0% found this document useful (0 votes)
16 views

DFT& FFT

This document discusses the discrete Fourier transform (DFT) and fast Fourier transform (FFT). It provides examples of calculating the DFT and IDFT of signals. It also covers properties of DFT/IDFT like circular convolution and overlap-add and overlap-save methods.

Uploaded by

VIJETHA J
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
16 views

DFT& FFT

This document discusses the discrete Fourier transform (DFT) and fast Fourier transform (FFT). It provides examples of calculating the DFT and IDFT of signals. It also covers properties of DFT/IDFT like circular convolution and overlap-add and overlap-save methods.

Uploaded by

VIJETHA J
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 51

DFT & FFT

Revision Class (30-03-2020 to 3-04-2020)


By
M.Uttara Kumari
Find DFT using formula ( x[n] X[K])
Find X[K] ( Frequency Domain) from x[n] (time domain) signal

Find 4 point DFT j


1. x[n]= [ 1 6 7 2] K W4
X[K]=
0 1
-1 1
2. X[n]= [7 9] 1 -j
X[K]= 2 -1
3 j -j
Find 8pt DFT

• Find 8 point DFT


1. x[n]= [ 1,5,7,9,3,7,3,-5]
X[K]=
K W8
0 1
• 2. x[n]= [ 3 4 7 9 1] 1 0.707- 0.707j
2 -j
X[K]=
3 -0.707- 0.707j
4 -1
5 -0.707 + 0.707j
Note: If x[n] is a real seq , them X[k] = X*[ N-K]
6 j
X[7]= X*[1], X[6]=X*[2], X[5]=X*[3], X[0] and X[4] are always
real 7 +0.707 + 0.707j
Magnitude and Phase plots of X[k]
k X[K] Mag X[K] Angle of X[k]
0
1
2
3
4
5
6
7
Find IDFT using formula ( X[K] x[n] )
Find x[n] (time domain) signal from X[K] ( Frequency Domain)

K W8 (-k)
K W4(-k) 0 1
Find 4 point DFT 0 1 1 0.707+ 0.707j
1. X[K]= [ 6 , 1+7j, 5, 1-7j] 2 -j
X[n]= 1 +j
3 -0.707+ 0.707j
2 -1
2 X[K]= [2 3 4 5 6 8 9], find 8pt IDFT 4 -1
x[n]= 3 -j
5 -0.707 -0.707j
6 j
7 +0.707 - 0.707j
4 Pt DFT & IDFT using Transformation Matrix

Find DFT of x[n] using transformation matrix


Find x[n] using transformation matrix
X[K]= [4, 1+2j, 4, 1-2j]

1 1 1 1 4
1 1 𝑗 −1 −𝑗 1 + 2𝑗
4 1 −1 1 −1 4
1 −𝑗 −1 𝑗 1 − 2𝑗
• Find N pt DFT of x[n]= 𝑎𝑛
3. Find DFT of x[n] = (−1)𝑛
(i) N=3, (ii) N Odd(iii) N=4, (iv)N Even
1.DFT problems(Significance of
N>>>L) L=10, N=10
10

b
2.Find DFT
• X[n]= u[n]- u[n-5] using 5 point and 10 point
Similar to the previous problem and plot Magnitude and Phase plot

Note: Similar problem Example 3.7 pg No 99 , N=5 , 3


DFT{ δ[n]}= 1
−𝑗2π𝑚
DFT{ δ[n-m]} = 𝑒 𝑁

DFT{1} = N δ[K]

𝑗2π𝑚𝑛
DFT{ 𝑒 𝑁 ) = N δ[k-m]

2π𝑚𝑛 𝑁
DFT{Cos( )} = {δ[k-m] + δ[k+ m]}
𝑁 2

2π𝑚𝑛 𝑁
DFT{Sin( )} = {δ[k-m] - δ[k+ m]}
𝑁 2𝑗
Properties of Twiddle Factor
Circular Convolution
Revision class -2
Using
1. Direct Formula
2. Matrix multiplication
3. Graphical Method
4. Properties
5. DIT FFT
6.DIF FFT
Circular Convolution Direct Formula

