Fourier Series

主講者：虞台文
Content
 Periodic Functions
 Fourier Series
 Complex Form of the Fourier Series
 Impulse Train
 Analysis of Periodic Waveforms
 Half-Range Expansion
 Least Mean-Square Error Approximation
Fourier Series

Periodic Functions
The Mathematic Formulation
 Any function that satisfies

f (t )  f (t  T )
where T is a constant and is called the period
of the function.
 Example:
t t
f (t ) cos  cos Find its period.
3 4
t t 1 1
f (t )  f (t  T ) cos  cos cos (t  T )  cos (t  T )
3 4 3 4

Fact: cos  cos(  2m)

T
2m T 6m
3 T 24 smallest T
T
2n T 8n
4
 Example:
f (t ) cos 1t  cos 2t Find its period.

f (t )  f (t  T ) cos 1t  cos 2t cos 1 (t  T )  cos 2 (t  T )

1T 2m 1 m 1 must be a


2 n 2 rational number
2T 2n
Example:
f (t ) cos10t  cos(10  )t
Is this function a periodic one?

1 10 not a rational

2 10   number
Fourier Series

Fourier Series
Introduction
 Decompose a periodic input signal into
primitive periodic components.

AAperiodic
periodicsequence
sequence
f(t)

t
T 2T 3T
Synthesis
a0  2nt  2nt
f (t )    an cos   bn sin
2 n 1 T n 1 T
DC Part Even Part Odd Part

T is a period of all the above signals

Let 0=2/T.
a0  
f (t )    an cos( n0t )   bn sin( n0t )
2 n 1 n 1
Orthogonal Functions
a set of functions {k} orthogonal
 Call

on an interval a < t < b if it satisfies

b 0 m n
a m (t )  n (t )dt 
rn m n
Orthogonal set of Sinusoidal
Functions

Define 0=2/T.
T /2 T /2

 T /2
cos(m0t )dt 0, m 0  T /2
sin( m0t )dt 0, m 0

T /2  0 m n
 T / 2 cos(m0t ) cos(n0t )dt T / 2 m n We now prove this one
T /2  0 m n
 T / 2 sin(m0t ) sin(n0t )dt T / 2 m n
T /2

 T /2
sin( m0t ) cos(n0t )dt 0, for all m and n
 1
cos  cos   [cos(  )  cos(  )]
2

Proof
mn
T /2
 T /2
cos(m0t ) cos(n0t )dt
1 T /2 1 T /2
  cos[(m  n)0t ]dt   cos[(m  n)0t ]dt
2  T /2 2  T /2
1 1 T /2 1 1 T /2
 sin[(m  n)0t ]  T / 2  sin[(m  n)0t ]  T / 2
2 (m  n)0 2 (m  n)0
1 1 1 1
 2 sin[( m  n)]  2 sin[( m  n)]
2 (m  n)0 2 (m  n)0
0
0 0
 1
cos  cos   [cos(  )  cos(  )]
2

Proof 2 1
cos   [1  cos 2]
2

m=n
T /2

 T /2
cos(m0t ) cos(n0t )dt
T /2 1 T /2
 cos (m0t )dt   [1  cos 2m0t ]dt
2
 T /2 2  T /2
T /2 T /2
1 1
 t  sin 2m0t ]
2  T / 2 4m0  T /2

0
T
 T /2  0 m n
2
 T / 2 cos(m0t ) cos(n0t )dt T / 2 m n
 Orthogonal set of Sinusoidal
Functions

Define 0=2/T.
11m,, t )dt 0, m 0  sin(m t )dt 0,
T /2 T /2
 cos( 0 0 m 0
 
 T /2  T /2

cos  22   
 cos  00tt,n,cos
cos  tt,,cos
cos 3
3  t
0 
t,,
 0 m  n
 
T /2
cos( m  t ) cos(
0  t )
0dt 00 0 
sin  t , sin 2Tt ,/ sin
 T /2 2 m n
3  t ,  
sin 00t , sin 200t0, sin
T /2
300t ,  
m n
 sin(m t ) sin(n t )dt T / 2 m n
 T /2
0 0

T /2 an
an orthogonal
orthogonal set.
set.

