Ian Fritz ijfritz at
Sun Mar 12 05:07:18 CET 2000

Hi all --

I've had some time over the past couple of weeks to work on developing the
new tri VCO core I posted a while ago. (Also see EN #111). It has turned out
to make a really nice VCO.

The main drawbacks of the original design were (1) a limited frequency range
and (2) a large and varying common-mode voltage presented to the
differential integrator. These were easily fixed by replacing the 100 k
resistors connected to the power supply with 1 k resistors connected to a
fixed 3.3 V reference.

For the hysteretic switch I used half of an LM319 comparator followed by a
symmetric zener-diode clamp. The second half of the 319 is slaved the first,
but out of phase. These two outputs are fed back push-pull to the transistor
switches. This makes a totally symmetric design that produces a clean,
symmetric triangle.

The oscillator easily runs to well over 100 kHz (tested with 500 pF caps and
1 ma of control current). With 1 nF caps and a current running up to 1.5 mA
maximum, the oscillator can be controlled from about 0.1 Hz to 70 kHz. BTW,
the caps don't have to be matched (exercise for the reader <g>).

As with my earlier experiments using OTA-based tri VCOs, I found that the
exponential tracking curve depends on the transistor drive level. Again in
this case, a drive level of 60 mV gives good tracking, whereas driving near
saturation (100 mV) gives a warped curve with severe overtracking at the
high end.

For the expo converter I used transistors from a CA3083 array. These have
much better performance than indicated in the spec sheets. The pair was
matched to delta Vbe < 0.2mV and had betas over 200. As I noted some time
age, they also have low series emitter resistance compared to the CA 3046
array or 2N3904, etc.

Tracking is within +/- 0.1% from 50Hz to 40kHz. Temperature stability has
yet to be investigated.


More information about the Synth-diy mailing list