4 Pt Circular Convolution

Ans: [4 14, 24 18]


Matrix multiplication
Circular Convolution
using
Graphical Method
Circular Convolution

• X1[n]=[2 1 2 1]
• X2[n]=[1 2 3 4]

• X3[n]= [14,16,14,16]
Using properties (i)
1 1 1 1 4
x1[n] = [ 4 2 7 9], x2[n] =[ 2 3 1] 1 −𝑗 −1 𝑗 2
1 −1 1 −1 7
Find 4 pt circular convolution using properties 1 𝑗 −1 −𝑗 9

Solution: Steps: X1[K]=

( Ensure both x1[n] and x2[n] are 4 points)

New X2[n]= [2 3 1 0] 1 1 1 1 2
(ii) 1 −𝑗 −1 𝑗 3
(i) Find X1[K] using DFT matrix method 1 −1 1 −1 1
1 𝑗 −1 −𝑗 0
(ii) Find X2[K] using DFT matrix method
X2[K]=
(iii) Find X3[K] = X1[K] * X2[K]
(iv) Find x3[n] using IDFT matrix method. To find x3[n]
1 1 1 1
( Refer slide 6 of DFT Revision 1 PPT)
1 1 𝑗 −1 −𝑗
4 1 −1 1 −1
1 −𝑗 −1 𝑗

X3[n]= [
Using DIT FFT Use this Diagram to find X1[K] and X2[K]
x1[n] = [ 4 2 7 9], x2[n] =[ 2 3 1]
Find 4 pt circular convolution using DIT FFT
Solution: Steps:
( Ensure both x1[n] and x2[n] are 4 points)

X2[n]= [2 3 1 0]
(i) Find X1[K] using DIT FFT matrix method
(ii) Find X2[K] using DIT FFT matrix method
(iii) Find X3[K] = X1[K] * X2[K]
(iv) Find x3[n] using DIT FFT matrix method.

Use this Diagram to find x3[n]


Using DIF FFT Use this Diagram to find X1[K] and X2[K]

x1[n] = [ 4 2 7 9], x2[n] =[ 2 3 1]


Find 4 pt circular convolution using properties
Solution: Steps:
( Ensure both x1[n] and x2[n] are 4 points)
X2[n]= [2 3 1 0]
(i) Find X1[K] using DIF FFT matrix method
(ii) Find X2[K] using DIF FFT matrix method
(iii) Find X3[K] = X1[K] * X2[K]
(iv) Find x3[n] using DIF FFT matrix method.

Use this Diagram to find x3[n]


