I’ve finally completed something I started close to a year and a half ago.
This is a Kosmo format module based on the Barton Musical Circuits 4046 Wave Shaper. That’s a module that takes an input audio signal, turns it into a square wave, and then, using the magic of a CD4046 phase locked loop and a CD4040 binary counter, creates seven square waves: One at the same frequency as the input, one each one, two, and three octaves lower, and one each one, two, and three octaves higher. These seven octave signals go into a mixer to form the module output; an attenuator pot on each mixer input lets you vary the amount of each octave in the output mix.
So you can put in even a simple sine wave, one single frequency, and what you get out is a big fat pile of harmonics and subharmonics.
“That’s cool,” said I to myself. “You know what would be even cooler? If that output mixer were voltage controlled. You could use CVs to vary the timbre of the output on the fly.”
I thought about that and I realized basically what you’d have to do to make that happen would be to build seven voltage controlled amplifiers and sum their outputs. Which would be ridiculous — more complicated than it’d be worth. Not worth contemplating.
So I contemplated it.
And it occurred to me, you don’t need a general purpose VCA, because the only signals you’d be amplifying would be the square waves produced by the binary counter. And you can build a “VCA” for square waves (or pulse waves), as I did in the Noise Bells module, using a quarter of a CD4066 analog switch and one resistor. What you’re doing is looking at the square wave as a series of logic pulses, alternating off and on, and using it to disconnect and connect a control voltage to an output. Two CD4066s and seven resistors would be all you’d need for seven “VCAs”. Well, that and seven control voltage inputs with attenuators. Anyway, I was right, the idea was ridiculous. Ridiculously easy. I thought.
Over a year later, I’ve got one working.
I initially thought about drawing up a schematic based on Barton’s but with the voltage controlled mixer and making a PCB for it. Then I decided I wasn’t really changing Barton’s design so much as just stealing seven signals out of it, putting them through “VCAs”, and sending them back. So why design a whole new board? I could just take Barton’s board and build it slightly altered to get the seven signals out and back. I had an idea for how to do that using pin headers and it turned out not to work for mechanical reasons. So I drew up another design using a ribbon cable, and then put it aside while I pursued the Kosmodrome project. Recently I decided it was time to finish it up.
Here’s the scheme: The signals coming out of the CD4040, in Barton’s design, go through capacitors for removing the DC components, then through attenuator pots, then into a summing output stage. I could steal the signals by not putting the capacitors in, instead using the capacitor footprints to connect a 14 conductor ribbon cable. On the other end an IDC connector plugs into a header on an auxiliary PCB where the seven inputs go into the “VCAs”.
That board turned out to have complications. The CV inputs need protection diodes, which need series resistors. They need normal connection to about 10 V. And it turns out, to make it work, the “VCA” outputs need to be buffered with op amps. Still fairly simple but less so than I’d initially thought.
Those “VCA” outputs go back to the Barton board on the ribbon cable’s other seven conductors. But what about the capacitors? You remember the capacitors? Where do they go? Well, the attenuator pots on the main PCB aren’t needed (because we have CV attenuators on the auxiliary board) — and their footprints are right there, not being used. So the caps get soldered to the pot footprints. It looks bizarre but it works.
The Barton board has solder pads for wires to connect panel mounted input and output jacks and a slew switch. My version board mounts these, and the wires get Molex connectors to plug into Molex headers on the auxiliary boards. There’s one more Molex header connected to another pot: The wires leading to that one connect to the footprint for the feedback resistor in the output stage, so that stage’s gain is variable instead of fixed. That enables you to easily mix seven large signals and not have to worry about clipping.
The other change to the main PCB is the power input. The 10 pin power header — an unshrouded one — goes on the back of the board, and the 10 µF bypass caps and the 10 Ω resistors (twitch, twitch) are omitted — jumpers or 0 Ω resistors taking the latter’s place. The header plugs into a socket on the auxiliary PCB, which has the actual power header, bypass caps, and power reversal diodes.
Schematics, KiCad design files, Gerbers, and documentation may be found here: https://github.com/holmesrichards/WaveShaper