One of the most straightforward ways to make the sound of an analog oscillator thicker and more prominent is to add a suboscillator to the circuit. You take the waveform, derive a square wave an octave lower by using a frequency divider, and mix it back in. Since a basic รท2 frequency divider can be made from a single CMOS flip-flop, and since those come two to the CD4013 chip, it's nearly as easy to divide by two again, yielding another square wave two octaves down from the original. The classic SH-101 synthesizer did this, with the added filip that by using diodes to mix those together, you can get a -2 octave pulse wave at 25% duty cycle, where the harmonics emphasize the -1 octave. The circuit for doing all this, buffered and offset for a modular synth, is given in a blog post on the Electric Druid site, and I was thinking of adapting it to add the circuit as a sidecar to my Kassutronics VCO 3340.
Except. I decided to test it before trying to build it by simulating the circuit inside the Falstad circuit emulator, where the circuit I was working on just didn't produce the right waveforms when all the outputs are connected. I was able to tweak my version into operation one output at a time, but when I tried to sanity check by inputting the original, it didn't work as drawn. I think I'm doing something wrong, rather than the blog author โ or Roland! โ having messed it up, and it may just be that I'm using the circuit simulator wrong. But it's frustrating to think, "Oh, I'll save time by doing this in software first," only to be caught by implementation problems.
