[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Discuss-gnuradio] designing filters: remez?
From: |
Eric Blossom |
Subject: |
Re: [Discuss-gnuradio] designing filters: remez? |
Date: |
Fri, 11 Jan 2008 09:05:08 -0800 |
User-agent: |
Mutt/1.5.17 (2007-11-01) |
On Fri, Jan 11, 2008 at 01:34:18PM +0100, Martin Braun wrote:
> On Thursday, 10. January 2008 20:01, Patrick Strasser wrote:
> > method functions. Is there something bad about designing filters with
> > the Remez method? I thought it would be clever to constraint the filter
> > to a number of coefficients, as this parameter is most important for
> > processing speed.
>
> I haven't had a look at the code, but in general the Remez method is always
> superior. In fact, the last course I had on filter design taught me the
> window method is only really used in quick-and-dirty filter design and in
> teaching, because it's so simple to explain (plus, it's easier to code).
>
> > Moreover, what would bring faster results, a designer using the window
> > method or one with the remez method?
>
> I doubt the window method can be beaten in terms of speed, as it's only an
> FFT
> and a multiplication, but I don't know how much faster it is.
> However, usually the filter design is not in the time critical part of an
> application, whereas the number of taps has direct influence on performance.
> Given the choice, I'd say use remez when possible.
>
> martin
>
> --
> Martin Braun
> Institut fuer Nachrichtentechnik
> Universitaet Karlsruhe
>
> http://www.int.uni-karlsruhe.de
The existing GNU Radio wrapper around the remez primitive needs help.
IIRC only the low_pass case works. Perhaps one of you guys can take a
look at it. Creating a few test cases would be great too ;)
Eric