[sdiy] Cheap frequency counter for oscillator calibration

[David G. Dixon]

> VCO precision is better expressed in cents (or PPM if you must)
> than in percents. It's like THD -- I wish they'd switch to
> something sensible like PPM or dB...

Yes, I see that now.

> > In any case, +/- 5 cents seems like a lot or tracking error,
> > doesn't it?  Particularly in the midrange.
> 
> There's hoping the error is concentrated below 50 Hz and above 3 kHz.
> "+/-quite-a-few-cents over many octaves" is not a very meaningful
> spec.

That's what I would have to assume.  Otherwise, the VCO would be damn near
useless.

> Have you managed to get rid of the 0.8 Hz error at 880 Hz ?

My latest version of the 2164 Expo VCO uses a hand-selected Coolaudio V2164
(about half of them in my latest batch of 25 are suitably good log
conformers), 2.2nF integrator cap, 5V reference voltage fed through a 453k
current limiting resistor (to give a reference voltage of 1003Hz, which
helps with HF tracking) and a 220R Franco resistor.  With this combination
of components, when properly calibrated, the tracking is roughly as follows
(two calibration trims set to match precisely at -3V and +2V by octaves from
1kHz at 0V):

-5V	   31.25	   31.05	-11.1 cents
-4V	   62.50	   62.39	 -3.1 cents
-3V	  125.0	  125.0	  0.0 cents
-2V	  250.0	  250.3	 +2.1 cents
-1V	  500.0	  500.3	 +1.0 cents
 0V	 1000		 1000		  0.0 cents
 1V	 2000		 2000		  0.0 cents
 2V	 4000		 4000		  0.0 cents
 3V	 8000		 8010		  2.2 cents
 4V	16000		15670		-36.1 cents

If we ignore the first and last entries, then that's tracking over 7 octaves
with +/- 3 cents of error, or an average absolute-value error of 1 cent.

The really nice thing is that I can get the Lissajous patterns from any two
VCOs to sit almost perfectly still over that entire range of octaves
(including the highest one), which suggests that all of the VCOs have
virtually identical tracking errors.  Indeed, two of my four VCOs are
essentially identical, and the Lissajous patterns are dead still over 9
octaves!  Otherwise, there is a little bit of movement at 2kHz which I can
minimize by distributing it over the other octaves such that there is less
than 1Hz of beating at any frequency.  All in all, I'm pretty pleased.