 T /2
sin( m t ) cos(n t )dt 0, for all m and n
0 0
Decomposition
a0  
f (t )    an cos( n0t )   bn sin( n0t )
2 n 1 n 1

2 t0 T
a0   f (t )dt
T t0
2 t0 T
an   f (t ) cos n0tdt n 1,2, 
T t0
2 t0 T
bn   f (t ) sin n0tdt n 1,2, 
T t0
 Proof
Use the following facts:
T /2 T /2

 T /2
cos(m0t )dt 0, m 0  T /2
sin( m0t )dt 0, m 0

T /2  0 m n
 T / 2 cos(m0t ) cos(n0t )dt T / 2 m n
T /2  0 m n
 T / 2 sin(m0t ) sin(n0t )dt T / 2 m n
T /2

 T /2
sin( m0t ) cos(n0t )dt 0, for all m and n
 Example (Square Wave)
f(t)
1

-6 -5 -4 -3 -2 -  2 3 4 5

2 
a0  1dt 1
2 0
2  1 
an   cos ntdt  sin nt 0 0 n 1,2, 
2 0 n
2  1  1 2 / n n 1,3,5, 
bn 
2 0 sin ntdt  n cos nt 0  n (cos n  1) 0 n 2,4,6, 
 11 22 11 11 
ff ((tt)) 
  sinsintt  sin
sin33tt  sin
sin55tt 

22  33 55 
Example (Square Wave)
f(t)
1

-6 -5 -4 -3 -2 -  2 3 4 5

2 
a0  1dt 1
2 0
2  1 
an   cos ntdt  sin nt 0 0 n 1,2, 
2 0 n
1  1  1 2 / n n 1,3,5, 
bn  sin ntdt  cos nt 0  (cos n  1) 
2 0 n n 0 n 2,4,6, 
 11 22 11 11 
ff ((tt)) 
  sinsintt  sin
sin33tt  sin
sin55tt 

22  33 55 
Example (Square Wave)
f(t)
1

-6 -5 -4 -3 -2 -  2 3 4 5

2 
a0  1dt 1
1.5 2
0

1
2  1 
an   cos ntdt  sin nt 0 0 n 1,2, 
0.52 0 n
0
1  1  1 2 / n n 1,3,5, 
 sin ntdt 
bn -0.5 cos nt 0  (cos n  1) 
2 0 n n 0 n 2,4,6, 
Harmonics
a0  2nt  2nt
f (t )    an cos   bn sin
2 n 1 T n 1 T
a0  
f (t )    an cos( n0t )   bn sin( n0t )
2 n 1 n 1

DC Part Even Part Odd Part

T is a period of all the above signals

Harmonics
2
Define 0 2f 0  , called the fundamental angular frequency.
T
Define n n0 , called the n-th harmonic of the periodic function.

a0  
f (t )    an cos n0t   bn sin n0t
2 n 1 n 1

a0  
f (t )    an cos nt   bn sin nt
2 n 1 n 1
 Harmonics
a0  
f (t )    an cos nt   bn sin nt
2 n 1 n 1

a0 
   (an cos nt  bn sin nt )
2 n 1
a0   an bn 
2
   an  bn
2
cos nt  sin nt 
2 n 1  a2  b2 a 2
 b 2 
 n n n n 
a0 
   an2  bn2 cos  n cos nt  sin  n sin nt 
2 n 1

C0   Cn cos(nt   n )
n 1
Amplitudes and Phase Angles


f (t ) C0   Cn cos(nt   n )
n 1

harmonic amplitude phase angle

a0
C0 
2 1 bn 
 n tan  
Cn  an2  bn2  an 
Fourier Series

Complex Form of the

Fourier Series
Complex Exponentials
jn0t
e cos n0t  j sin n0t
 jn0t
e cos n0t  j sin n0t

1 jn0t
cos n0t  e
2

 e  jn0t 
sin n0t 
1 jn0t
2j
e e 
 jn0t
 e
2

j jn0t

 e  jn0t 
Complex Form of the Fourier Series

a0  
f (t )    an cos n0t   bn sin n0t
2 n 1 n 1

a0 1  j 
   an e
2 2 n 1

jn0t
e  jn0t
 
  bn e jn0t  e  jn0t
2 n 1

a0   1 1 
    (an  jbn )e jn0t
 (an  jbn )e  jn0t 
2 n 1  2 2  aa0
cc0 
0 2
0

 

c0   cn e jn0t  c n e  jn0t 2
11
n 1 ccn  (a  jbn ))
n 2 ( ann  jb n
2
11
cc n  (a  jbn ))
 n 2 ( ann  jb n
2
Complex Form of the Fourier Series

