DSPRelated.com
Forums
   Forums    More Forums    comp.dsp  

Complex FFT in DSP vs Matlab

Started by Unknown ●February 14, 2007
Reply by Scott Seidman ●February 15, 20072007-02-15
Tim Wescott <tim@seemywebsite.com> wrote in 
news:sJGdnYoniK7nXU7YnZ2dnUVZ_vLinZ2d@web-ster.com:


> and you can bet that you'll never get the MathWorks to 
> cough up their code.
> 


Actually, Mathworks makes no bones about using FFTW.  Very well known, and 
well tested.

-- 
Scott
Reverse name to reply
Reply by Scott Seidman ●February 15, 20072007-02-15
Scott Seidman <namdiesttocs@mindspring.com> wrote in
news:Xns98D8899C85DDEscottseidmanmindspri@130.133.1.4: 


> Tim Wescott <tim@seemywebsite.com> wrote in 
> news:sJGdnYoniK7nXU7YnZ2dnUVZ_vLinZ2d@web-ster.com:
> 
>> and you can bet that you'll never get the MathWorks to 
>> cough up their code.
>> 
> 
> Actually, Mathworks makes no bones about using FFTW.  Very well known,
> and well tested.
> 


Oops, should read a bit further down before replying.

-- 
Scott
Reverse name to reply
Reply by marc ●February 15, 20072007-02-15
Do both implementations use a rectangular window?  That would effect
your main lobe width and side lobe heights.  There is also the effect
of zero-padding. If I'm not mistaken, increasing the FFT length by
zero-padding will create interpolative effects in the frequency
resolution.  Are you using the same fft lengths?

-marc
Reply by Fred Marshall ●February 15, 20072007-02-15
"marc" <marc.robins@gmail.com> wrote in message 
news:1171577308.283470.185100@m58g2000cwm.googlegroups.com...

> Do both implementations use a rectangular window?  That would effect
> your main lobe width and side lobe heights.  There is also the effect
> of zero-padding. If I'm not mistaken, increasing the FFT length by
> zero-padding will create interpolative effects in the frequency
> resolution.  Are you using the same fft lengths?
>
> -marc


In Matlab "fft" has more than one form:

fft(x) is of the length of x

fft(x,n) is of the length n, no matter the length of x.

So, if the second form is used then it could well be different than another 
implementation on a vector of the length of x.

Also, "fft" often means or requires the length be a factor of 2 and a 
shorter length vector might be zero padded to make up the difference.  I 
don't know what the TI DSP library does...

Fred

Fred