Digital filter simulators?

Hello Folks,

Is there a low/no cost simulator tool for non-ideal digital filters, 
maybe from a university?

What I want to do is briefly simulate the effects of reduced coefficient 
granularity in filters such as WDF. IOW, play around a bit. I know I 
could do this with Matlab but that is an expensive piece of software 
that I wouldn't really need for anything else. Writing code for the PC 
or writing a huge spreadsheet routine is another option but takes too 
long. I am looking for something equivalent to what SPICE is for the 
analog world. Ideally where you could use test tapes piped in through 
the sound card.

I have some filter design software that came with DSP books but they 
don't go that far. They assume 'ideal' coefficient granularity base on 
the available word width. Didn't find anything with Google but maybe I 
hadn't used optimal search phrases.

Regards, Joerg

http://www.analogconsultants.com

Joerg wrote:

> Hello Folks,
> 
> Is there a low/no cost simulator tool for non-ideal digital filters, 
> maybe from a university?
> 
> What I want to do is briefly simulate the effects of reduced coefficient 
> granularity in filters such as WDF. IOW, play around a bit. I know I 
> could do this with Matlab but that is an expensive piece of software 
> that I wouldn't really need for anything else. Writing code for the PC 
> or writing a huge spreadsheet routine is another option but takes too 
> long. I am looking for something equivalent to what SPICE is for the 
> analog world. Ideally where you could use test tapes piped in through 
> the sound card.
> 
> I have some filter design software that came with DSP books but they 
> don't go that far. They assume 'ideal' coefficient granularity base on 
> the available word width. Didn't find anything with Google but maybe I 
> hadn't used optimal search phrases.
> 
> Regards, Joerg
> 
> http://www.analogconsultants.com

SciLab (http://www.scilab.com).  The last time I recommended it the 
fellow responded with "boy, it's about as user friendly as a chainsaw". 
  I had to disagree because chainsaws come with instruction manuals and 
finger guards.

But it's as powerful as MatLab, it's very MatLab-like in the way it 
works (although the details are quite different), it's free, it's being 
maintained, and it has good newsgroup support (on 
comp.soft-sys.math.scilab).  I find that for control systems work its 
better than Matlab because it comes with a native type for polynomial 
ratios that gives you transfer functions in s and z, and another one for 
representing state-space systems.

For your stuff you can use the z-domain transfer functions and force 
granularity on the coefficients.  If you need to investigate fixed-point 
arithmetic it has a discrete-time ODE solver that works well.

You may also want to look into Octave, which is also free but apparently 
no longer well maintained.  I keep it around because its nearly 100% 
MatLab compatible, but the version that comes with Cygwin doesn't really 
work right and the Cygwin-independent version doesn't play well with Cygwin.

-- 

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

Joerg wrote:

> Hello Folks,
> 
> Is there a low/no cost simulator tool for non-ideal digital filters, 
> maybe from a university?
> 
> What I want to do is briefly simulate the effects of reduced coefficient 
> granularity in filters such as WDF. IOW, play around a bit. I know I 
> could do this with Matlab but that is an expensive piece of software 
> that I wouldn't really need for anything else. Writing code for the PC 
> or writing a huge spreadsheet routine is another option but takes too 
> long. I am looking for something equivalent to what SPICE is for the 
> analog world. Ideally where you could use test tapes piped in through 
> the sound card.
> 
> I have some filter design software that came with DSP books but they 
> don't go that far. They assume 'ideal' coefficient granularity base on 
> the available word width. Didn't find anything with Google but maybe I 
> hadn't used optimal search phrases.
> 

Joerg,

The good book where you can find the treatment of the digital filter 
limited precision artifacts is:

Dietrich Schlichtharle. Digital Filters: Basics and Design.
Springer ISBN 3-540-66841-1

In most cases the analysis is not very complicated; there is really no 
point to do the simulation.


Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

There is also Ptolomey.  It's free but it may be more than you want.  I 
haven't used it.



In article <stydnUmu99f5dDjeRVn-gw@web-ster.com>, Tim Wescott 
<tim@seemywebsite.com> wrote:
>Joerg wrote:
>
>> Hello Folks,
>> 
>> Is there a low/no cost simulator tool for non-ideal digital filters, 
>> maybe from a university?
>> 
>> What I want to do is briefly simulate the effects of reduced coefficient 
>> granularity in filters such as WDF. IOW, play around a bit. I know I 
>> could do this with Matlab but that is an expensive piece of software 
>> that I wouldn't really need for anything else. Writing code for the PC 
>> or writing a huge spreadsheet routine is another option but takes too 
>> long. I am looking for something equivalent to what SPICE is for the 
>> analog world. Ideally where you could use test tapes piped in through 
>> the sound card.
>> 
>> I have some filter design software that came with DSP books but they 
>> don't go that far. They assume 'ideal' coefficient granularity base on 
>> the available word width. Didn't find anything with Google but maybe I 
>> hadn't used optimal search phrases.
>> 
>> Regards, Joerg
>> 
>> http://www.analogconsultants.com
>
>SciLab (http://www.scilab.com).  The last time I recommended it the 
>fellow responded with "boy, it's about as user friendly as a chainsaw". 
>  I had to disagree because chainsaws come with instruction manuals and 
>finger guards.
>
>But it's as powerful as MatLab, it's very MatLab-like in the way it 
>works (although the details are quite different), it's free, it's being 
>maintained, and it has good newsgroup support (on 
>comp.soft-sys.math.scilab).  I find that for control systems work its 
>better than Matlab because it comes with a native type for polynomial 
>ratios that gives you transfer functions in s and z, and another one for 
>representing state-space systems.
>
>For your stuff you can use the z-domain transfer functions and force 
>granularity on the coefficients.  If you need to investigate fixed-point 
>arithmetic it has a discrete-time ODE solver that works well.
>
>You may also want to look into Octave, which is also free but apparently 
>no longer well maintained.  I keep it around because its nearly 100% 
>MatLab compatible, but the version that comes with Cygwin doesn't really 
>work right and the Cygwin-independent version doesn't play well with Cygwin.
>