𝑁 𝑁
• X1[k] = {δ[k-1] + δ[k+ 1]} , X2[K]= {δ[k-1] - δ[k+ 1]}
2 2𝑗
• a) Circular Convolution x3[n] = IDFT{X3[K}}= IDFT{ X1[K] × X2[K]}
𝑁 𝑁
• = IDFT( × {δ[k-1] - δ[k+ 1]} )
2 2𝑗
𝑁
=
2
b) Circular correlation x3[n] = IDFT{X3[K}}= IDFT{ X1[K] * X2*[K]}
𝑁 𝑁
= IDFT( × {δ[k-1] - δ[k+ 1]} )
2 −2𝑗
𝑁
=-
2
c)Let x3[n]= Auto corr of x1[n] = IDFT{X3[K}}= IDFT{ X1[K] × X1*[K]}
𝑁 𝑁
= IDFT( × {δ[k-1] + δ[k+ 1]} )
2 2
𝑁
=
2
d) Let x3[n] = Auto corr of x2[n] = IDFT{ X2[K] × X2*[K]}
𝑁 𝑁
= IDFT( × {δ[k-1] - δ[k+ 1]} )
−2𝑗 2𝑗
𝑁
=
2
Overlap add and save
methods
Revision class-3
Overlap Add Method 1: When N ( number of
Given x[n]= [ 1,6,8,2,5,3,9,2,7,0,1,3,6]
circular points is not given)
h[n]= [ 1 4 3]
Procedure:
1. Let ‘L’ is the length of h[n]( L=3]
2. Divide x[n] into various blocks.
Let ‘M’ is the length of each block x1[n], x2[n]…..
In this method M is taken as L ( M=L)
Then x1[n]= [1,6,8], x2[n]=[2 5 3]………………..
3.The no of circular point convolution required is
N= L+M-1= 3+3-1= 5
4. As N=5, Make h[n] and x1[n],x2[n]……… to length ‘N’
(here N=5) by appending 2 (N-M)zeros
5. So, the new sequences are
h[n]= [ 1 4 3 0 0]
X1[n]=[1 6 8 0 0 ], X2[n]=[2 5 3 0 0 ] ,X3[n]=[9 2 7 0 0 ],X4[n]=0 1 3 0 0 ],X5[n]=[ 6 0 0 0 0]
6. Find Circular convolution of h[n] with x1[n] , x2[n] etc…, (5 point circular convolution)
7. Find y1[n], y2[n]………
8. Add last 2 (N-M) samples of y1[n] to first 2 (N-M) samples of y2[n] and continue in the same for other blocks also as
shown in fig
9. The final output Y[n]= [1 10 35 52 37 29 36 47 42 34 22 7 21 33 18]
Overlap Add Method 2: When N ( number of circular points is given)
Given x[n]= [ 1,6,8,2,5,3,9,2,7,0,1,3,6]
h[n]= [ 1 4 3] Use 6 point circular convolution (N=6)
Procedure:
1. Let ‘L’ is the length of h[n]( L=3]
2. Divide x[n] into various blocks.
Let ‘M’ is the length of each block x1[n], x2[n]…..
3.The no of circular point convolution required is
N= L+M-1= 6 Then, M= 6-3+1=4
4. x1[n]=[1 6 8 2], x2[n]=[5 3 9 2], x3[n]= [7 0 1 3], x4[n]= [6 0 0 0]
5. Make h[n] to 6 length (N) by adding zeros New h[n]=[ 1 4 3 0 0 0]
6. Make x1[n],x2[n]……… to length ‘N’ (here N=6) by appending 2 (N-M) zeros.
7. So, the new sequences are
x1[n]=[1 6 8 2 0 0 ], x2[n]=[5 3 9 2 0 0 ], x3[n]= [7 0 1 3 0 0], x4[n]= [6 0 0 0 0 0]
8. Find y1[n], y2[n]……… by Circular convoluting h[n] with x1[n] , x2[n] etc…, (6 point circular
convolution)
9 Add last 2 (N-M) samples of y1[n] to first 2 (N-M) samples of y2[n] and continue in the same for other
blocks also as shown in fig
10. The final output Y[n]= [1 10 35 52 37 29 36 47 42 34 22 7 21 33 18]
Overlap Save Method 1: When N ( number of circular points is not given)
Given x[n]= [ 1,6,8,2,5,3,9,2,7,0,1,3,6]
h[n]= [ 1 4 3] Use 6 point circular convolution (N=6)
Procedure:
1. . Let ‘L’ is the length of h[n]( L=3]
2. Divide x[n] into various blocks.
Let ‘M’ is the length of each block x1[n], x2[n]…..
In this method M is taken as L ( M=L)
Then x1[n]= [1,6,8], x2[n]=[2 5 3]………………..
3.The no of circular point convolution required is
N= L+M-1= 3+3-1= 5
4. Make h[n] to 5 length (N) by adding zeros New h[n]=[ 1 4 3 0 0 ]
6. Make x1[n],x2[n]……… to length ‘N’ (here N=5) by appending (N-M) samples of previous block.
5. So, the new sequences are
x1[n]=[0 0 1 6 8 ], x2[n]=[6 8 2 5 3], x3[n]= [5 3 7 0 1 ], x4[n]= [0 1 3 6 0 ]
6. Find y1[n], y2[n]…… by Circular convoluting h[n] with x1[n] , x2[n] etc…, 5 point circular convolution)
7. Discard first 2 (N-M) samples of y1[n] ,y2[n] etc….
9. The final output Y[n]= [1 10 35 52 37 29 36 47 42 34 22 7 21 33 18]
Overlap Save Method 2: When N ( number of circular points is given)
Given x[n]= [ 1,6,8,2,5,3,9,2,7,0,1,3,6]
h[n]= [ 1 4 3] Use 6 point circular convolution (N=6)
Procedure:
1. Let ‘L’ is the length of h[n]( L=3]
2. Divide x[n] into various blocks.
Let ‘M’ is the length of each block x1[n], x2[n]…..
3.The no of circular point convolution required is
N= L+M-1= 6 Then, M= 6-3+1=4
4. x1[n]=[1 6 8 2], x2[n]=[5 3 9 2], x3[n]= [7 0 1 3], x4[n]= [6 0 0 0]
5. Make h[n] to 6 length (N) by adding zeros New h[n]=[ 1 4 3 0 0 0]
6. Make x1[n],x2[n]……… to length ‘N’ (here N=6) by appending (N-M) samples of previous block.
5. So, the new sequences are
x1[n]=[0 0 1 6 8 2], x2[n]=[8 2 5 3 9 2], x3[n]= [9 2 7 0 1 3], x4[n]= [1 3 6 0 0 0]
6. Find y1[n], y2[n]…… by Circular convoluting h[n] with x1[n] , x2[n] etc…,
7. Discard first 2 (N-M) samples of y1[n] ,y2[n] etc….
9. The final output Y[n]= [1 10 35 52 37 29 36 47 42 34 22 7 21 33 18]
DIT & DIF FFT
Revision class-4
• Find X[K] using 8 point DIT FFT ( Input Bit reversed and out put Normal Order)
X[n]= [1 2 7 4 0 1 8 2]
A B C D
• If X[K] = [1 4 2 6 2 8 1 9] Find IDFT ( X[k]) or x[n] using 8 point DIT FFT
Using 4 point DIT FFT
x[n] = [ 4 2 7 9],.

