[sdiy] D50 a VA or not?

[Tom Wiltshire]

This sounds like synthesis of pulse waves using BLEPs, give or take. If so, the D50 probably predate the papers on the idea?

T.

On 27 Dec 2010, at 07:12, Juergen Haible wrote:

> I'm not a digital guy (and even less a guru), but I thought I'd understood that clever TVF in the D50 working like this:
> For an approximated filter function on a pulse wave, you don't need a real filter. You just need the step response of the filter you want to emulate (lookup table, surely), and trigger it with each upward and each downward slope of the ("filter incoming") slope of the oscillator.
> This only works on a single oscillator, of course - thta's why wach oscillator has it's own filter in the D50. (Clever marketing gives 2 filters for Single voices, or 4 filters for Dual voices, as an added benefit.)
> There are no saw waves either - they look like a sine multiplied with the pulse waveform. That "filter" step response table method won't work on saw waves, either. So I think what they did is performing the "Filter" function on the pulse waveform *first*, and then multiply it with the sine.
> Also fitting in here: No filter function availavle when Ring modulator is used. They use up the same multiplier for ringmod, that would otherwise be used for the "filter" on saw waves.
> That's what I conclusded from the description in the user manual. No idea how accurate or inaccurate it is - does it fit to what you found out, somehow?
> 
> As for "VA": It predates the terminus (or at least the marketing use of the terminus), so it's an anachronism. But from it's purpose, it clearly was intended to give an emulation of an analog synth (plus added features like attack PCMs), so if it's not a virtual analog, I wouldn't know what else to call it.
> 
> JH.
> 
> 
> 
> ----- Original Message ----- From: "karl dalen" <dalenkarl at yahoo.se>
> To: <synth-diy at dropmix.xs4all.nl>
> Sent: Monday, December 27, 2010 4:24 AM
> Subject: [sdiy] D50 a VA or not?
> 
> 
> Little are known to the public on how the D50 works, so i had this brief
> conversation with a former Roland R/D manager on that item i asked as a
> shot in the dark if it was just table reading for the osc's and some
> kind of simple ASIC DSP for the filter TVF.
> 
> He replied that the D50 was a real VA based upon clever table manipulation
> to simulate the filter effects, that was also the reason PCM samples could
> not go trough the table process (TVF). (Presuming the entire wave would be
> screwed up!)
> 
> So the TVF in the D50 appears not to be a filter at all, rater relabeled
> to set the user into believing it was! So assuming the following (might be wrong or right i don't know) the D50 has two cyclic waves, Pwm and Saw, the
> Pwm are created out of a sine lookup table, Xor'ing or something then goes
> trough this TVF thing, then the Pwm are multiplied and shifted with the
> Sine to create a double freq SAW and that's it? Since the Saw are flipped
> up so are also the Pwm's lower edge resonance emulation and everything
> looks neat and tidy.
> 
> So i did some initial tests, i took 2 oscillators one Square and one Sine
> 2x- freq and phase locked to the Square front edge who also triggered an
> AD who shaped the amp of the Sine to get the log resonance curve and then
> simply added these two and then swept the freq of the Sine and sure it
> sounds exactly like a mid open filtered Square with resonance.
> 
> I assume some correlation between the amplitudes of the Square and Sine
> during play time to satisfy the emulation of a closing filter effect
> must also be in place somehow, or some phase shifting thing to get
> Sine dominat in amplitude over square during lowering of center
> frequency to emulate a closing filter?
> 
> So a partial in the D50 could be made up of one table osc creating
> the actual tone and one sine table osc to create the resonance/filter?
> 
> 1:So digital guru's what do you think about that D50 thing?
> 
> 2:And what really defines à VA instrument? I had thought a mirco running
> a algorithm that simulates analog functions was regarded VA, i.e computes
> the function, not by fetching data from predefined tables, or similarly?
> 
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy