I don’t have the final front panel yet, but this module seems to be working:
It’s a multimode VCF based on a (partial) design by Electric Druid. I made a number of modifications but the basic design is not really changed.
It uses pole mixing to achieve different filter responses. The Shape rotary switch allows selection of two different low pass, two high pass, two band pass, and two notch filter shapes. The latter two can be altered by changing resistors on the PCB; enough footprints are provided to get just about any possible filter shape. In addition there’s an expansion header potentially allowing use of a daughterboard or expansion module for additional filter shapes or other features (though this has not been tested yet).
The design is based around the Alfa AS3320 filter chip.
Schematics, layout, Gerbers, documentation, and a Python script to generate Bode plots for different pole mixes may be found at:
Ohhh, interested to see the resonance control, will look at lunch time
I built the pole mixing filter from the electric druid website about 3 years ago, it wasn’t very successful and I questioned Tom regarding some of the values used. Needless to say I abandoned it and found a better version I think by Yusynth that worked perfectly well. I built 6 of them for my wavetable synth.
So I’m glad you built this and fixed it up so it’s usable. I’m going back through some of my early stuff and realizing how little I actually knew then and now.
Wow, I’ve been looking for a home for my AS3320 forever, and looks like my procrastination has paid off!
I have now redesigned my main resonance input, thanks.
I am following the ELKA design primarily, the stage 3 block in the design has a 820K where as all the other stages are 82k, and all other designs including yours has uniform values for the stages…
Thanks, proves I am not totally stupid and following the diagram blind :)…
Every time I come back to this it’s been over 6 months and have to remember what’s what…
On yours you invert with U3C, then had U3D to restore the phase, but then you have U2B at the end, does this not invert it back out of phase?
Hmm, yes, looks that way. Not sure what I was thinking. Oh well.
OMG, I must be getting somewhere in this whole electronics malarky.
That’s another one I tried and failed at. It seems filters are a bit hit and miss for me.
I put the fabbed panel and knobs on:
Extra boards and panels are for sale on Tindie:
is it performing as expected?
Have you thought about how the sensitivity to component variation might be eliminated without needing 5 trimpots turned to within .1% of perfect? I wonder if there’s some control theory tricks that could be played to get convergence, or perhaps some sort of calibration routine via a microcontroller that’s setting the mixing levels. That could also compensate for CV variations.
Even with a 20k pot out of a 100k total resistance, you’re getting (by spec) a minimum of 200R change (1% CRV) with each time you adjust the pot. That’s .2%. And I wonder if they were available in the 70s at that spec. One could use two trimpots or hand make wirewound resistors, but did they? I should look at the old circuits.
Maybe perfectly cutting off things wasn’t important musically anyway. The important thing is each setting sounds different.
Also, @craigyb I’d be interested in seeing that yusynth design. I haven’t been able to find it, could you link it?
I think so. Certainly it seemed to me that getting minimum bleed through at low frequency wasn’t all that touchy, probably in part because various imperfections and distortions mean you’re not going to get ideal cutoff anyway and partly because you’re just not going to hear the difference.
It wasn’t Yusynth, it was SynthR, sorry my mistake
Interesting design, somewhat simpler than the ED one. My first reaction was that some of the problems I ran into might be present here too but on closer inspection maybe not; simpler may not be too simple. I’d definitely breadboard it before committing to it, though, and might add trimmers to make sure all the filter stage gains are unity.
One reason it’s simpler is because the input signal is not separately sent to the mixers. Looks to me like the “pole” switch effectively switches the “LP1” through “LP4” signals to input plus LP1 through LP3. That limits the range of filter types you can get, in particular HP4 and the broad notch filter I used require input and all four LP so aren’t available.
The mixers tend to use more resistors than ED because they are based on divisors of 100k rather than 30k. But 100k has fewer even divisors available as standard values than 30k does, so more series pairs need to be used — and several 49.9k, which are not so cheap and easy to come by as E12 values. (ED used a couple of 4.99k, but I replaced them with 2k+3k.)