SW to calculate Wave Digital Filters?

[Joerg]

Ok, I had asked this question in the DSP newsgroup and in a German one
because there used to be a lot of academic activity regarding wave
digital filters over there. Seems to have fizzled a bit though. Anyway,
I did not get very far.

Is there any simple software that can calculate the architecture and
most of all the coefficients for wave digital filters (WDF)? Maybe even
some kind of Excel routine? Maybe a book with a CD?

Background: Many moons ago I bought a DOS program that can do that.
However, it is lacking in two important areas: I can't easily play with
the coefficient granularity to see what the stop band behavior would be
afterwards, and staggered-decimation filters are next to impossible.
This is important when you have to make do with a processor of limited
horse power that does not contain a multiplier. Secondly, with this SW
one has to pick from a common architecture (Bessel, Butterworth,
Chebychev, Cauer etc.), none of the more esoteric stuff.

 

[Deefoo]

Ok, I had asked this question in the DSP newsgroup and in a German one
because there used to be a lot of academic activity regarding wave
digital filters over there. Seems to have fizzled a bit though. Anyway,
I did not get very far.

Is there any simple software that can calculate the architecture and
most of all the coefficients for wave digital filters (WDF)? Maybe even
some kind of Excel routine? Maybe a book with a CD?

Background: Many moons ago I bought a DOS program that can do that.
However, it is lacking in two important areas: I can't easily play with
the coefficient granularity to see what the stop band behavior would be
afterwards, and staggered-decimation filters are next to impossible.
This is important when you have to make do with a processor of limited
horse power that does not contain a multiplier. Secondly, with this SW
one has to pick from a common architecture (Bessel, Butterworth,
Chebychev, Cauer etc.), none of the more esoteric stuff.

Matlab? Check this one:
http://ens.ewi.tudelft.nl/~huib/mtbx/index.php

--DF

 

[Tim Wescott]

Ok, I had asked this question in the DSP newsgroup and in a German one
because there used to be a lot of academic activity regarding wave
digital filters over there. Seems to have fizzled a bit though. Anyway,
I did not get very far.

Is there any simple software that can calculate the architecture and
most of all the coefficients for wave digital filters (WDF)? Maybe even
some kind of Excel routine? Maybe a book with a CD?

Background: Many moons ago I bought a DOS program that can do that.
However, it is lacking in two important areas: I can't easily play with
the coefficient granularity to see what the stop band behavior would be
afterwards, and staggered-decimation filters are next to impossible.
This is important when you have to make do with a processor of limited
horse power that does not contain a multiplier. Secondly, with this SW
one has to pick from a common architecture (Bessel, Butterworth,
Chebychev, Cauer etc.), none of the more esoteric stuff.

Y'know, when you posted that originally I looked at DWF stuff on the web
-- it seemed like a bit of a gimmick, just to get a guaranteed-stable
sampled-time filter.

Why not just fit your transfer function using first principals, without
starting in the continuous-time domain?

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

 

[Joerg]

Hello Tim,

Y'know, when you posted that originally I looked at DWF stuff on the web
-- it seemed like a bit of a gimmick, just to get a guaranteed-stable
sampled-time filter.

Why not just fit your transfer function using first principals, without
starting in the continuous-time domain?

Well, they aren't a gimmick. They are used in many products where the
ratio of clock to highest sampling frequency is uncomfortably close, or
where cost was the predominant factor.

It isn't just about guaranteed stability. There are tricks for other
filters that can achieve that as well. But when you have a need for
continuous output all the "burp mode" schemes employing DFT, Goertzels
and so on don't quite cut it. I guess our kids would now say "streaming
output". Also, most other schemes are very RAM hungry for all the
coefficients and the usual micro won't have more than 256 bytes, if that.

App note SLAAE16 from TI describes an implementation. TI was willing to
share the design routines with me which was very nice of them. However,
this was more of a loose collection of university research efforts and
not in any way a practical design environment, IMHO.

If there was a powerful DSP that would be very economical in power
consumption all this would be easy. You could do a few running Goertzels
or maybe even staggered FFTs. But there ain't, even a Blackfin would
gobbled up a couple of AA cells in a jiffy. Plus you can't really get a
decent DSP for under $3.

 

[Apostrophe Police]

Why not just fit your transfer function using first principals,

I'm sure they're all pushing up daisies by now.

Did you mean "principles?" ;-)

Cheers!
Rich Grise, self-appointed Chief,
Apostrophe Police, and apparently, spelling nitpicker. ;-)