Alternative to MIDI-CV revisited

[Don Tillman]

   Date: Fri, 10 May 1996 16:03:39 -0700
   From: Tom May <ftom at netcom.com>

   (Wow, I got quite an oscillation going by applying a small shock to
   Don & Gene :-)

Sure!

   Don Tillman writes regarding the cap-discharge F-V converter:
   >It's not too bad though; if you model it you'll see that it can come
   >remarkably close over a range of a couple octaves.

   I checked this out.  I've got a starting point which I determined by
   matching the ideal vs. cap-discharge curves at three somewhat
   arbitrary points between 82Hz (hmm) and 1000Hz to come up with some
   circuit parameters.  The match is actually pretty terrible (except at
   the three points where it is forced to be equal), but I'll let my
   computer loose on it and see if it can't optimize it a bit.  

Hmm, my match was pretty good.  I wonder what we're doing differently. 

								Maybe
   reduce the number of octaves I'm trying to match as well.  Just
   looking at it, the -log curve looks more like the sum of *two*
   decaying exponentials, but that would be a hard circuit to calibrate
   in real life.

(That suggests an interesting approach!)

   >The curve runs higher than it ought to be for lower pitches, lower
   >than it ought to be for higher pitches, and very reasonable for a
   >couple octaves in between.  If you use a fixed-time portion of the
   >cycle to charge the cap, say 1/3 the period of the highest expected
   >pitch, the high pitch end of the curve gets corrected up a little bit
   >and you maybe get a three octave range.

   That's too cryptic for me . . . are you talking about not charging the
   cap all the way?  

Sorry, yeah, it is cryptic.

I noticed that the voltage of high pitches was lower that it ought to
be, and I also needed some time to charge up the cap, sooooo, if I
take a constant-time chunk off each cycle I address both problems.
Say your highest input frequency is 1000Hz.  If you take a constant
333uSec off each cycle to charge the cap up to a reference voltage,
you spend the rest of the cycle discharging, and that effectively
warps up the output voltage for the higher pitches, mostly
compensating and extending the range to about 3 octaves.

I'm not claiming it's competely accurately though.

  -- Don