Use this Diagram to find X[K]


Using 4 point DIT FFT
X[K] = [ 2, 2+3j,1,2-3j]. Find x[n]
• x[n] = [1 3 5 7 2 8 3 9 ] Find X[K] using DIF FFT
• Find IDFT of X[K] using 8 point DIF FFT X[k]= [ 1 2 3 4 5]
New X[K]= [1 2 3 4 5 0 0 0]
Point A Point B Point C Point D Final Ans 8*x[n] x[n]
1 6 3 15 X[0]=15/8
2 2 2-4j 3 X[4]=3/8
3 3 -3 3-2j X[2]
4 4 2+4j 3+2j X[6]
5 -4 -1 -5.414+7.242j X[1]
0 2 2-4j - 2.586-1.242j, X[5]
0 3 -7 -2.586+1.242j, X[3]

0 4 2+4j - 5.414-7.242 X[7]

Final ans=15,-5.414+7.242j, 3-2j, -2.586+1.242j, 3, - 2.586-


1.242j, 3+2j, - 5.414-7.242j Divide by 8
Using DIF FFT
x[n] = [ 4 2 7 9] find X[K]

Use this Diagram to find X[K]


Using DIF FFT
• X[K]= [2 , 3-5j, 1, 3+5j]
• Find x[n]
Problems on FFT applications
Revision class-5
Find DFT of a real sequences using 4 point DIT FFT
For the given X[n] =[ 2, 6, 8, 1] Find X[K] 4 point DIT FFT

Solution procedure:
A B
Input Point ‘A’ Point ‘B’ Out put X[K]
X[n]

x[0] 2 2+8=10 10 X[0]= 10+7=17

x[2] 8 2-8=-6 -6 X[1]=-6-5j

x[1] 6 6+1=7 7 X[2]=10-7=3

x[3] 1 6-1=5 -5j X[3]=-6+5j

X[K]=[ 17, -6-5j, 3, -6+5j]


Find DFT of two 4 point real sequences using single 4 point DIT FFT
For the given X1[n] =[ 2, 6, 8, 1] and X2[n]= [3 1 9 5]
Find X1[K] and X2[K] using single 4 point DIT FFT
Solution procedure:

1. x[n]= x1[n] +j x2[n]


