[sdiy] "Synth design questions" or "Learning from Dave Smith"

[Tom Wiltshire]

Hi all,

I've been looking at the very interesting guts-out photos of the new  
DSI Prophet 08:

http://prophet5.org/prophet08/

I'm always very interested in this kind of thing since it allows me  
to see a little of how other people have tried to solve design  
problems that I'm looking at myself as part of my monosynth project.

Before I mention that, I'd just like to point out the RJ11  
programming connectors for each of the processors on the voice board/ 
main board - custom firmware for your DSI Prophet, anyone? The mind  
boggles!

Ok, now my questions:

1) Should the main control processor talk to the voice board  
digitally or using control voltages?

Originally, I thought I'd use control voltages, since it's simple and  
keeps the analogue spirit of the voices. I had this in mind when I  
did my PIC-based LFO and ADSR designs.
However, it does seem a little bit daft to have one processor take  
digital information (patches from memory) and use an D/A to convert  
it to a control voltage, just so that another processor (say, a PIC- 
based envelope generator) can use a A/D to convert it back to a  
digital parameter. Consequently, I'm currently wondering about using  
a SPI connection instead.

2) Should the front panel of a programmable synth use pots or rotary  
encoders?

Again, originally I'd thought pots. This works reasonably well whilst  
you have a individual pot for each parameter, although even this is a  
bit of a pain with a programmable synth, since as soon as you change  
program, none of the knobs tell you anything. When programming my  
Korg Polysix, I always have the 'manual' button pressed, so that the  
sound I hear is the one I can see on the panel. And that's a simple  
instrument.
However, it would be nice to have multiple LFOs or envelopes that  
share controls, since this gives much more flexibility without making  
the panel enormous . The Prophet 08 is an example of what I have in  
mind - its four LFOs share the same group of controls, with simple  
buttons to select which one to edit. The trouble with this is that as  
soon as you switch to the next LFO, the knobs don't tell you anything  
again. Given that they don't, are rotary encoders easier to work with  
since you can just pick up the value from where you are without  
having to worry about the end of the track? As a technical issue, how  
does one go about monitoring 64 rotary encoders?
So far, I feel the only really convincing solution to this is  
encoders with LED rings like Clavia use, but resolution is a problem,  
so you still need a LCD to see the true value, although the lights  
might give a nice guide. Also, building a serious synth panel with as  
many knobs as the Prophet 08 has would require some serious number of  
LEDs, and similarly serious amount of current to light them all.

At one point, I'd made decisions about many of these things, but as I  
learn more, I keep finding more sophisticated ways to do things, and  
then wonder if the earlier decision was really so wise in the light  
of the new information. I guess I should just get on and build the  
simpler instrument I designed originally and save the clever stuff  
for the Mk2.

Thanks!

Tom