carrier power

65
 2.13 Noise in FM Receivers

The discriminator consists of a slope network and an

envelope detector.
Let n(t )  nI (t ) cos(2 f ct )  nQ (t ) sin(2 f ct )
n(t )  r (t ) cos(2 f ct )  (t ) (2.130)

The envelope is r (t )  nI (t )  nQ (t )
2 2
1
2
(2.131)
 nQ (t ) 
The phase is  (t )  tan  1
 (2.132)
 I 
n ( t )

where r (t ) is Rayleigh distributed, and Ψ (t ) is uniform distributed over 2 .

r r2
f R (r )  2 exp(  2 ), r  0 (1.115)
 2
1
fΨ ( )  , 0    2 (1.114) 66
2
The incoming FM signal s(t) is defined by
s(t )  Ac cos 2 f ct  2 k f  m( )d 
 t
(2.133)
 0 
 Ac cos2 f ct   (t ) (2.135)
t
where  (t )  2 k f  m( ) d (2.134)
0
At the bandpass filter output
x(t )  s(t )  n(t )
 Ac cos2 f ct   (t )  r (t ) cos2 f ct  (t ) (2.136)

 
  
 
Ac

where Ac  r (t )
 r (t ) sin (t )   (t ) 
 (t )   (t )  tan 
1
 (2.137)
 Ac  r (t ) cos (t )   (t ) 
67
Note that the envelope of x(t) is of no interest to us (limiter)
Because Ac  r (t )

sin (t )   (t )
r (t )
 (t )   (t )  (2.138)
Ac
sin (t )   (t ) (2.139)
t r (t )
 2 k f  m( )d 
0 Ac
The discrimina tor output is (Fig 2.40)
1 d (t )
v(t ) 
2 dt
 k f m(t )  nd (t ) (2.140)
message additive noise
where

n d (t ) 
1 d
 r (t ) sin  (t )   (t )   (2.141)
2 Ac dt
68
Assume  (t )   (t ) is uniformly distributed over (0, 2 ),
then nd (t ) is independent of message signal.
We may simplify nd (t ) as

nd (t ) 
1 d
r (t ) sin (t ) (2.142)
2 Ac dt
From definition of r (t ) and  (t ) , we have
nQ (t )  r (t ) sin (t ) (2.143)
1 dnQ (t )
nd (t )  (2.144)
2 Ac dt
The quadrature component appears
69
From (2.140)
The average output signal power = kf2P
Recall d F .T
 j 2f
dt

nQ(t) 1 d nd(t)
S NQ ( f ) 2 Ac dt S Nd ( f )

f2
S N d ( f )  2 S NQ ( f ) (2.145)
Ac
noise is enhanced at high frequency
70
Assume that nQ(t) has ideal low-pass characteristic
with bandwidth BT

N0 f 2 BT
S Nd ( f )  2
, f  (2.146)
Ac 2
If BT W
2
At the receiver output
N0 f 2
S N0 ( f )  2
, f W (2.147)
Ac

71
 N0 W 2
Average power of n0 (t )  2  f df
Ac W
2 N 0W 3
 (2.148)
3 Ac2
1
 2 noise quieting effect
2 2 Ac when increasing carrier power
3 Ac k f P
(SNR )O ,FM  3
(2.149)
2 N 0W Ac2
The average power of s(t ) is ,
2
the average noise power in message bandwidth is WN0
Ac2
 (SNR )C ,FM  (2.150)
2WN0
(SNR )O 3k 2f P
  2
(2.151)
(SNR )C W
f  k f Am ( SNR )o,FM  ( f )2
FM
(2.29) 72
 Example 2.5 Single-Tone Modulation
 f 
s(t )  Ac cos 2 f ct  sin(2 f mt )
 t
fm
f 
We may write, 2 k f  m( )d  sin(2 f mt )
0 fm
d f
both side  m(t )  cos(2 f mt )
dt kf
(f ) 2
The average power of m(t ) (across 1 load) is P 
2k 2f
3 Ac2 ( f )2 3 Ac2  2 f
From (2.149), (SNR )O ,FM   ,  
4 N0W 3 4 N0W W

(SNR )O 3 f 3
  ( )2   2 (2.152)
(SNR )C FM 2 W 2
( SNR)O 1
compare to AM ,  (from Example 2.4)
(SNR )C AM 3
3 2 1
When   , FM has better performance.
2 3
2
   0.471
3
Define   0.5 as the transition between narrowband FM and wideband FM . 73
 FM Threshold Effect (When CNR is low)

When there is no signal, i.e., carrier is unmodulated.

The composite signal at the frequency discriminator input
x(t )   Ac  nI (t ) cos(2 f ct )  nQ (t )sin(2 fct ) (2.153)
nQ (t )
 ( t )  tan 1

Ac  nI (t )
Occasionally, P1 may sweep around the origin , ( r (t )> Ac )
 (t) increases or decreases 2
 ' (t )
The discriminator output is equal to
P 1 2
r(t)
nQ(t) x(t)

 P2
0
nI(t) Ac 74
Figure 2.44 Illustrating impulselike components in  (t) 
d (t)/dt produced by changes of 2 in  (t); (a) and (b)
are graphs of (t) and  (t), respectively.
75
A positive-going click occurs , when
d (t )
r (t )  Ac ,  (t )     (t )  d (t ) ,  0
dt
A negative-going click occurs when
d (t )
r (t )  Ac ,  (t )     (t )  d (t ) ,  0
dt
The carrier-to -noise ratio is defined by
2
Ac
 (2.154)
2 BT N 0
The output signal-to-noise ratio is calculated as
1. The average output signal power is calculated assuming
BT
a sinusoidal modulation which produces f  . (noise free)
2
2. The average output noise power is calculated when no
signal is present (The carrier is unmodulated). 76
 Figure 2.45 Dependence of
output signal-to-noise ratio on
input carrier-to-noise ratio for
FM receiver. In curve I, the
average output noise power is
calculated assuming an
unmodulated carrier. In curve
II, the average output noise
power is calculated assuming
a sinusoidally modulated
carrier. Both curves I and II
are calculated from theory.

