PWM on Wavetables

infuzion · 6 Nov 2011

Seems neat! But I wish you would have attached the OSM; I suspect there is something wrong with the waveforms. Ideally, they should be +1 -1 ballanced, so that the sum of a full cycle nets 0. I don't think your system does that, but it does seem close.

rl · 6 Nov 2011

I think there will be a little DC offset when the areas below and above the x-axis are not the same (integral of positive parts vs. negative parts). That could be improved using a DC blocking filter (high-pass with very low cutoff freqency)

Cyforce · 7 Nov 2011

From the optical it looks as already said more "neat" with the modded version, but it`s not excactly balanced on 0.0(+/-) but still better as via the stocked version. The reason why no osm - it`s just a very messy osm inside, just for testing etc, and the tweak is only to set the dc-offset from the basic valume to a specific value, which belows in the fine-setting on the waveform which you are using. 😀 So on the optical view, the changing of the dc value looks neat, but technical i`m not sure....

On the first 2 screens in the first post you see the "going down to -1" effect by using the normal stocked pwm with analog waveforms, and every basic waveform(saw) from old synths is different, someones are more in balance, someons are more on the -1 area, so the new dc value belongs everytime to the waveform.

Later the day i will upload a newer osm of that 😉

/btw, is the dc really so important? Like for example "preserve dc" of the wavetable elements, because in the end, it counts only that the sound sounds good to the users, and the little preserve function eats additional cycles, and by using a little army of wavetable oscs, well this could be a point to save cpu usage. By using 10 wavetable osc, without dc preserve i save enough cycles on my laptop to have the "cou cost" for the filter in 😀

trogluddite · 7 Nov 2011

Cyforce wrote: /btw, is the dc really so important?

I would certainly include one, as the effect of DC will depend on how the end user connect the synth. A offset may not seem to seriously affect your plugin in isolation, but it may have the following negative effects once it is in 'real use' ...

- Disrupt the behaviour of other plugins (either on the synth track or master effects), especially if they are peak-level dependent (e.g. compression,limiting, saturation) - It will eat up headroom - leading to distortion happening at lower volume levels than normal. - It could mess up the readings shown on DAW level meters etc. - More chance of clicks, pops, thumps when using fast attack/release times, or when changing presets etc.

A single DC remover after the oscillators are summed may do the job just as well as one per oscillator, so the CPU load needn't be too high.

Cyforce · 7 Nov 2011

So here now an expanded osm, for better comparision.

- stock pwm module - modded pwm module

the modded pwm module is based on an older posted file from trogluddite if i remind me rightly. there is a dc switch in there, the basic setting was -0.85, by setting it to -0.22 it looks "optical" right, but this value belongs mostly to the used saw waveform(in this case nordlead 2). Behind the osc is also a dc filter(dev 0.18 kit) which makes a quite interesting effect on the analyser by using the mod pwm with -0.22 value... or specially on the mod pwm by using the normal 50% width, which "looks" through the analyser not so ok, what the dc filter is doing...^^

So mhmm anyone with an idea how to make pwm for wavetable/analog waveforms working without this ugly offset/moving down and high...?

infuzion · 8 Nov 2011

Cyforce wrote: So mhmm anyone with an idea how to make pwm for wavetable/analog waveforms working without this ugly offset/moving down and high...?

I'm a search ninja (actually took 20min to find the stupid thing): [old link]

rl · 8 Nov 2011

Cyforce wrote: So mhmm anyone with an idea how to make pwm for wavetable/analog waveforms working without this ugly offset/moving down and high...?

I think there is something wrong with your PWM logic. For a test remove all additional offset additions, so that the pwm only adds the original wave with a phase shifted inverted wave: the result looks much better (at least for pulsewidth values < 0.9)

Acrobat · 27 Dec 2011

Cyforce wrote: So mhmm anyone with an idea how to make pwm for wavetable/analog waveforms working without this ugly offset/moving down and high...?

I think the offset of course depends on the original wave offset, if you downloaded the OSM from this post download/file.php?id=6511 you'll see that choosing the loop sample start/end of the captured wave I setted an arbitrary offset (that incidentally made me in trouble when adding one another, ATM). I think we should centre the wave BEFORE modulating it... at least that's what I got.. 😉

cyto · 27 Dec 2011

I believe there is an error in the "DC Filter" module of the "test-example_wavetable-pwm.osm" file posted a few replies back. Correct me if I'm wrong but shouldn't it read...

//out = in - inOld * f * out;//ORIGINAL out = in - inOld + f * out;//CORRECTED? inOld = in;

-cyto

Acrobat · 28 Dec 2011

cyto wrote: I believe there is an error in the "DC Filter" module of the "test-example_wavetable-pwm.osm" file posted a few replies back. Correct me if I'm wrong but shouldn't it read...

//out = in - inOld * f * out;//ORIGINAL out = in - inOld + f * out;//CORRECTED? inOld = in;

-cyto

yeah , your are right, ADD instead of MUL, now it works. 😀

Cyforce · 3 Jan 2012

The DC Blocker was from Dev Kit 0.18