This design is taken from the electric druid multi mode filter discussion pages. Here Tom Wiltshire outlines various filter types used in several polysynths and monos. Amongst them are a pole mixing filter similar to the Matrix 12, but rather than use the CEM3372 or the modern AS3372E, Tom as put two designs together, one that uses the AS2164 and the other that uses the CEM3320 or AS3320. I decided to go with the slightly more complicated AS2164 design as I had a bucket load spare, so I have built a dual voice card with the same pin out as my other AS3372E dual cards. This should allow me to test the card in an existing synth.
My high accuracy 330pF caps arrived today so I added a lot more components and realized my mistakes and added 2 more TL074 to cope with various functions.
Left to right
ADSR chips, ADSR inversion, buffering.
Input VCAs
Control switching, EG invert, velocity, key track etc
Filter cores
Resonance and final VCAs
Pole mixing resistors and pole selection and output buffers.
Here’s the finished version of the AS2164/V2164 filter card. Ready to be tested in my existing PolyKit-16 chassis. Hopefully it will work ok without too much fuss. I might build an extension card so I can debug it out of the synth. I’ve got two spare VCAs and two spare opamps.
Well I’m not sure what I’ve done wrong but both filters sound absolutely awful, like distortion and full resonance.
I calibrated the frequency ranges, -200 mV to 2V swing high to low. No problems there.
I found issues in my VCA design where the voltage was not swinging 2-0V as desired, looking around at various VCA designs using the AS2164/V2164 I corrected that and now my VCA’s open when they should but don’t fully close. I expected a swing of 2 to 0V but they are sitting around 0.6v, so thats something I need to look into.
But I don’t get any waveforms into the filters at all from the VCA’s that control SAW, PULSE levels, I can see the voltage swing from 0-2V low to high and I know it needs to be 2-0V from low to high so I can just turn the pot anticlockwise to get an input, but I cannot hear a corresponding wave in the filter output. Filter sweeps seem to work though as I have a noise source on the same inputs, it sort of works with lots of cracks and pops and certainly stepping of the CV voltage even with 1023 steps. The resonance just goes crazy though if you touch it so I get a lot of whistles out of this thing. I might disconnect the resonance output and just listen to the filter with any feedback to see if it’s behaving itself.
I have built pole mixing filters before based on AS3320 as per electric druid and AS3372E as per polykit with the addition of the pole switching on the first pole. These have worked perfectly.
In fact this filter is basically a copy of the Shruthi pole mixing filter so I can reference that too. I also have a PCB for the Shruthi that I might construct just to test the functionality and my components on that board and see if its sounds better than my version although there is very little difference.
So I think I’ve found a fault in the VCA control circuit, its the type of offset non inverting opamp config I’ve used many times, so I was confused, probably by my confidence in the design, as to why it wouldn’t offset correctly. Seems I used 1K resistors for the feedback and 1 leg of the input and a 10K resistor for the other leg to the offset trim pot, no wonder it struggled to adjust. So R1 & R3 were 1K, not a good combination.
I also could probably lose the voltage follower altogether, but it’s there now.
So I breadboarded the pinouts of the filters and connected the EG of the amps to 0v for A,D & R and +5v for Sustain. Set it to non velocity mode so the EG level is full +5v and triggered the gates and sure enough I get a perfect 2v to 0v swing for the VCA, so that fixes that issue. Now I just have to figure out the rest including this terrible resonance issue.
