Digital Signal Processing Digital Signal Processing

Prof. Nizamettin AYDIN

Lecture 7
naydin@[Link]

[Link] Fourier Series & Spectrum

1 2

READING ASSIGNMENTS
License Info for SPFirst Slides
• This Lecture:
• This work released under a Creative Commons License with the – Fourier Series in Ch 3, Sects 3-
3-4, 3-
3-5 & 3-
3-6
following terms: • Replaces pp. 62-66 in Ch 3 in DSP First
• Attribution • Notation: ak for Fourier Series
• The licensor permits others to copy, distribute, display, and perform
the work. In return, licensees must give the original authors credit.
• Non-Commercial
• The licensor permits others to copy, distribute, display, and perform
• Other Reading:
the work. In return, licensees may not use the work for commercial
purposes—unless they get the licensor's permission. – Next Lecture: Sampling
• Share Alike
• The licensor permits others to distribute derivative works only under
a license identical to the one that governs the licensor's work.
• Full Text of the License
• This (hidden) page should be kept with the presentation

3 4

LECTURE OBJECTIVES
SPECTRUM DIAGRAM

• ANALYSIS via Fourier Series • Recall Complex Amplitude vs. Freq

– For PERIODIC signals: x(t+T0) = x(t)
ak∗
10 ak = 12 Ak e jϕ k
7 e jπ / 3 7e − jπ / 3
4e − jπ / 2 4e jπ / 2
T0

ak = 1
T0 ∫ x (t )e − j ( 2π k / T0 ) t dt a0
0
–250 –100 0 100 250
f (in Hz)
• SPECTRUM from Fourier Series
{ }
N
– ak is Complex Amplitude for k-th Harmonic x (t ) = a0 + ∑ ak e j 2π f k t + ak∗ e − j 2π fk t
k =1
5 6

1
 Harmonic Signal Example
x(t ) = sin 3 (3π t )
∞
x (t ) = ∑ ak e j 2π k f t 0

k = −∞
PERIOD/FREQUENCY of COMPLEX EXPONENTIAL:

2π
2π ( f 0 ) = ω0 =
1
or T0 =
T0 f0  j  − 3 j  j 3π t  3 j  − j 3π t  − j  − j 9π t
x (t ) =  e j 9π t +  e +  e + e
8  8   8   8 
7 8

Example STRATEGY: x(t)
ak

x(t ) = sin (3π t ) 3

 j  − 3 j  j 3π t  3 j  − j 3π t  − j  − j 9π t • ANALYSIS
x ( t ) =   e j 9π t +  e +  e + e
8  8   8   8  – Get representation from the signal
In this case, analysis – Works for PERIODIC Signals
just requires picking
off the coefficients.
ak • Fourier Series
– Answer is: an INTEGRAL over one period
T0

k = −3
k = −1 k =1
k =3
ak = 1
T0 ∫ x (t )e − jω0k t dt
0

9 10

FS: Rectified Sine Wave {ak} FS: Rectified Sine Wave {ak}
T
1 0
x (t )e − j ( 2π / T0 ) kt dt
T0 ∫0
ak = ( k ≠ ±1)
T0 / 2 T0 / 2
Half-Wave Rectified Sine e − j ( 2π / T0 )( k −1)t e − j ( 2π / T0 )( k +1)t
T0 / 2 ak = j 2T0 ( − j ( 2π / T0 )( k −1))
− j 2T0 ( − j ( 2π / T0 )( k +1))
− j ( 2π / T0 ) kt
ak = T1 ∫ sin( 2Tπ t ) e dt 0 0

