Is there any synth that can do roland's Saw PWM?

[Ingonator]

Anybody tried to use wavetables in either PPg Wave 2.V or PPg Wave 3.v for this?
Maybe i will give that a try soon.


Ingo

[zorniko]

cytospur's DUNE exapmle:
EDIT: Look down.

[cytospur]

Thanks for the pic zorniko, but it doesn't show the whole story and neither do the other pics I'm afraid. The waves are very dynamic and that doesn't come across in these pictures.

[zorniko]

Hm... anywhere I choose to take a pic, I get almost the same waveshape results. In your case, I guess you're talking about the lfo "spots"?

[bstageboss]

+1 for that

1
2
3
4

NIICE! And very revealing! Thank you for the illustrations, zorniko!

I had a revelation after my last post about the means of obtaining a sawtooth wave in the D-50, and just as luck would have it, the example 2 picture proved it correct.

First up, another sound example:

http://www.box.net/shared/d8nx3unh1zrjdqzunrv5

Zorniko, if you would be so kind as to compare it to the example 2 image and post it for the folks on this thread, I would be much obliged.

As far as an explanation goes, it's an interesting story about the D-50 oscillators (or wave generators, as Roland calls them):

There are indeed only two waveforms in the WG, but they're not square and sawtooth, but square and sine. The tricky thing is that the sine wave is 'hidden' and never heard directly. The sawtooth wave you hear is actually a sine wave ring modulated by the square wave. The result is that half of the sine wave gets flipped upside down.
You might expect that to result in a waveshape that resembles one of the TX81Z operator waveforms (or the impression Pamela Anderson leaves in the sand after falling face down), but Roland has another trick: The sine wave is offset by a phase of 90 degrees, so that the flip occurs at the maximum and minimum amplitude. Now the result looks a lot like a sawtooth wave with a bit of curvature to it. It really is just the first half of the offset sine wave over and over again, resulting in a sawtooth wave at half the wavelength i.e. twice the frequency. That's why the sound goes up an octave when SAW is selected.

What this means is that the D-50 has 4 additional hidden ring modulators used for nothing more than to generate a sawtooth wave from 2 simpler components. There is always a square/pulse wave involved in making the sound, which means the somewhat cryptic statement (about only square waves being used in the sound generators) in the D-50's manual is now starting to make sense.

Now you can also see what happens when the square wave's pulsewidth is modulated: the flip becomes asymmetrical and at extreme settings, sounds closer to a narrow pulse wave than a sawtooth, thus accounting for the nasal quality of the sound.

With that information, I decided to replicate this process with one of my Plugin synths and got the above example.
As you'll hear, the nasal quality is definitely there, and, with a bit of tweaking, could probably be made to sound very close to example 2 in the first sound file.

Now all that's left is me letting all of you guess which VST I used for the mystery example. No prize, unfortunately, but I hope you guys have fun with it anyway.

STV

[cytospur]

Well by any definition saw or pulse width modulation is dynamic. It's not a criticism of you, I'm just pointing out that your snippets aren't showing the whole picture. I know from looking at my own creations with Signal Analyser that they are changing quite a bit from the snap that you showed. Again it's not a criticism and thanks for showing the pics you have done so far.

[zorniko]

In bstageboss's erxamples the pics are exatcly the same as he described it. So... I think the pics ARE ok. However, maybe I can post a longer segment of your sample?

[cytospur]

+1 for that

1
2
3
4

NIICE! And very revealing! Thank you for the illustrations, zorniko!

I had a revelation after my last post about the means of obtaining a sawtooth wave in the D-50, and just as luck would have it, the example 2 picture proved it correct.

First up, another sound example:

http://www.box.net/shared/d8nx3unh1zrjdqzunrv5

Zorniko, if you would be so kind as to compare it to the example 2 image and post it for the folks on this thread, I would be much obliged.

As far as an explanation goes, it's an interesting story about the D-50 oscillators (or wave generators, as Roland calls them):

There are indeed only two waveforms in the WG, but they're not square and sawtooth, but square and sine. The tricky thing is that the sine wave is 'hidden' and never heard directly. The sawtooth wave you hear is actually a sine wave ring modulated by the square wave. The result is that half of the sine wave gets flipped upside down.
You might expect that to result in a waveshape that resembles one of the TX81Z operator waveforms (or the impression Pamela Anderson leaves in the sand after falling face down), but Roland has another trick: The sine wave is offset by a phase of 90 degrees, so that the flip occurs at the maximum and minimum amplitude. Now the result looks a lot like a sawtooth wave with a bit of curvature to it. It really is just the first half of the offset sine wave over and over again, resulting in a sawtooth wave at half the wavelength i.e. twice the frequency. That's why the sound goes up an octave when SAW is selected.

What this means is that the D-50 has 4 additional hidden ring modulators used for nothing more than to generate a sawtooth wave from 2 simpler components. There is always a square/pulse wave involved in making the sound, which means the somewhat cryptic statement (about only square waves being used in the sound generators) in the D-50's manual is now starting to make sense.

Now you can also see what happens when the square wave's pulsewidth is modulated: the flip becomes asymmetrical and at extreme settings, sounds closer to a narrow pulse wave than a sawtooth, thus accounting for the nasal quality of the sound.

With that information, I decided to replicate this process with one of my Plugin synths and got the above example.
As you'll hear, the nasal quality is definitely there, and, with a bit of tweaking, could probably be made to sound very close to example 2 in the first sound file.

Now all that's left is me letting all of you guess which VST I used for the mystery example. No prize, unfortunately, but I hope you guys have fun with it anyway.

STV

Spot on, I think, from what I've managed to read on the subject this evening. It sounds like my attempt wasn't too far off.

[zorniko]

Here they are:
Mystery:

D-50:

[bstageboss]

Here they are:
Mystery:

D-50:

OK, looks like my phase offset is going other way, but it's pretty clear that they're both derived from sine waves. Just take the portion that looks chopped, flip it upside down and you'll get the original sine wave back (or something close to it).

Thank you so much, zorniko.

STV

[zorniko]

cytospur, I think this pic of your sample may be better:

Not the whole story (we don't have that much space , but close enough, I hope.

[cytospur]

Yeah, it's difficult to show what's going on in such a limited space. A lower frequency would be better though, but as you say/show there are limitations. Can you do FFTs of the waves? That would be an interesting comparison.

[Muzik 4 Machines]

EDIT: Here's a question maybe for Urs or anyone else interested in obscure Roland waveforms that has a JP-8000. It's Triangle waveform has a PWM parameter (NOT the triangle mod waveform) which seems like it's doing some gentle cross between FM and PWM to the triangle wave... Anyone got any clue what's going on with it? (I can make a waveform diagram if needed)

isnt it like from up saw at 1 end, triangle in the middle and down saw at the other end?
my fav wave from my jp8080 was the feedback osc, but that's deserves it's own thread lol

[zorniko]

Yeah, it's difficult to show what's going on in such a limited space. A lower frequency would be better though

Yes, I choose the lowest.
FFT - you must be more concrete

[cytospur]

Yeah, it's difficult to show what's going on in such a limited space. A lower frequency would be better though

Yes, I choose the lowest.
FFT - you must be more concrete

I see. What program did you use to grab the waves? They don't look like what I see with Signal Analyser. What I mean by FFT is Fourier transforms of the wave files. I don't have time to do this myself atm, maybe tomorrow.