This has been on the back burner for quite some time.
Doing a MS-20 like LPF is easy. Doing a PT2399 delay is easy. But making the delay controllable from a Shruthi-1 filter board and/or getting the whole thing to fit in the Shruthi-1 filter board format was much harder than I thought.
The Shruthi-1 emits 2 CVs. This is not enough for controlling a delay for which we usually want at least 3 parameters (time, feedback, level). My first try with just a time + feedback control (output signal taken out of the feedback loop) kept the length of the delay “tail” coupled with its level…
So I had to find a way of cramming more controllable parameters out of the Shruthi-1 system. My very first intuition was to do shitty PWM (a few bits of resolution, and a carrier frequency of 200 Hz) on the digital expansion port pins. This required super big filter caps to get reasonably smooth CVs, or an even bigger 2 or 4-pole active filter. The efforts that followed - and I was stuck there with no improvement over 2 or 3 months - focused on using the same approach as on the other filter boards (74hc595 receiving data through the expansion port), and toggling resistor networks through 4051s or 4053 in the feedback path - with CV1 controlling delay time and CV2 controlling a VCA on the delay output. The original design gave 8 unique feedback values, and another 4053 provided 4 different feedback routes (with fixed LP / HP filters)… not too bad but still it did not fit on the board
At some point I considered doing the whole filter board in SMT so that everything would fit there. Another option I considered was to use good old pots, and make a dedicated bigger case for this version. with an extra column of controls on the side for the delay.
Finally I had two epiphanies. The first one was that at the exception of the “main” VCA which has to be dead silent, a tiny bit of bleeding was allowed for whatever other VCA was used in the circuit (resonance control, delay mixing, delay feedback) ; and that some amount of compensation could be done software-side. So I could use very very simple, non-ideal V to I converters for those VCA (a single transistor), trading an op-amp for a bit of software-side calibration. The second epiphany was that I could use the TX> pin, normally used for serial communication with the digital filter board, as an extra GPIO to do proper SPI communication to a SPI device with the expansion port. This is a bit strange to do because the expansion port is really designed for chaining 595s, so everything we want to do with it has to go through the LEDs 595 and has to be done in sync with it, but I successfully managed to talk to a SPI slave through the LEDs 595 and the spare TX> pin. Yeah! At the same time I found about Microchip’s dual digital pots chips, and this was exactly what I was looking for!
The 4251s were delivered on friday morning, I got a proof of concept in the afternoon, and finished the whole thing over the week-end.
The last hurdle was that even with a minimal circuit for the delay, the board was too small. I managed to make it fit by putting to work the LM13700 Darlingtons as buffers / mixers and getting rid of 2 TL072.
The final feature list is:
- OTA MS20 LPF (not the original Korg35 design, the later version)
- PT2399 delay with adjustable delay time (from a few ms to 550ms), delay level, feedback level, and feedback tone control (the echoes can get darker and darker à la cave dweller or brighter and brighter as often heard in dub…)
There’s a lot of noise but there’s a very good chance it’s just due to the messiness of the proto and poor grounding of breadboards…