(New page: Category:ECE438 (BoutinFall2009) =ECE 438 Fall_2009 Prof. Boutin: graph of the magnitude of the DFT of a windowed filter= This is an approximation of the graph of the magniture of <m...)
 
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
 
[[Category:ECE438 (BoutinFall2009)]]
 
[[Category:ECE438 (BoutinFall2009)]]
  
=ECE 438 Fall_2009 Prof. Boutin: graph of the magnitude of the DFT of a windowed filter=
+
=[[ECE438|ECE 438]] Fall_2009 [[user:mboutin|Prof. Boutin]]: graph of the magnitude of the DFT of a windowed filter=
  
This is an approximation of the graph of the magniture of <math>H_{FIR}(\omega)=W(\omega)*\bar{H}(\omega)</math> for <math>M=1,000</math> and <math>\omega_c=\frac{\pi}{2}</math>.
+
Consider the ideal low-pass filter
  
[[Image:hFIR.png|500px]]
+
<math>H(\omega)=\left\{ \begin{array}{ll}1, & \text{ when } |\omega|< \omega_c\\
 +
0, & \text{ when } \omega_c< |\omega|< \pi \end{array}\right.,</math>
  
This is the graph of the magniture of <math>W(\omega)</math> for <math>N=100</math>.
+
which corresponds to the DT signal
 +
 
 +
<math>h[n]=\left\{
 +
\begin{array}{ll}
 +
\frac{\omega_c}{\pi}, & \text{ if }n=0,\\
 +
\frac{\sin (\omega_c n)}{\pi n}, & else.
 +
\end{array}
 +
\right.</math>
 +
 
 +
First we shift <math>h[n]</math> by <math>(M-1)/2</math> units.
 +
 
 +
<math>\bar{h}[n]=h[n-\frac{M-1}{2}]</math>
 +
 
 +
Then we multiply the result by the window function w[n] where
 +
 
 +
<math>w[n]=\left\{
 +
\begin{array}{ll}
 +
1, & 0 \leq n < M,\\
 +
0, & else.
 +
\end{array}
 +
\right.</math>
 +
----
 +
Below is an approximation of the graph of the magniture of  
 +
 
 +
<math>H_{FIR}(\omega)=W(\omega)*\bar{H}(\omega)</math>  
 +
 
 +
for <math>M=1,000</math> and <math>\omega_c=\frac{\pi}{2}</math>.
  
 
[[Image:DFT_hFIR.png|500px]]
 
[[Image:DFT_hFIR.png|500px]]
 +
----
 +
This is a zoom on of of the the little "spiky corners"  of the graph of the magniture of <math>W(\omega)</math>
  
This is the graph of the magniture of <math>W(\omega)</math> for <math>N=10000</math>.
+
for <math>N=1000</math> <math>\omega_c=\frac{\pi}{2}</math>..
  
 
[[Image:DFT_hFIRzoom.png|500px]]
 
[[Image:DFT_hFIRzoom.png|500px]]

Latest revision as of 11:50, 24 October 2011


ECE 438 Fall_2009 Prof. Boutin: graph of the magnitude of the DFT of a windowed filter

Consider the ideal low-pass filter

$ H(\omega)=\left\{ \begin{array}{ll}1, & \text{ when } |\omega|< \omega_c\\ 0, & \text{ when } \omega_c< |\omega|< \pi \end{array}\right., $

which corresponds to the DT signal

$ h[n]=\left\{ \begin{array}{ll} \frac{\omega_c}{\pi}, & \text{ if }n=0,\\ \frac{\sin (\omega_c n)}{\pi n}, & else. \end{array} \right. $

First we shift $ h[n] $ by $ (M-1)/2 $ units.

$ \bar{h}[n]=h[n-\frac{M-1}{2}] $

Then we multiply the result by the window function w[n] where

$ w[n]=\left\{ \begin{array}{ll} 1, & 0 \leq n < M,\\ 0, & else. \end{array} \right. $


Below is an approximation of the graph of the magniture of

$ H_{FIR}(\omega)=W(\omega)*\bar{H}(\omega) $

for $ M=1,000 $ and $ \omega_c=\frac{\pi}{2} $.

DFT hFIR.png


This is a zoom on of of the the little "spiky corners" of the graph of the magniture of $ W(\omega) $

for $ N=1000 $ $ \omega_c=\frac{\pi}{2} $..

DFT hFIRzoom.png


Back to ECE438 (BoutinFall2009)

Alumni Liaison

Correspondence Chess Grandmaster and Purdue Alumni

Prof. Dan Fleetwood