[sdiy] Dual oscillator cores?

[David G. Dixon]

> Regarding the triangle core designs: I'm not sure how triangle cores
> switch the direction, but I presume that for example when switching
> from rising to falling that the core has to be *disconnected* from the
> positive current source and only when that is done connected to the
> negative current source, right? I would understand that in real-world
> terms this means some form of delay between the 'disconnect' moment
> and the 'connect' moment, but of course I'm not a solid state
> engineer. It would seem like this could be a much shorter time than a
> saw core flush though. This is just pure speculation on my part..

One example from Thomas Henry, of which I've built two, is given here:

http://mypeoplepc.com/members/scottnoanh/birthofasynth/id33.html

The way this one works is that a schmitt trigger (comparator with
significant hysteresis -- a discrete version based on two NPN transistors in
this case) sends +/-5V signals into one input of an OTA (the other input is
grounded) at a high enough differential voltage to saturated it, with the
expo current coming into the Iabc pin.  The OTA feeds an integrator.  This
gives reasonably nice triangles, although judging by the output from the saw
shaper I added to it, the two slopes are not exactly equal and opposite.
The speed of the schmitt trigger is such that it still requires HF trim, as
well.

With my new 2164-expo VCO, I'm thinking that a triangle-core VCO could be
done very easily simply by switching the polarity of the
reference-current-determining voltage with an opamp and a JFET wired as a
buffer/inverter fed by a schmitt trigger at +/-5V.  I don't think that the
switching speed would be as critical as it is in a saw-core VCO.  However,
the gain-determining resistors on the opamp should probably be hand-matched
to 0.1%, which is usually no big deal with a batch of good 1% resistors.