[sdiy] 2164 Expo VCO tracking

[David G. Dixon]

> I wonder if the last Ota are driven into gain mode and that causes
> internal heating and/or creates current offsets thats large enough to
> cause more current into core then needed? But yet your breadboard
> figures seams OK....

The 2164 should be OK up to at least 20dB gain, or 660mV control voltage,
which would correspond to a frequency of about 4.6kHz in my VCO.  I did
replace the 1nF with a 2.2nF cap, and doubled the reference current from
2.5uA to 5uA.  However, if I dropped the cap back down to 1nF, I could move
the 20dB point to 10kHz.  That might help...?

> How did you come to the 500k on the AB mode pin? By calculating internal
> beta of the core? What if this are set wrong and causes a shift at a
> certain point?

Actually, I leave the mode pin open on the VCO.  However, I came to 560k by
looking at Fig. 13 in the datasheet, which shows that CV feedthrough goes to
0 at roughly this resistance.

I'm going to try running the 2164 in Class A mode (7.5k resistor) to see if
this helps the tracking, as per Achim's suggestion.

> One thing i have seen are if a 2164 ota are used in a triangle core loop
> a increase in gain of wave occur just before 0V CV.I figure the core enter
> active/bost mode which should be avoided when used for osc applications.
> But i might be wrong in my assumptions.But it did go very unlinear.

I'll have to keep that in mind.  That may explain why the tracking goes
astray above 0V.

I could move the entire audio range of the VCO into the attenuation range of
the 2164 if I were willing to sacrifice use of the VCO as an LFO.  There
would be 15 octaves available below 0V.  Hence, if I set the top frequency
at 0V to, say, 20kHz, then the lowest frequency would only be about 0.6Hz.
Still, this might be an option, and it would only require changing one
resistor in the circuit.  Hmmm....  I guess I could put a frequency range
switch on to run the VCO in LFO mode....

However, it would be more difficult to add HF trim under these conditions,
as there would be no positive control voltages.

> Have you tried to simply shift the operating frequency of the core by swap
> the charge cap up or down, lets say from 1n to 470p.

See above.