Hello Tim,

> SciLab (http://www.scilab.com).  The last time I recommended it the 
> fellow responded with "boy, it's about as user friendly as a chainsaw". 
>  I had to disagree because chainsaws come with instruction manuals and 
> finger guards.
> 

Thank you, I will try that out. If it's like a chainsaw I'll be comfy. 
The only somewhat tricky task with chainsaws is the correct sharpening 
of the chain when the fingers are almost frozen. Also, I don't trust 
those 'automatic' oilers anymore and always carry a pot of chain oil 
with me.

Just in case others are also interested, it's: http://www.scilab.org

Regards, Joerg

http://www.analogconsultants.com

Hello Vladimir,

> The good book where you can find the treatment of the digital filter 
> limited precision artifacts is:
> 
> Dietrich Schlichtharle. Digital Filters: Basics and Design.
> Springer ISBN 3-540-66841-1
> 

Thanks. I'll play around with Scilab and when that doesn't get me going 
I'll order the book. Luckily it is still available on Amazon, about $90. 
I was kind of surprised that IEEE doesn't carry it.


> In most cases the analysis is not very complicated; there is really no 
> point to do the simulation.
> 

I'll have to try it out somehow by piping in real test data and see what 
the noise levels will be. IOW how much of unwanted response each filter 
produces. It is hard to forecast that when things in the signal path can 
go non-linear as in my case.

Regards, Joerg

http://www.analogconsultants.com

Joerg wrote:
> Hello Folks,
> 
> Is there a low/no cost simulator tool for non-ideal digital filters, 
> maybe from a university?
> 
> What I want to do is briefly simulate the effects of reduced coefficient 
> granularity in filters such as WDF. IOW, play around a bit. I know I 
> could do this with Matlab but that is an expensive piece of software 
> that I wouldn't really need for anything else. Writing code for the PC 
> or writing a huge spreadsheet routine is another option but takes too 
> long. I am looking for something equivalent to what SPICE is for the 
> analog world. Ideally where you could use test tapes piped in through 
> the sound card.
> 
> I have some filter design software that came with DSP books but they 
> don't go that far. They assume 'ideal' coefficient granularity base on 
> the available word width. Didn't find anything with Google but maybe I 
> hadn't used optimal search phrases.

Octave. Scilab. Any program that plots filters with coefficients you 
give it -- modify ideal filters to suit your test.

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

Hello Jerry,
> 
> Octave. Scilab. Any program that plots filters with coefficients you 
> give it -- modify ideal filters to suit your test.
> 

After Tim mentioned Scilab I played with it a bit. Looks like quite a 
learning curve so maybe I just fire up the solder iron, get a 16bit ADC 
plus a good uC and try all this out in the lab.

Problem is, the lab is also where the Christmas cookie jar is and I 
can't stay away from it.

Tim: Scilab is more complicated than a chainsaw, at least for a hardware 
guy...

Regards, Joerg

http://www.analogconsultants.com

Joerg wrote:
-snip-

> 
> Just in case others are also interested, it's: http://www.scilab.org
> 
> Regards, Joerg
> 
> http://www.analogconsultants.com

Oops.

-- 

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

Joerg wrote:

> Hello Vladimir,
> 
>> The good book where you can find the treatment of the digital filter 
>> limited precision artifacts is:
>>
>> Dietrich Schlichtharle. Digital Filters: Basics and Design.
>> Springer ISBN 3-540-66841-1
>>
> 
> Thanks. I'll play around with Scilab and when that doesn't get me going 
> I'll order the book. Luckily it is still available on Amazon, about $90. 
> I was kind of surprised that IEEE doesn't carry it.
> 
> 
>> In most cases the analysis is not very complicated; there is really no 
>> point to do the simulation.
>>
> 
> I'll have to try it out somehow by piping in real test data and see what 
> the noise levels will be. IOW how much of unwanted response each filter 
> produces. It is hard to forecast that when things in the signal path can 
> go non-linear as in my case.
> 
> Regards, Joerg
> 
> http://www.analogconsultants.com

If you think that Scilab has a future with you get a copy of 
"Engineering and Scientific Computing with Scilab with CDROM"
by Claude Gomez.  Neither it nor the Scilab on-line documentation give 
you the complete picture, and even with both there are still holes, but 
it'll give you a good step up.  I found it particularly helpful with 
making decent graphs.

If someone would fund me, and if I thought the Scilab team would pay 
attention, I'd love to dig in and write _good_ documentation for that 
package.

-- 

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