=?ISO-8859-1?Q?Re: [sdiy] Analog polyphony-how does it work=3F?=again-sorry.

[jhaible at debitel.net]

> > Particularly, how a single knob can control the CV of
> > four, six, eight... voices?
> 
> The key magic is multiplexing. A single analog voltage source (a DAC
> controlled by a uC) is routed through (a bunch of) analog multiplexers
> or S&H multiplexers like 4051, SSM2300 or similar to the destination.

Not really necessary. You can route a single CV source (buffered front
panel pot) to any number of CV inputs of several voices. CS-80 does it,
PS-3100 does it.

Multiplexing comes with digital patch storage - you don't want to uses
an individual DAC and individual RAM for each parameter, so you
multiplex. 

So far I'm speaking of multiplexing *parameters* (not voices): You can still
connect all the voices' corresponding CV inputs together. (I think the
PS-3200 does this.)

You only need multiplexing between voices if you transfer a combination
of CVs to each voice. Example:

Parameter 1: VCF Cutoff, Parameter 2: VCF keyboard tracking

You can implement a VCA (or a switch, if it's just discrete steps)
for keyboard tracking in hardware (in each voice). Then you can send
Parameter 1 (Cutoff) to all modules, and Parameter 2 (Tracking)
to all modules. 2 Wires for any number of voices. Parameter 2 will
just tell each voice "use half of the VCO's CV for VCF" and the hardware
will do as told, with the same input value for each voice ("half"),
but with different results for each voice, depending on the VCO's
key CV of each voice.
No MUX between voices required.

Alternative: Send a *combined* CV to the VCFs, which inludes both,
Cutoff (Parameter 1) and Tracking (Par 2) information. Only one
wire will be needed, but this must be multiplexed for each voice,
as now each voice must get a different value. (As calculated in some
central processor). 

It's easy to see that the 2nd method makes more sense if you can
combine more parameters: LFO (software) to VCF, ENV (if they are
made in software), and so on.

On the other extreme, if you make all modulation routings of your
sound engine in hardware, and if you use analog storage (;->),
you don't need a MUX for your parameters at all. (CS-80).

JH.