About
This module is a very versatile audio processor. Essentially, it's four discrete core filters bundled in one neat package, but its applications go way beyound just being filters. In my case, 3 of them are bandpass, and the fourth one is lowpass - there are instructions for both modes on the schematic, and you can have it your way. I called them 'resonators' because my main aim was to create resonant artifacts in sound with it, and to trick my (and your) mind into percieving them not like VCFs, but like resonant black boxes - and to be fair, they don't quite behave like your usual VCF! so for this text wall only, think of 'filter' and 'resonator' as synonims.
This module is based entirely on BARD Synthesizers' "Simple Discrete VCF" design - an ingenious little circuit with very particular sound. Big up for BARD Synthesizers!
The first resonator's input is normalised to other 3 inputs, and so is CV. Outputs which have no patch cord plugged into the socket get auto-summed into the last resonator's output. This brings us to the most obvious use of it - 4 parallel VCFs. Just plug a signal into resonator A input, your CV to resonator A CV input, and take the summed output from D output, and it gets insane right off - just trust me, it turns a sawtooth into literal drone chords. It gets even better when we do all sorts of cascading, put them in series, feedback, and whatnot. The filters are also self-oscillating with some resonance setting, so you can use this as 4 very rough VCOs, too! On low resonance settings, this thing can be a nice mixdown tool: put your hi hats on high layer, leads and noises on the middle, cut the unnecessary high frequencies of the droning bass with the remaining lowpass - applications are pretty much limited only by imagination.
An observant netizen might've noticed how the factplate is terribly similar in design and placement to the Linear Oscillator Clsuter - that's cause it literally is the same exact panel repurposed along with half of the passive parts and semiconductors, and all of the mechanical parts of the LOC. Reduce, reuse, recycle!
Schematic
The filter is a basic Sallen-Key topology with resistive elements ingeniously replaced with BJTs. BARD Syntheziers themselves describe how exactly it works very well on their page for the design - so check it if you want to dig deeper: i'm gonna be lazy and won't re-explain it here. My changes include an output amplifier compensating for gain loss and a voltage control buffer/summator op amp, which means the filter is no longer discrete, but discrete core. But hey, that improves its characteristics a lot! Also, i swapped around some of the values for more harsh resonant peak, and to accomodate for 50K resonance potentiometer. If you have a bunch of 2k pots, i suggest you use the original setup instead, but this works almost the same.
Mind the signal normalization! Input A is normalized to inputs B, C and D. Likewise, CV A is normalized to CV B, C and D. This way, one input and one CV goes to all 4 resonators at once. The output jacks' of filters A, B and C are normalized signal pins go to a simple double-invert unity gain op amp summator, so, if you don't stick a patch cord into the output jack of first 3 filters, they will appear at the last one's output jack. Note how Resonator D doesn't have an output jack and always goes to the summator stage - that's because the summator's output is output D, + the un-re-routed outputs of other resonators.
Media
Single bandpass resonator sweeps with a sawtooth input.
Single bandpass resonator sweeps with an A-118 white noise (loud AF) input.
Two resonators processing an FM tone, panned to extreme left/right, sequenced by different tracks of the step phrase programmer for wicked stereo.
Sawtooth put into the first resonator input and normalled to all others, sum taken out from the last reso output. All resonators' frequencies controlled by four asynchronous external LFOs.
Same patch as above, but reso 1 goes to reso 2 input (hence taken out of the mix), reso 2 to reso 3, and reso 3 to reso 4 - all four resonators are in series now, and the sum output only has reso D output in it.
Resonators A and B are in series, resonators C and D are also in series, hence output B and D are in the final summator mix/out D. We get a "2x2" mode - two blocks in parallel, each block is two resonators in series. Here they are processing a slightly FMmy tone.
Obligatory "will it ping" test: it kinda will.