 

f (t ) c0   cn e jn0t  c n e  jn0t
n 1

 1
c0   cn e jn0t
  cn e jn0t
n 1 n 
aa0
 cc0 
0 2
0

  cn e jn0t 2
n 
11
ccn  (a  jbn ))
n 2 ( ann  jb n
2
11
cc n  (a  jbn ))
 n 2 ( ann  jb n
2
Complex Form of the Fourier Series

a0 1 T /2
c0  
2 T 
 T /2
f (t )dt

1
cn  (an  jbn )
2
1  T /2
  f (t ) cos n0tdt  j  f (t ) sin n0tdt 
T /2

T   T / 2  T /2 
1 T /2 aa0
  f (t )(cos n0t  j sin n0t )dt cc0 
0 2
0
T  T /2 2
1 T /2 11
  f (t )e  jn0t dt ccn  (a  jbn ))
T  T /2 n 2 ( ann  jb n
2
1 1 T /2 11
cc n  (a  jbn ))
c n  (an  jbn ) 
2 T  T /2
f (t )e jn0t dt  n 2 ( ann  jb
2
n
 Complex Form of the Fourier Series
 aa0
c 
 00 
0
jnjn0t t c
ff((tt)) 
 ccnnee
nn


0 22
11
ccn  (a  jbn ))
n 2 ( ann  jb
11 TT/ /22 2
n


  jnjn0t t
ccnn   T / 2 f (t )e 0 dt
f ( t ) e dt 11
TT  T / 2 cc n  (a  jbn ))
 n 2 ( ann  jb n
2
cn | cn | e j , c n cn* | cn | e  j
n n

If f(t) is real,
1 2
* | cn || c n | an  bn2
c n c n
2 n 1,2,3, 
 b  1
 n tan  1   n  c0  a0
 an  2
Complex Frequency Spectra
cn | cn | e jn , c n cn* | cn | e  jn  b 
 n tan  1   n  n 1,2,3, 
1  an 
| cn || c n | an2  bn2
2
|cn| amplitude
1 spectrum
c0  a0
2


phase
n spectrum


 Example
f(t)
A

t
T T d d T T
 
2 2 2 2

A d /2 A 1
cn   e  jn0t dt  ( 2 j sin n0 d / 2)
T  d /2 T  jn0
d /2 A 1
A 1  1 sin n0 d / 2
 e  jn0t T 2 n0
T  jn0  d /2
 nd 
sin  
Ad  T 
A 1  jn0 d / 2 1 jn0 d / 2  
  e  e  T  nd 
T   jn0  jn0   
 T 
Example
A/5

-120 -80 -40 0 40 80 120

-150 -100 -50 50 100 150

 nd  1 1 d 1
sin   d , T , 
Ad  T  20 4 T 5
cn 
T  nd  2
  0  8
 T  T
Example
A/10

-120 -80 -40 0 40 80 120

-300 -200 -100 100 200 300

 nd  1 1 d 1
sin   d , T , 
Ad  T  20 2 T 5
cn 
T  nd  2
   0  4
 T  T
Example
f(t)
A

t
T 0 d T

A d A 1
cn   e  jn0t dt  (1  e  jn0 d )
T 0 T jn0
d A 1  jn0 d / 2 jn0 d / 2  jn0 d / 2
A 1  e (e e )
 e  jn0t T jn0
T  jn0 0
 nd 
sin  
Ad  T  e  jn0 d / 2
A 1  jn0 d 1  
  e   T  nd 
T   jn0  jn0   
 T 
Fourier Series

Impulse Train
Dirac Delta Function
 0 t 0 
(t ) 
 t 0
and  (t )dt 1


Also called unit impulse function.

