[sdiy] M110 with more outputs

[Scott Nordlund]

> >> No idea if it's technically possible, but if the next revision
> >> of your M110 clone could have outputs for more than 4 octaves of
> >> squares, that would be I/O pins well spent. That annoying whine
> >> four octaves above is probably the worst feature of the Siel
> >> Mono. Really limits its field of application.
> >
> > It will likely need a higher clock rate if you want to preserve pitch resolution.
> > Those buzzy sawtooth waves are really irritating, it's the bane of divide-down architectures. I like it on triangle waves, though (as found in the Crumar Bit One, Casio HT series, NES...). The Polymoog had a reasonable solution...
>
> Why do you get a whine four octaves above? What did the Polymoog do about it? Why are the buzzy sawtooth waves irritating? Aren't sawtooth waves supposed to be buzzy?
>
> I'm getting a coming-in-in-the-middle-of-a-conversation feeling here. Did I miss something?
>
> Thanks,
> Tom

You didn't see anything that I didn't, I just inferred what he was talking about. Lots of divide-down-style organs, synths, etc. mix multiple octaves of binary weighted square wave tones (i.e. Rademacher functions) to approximate a sawtooth. You usually see 16', 8', 4', 2' mixed together to make a 16 stepped ramp waveform. The spectrum is correct through the first 8 (?) harmonics, but after that you get a bunch of extraneous stuff. Picture it as either image frequencies (due to "sampling" of an ideal sawtooth), or an additional "phantom" sawtooth wave, 4 octaves higher and inverted. It sounds buzzy and crappy. Some companies found ways to make real sawtooth waves, or at least better fake ones. The ARP Omni puts a single square wave through a high pass filter and diode. The Polymoog (as far as I can recall from a brief glimpse at the schematic)  adds a DCO-like ramp waveform to square waves, producing sawtooth waves at arbitrary octaves (a huge improvement since it permits glorious PWM). Anyway, the proposal is to make a clone of a popular divide-down chip that does a better job of this, but simply adding more square waves to make a higher resolution ramp (i.e. moving the "phantom sawtooth" to a less offensive higher octave) will require a higher input clock frequency.