(e ) ( )
0 0
− j ( 2π / T0 )( k −1)T0 / 2
= − 1 − 4π (1k +1) e − j ( 2π / T0 )( k +1)T0 / 2 − 1
0 1
4π ( k −1)
e j ( 2π / T0 ) t − e− j ( 2π / T0 )t − j ( 2π / T0 ) kt
T0 / 2
= T1
0 ∫
0
2j
e dt = 1
4π ( k −1) (e − jπ ( k −1)
−1 − ) 1
4π ( k +1) (e − jπ ( k +1)
−1 )
T0 / 2
− j ( 2π / T0 )( k −1) t
T0 / 2
− j ( 2π / T0 )( k +1) t
 0 k odd
= ∫e dt − ∫e ( )( 
1 1

)
dt
− (−1) − 1 =  ±?j14
j 2T0 j 2T0 k +1− ( k −1)
0 0 = 4π ( k −1)
2
k
k = ±1
e− j ( 2π / T0 )( k −1)t
T0 / 2
e − j ( 2π / T0 )( k +1)t
T0 / 2  −1 k even
= −  π ( k 2 −1)
j 2T0 ( − j ( 2π / T0 )( k −1)) j 2T0 ( − j ( 2π / T0 )( k +1))
0 0 11 12

2
 SQUARE WAVE EXAMPLE
FS for a SQUARE WAVE {ak}
T0
1 0 ≤ t < 12 T0 1 − j ( 2π / T0 ) kt
x (t ) =  ak =
T0 ∫ x ( t )e dt ( k ≠ 0)
0 12 T0 ≤ t < T0 0

for T0 = 0.04 sec. 1

.02
1 e − j ( 2π / .04) kt dt = e − j ( 2π / .04 ) kt
.02

.04 ∫0
ak = 1
.04 ( − j 2π k / .04 )
x(t) 0
1
1 1 − ( −1) k
= ( e − j (π ) k − 1) =
–.02 0 .01 .02 0.04 t ( − j 2π k ) j 2π k
13 14

DC Coefficient: a0 Fourier Coefficients ak

T
1 0 • ak is a function of k
ak = ∫ x( t )e − j ( 2π / T0 ) kt dt ( k = 0) – Complex Amplitude for k-th Harmonic
T0 0
– This one doesn’t depend on the period, T0
T0
1 1  1
a0 =
T0 ∫ x(t )dt = T0 (Area )  jπ k k = ±1,±3,K
0 1 − ( −1) k 
ak = = 0 k = ±2,±4,K
1
.02
1 j 2π k 
a0 = ∫
.04 0
1 dt =
.04
(.02 − 0) = 1
2  12 k =0

15 16

Fourier Series Synthesis

Spectrum from Fourier Series
− j
 πk k = ±1,±3,K • HOW do you APPROXIMATE x(t) ?
ω0 = 2π /(0.04) = 2π (25)
 T0
ak =  0 k = ±2,±4,K
− j ( 2π / T0 ) k t

 12 k =0
ak = 1
T0 ∫ x (t )e dt
 0
• Use FINITE number of coefficients
N
x (t ) = ∑ ak e j 2π k f t 0
a−k = ak* when x ( t ) is real
k =− N
17 18

3
 Fourier Series Synthesis Synthesis: 1st & 3rd Harmonics
1 2 2
y (t ) = + cos(2π ( 25)t − π2 ) + cos(2π (75)t − π2 )
2 π 3π

19 20

Synthesis: up to 7th Harmonic Fourier Synthesis

1 2 2 2 2
y (t ) = + cos(50π t − π2 ) + sin(150π t ) + sin( 250π t ) + sin( 350π t ) 1 2 2
2 π 3π 5π 7π
x N (t ) = + sin(ω0t ) + sin( 3ω 0t ) + K
2 π 3π

21 22

Gibbs’ Phenomenon Fourier Series Demos

• Convergence at DISCONTINUITY of x(t)

• Fourier Series Java Applet
– There is always an overshoot
– Greg Slabaugh
– 9% for the Square Wave case • Interactive

– [Link]

• MATLAB GUI: fseriesdemo

– [Link]

23 24

4
 fseriesdemo GUI Fourier Series Java Applet

25 26

Harmonic Signal (3 Freqs)

a1
a3 a5

T = 0.1

27