2. Then x[n]= [2+3j, 6+j, 8+9j, 1+5j]
3. Find X[K] using DIT FFT
4. X[K]= [17+18j, -10-11j,3+6j,-2-j]

5. X1[K]= {X[K] + X*[N-K]}/2


X1[0]= {X[0] + X*[0]}/2 = [17+18j +17-18j]/2 = 17
X1[1]= {X[1] + X*[3]}/2 ={ -10-11j -2+j] /2 = -6 – 5j
X1[2]=[3+6j +3-6j]/2 =3
X1[3] = -6+5j
X1[K]= [[17, -6-5j, 3 , -6+5j ]

1. X2[K] = {X[K] - X*[N-K]} /2j

2. X2[ K] = [ 18, -6 +4j, 6, -6-4j]


• where N=4
Find the DFT of the following 8point real sequence using single 4 pt DIT FFT
Given sequence is X[ n] =[ 2, 5,6,1,0,3,4,2] Find X[K] using single 4 pt FFT use DIT FFT

Solution Procedure:

1. The new seq is x3[n]= x1[n] +j x2[n] ( x1[n] even seq, x2[n] is the odd seq )
2. Where x1[n]=[ 2,6,0,4] and x2[n]=[ 5,1,3,2]
3. Then x3[n]= [2+5j, 6+j, 3j, 4+2j]

4. Find X3[K] using DIT FFT X3[K]= [ 12+11j, 1, -8+5j, 3+4j]


5. X1[K]= {X3[K] + X3*[N-K]}/2 X1[K]= [ 12, 2-2j, -8, 2+2j]
6. X2[K] = {X3[K] – X3*[N-K]}/2j X2[K] =[ 11, 2+j, 5, 2-j]

7. X[0] = X1[0] + W8 0 X2[0] , X[1] = X1[1] + W8 1 X2[1]


X[2] = X1[2] + W8 2 X2[2] , X[3] = X1[3] + W8 3 X2[3]
8. X[4] = X1[0] - W8 0 X2[0] , X[4] = X1[1] - W8 1 X2[1]
X[6] = X1[2] - W8 2 X2[2] , X[7] = X1[3] - W8 3 X2[3]
9. X[K] =[ 23, 4.121-2.7j, -8-5j, -0.12+1.29j, 1, -.121-1.29j, -8+5j, 4.121+2.7j ]
Find DFT of two 4 point real sequences using single 4 point DIF FFT
For the given
X1[n] =[ 2, 6, 8, 1] and X2[n]= [3 1 9 5]
Find X1[K] and X2[K] using single 4 point DIF FFT

• Solution procedure:

1. x[n]= x1[n] +j x2[n]


2. Then x[n]= [2+3j, 6+j, 8+9j, 1+5j]
3. Find X[K] using DIF FFT

4. X1[K]= {X[K] + X*[N-K]}/2


5. X2[K] = {X[K] - X*[N-K]}/2j
• where N=4
Find the DFT of the following 8point real sequence using single 4 pt DIF FFT
Given sequence is X[ n] =[ 2, 5,6,1,0,3,4,2] Find X[K] using single 4 pt FFT use DIF FFT

Solution Procedure:

1. The new seq is x3[n]= x1[n] +j x2[n] ( x1[n] even seq, x2[n] is the odd seq )
2. Where x1[n]=[ 2,6,0,4] and x2[n]=[ 5,1,3,2]
3. Then x3[n]= [2+5j, 6+j, 3j, 4+2j]

4. Find X3[K] using DIF FFT


5. X1[K]= {X3[K] + X3*[N-K]}/2
6. X2[K] = {X3[K] – X3*[N-K]}/2

7. X[0] = X1[0] + W8 0 X2[0] , X[1] = X1[1] + W8 1 X2[1]


X[2] = X1[2] + W8 2 X2[2] , X[3] = X1[3] + W8 3 X2[3]
8. X[4] = X1[0] - W8 0 X2[0] , X[4] = X1[1] - W8 1 X2[1]
X[6] = X1[2] - W8 2 X2[2] , X[7] = X1[3] - W8 3 X2[3]
9. X[K] =[ ………………………………………]
Thank you

You might also like