Model F

Here we go – my first active* Eurorack Module design.

I’ve always had a thing for the Roland SH101 filter, so it was happy times when Alfa Par announced that they were going to reissue the IR3109 chip. Of course, Robert Keeble has an excellent implementation with his AM8101 VCF module, and Ericasynths have a variant as well (with VCA and resonance CV, but without diode clipping in the feedback path). And, if you’re into DIY, you can still source the Pharmasonic SH module files, which are based on the ace 3109 Shruti board developed back then by Emilie Gillet.

Why a maffez version? I guess I wanted a straightforward implementation that also harnesses my experiences with modifying Roland filter clones. So, this means in addition to a simple build (two ICs and some passives) you get an extra bandpass mode for yummy acid vibes! Listen to some impressions:

Saw and Sub Smack
Characteristic level loss compensation and diode clipping in the filter resonance
Bandpass through pole-mixing

*First “active” simply means that this is my first design with ICs, Gerbers and the whole load, though it is not my first module design ever (look at the mysterious Demix Module here, something you will hear in the upcoming Mandalorian season, whoop!).

The Module

Here’s a look at the beta version with some kinks ( ironed out in the version linked below).

Model F and a first Model A version

In terms of controls, there are cutoff, resonance, modulation depth, audio input signal level, and filter mode. There is a second audio input with fixed level, an extra “key” modulation input that tracks your cutoff to 1V/octave, so you can play sine wave solos with the self oscillating filter. The inputs and output are buffered.

self oscillating filter – octave tracking
first proto boards – worked well but still I wanted some things changed

“Inside” that little thing you have two ICs, the AS3109 and a quad op-amp. The original circuit uses two transistors in the feedback path that do clever signal polarity tricks for level loss compensation (that’s when the passband loses level when you dial in resonance – we often also call this “bass drop”) and resonance itself. In my version I simplified this by means of an inverting op-amp buffer.

Simple sound with reso at zero, half and full.

The second that’s a bit different is an added bandpass filter response. This basically just harnesses well known principles of filter pole mixing (check electricdruid’s excellent site and Gillet’s papern on that topic), yet in the 101 realm, that’s less common as it should be 🙂

In this example you hear lowpass, bandpass and pseudo-highpass.

Theoretically, you can create a whole lot of more filter responses, and some folks are working on a 12hp through hole version of the Model F, if I’m correctly informed. That one will also sport -6db, -12db outputs and such. For 8HP, however, a rotary switch would have taken up quite some real estate, so something had to give. Well, still a better love story than the stock 24db VCF… and if you don’t like the bandpass, you can easily rewire for -6db out or -12db out.


If you build the version linked below, put wire in place of C15. With C15 in place, resonance drops too much below 100hz. Will update schem/gerbers soon.

Production Files and Schematic

So far, the module is something you have to have made for you or make yourself. All necessary files you need are linked right here:


Bill of Materials


This Gerber (and the one for the panel) is not yet tested – I did a proto that turned out well but wanted some final polishing. Will order that one very soon though!

Latest iteration 11 Oct 22 – all neat n tidy


Build Notes

As mentioned, the Gerbers upped here are an updated version and I need to order these as PCBs first. Once they’re in, I’ll post pictures of the build process. Should you want to build one in the meantime, do as always – components with lowest profile first, then the more spacious ones at the end. Only solder in pots and jack sockets with the panel fitted unless you want uneven results.

Modwiggler Thread