[sdiy] Prophet VS a phase accumulator design?

[karl dalen]

Thanks to everyone participating into these questions.

I took the liberty to ask Dave directly to see if he knew:

* Dave Smith said:
* "Dont remember much of the details. I think we had custom chips
* that shifted the sample rate, but not sure. The Evolvers use the
* same waveshapes as the VS, otherwise it's different."

Perhaps something can be figured out of that?

The Evolver uses a ADSP-2191M for the oscillators and delays. 

> Right, the phase increment is larger than one sample.

> The ESQ should be similar to the DW8000, only with a sample
> every 8 semitones or so, depending on waveform, and much shorter
> waveforms. 
 
> The ESQ has a lot more audible aliasing than the DW-8000.

Yes remember that to. 
An instrument who failed misserably using the DW8000 principles where
the Kawai K3, serious octave miss match, one could clearly hear that a
different table was used. 
 
> My understanding of it is this: I believe the PPG and
> Prophet VS are fixed sample rate, but with much higher rates
>(in the hundreds of kHz range), so they can more or less get
> away with a single waveform over the entire keyboard without
> too much aliasing. Synths like the DW-8000 and ESQ use a lower
> sample rate (50 kHz or so), so they need to use multisamples. 

I vaguely remember Paul M once said 40Mhz split by the oscs?
 
> But the difference between fixed and variable sample rate
> becomes a little arbitrary at high rates. Variable sample
> rate systems that aren't clocked by a VCO derive their clock 
> from a programmable  divider, in which case you can say that
> the clock source (crystal oscillator or whatever) is the
> "true" sample rate, only the phase increments are always
> one sample and constrained  to integer divisions of the
> clock frequency. It's trading aliasing and jitter for high
> frequency pitch resolution.

Hmm, and then we have te mysterious DSS1 oscillator 18pieces of
8253 timers and a custom DMA chip.It seams that the band limitation 
software Chriss mentions are included into the DDS1 to directly
compute the band limited octaves(waveform mode only or does it
also do that on samples?) DSS1 schematics says variable sample rate
would that suggest a timer rollover tick to do a DMA sample transfer
but that sounds very slow and would limit replay range?!

Regards
Kd