How to convert period CV to pitch CV

[Mark Smart]

>      Sounds like a fun thing to experiment with. Here's the way I'd 
>      approach it:
>      
>      Since the pitch VCO puts out frequency F for 10 volts, and frequency 
>      2F for 5 volts, this isn't enough data points to conclude that the 
>      scale is linear or exponential. 

It's linear. The way the converter works is pretty cool. First the guitar
string sound is filtered through a pretty elaborate adaptive filter whose
response shape changes depending on where you play on the neck. This
compensates for the fact that a guitar string's harmonic content is
different as you go up the neck. This signal is converted to a square wave
with a comparator. The square wave drives a monostable multivibrator to
produce a short trigger pulse. This pulse drives another MM to produce
another pulse slightly later. The second pulse is used to reset a
constant-current integrator. The constant-current integrator produces a
sawtooth whose slope is always the same, but whose height depends on the
guitar string's frequency. Lower frequency-->more time for integrator to
charge-->higher sawtooth wave.

The FIRST pulse is then used to trigger a sample and hold which samples the
sawtooth wave right before it gets reset. The output of this sample and hold is the control voltage which is proprtional to the guitar string's frequency!!
 Pretty cool! 

So then the sawtooth wave is put through a VCA which follows the guitar
string's volume, and also compensates for the fact that the sawtooth wave
is different heights for different frequencies. This is the sound you hear.

I know I've ranted about this on here before, but the GR300 has the most
reliable tracking of any guitar synth I've ever played, and it's all done
with analog! I think the secret is the pitch-tracking filter at the input.
It consists of two band-pass filters whose cutoff frequencies move apart as
you go up the neck. I don't really understand how that part works. In
addition to that, there is extra circuitry which squelches the envelope
follower's output when the second harmonic starts to become dominant in the
guitar signal. This prevents the octave-jumping problem that so many guitar
synths have. 

>However since the voltage halves for a 
>      double in frequency, there may be a chance that this curve will do 
>      without having to use a liner-to-exponential converter.
>      
>      First I'd invert the CV. This means that after inversion the range 
>      would be from -10 volts to -5 volts, for open string to 12th fret. Now 
>      things are traveling in the right direction. Next I'd level shift with 
>      a -10 bias into the inverter (same op-amp) so that the range is now 0 
>      to +5, open string to 12th fret. Plug this into a standard VCF with a 
>      5:1 attenuator on the front end (like the appropriate series resistor 
>      or an adjustable attenuator) so that the VCF will see a change of 1 
>      volt in the positive direction for the octave change up on the guitar 
>      string.
>      
>      If the curve is not linear, then the VCF will not track the VCO 
>      exactly but it will probably be pretty darn close (like the 5th fret 
>      on the VCO will track like the 6th fret on the VCF) so that the 
>      timbral quality of the filtered VCO's sound may be constant over the 
>      playing range. The only time that exact tracking of the VCF would be 
>      critical is if you were trying to run the VCF's in self-oscillation 
>      mode and the intonation would need to be precise. This is where the 
>      above may not be good enough and more experimentation would be needed.
>      
>      - Gene
>      gstopp at fibermux.com

Wow. That's a good idea! The simple solutions always elude me. But I think
I would like to be able to self-oscillate the filters and have it in tune,
if possible. I think I'll try the lazy way and use a divider chip. 

Thanks for the great ideas, guys!!!!
I still don't know when I'll have time to actually build this stuff.

************************************************
*     Mark Smart                               *
*     Network Technician                       *
*     University Communications Inc. (UCI)     *
*     smart at medusa.nn.com                      *
************************************************