Hi,
On the weekend I have finally tested the wasp filter clone, and
made some modifications. Results so far:
It sounds considerably different than the ordinary SEM-type state variable filter. The maximum Q is lower on the Wasp version. And there is an additional distortion coming from the CMOS inverter nonlinearities. This distortion is gradually increasing with input level, and you can slightly hear it way before the circuit actually clips. The CMOS inverters seem to be the dominant source of distortion; the CA3080 input dividers are rather on the save side. (100k / 1k ; but remember the absolute maximum voltage swing is limited to 5V anyway.) I have replaced the 100k resistors with 51k without noticeable increase of distortion.
Ok, the Wasp Filter sounds different than other state variables. But what is it good? IMO, every filter's overdrive characteristic has its own special applications. The SSM2040 in LP configuration, for example, is unbeatable when you want to process a full chord of buzzy "Jump"-type saw voices. The Wasp Filter works best on Farfisa-type *organ* sounds. I tried various sounds from my OB-8, and really, the best results came with bright organ sounds. My favorite patch is the filter in LP mode with medium cutoff, and then an envelope with slow attack opening the filter (resonance quite low). This certain "edge" that is added by the filter's distortion is hard to describe, but very pleasant.
I had this circuit at the breadboard first, and I was so pleased with its sound that I build it again on a tiny veroboard. The whole filter consists of 3/6 CD4069 and two 3080's, counting the active components. I built a notch filter from another 1/6 CD4069 (add HP and LP), a heavy distortion section (1/6 4069 with 1meg feedback resistor, and 100nF input capacitor without series resistor), and a fixed 2pole, 5kHz LPF from the remaining 1/6 4069 as a speaker simulator for the distortion section. Now use the SVF to preshape the frequency response of some input signal, and then go into the overdrive / speaker simulator. LPF, BPF and even HPF settings sound pleasant thru the overdrive, with or without resonance. For some reason the notch filter doesn't work well in this configuration. (But it makes a great "one notch phaser" without the overdrive, and controlled by an LFO.)
I was really astonished what you can get out of one single CMOS chip (the OTAs just act as variable resistors, so they don't get much credit here.)
I've also made some other experiments with the 4069. You can build a little saw VCO just with one 4069 and a npn pair: One inverter as integrator, two more for the hysteretic switch, another one as opamp that regulates the current in the exponential converter. Use another one for scaling / level shift and drive a pnp transistor with equal collector and emitter resistors, and you have a poor man's version of the VCS-3 variable triangle/saw VCO. Or overdrive another inverter with a triangle wave, and add a DC offset, and you have a poor man's VCS-3 - "sine with waveform control". Ok, these things have limitations. You *always* get a glitch when the Integrator in the VCO changes direction. And the waveshapers work far from ideal, i.e. bend a positive slope more than a negative one, and I have not even tested the accuracy of the expo converter. But for one little CMOS chip plus a few transistors this makes a hell of a utility and modulation VCO.
Later, I decided to build a special ADSR / LFO circuit for this filter. This is surely not the best ADSR circuit around (time range is limited, offset voltages due to single supply ...), but my design goals were a cheap circuit and 5V single supply operation. I don't think it makes much sense to build it as a standalone envelope generator, but together with the Wasp Filter it makes a nice little processing box.
The whole combination of the separate parts has some unusual aspects that one might consider suboptimal design, but which I have chosen deliberately to have something different from my other modules:
