Pitch -> CV conversion

[Don Tillman]

   Date: Wed, 2 Apr 1997 13:34:42 -0500 (EST)
   From: Tony Clark <clark at andrews.edu>

   Would a better way be to have a fairly fast counter running that keeps 
   track of the zero crossing times and then have some processor generating 
   CV voltages in relation to the final value before the counter resets?  In 
   that case it'd be generating the correct CV voltage but lagged by only 
   half the input waveform's cycle.  Right?  I'd imagine that'd work fine 
   for higher frequencies but low frequencies might have a noticeable lag?

What's the counter for?  (You digital folks, I dunno...)  

An analog approach that I've tried is to have three identical timing
capacitors switched with CMOS switches controlled by a counter.  The
capacitors are alternately connected to a voltage reference, a current
sink, and the output (buffered).  At any given time one cap is being
charged to the reference, one is being discharged at an appropriate
rate, and the third is offering that voltage as an output.  And at
each clock cycle they switch responsibilities.

But however you choose to measure the cycle time is probably fine.
Then you have to perform a 1/x on it.  

A 1/2 cycle lag for low frequencies is pretty nice compared to what it
would take for a PLL to settle in on the pitch.  For an 82Hz low E
string that's 6 mSec or the equivalent of placing your amp 6 feet
behind you.  

Assuming of course the first 1/2 cycle of the transient waveform
allows such a thing.

  -- Don