PWM Questions

[mmxgn]

Hello,

I recently joined this forum (first post).
I really like zynaddsubfx and recently started to see it more seriously, although I am a total sound synthesis noob.

One thing I want to do, is start with a PWM wave. I know that zynaddsubfx (in my case, yoshimi) does not directly provide it, so I tried to emulate it using two saw waveforms at different frequencies (one with parameter 63 and the other at -64) as described in this article:

http://www.soundonsound.com/sos/Mar03/a ... rets47.asp

And playing around with the frequency of the second oscillator (using ADsynth).

The thing is, the resulting waveform (I view the waveforms using baudline) do not look quite like ones in figure 3 in that article. Aside from the different amplitude (which is not a real concern) I get some bad "peaks" at the edges of each pulse.



Uploaded with ImageShack.us

I don't really know but my intuition says that these peaks include some very high frequencies, that, especially when playing low pitched notes, sound like "bad" noise.

Also, for a ramp and a saw (like in the article) at the same frequencies, one should expect that one would cancel the other. Still, these high pitches are still there.

I also notice that a saw with a changed parameter shows an anomaly in the oscilloscope, at its "edge". I am guessing this is the cause of the weird peaks at the resulting pulse waves.

Questions:

1. Is there a way to make a perfect "saw" and "ramp" wave, as in the article?
2. Is there a way to erase these peaks afterwards?
3. Is there another way for PWM in zynaddsubfx/yoshimi?

Thank you for your answers

Edit: Managed to create a cancellation by using a "bent" sawtooth waveform (parameter -63) and the exact same, but "reversed".

[fundamental]

1) It seems you have found a work-around
2) I would assume you are seeing gibbs phenomenon, which is harmless (google for more info)
3) In the adnote voice parameters, there are modulator, like PWM.
I am not sure on the exact signal processing that occurs with them, but try fooling around with them.

[mmxgn]

Thank you for your answers.

1) yes, and it works like a charm.
2) It was not the Gibbs phainomenon I was talking about (I'm familiar with it).
What I was seeing was actually consequence of the fact that a saw with parameter -64 is not a "reverse" saw with parameter 63, there were some moments in the beginning that would not cancel each other.
3) Can you direct me exactly? The only modulators I see are "MORPH (I don't know what this does)" "RING" "PM" and "FM".

[FreeFull]

You can get a perfect saw by using the power waveform with a parameter of 0.

[Chris-S]

Hi,
you can also get PWM by using PM:



You can then use the modulator envelope to change the Pulse width over time.

Note that I've put the F.Damp to zero (otherwise the amount of modulation
is depending from the note number).