[sdiy] 2164 Expo VCO tracking

[cheater cheater]

David,
that's great news!

On Tue, Apr 13, 2010 at 06:57, David G. Dixon <dixon at interchange.ubc.ca> wrote:
> OK, here's the deal:
>
> The reason why the 2164 Expo VCO's tracking falls apart at high frequencies
> is because the 2164's advertised 33mV/dB response falls apart above +20dB
> (as advertised).

so what's the reason it worked on the breadboard?

> I doubled the reference current going into the 2164 (to double the frequency
> at any given voltage) and immediately got excellent tracking up to 7kHz, and
> only about -0.2% error at 14kHz.  In other words, I obtained virtually the
> same tracking performance as lastnight, only at double the frequencies.
>
> Other observations:
>
> The Franco resistors are extraneous, but may help slightly.  Only a tiny bit
> of HF trim is needed without them, and only at 7kHz (which needed about a
> 40Hz boost).  My reference (0V) frequency would have been about 920Hz, so
> the +20dB frequency would have been about 9.2kHz.
>
> The only "weird" thing about using the 2164 is that (at least with the
> PN4391 JFET and an LM311 comparator for reset) the comparator output to the
> gate must be filtered.

Where is that filter located?

> 100pF or slightly smaller is ideal.  Otherwise, I
> think that the reset spikes are injected into the ramp, and this is probably
> what is causing the overshoot at the frequencies from 0 to +20dB.  With the
> 100pF cap, I can limit the error at those frequencies to about +0.1%, and
> the edge of the ramp looks a lot cleaner.  Tracking below 0dB is excellent
> regardless.
>
> Also, I tried Achim's suggestion of running in Class AB mode, and this was a
> total disaster.  I was correct when I said that I had been down that path
> before.  Any resistor between the MODE pin and the positive rail distorts
> the gain response, and the recommended 7.5k resistor sends it completely out
> of whack (errors of about 100%).  This pin must remain unconnected.
>
> Essentially, what all of this means is that one can get about 18 good
> octaves out of this VCO.  However, tracking will only be precise about 3 to
> 4 octaves above the 0V (0dB) frequency.  Hence, that frequency should be set
> to about 1/10 of the maximum frequency you care about tracking accurately.
> If, for example, that frequency is set at 1kHz, then tracking should be
> excellent up to 8kHz, and serviceable up to 16kHz.  However, for a +/-15V
> system, the lowest frequency will be about 0.03Hz, which gives a period of
> about 33s.  This is plenty low enough for most LFO purposes, but to get
> lower would require a range switch.

18 octaves is more than you'd ever want, I think. I've never composed
a synth part which spanned 18 octaves.

> Now I can finally put this naughty little puppy to bed and get on with my
> life!  Thanks to everyone for your helpful suggestions and encouragement.

Well done! :-) Now it's time to think about features on the PCB.

Cheers
D.