[sdiy] Bandpass Filter Recommendations

[Scott Juskiw]

Thanks for the info Tom. I think I'll stick with the cascaded OTA  
stages for now because: I know how to build it, it gives me the cutoff  
slopes I need. There doesn't seem to be any easier way to build what I  
want. If there are four poles in the filter response, I'm going to  
need four OTAs (or equivalent). Plus one more for resonance, plus  
another to control overall gain. I was hoping to cut down a bit on the  
circuitry, but not at the expense of quality.

On 25-May-09, at 3:53 AM, Tom Wiltshire wrote:

> Hi Scott,
>
> I did some work towards a similar thing, but never got to a point I  
> was happy with. I decided to use an SVF based on the SSM2164. I  
> started out with Roman Sowa's design, but only used the filter part  
> of his circuit:
> http://www.sowa.synth.net/modular/vcf.html
>
> This uses two OTA stages for the filter and one for the resonance. I  
> used the spare stage for the gain control. This makes a convenient  
> total of 1 SSM2164 and one quad op-amp per filter.
>
> I had some difficulty with the resonance. You'll find the discussion  
> on this in the archives. Ian Fritz and Neil Johnson gave me a lot of  
> help, but I never did get it going how I wanted, although I think  
> this was a fault, not the design per se. It's currently a project on  
> my "things to go back to" shelf.
>
> If you want more details, ask me and I'll mail you the schematic.
>
> Hope this helps,
> Tom
>
>
> On 24 May 2009, at 19:23, Scott Juskiw wrote:
>
>> I'm working on a programmable filter bank and am asking the group  
>> for recommendations on a suitable bandpass filter. The filter  
>> requirements are:
>>
>> 1. voltage controlled frequency
>> 2. voltage controlled resonance
>> 3. voltage controlled gain
>> 4. analogue circuit
>>
>> Some solutions I've come up with:
>>
>> 1. state variable filter
>> 2. cascaded lowpass/highpass filters (using cascaded OTA stages)
>>
>> #1 is relatively easy but I'm less than thrilled with the 6db/oct  
>> slopes. #2 seems to be better for my needs since I can get 12 db/ 
>> oct slopes.
>>
>> Are there any other filter topologies that anyone can recommend?  
>> Hopefully something that won't require 400 components per filter  
>> since I'd like to have a bank of these on one PCB. But maybe that's  
>> asking for too much. I would also consider a digital filter  
>> solution, if it would be easier to build and if (big IF) I could  
>> figure out how to do it digitally.
>>
>> Thanks
>>