Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sdd.hp.com!hplabs!well!dmassie
From: dmassie@well.sf.ca.us (Dana C. Massie)
Newsgroups: comp.dsp
Subject: Re: Check of an FFT routine...
Summary: I agree - but use a simpler test case...
Keywords: fft test signal
Message-ID: <24112@well.sf.ca.us>
Date: 9 Apr 91 17:45:14 GMT
References: <1991Apr4.004750.15969@en.ecn.purdue.edu> <SRF.91Apr4100830@claudius.juliet.ll.mit.edu>
Distribution: usa
Organization: E-mu Systems, c/o Whole Earth 'Lectronic Link, Sausalito, CA
Lines: 22

In article <SRF.91Apr4100830@claudius.juliet.ll.mit.edu> srf@claudius.juliet.ll.mit.edu ( Steve Feinstein) writes:
>
>Your values are wrong and there aren't enough of them, either.
>You should be generating at least N FFT points for an N-point
>array.  The example you've posted is a 16-pt FFT on 32-pt
>data.  Here is what MATLAB generated:
>
>  -0.0023  		-pi
...
>		Steve Feinstein

I got the same results with matlab that Steve got.
When i wrote an FFT i found it *much* easier to use simple test signals
like a sine wave to test the fft.  I would strongly addvise
using a simple impulse in bin 0 of the input, then switch to an impulse in
bin 1, etc. and check the output of each case.  This is a very
effective method.  Also, once this is working, put an impulse in
bin 0 and bin 1 - this will test superposition.

good luck!

dmassie@well.sf.ca.us