[sdiy] SH-3A: so it's Pwm, but ...

[jhaible at t-online.de]

Thanks for the pointers to the SH-3 schemos.
The zipped ones at music machines are somewhat readable.
Unfortunately, parts of the diagrams at the edges of the
sheets are lost, so I could not figure out everything.

What I did figure out so far:
PWM yes, ordinary PWM no.

There's a fairly ordinary triangle VC-LFO (slightly 
unsymmetrical wave, but just a little), and a comparator.
Between the comparator and the LFO, a LPF.
So the higher the "chorus rate", the lower the effective
LFO amplitude. Also, the waveform is almost sine-shaped
for high rate, soft triangle for medium rate, and almost
triangle for low rate.

This is a quite sensible choice already, but there might
be more. There's a "Bias" circuit built around another
opamp and a transistor, adjustable with two trimpots,
which apparently adds an offset to whatever signal is used
on the audio side to feed the comparator.
Unfortunately, I can't see what feeds the bias generator,
nor how the audio signal is processed before it runs into
the comparator. Square waves obviously will not work,
so maybe it's a signal shaped by C251, TR251, and the caps
at the "transpose" switch ??
Too bad the connections are not on the page - this looks 
interesting.
So if anybody has a more complete scan of this, and some
interest to solve the mystery, mail me a copy.
However, don't make scans or even send paper copies. I cannot
return such favors at the moment. When my interest rises
- which is unlikely - I can still get the schemos from Roland.
Just wanted to give you some return of my question - it
*does* look interesting.

JH.