AW: tri 2 saw modualtion

[Haible Juergen]

Great idea!
Works fine in Simulation, as long as the rising/falling slope ratio
is not too large.
For an almost-sawtooth wave you'd need very large CV's, however.
(the old 1/x problem raises its ugly head ...)
Maybe this circuit would still work with exponential VCAs instead
of linear ones? As the 4 VCAs only have to be matched, and the
two CV-generating ones are in a feedback loop, the CV sensitivity
might be exponential as well ... haven't tried this yet.

JH.