0 t
Property

 ((tt))((tt))dtdt ((00))

(t): Test Function

  
 (t )(t )dt  (t )(0)dt (0) (t )dt (0)
  
Impulse Train

3T 2T T 0 T 2T 3T t




TT ((tt))  ((tt 
 nT
nT )
)
nn


 Fourier Series of the Impulse Train



TT ((tt))  ((tt  nT ) 2 2
T /2
nT ) a0   T (t )dt 
nn T  T /2 T


2 T /2 2
an   T (t ) cos(n0t )dt 
T  T /2 T
2 T /2
bn   T (t ) sin(n0t )dt 0
T  T /2
11 22 
  
TT ((tt))   cosnn
cos 00tt
TT TT nn
 Complex Form
Fourier Series of the Impulse Train



TT ((tt))  ((tt  nT
nT))
nn

 a0 1 T /2 1
c0  
2 T  T / 2 T (t )dt T
1 T /2 1

 jn0t
cn  T (t )e dt 
T  T /2 T
11  jnjn00tt
 
TT ((tt))   ee
TT nn
Fourier Series

Analysis of
Periodic Waveforms
Waveform Symmetry
Even Functions
f (t )  f ( t )
Odd Functions

f (t )  f ( t )
Decomposition
 Any function f(t) can be expressed as the
sum of an even function fe(t) and an odd f
unction fo(t).
f (t )  f e (t )  f o (t )
f e (t )  12 [ f (t )  f ( t )] Even Part

f o (t )  12 [ f (t )  f ( t )] Odd Part
Example
e  t t 0
f (t ) 
0 t 0

 12 e  t t 0 Even Part
f e (t )  1 t
 2e t 0

 12 e  t t 0 Odd Part
f o (t )  1 t
 2 e t 0
Half-Wave Symmetry
f (t )  f (t  T ) and f (t )  f t  T / 2 

T/2 T/2 T
Quarter-Wave Symmetry
Even Quarter-Wave Symmetry

T/2 T/2 T

Odd Quarter-Wave Symmetry

T/2 T/2
T
Hidden Symmetry
 The following is a asymmetry periodic function:
A

T T
 Adding a constant to get symmetry property.
A/2

T T
A/2
Fourier Coefficients of
Symmetrical Waveforms

 Theuse of symmetry properties simplifies the

calculation of Fourier coefficients.
– Even Functions
– Odd Functions
– Half-Wave
– Even Quarter-Wave
– Odd Quarter-Wave
– Hidden
Fourier Coefficients of Even Functions

f (t )  f ( t )

a0 
f (t )    an cos n0t
2 n 1
4 T /2
an   f (t ) cos(n0t )dt
T 0
Fourier Coefficients of Even Functions

f (t )  f ( t )

f (t )  bn sin n0t
n 1

4 T /2
bn   f (t ) sin(n0t )dt
T 0
Fourier Coefficients for Half-Wave Symmetry

f (t )  f (t  T ) and f (t )  f t  T / 2 

T/2 T/2 T

The
The Fourier
Fourier series
series contains
contains only
only odd
odd harmonics.
harmonics.
Fourier Coefficients for Half-Wave Symmetry

f (t )  f (t  T ) and f (t )  f t  T / 2 

f (t )  (an cos n0t  bn sin n0t )
n 1

0 for n even

an  4 T / 2

T 0 f (t ) cos( n 0 t ) dt for n odd

0 for n even

bn  4 T / 2
T 
0
f (t ) sin( n0t )dt for n odd
Fourier Coefficients for
Even Quarter-Wave Symmetry

T/2 T/2 T


f (t )  a2 n  1 cos[(2n  1)0t ]
n 1

8 T /4
a2 n  1 
T  0
f (t ) cos[(2n  1)0t ]dt
Fourier Coefficients for
Odd Quarter-Wave Symmetry

T/2 T/2
T


f (t )  b2 n  1 sin[(2n  1)0t ]
n 1

8 T /4
b2 n  1 
T  0
f (t ) sin[(2n  1)0t ]dt
 Example
Even Quarter-Wave Symmetry
1
T/2 T/2
T T/4 T/4 T
1

8 T /4 8 T /4
a2 n  1 
T 
0
f (t ) cos[(2n  1)0t ]dt 
T 
0
cos[(2n  1)0t ]dt

T /4
8 n 1 4
 sin[(2n  1)0t ] ( 1)
(2n  1)0T 0
(2n  1)
 44 11 11 
ff ((tt)) 
 coscos
00tt  cos
cos33
00tt  cos
cos55
00tt 

 33 55
Example 

Even Quarter-Wave Symmetry

1
T/2 T/2
T T/4 T/4 T
1

8 T /4 8 T /4
a2 n  1 
T 
0
f (t ) cos[(2n  1)0t ]dt 
T 
0
cos[(2n  1)0t ]dt

T /4
8 n 1 4
 sin[(2n  1)0t ] ( 1)
(2n  1)0T 0
(2n  1)
 Example
Odd Quarter-Wave Symmetry
1
T/2 T/2
T T/4 T/4 T
1

