[sdiy] Synthex Oscillator

[Tom Wiltshire]

> What follows is a pure wild-ass guess, first-order conjecture of the
> worst kind.

Hey, don't worry, we're amongst friends.

> There's no way it's a digital current source because there's no way to
> preload the counter.  The only way I could even vaguely see with that
> layout was if there was some way to reset IC11/IC21 and then feed them a
> predetermined number of clocks.

My thoughts exactly. So what the hell is Mr Maggi on about? I really don't want to disagree with the designer of a famous synth about his own work, but I'm not left with much choice in this case.

> What I'm seeing here is in the "top half" of the diagram, a 12-bit
> counter, which would give pretty decent frequency control.

Where are you seeing 12 bits? There's a 8-bit counter on the left, and another 8-bit counter on the right. That's a simplification, a bit, but I don't think it's 12-bit.

> In the bottom half is an eight-bit counter with a simple DAC.  C93 and
> Q90 do indeed form a kind of output filter with Q90 discharging C93 when
> the saw resets by driving its base low.  I'm guessing that R70/R72 keep
> pins 5 and 2 of IC1H approximately low, with C69 allowing them to flick
> high when the MSB goes high?  That would give you a quick reset pulse if
> "SAW 1" was high.  I guess the DAC outputs are inverted by the XOR gates
> or something, but 3H has to be doing something or one leg of each gate
> is left floating.
> 
> I don't know what H2 does or MR2 but they seem related to the "DAC
> rolled over" pulse too.  Something to do with syncing sawtooths?

Yes. MR2 is the Master Reset for Osc2, here used as a Hard Sync pulse.

> Looks like the bottom 4013 is wired as a flipflop so depending on how
> EOR1 is set, it'll either take the "other" input to the XOR gates from
> /Q on the flipflop (that would be a triangle wave, I'd think - count up
> then down - but wouldn't that make the triangle wave be an octave down?)
> or from VCO2 for the "ring mod" effect.

Yes, exactly. Triangle wave being an octave lower can be corrected by switching the octave divider up one, I guess. Dunno what happens in the top octave. Maybe the ramp doesn't use the highest setting.

> The 12-bit frequency divider followed by the 8-bit DAC counter would
> give you a range from about 4Hz to 15625Hz, which sounds about right.

It would, but…

Tom