Choosing a FIR filter windows type

Wen inplemening a FIR filter, what considerations are given when choosing 
between, rectangular, Bertlet, Hanning, Hamming, Blackman, kaiser, etc 
windows for the coefficients?

Basically, aside from the attenuation differences at the first hump in the 
stop band, and transition band width, what other characteristics (trade 
offs) are to considered?


thanks 

Mook Johnson wrote:
> Wen inplemening a FIR filter, what considerations are given when choosing 
> between, rectangular, Bertlet, Hanning, Hamming, Blackman, kaiser, etc 
> windows for the coefficients?
> 
> Basically, aside from the attenuation differences at the first hump in the 
> stop band, and transition band width, what other characteristics (trade 
> offs) are to considered?

"Ringing" in the passband, for one. Other humps in the stop band, for 
another. (Compare Hamming to von Hann.)

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Mook Johnson wrote:
> Wen inplemening a FIR filter, what considerations are given when choosing
> between, rectangular, Bertlet, Hanning, Hamming, Blackman, kaiser, etc
> windows for the coefficients?
>
> Basically, aside from the attenuation differences at the first hump in the
> stop band, and transition band width, what other characteristics (trade
> offs) are to considered?

Filter lengths.

The filters designed using high attenuation windows like the
Blackman window, tend to be longer than the low-attenuation
windows for the same transition bandwidth.

Rune

"Mook Johnson" <mook@mook.net> wrote in message 
news:cbdWe.12204$S26.519@tornado.texas.rr.com...
> Wen inplemening a FIR filter, what considerations are given when choosing 
> between, rectangular, Bertlet, Hanning, Hamming, Blackman, kaiser, etc windows 
> for the coefficients?

My consideration is to choose the Kaiser and ignore the rest!  Since it has an 
adjustable parameter, you can taylor it to get the response you want.  One 
reason that you would NOT want to do this is if you had to compute the window 
coefficients in real-time, in which case Kaiser would be the most 
computationally intensive option.  (I have heard there is a computationally 
cheaper approximation of Kaiser, but haven't tried it myself.)  But for 
pre-computed filters, this is not a problem.

> Basically, aside from the attenuation differences at the first hump in the 
> stop band, and transition band width, what other characteristics (trade offs) 
> are to considered?

The two things you mention are what I usually look at when tuning a Kaiser 
window.