8 T /4 8 T /4
b2 n  1 
T 
0
f (t ) sin[(2n  1)0t ]dt 
T 0
sin[(2n  1)0t ]dt

T /4
8 4
 cos[(2n  1)0t ] 
(2n  1)0T 0
(2n  1)
 44 11 11 
ff ((tt)) 
 sinsin
00tt  sin
sin33
00tt  sin
sin55
00tt 

 33 55
Example 

Odd Quarter-Wave Symmetry

1
T/2 T/2
T T/4 T/4 T
1

8 T /4 8 T /4
b2 n  1 
T 
0
f (t ) sin[(2n  1)0t ]dt 
T 0
sin[(2n  1)0t ]dt

T /4
8 4
 cos[(2n  1)0t ] 
(2n  1)0T 0
(2n  1)
Fourier Series

Half-Range
Expansions
Non-Periodic Function Representation

A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Without Considering Symmetry

 T

A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Expansion Into Even Symmetry

 T=2

A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Expansion Into Odd Symmetry

T=2


A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Expansion Into Half-Wave Symmetry

 T=2

A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Expansion Into
Even Quarter-Wave Symmetry

T/2=2
 T=4

A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Expansion Into
Odd Quarter-Wave Symmetry

T/2=2 T=4


A non-periodic function f(t) defined over (0, )

can be expanded into a Fourier series which is
defined only in the interval (0, ).
Fourier Series

Least Mean-Square
Error Approximation
Approximation a function
a0 k
Use S k (t )    an cos n0t  bn sin n0t 
2 n 1
to represent f(t) on interval T/2 < t < T/2.

Define  k (t )  f (t )  S k (t )
1 T /2 Mean-Square

2
Ek  [ k (t )] dt
T  T /2 Error
Approximation a function
Show that using Sk(t) to represent f(t) has leas
t mean-square property.
1 T /2
Ek   [ k (t )]2 dt
T  T /2
2
1 T /2  a0 k

  T / 2  f (t )  2   an cos n0t  bn sin n0t  dt
T n 1 

Proven by setting Ek/ai = 0 and Ek/bi = 0.

 Approximation a function
1 T /2
Ek 
T   T /2
[ k (t )]2 dt
2
1 T /2  a0 k

  T / 2  f (t )  2   an cos n0t  bn sin n0t  dt
T n 1 

EEkk aa00 11 TT/ /22 EEkk 22 TT/ /22

  

 dt 
ff((tt))dt 00 aann  

 cosnn00tdt
ff((tt))cos tdt 
00
aa00 22 TT  T / 2
 T / 2 aann TT  T / 2
 T / 2

EEkk 22 TT/ /22

bbnn  
 ff((tt))sin nn00tdt 00
TT 
sin tdt 
bbnn  TT/ /22
Mean-Square Error
1 T /2
Ek 
T  T /2
[ k (t )]2 dt
2
1 T /2  a0 k

  T / 2  f (t )  2   an cos n0t  bn sin n0t  dt
T n 1 
22kk
11 TT//22 a
a 1
1
 
EEkk 

TT  T / 2
 T / 2
[[ ff (t
(t)]22
)] dt
dt   

00
44 22 nn11
((aa
22
nn

 b
bn
22
)
n)
Mean-Square Error
1 T /2
Ek 
T   T /2
[ k (t )]2 dt
2
1 T /2  a0 k

  T / 2  f (t )  2   an cos n0t  bn sin n0t  dt
T n 1 
22
kk
11 TT//22 a
a 1
1


TT  T / 2
 T / 2
[[ ff (t
(t)]22
)] dt
dt 
 
 

00
44 22 nn11
((aa
22
nn

 b
bn
22
)
n)
Mean-Square Error
1 T /2
Ek 
T   T /2
[ k (t )]2 dt
2
1 T /2  a0 k

  T / 2  f (t )  2   an cos n0t  bn sin n0t  dt
T n 1 
22

11 TT//22 a
a 1
1


TT  T / 2
 T / 2
[[ ff (t
(t)]22
)] dt
dt 
 
 

00
44 22 nn11
((aa
22
nn

 b
bn
22
)
n)