AW: Synth Programmer

[Haible_Juergen#Tel2743]

>  So, to keep the discussion going, what sort of input/output voltage range
>  should this thing have?
>  0-5v is the obvious choice - this would have a resolution of about
>  .02mV/bit for 8bit conversion.
>  Perhaps some of the values should be stored as 0-5 and some as -5 to +5?

I'd say go as high as you can, to make offset voltages in the MUX buffer 
opamps
less of a problem.
In my Polysynth project (yes, it is not dead, but I don't work much on it at 
the
moment) I chose a CV range of 0 ... 10V for the following reasons:

(1) Large CV swing to minimize offset problems
(2) A cheap CMOS mux can handle approx. 15V
(3) A cheap opamp with +/-15V supply can handle 10V without problems

For an universal programmer, I would make the voltage swing adjustable.
I'd make sure that the MUX/DEMUX can scan a 15 V range, and I'd make
the DAC scale adjustable (usually a single feedback resistor in the
opamp that converts DAC output current to a voltage.

As for bipolar voltages, I spent a long, long time on this problem.
Ok, you surely can convert an unipolar CV to a bipolar one with
another opamp. Question is, how can you do it with a minimum
part count in your whole system.
Now I found that *most* of the CVs in my Polysynth would not be used
as mere voltages, but rather control some kind of VCA. So often a
bipolar control range actually means positive/negative modulation
amount for some input. In other words, an inverting or non-inverting
VCA. Sort of 4-quadrant multiplying. But unlike a ring modulator, here
you would *want* some dead-band near zero !
Now here is the point: If you use OTAs for this, a 0 ... 10V CV range is
even better for positive / negative modulation than a bipolar CV. The
reason is that you need some current conversion anyway.
I develloped a little circuit that only needs two pnp transistors, three
resistors and a dual ota, that does exactly this:
CV = 10V means maximum positive modulation
CV = 5V means no modulation
CV = 0V means maximum negative modulation
There is a dead-band of a few hundred millivolts around 5V.
The control scale is linear between 0V and 4V,
and between 6V and 10V,
and it fades to Zero in an exponential way between 4 and 6 volts.
This is nearly ideal for filter bipolar envelope modulation and
similar things.

JH.