F-V converter and Ring Mod. stuff

[gstopp at fibermux.com]

     My failure to get Don's discharging-cap F-V converter to work properly 
     was probably just my getting lost between the scope period 
     measurements and the DVM, what with all the tweaks it takes to 
     calibrate something that way....
     
     I started to calculate what kind of circuitry it would take to make a 
     digital exponential counter with granularity comparable to a 
     continuous exponential converter, and it's a bitch. I'd need something 
     like 64 bits of exponent register just for six bits of final output - 
     I've concluded it's much easier just to use electrons as the "bits". 
     No wonder everybody does a linear period conversion and then runs it 
     through a log generator!
     
     Then I got to thinking - no matter what method is used to drive an 
     exponential VCO with a pitch tracker, it's still open loop - meaning 
     that the VCO will never track the input to within anything closer than 
     several hertz, at best.
     
     So the next logical step was to try a phase-locked loop, and use the 
     VCO loop control voltage as the recovered frequency voltage output. I 
     realized that if an exponential VCO is used in the PLL, then the 
     voltage output would be 1 octave per volt, since the loop filter will 
     put out whatever it has to so that the VCO is phase-locked to the 
     external signal input. So I built a simple exponential VCO based 
     around the guts of the EN tri-square VCO (input summer, matched PNP 
     pair, 3080 integrator, 3140 buffer, 3080 comparator). Then I built a 
     wide-band phase comparator with a 74LS00, followed by a loop filter. 
     The output of the loop filter controls the VCO frequency.
     
     The capture range of this PLL is from about 10 hertz to about 5 
     kilohertz, which should be perfect for musical use. The loop filter is 
     highly damped and extremely fast-responding, with hardly any jitter. 
     An additional op-amp is used for a little extra filtering and to 
     buffer the voltage output for driving external pitch sources.
     
     Next comes the input signal conditioning. Like I said before that's 
     the hard part of pitch recovery - wish me luck.
     
     - Gene
     gstopp at fibermux.com


______________________________ Reply Separator _________________________________
      Subject: Re: F-V converter and Ring Mod. stuff
      Author:  don at till.com at ccrelayout
      Date:    2/15/96 10:48 AM
      
      I then took Don Till's advice and tried the 
   exponential-discharge capacitor into a sample & hold idea. It looks 
   good on the scope, but I had trouble trying to get it to convert at 1 
   volt per octave using scope period measurement and a DVM. 
     
Hmmm, I wonder why.  It should work.
     
Try this instead: Clock the counter off of a VCO driven by the DAC 
from the counter.  The counter will start off clocking fast, and as it 
counts down (up, whatever) the DAC voltage comes down and slows the 
clock rate of the counter.  The result will be the digital equivalent 
of a capacitor discharging.  Total incremental cost is single linear 
VC0, so it's almost free.
     
Not very good.  It's just a one-shot, so your input signal gets 
converted to a fixed with pulse train that gets filtered.  So there's 
the obvious tradeoff between response time and ripple in the output, 
but also a single misshaped pulse will screw the output voltage signal 
badly.
     
Actually measuring the cycle time of the input signal (like with 
Gene's digtial method, my analog method, or something similiar, such 
as a charge pump), requires much less filtering, and fewer tradeoffs.