[sdiy] filtering out a saw-reset glitch in a triangle shaper

[David G. Dixon]

I've simulated this, and it works like a dream.  I connected the collectors
directly to +15V, connected the inverted and non-inverted saw through 22k
resistors to the bases (this resistor value is pretty arbitrary), and
connected the emitters through a 180k resistor and a 1nF cap to -15V.  The
only important values are the emitter resistor and cap, and, of course,
increasing one has the same effect as decreasing the other.  The selection
or RC is a compromise:  It must be large enough to filter out the glitch,
but not so large that it skews the resulting triangle wave, at least within
the useful audio range.  With and RC of 180us, the glitch is mostly gone,
and the triangle is symmetrical up to about 10kHz.  Beyond that I don't
really care.

The parts count of this technique is a little bit lower as well: one fewer
opamp, and only the one capacitor.  The glitch isn't cleaned up quite as
effectively, perhaps, particularly at high frequencies, and there is the
skew issue, but on balance, I think this might be the preferred approach.

> Connect two transistors as emitter followers with the emitters tied
> together. Feed the second transistor with inverted signal. By connecting a
> small capacitor to the output, you can easily control the slewrate
> asymmetrically. The transistors can charge the capacitor very fast, but
> the resistor/current sink at the output has only finite current available
> for discharging it. Now arrange input sawtooth polarity so that the glitch
> at triangle output goes down and adjust resistor/current sink and
> capacitor to taste.