Ac2 Ac2
When    20 or  20 BT N 0 (2.155),
2 BT N 0 2
threshold effects may be avoided 77
The procedure to calculate minimum Ac (   20)
1. Given  and W, determine BT
(using Figure 2.26 or Carson's rule)
2
A
2. Given N0 , we have c
 20 BT N 0
2
Capture Effect:
The receiver locks onto the stronger signal
and suppresses the weaker one.

78
 FM Threshold Reduction (tracking filter)
• FM demodulator with negative feedback (FMFB)
• Phase locked loop

Figure 2.46
FM threshold extension.

Figure 2.47
FM demodulator with
negative feedback.

79
 Pre-emphasis and De-emphasis on FM

Figure 2.48 (a) Power spectral density of noise at FM receiver o

(b) Power spectral density of a typical message signal.

Figure 2.49 Use of pre-emphasis and de-emphasis in an FM system.

80
 1
H de ( f )  , W  f  W (2.156)
H pe ( f )
The PSD at the discrimina tor output is
N0 f 2 BT
S Nd ( f )  2
, f  (2.146)
Ac 2
N0 f 2 BT
H de ( f ) S N d ( f )  H de ( f ) , f 
2 2
2
(2.157)
Ac 2
 Average output noise  N 0 W 2
   2  f H de ( f ) df (2.158)
2

 power with de - emphsis  Ac W

The improvement factor I is
2W 3
I w
(2.162)
3
2
f 2 Hde( f ) df
-w
81
Example 2.6 Figure 2.50 (a) Pre-emphasis filter.
(b) De-emphasis filter.

A simple pre - emphsis filter response is

jf
H pe ( f )  1 
f0
A de - emphsis filter response is
1
H de ( f ) 
1 j f
f0

2W 3 (W )3
f0
I  (2.161)
f 2 df 3 (W )  tan 1 (W )
W
3  f0 f 0 
W
1 ( f ) 2
f0 82
 Preemphasis for FM
 The main difference between FM and PM is in the
relationship between frequency and phase.
f = (1/2).d/dt.
 A PM detector has a flat noise power (and voltage) output
versus frequency (power spectral density). This is
illustrated in Figure 9-38a.
 However, an FM detector has a parabolic noise power
spectrum, as shown in Figure 9-38b. The output noise
voltage increases linearly with frequency.
 If no compensation is used for FM, the higher audio
signals would suffer a greater S/N degradation than the
lower frequencies. For this reason compensation, called
emphasis, is used for broadcast FM.

83
 Preemphasis for FM

Figure 9-38. Detector noise output spectra

for (a). PM and (b). FM.

84
 Preemphasis for FM
 A preemphasis network at the modulator input
provides a constant increase of modulation index mf
for high-frequency audio signals.
 Such a network and its frequency response are
illustrated in Figure 9-39.

Fig. 9-39. (a)Premphasis network, and

(b) Frequency response. 85
 Preemphasis for FM
 With the RC network chosen to give  = R1C = 75s in North
America (150s in Europe), a constant input audio signal will
result in a nearly constant rise in the VCO input voltage for
frequencies above 2.12 kHz. The larger-than-normal carrier
deviations and mf will preemphasize high-audio frequencies.

 At the receiver demodulator output, a low-pass RC network

with  = RC = 75s will not only decrease noise at higher
audio frequencies but also deemphasize the high-frequency
information signals and return them to normal amplitudes
relative to the low frequencies.

 The overall result will be nearly constant S/N across the 15-
kHz audio baseband and a noise performance improvement of
about 12dB over no preemphasis. Phase modulation systems
do not require emphasis.
86
 Pre-emphasis and De-emphasis on FM

Preemphasis and deemphasis: (a) schematic diagrams; (b) attenuation curves

87
Pre-emphasis and De-emphasis on FM

Example of S/N without preemphasis and deemphasis. 88

Pre-emphasis and De-emphasis on FM

Example of S/N with preemphasis and deemphasis. 89

Dolby dynamic preemphasis
90
Figure 2.55 Comparison of the noise performance of various CW modulation
systems. Curve I: Full AM,  = 1. Curve II: DSB-SC, SSB. Curve III: FM,  = 2.
Curve IV: FM,  = 5. (Curves III and IV include 13-dB pre-emphasis, de- 91
emphasis improvement.)
In making the comparison, it is informative to keep in
mind the transmission bandwidth requirement of the
modulation systems in question. Therefore, we define
normalized transmission bandwidth as
BT
Bn 
W

Table 2.4 Values of Bn for various CW modulation schemes

FM

AM, DSB-SC SSB 2 5

Bn 2 1 8 16

92
李家同教授-我的恩師