[sdiy] Legal issues of cloning

[rsdio at sounds.wa.com]

Those circuits may look like simple voltage followers, but there's  
more to it than just that. They're actually sample-and-hold circuits,  
and the difference is an isolated capacitor on each input to hold the  
voltage. The critical piece is the capacitor that holds the voltage,  
and the follower is just there to provide as much current as is  
necessary to buffer that voltage into an arbitrary load without  
draining any (at least not much) current from the capacitor.

With an analog switch that goes into very high impedance when it is  
off, and an op-amp input that is also high impedance, those  
capacitors will drain very slowly. Just determine the amount of time  
that you need to hold, then adjust the capacitance to be large enough  
to not droop too much during the necessary time. You don't want the  
capacitance to be any higher than the minimum needed, though, because  
that increases the time needed to charge to each new voltage. An  
analog multiplexer is a great for this, because it's basically a set  
of analog switches where the common input can be connected to the  
DAC, and each individual output connects to a S&H channel (capacitor  
and op-amp voltage follower). The best choice is the CMOS 4067, which  
has 16 channels and costs less than two 8-channel mux chips.

The serial-to-parallel shift register is a red herring. You can  
design a multichannel CV generator without that particular part. If  
your processor has a lot of parallel I/O ports, then the SI/PO is not  
needed. The absolute essentials are the DAC, mux, caps, and op-amps.  
You want to select JFET input op-amps with 1 teraohm to 10 teraohm  
input impedance to make the hold as long as possible given the  
capacitance chosen.

The PIC is a fine choice, but so is the AVR, MSP-430, and a few  
others. I'd pick a part based more on cost and I/O capabilities than  
on which family it belongs to. I have nothing against the PIC, since  
I've designed several products with it, but I also don't think  
there's anything special about it for the task of cloning a vintage  
synth that might have a 6502, 6809, Z-80, D7811G, 8049, 8051, or  
other 8-bit micro from that era.

Brian Willoughby
Sound Consulting


On Oct 6, 2013, at 13:30, Jack Jackson wrote:
> In terms of the patch memory, I'm looking forward to get cracking  
> with that. My day job is programming so it's not too daunting. Tom  
> Wiltshire has done a lot of really awesome stuff using PICs, so  
> that is quite encouraging and inspiring. It seems the JP8 achieved  
> all the analogue voltages out of the microprocessor using one DAC a  
> SI/PO shift register and a load of opamps as voltage followers. If  
> I understand correctly the microprocessor cycles through all the  
> outputs of the shift register and feeds each opamp the relevant  
> voltage in turn. In Ray Wilsons book he mentions that voltage  
> followers can hold a state for a few seconds before the voltage  
> begins to drop off, so I'm not sure at what rate the shift register  
> clocks through all the opamps, but could be once per second to  
> several times a second.