Filter causes oscillator's phase to change, multiple osc outputs

[Tausendberg]

Ok, simple experiment, you take a sawtooth oscillator and output the same sawtooth oscillator twice to two of the mixers and then link them together and output them into VCA 1. Predictably what happens when you adjust the mixer output is that they essentially just cleanly adjust the amplitude of the signal because they're in phase, they're duplicate signals basically, and so it's just more energy in the waveform.

But something I just noticed, while intending to mix a filtered oscillator output with an unfiltered oscillator output, is that if I take an oscillator and directly connect it to the mixer and then I take the same oscillator and put a second output through the default low pass filter and then put it to a 2nd mixer input and link it, I can clearly see on my oscilloscope (with the lowpass filter completely open) as I adjust the mixer levels between the clean and filtered that even though the source signal is the same oscillator just duplicated, the mixed signals are visible and audibly out of phase relative to one another.

I don't know if this is a bug per se, maybe a slight alteration of the signal's phase is some kind of byproduct of the DSP, but it would be nice of Arturia to perhaps put in some code to align the phases of the oscillator outputs when a single oscillator has more than one output.

Incidentally, I can connect two outputs of an oscillator to two low pass filters and then put the outputs back into two mixer channels and the signals are in phase again.

[Tausendberg]

I just did some further experiments...

There is good news. I just tested all of the filters including all of the settings on the multimode and they all shift the phase by the same exact amount so all I need to do is run one duplicate signal through a fully open low pass filter and the other duplicate signal through any filter I want or mix and match and they'll be in phase, not an ideal solution cause it costs me one of three filters, but it works.

There is a limitation to this trick though, apparently if you run a signal through more than one filter (osc 1 > Filter 1 > Filter 2 > Mixer) Then each filter applies more phase change and so you don't have enough pass through filters to offset 2 or 3 filters' phase changing in this manner.

Did a little further testing and apparently mixers can also cause phase change. (Sawtooth 1 > VCA 2 :: Sawtooth 1 > Mixer 2 > Mixer 1 > VCA1) and I can see when wired up this way that the signals are out of phase.

I don't know how big of a deal this might be to some people but the way I see it, part of the advantage of using multiple outputs from a single oscillator is that they will be perfectly in phase and this quirk (don't know if it deserves to be called a bug per se) does complicate things.

[Tausendberg]

In case anyone is wondering, I just tested it and the mixer shifts the phase by a slightly different amount (apparently slightly less) compared to the filter so I don't think the mixer can be used as a perfect offset for the filter unfortunately.

Update:

Alright, so, after a bunch of experimentation, long story short, the main filter module shifts the phase of the signal slightly more than, apparently, every other audio processing module including the formant filter, the filters on the noise module, the ring modulator, and the mixer output. What that means is, to compensate for the phase shift of a main filter module, you need to use another filter module and all the other modules can be compensated with any of the other modules.

Arturia, if you're reading this, if you could somehow streamline or remove these phase shifts entirely from the VST, that would go quite a long way to simplify things.

[LBH]

Are you sure it is'nt a sync behavior you get and have? Perhaps a sync issue?

How do you read/ see the phase issues in your scope?

Is it all waveforms or only when using Saw you see issues?

[Tausendberg]

Ok, I've added a couple examples of very simple patches in addition to oscilloscope images.

It clearly shows how if I make two outputs from the same exact sawtooth wave, one going straight into a mixer channel and the other one passing through a low pass filter and then to a linked mixer channel, the oscilloscope shows (and one can hear) that the signals are definitely not in phase. See what I mean?

While in the second image, if I have two outputs from the same oscillator and run them each through their own filter, and then back to the mixer, they both end up being in phase again.

This phase issue happens with all of the waveforms, not just the sawtooth.

[Tausendberg]

2nd image

[Tausendberg]

"dusted off" Modular V2 to see if it also had the same problem.

Apparently it does.

[LBH]

Thanks for the images.

On your images i see some kind of sync behavior.

What do your Phase meter tell you about phase? (Another plug-in.)

[Tausendberg]

Oh, I never used a phase meter...

So you would say it's actually a sync issue?

[jeffbart]

way interesting discovery!

It makes sense as you suggest the software may impose some latency to the start time of the cycle.
Maybe Arturia could, in future, have a setting that when ever anything goes through a filter  (or whatever causes this delay) to delay the other audio sources' cycle start ..or force the delayed wave 'back to the left'... ...like some DAWs can do to compensate for latency .e.g  'force waveform cycle sync' ?
I bet this is not quarantined just to Modular V.

There are a few other issues that I suspect may be related to the cycle starts of waves shifting around unpredictably. Matrix has some odd, perfectly in tune behaviours that I bet have got something to do with wave cycle starts...but I am no tech and am only guessing.

cheers jb

[LBH]

Oh, I never used a phase meter...

So you would say it's actually a sync issue?

Your Phase meter show pure mono in phase it looks like.
You can read about how to read the Phase meter in Studio Ones/ your DAWs manual.

Well - Perhaps i would'nt call it an issue at all. More a sync behavior. You get the same behavior if you hard sync 2 OSCs.

In the combinations where you don't get a sync behavior, i think you would get the same result raising the volume for one OSC, if the sound in each input is identical. But one can wonder why there's no sync behavior in those cases, but in others there are a sync behavior.

You know i have written about Sync in some threads. I can't tell if sync may give some phase issue. But i would not say, this at first hand has to do with phase. It's a sync behavior.
See the illustrations on page 93 in Modulars manual.

[LBH]

BTW: For the SAW waveform in an osc you get the same sync behavior using the PW parameter. That i have written about here -
http://forum.arturia.com/index.php?topic=88045.0
- as i see some troubles in this.

I also think using sync can  thin out the sound like for instance in MINI V too.  If this has to do with phase i can't tell.

[Tausendberg]

Ok, but it's not that the signal are out of sync because of the PW/Wave Shape setting, all of the oscillator wave shapes do that.

The signals appear to be out of sync because the modules put them out of sync, regardless of which oscillator is used.

[LBH]

Ok, but it's not that the signal are out of sync because of the PW/Wave Shape setting, all of the oscillator wave shapes do that.

The signals appear to be out of sync because the modules put them out of sync, regardless of which oscillator is used.

Yes i know. I just tell you that a sync behavior ALSO appear for a SAW wave, if you just use PW. Just like you also get sync when you hard sync 2 OSC's. Check it in your scope.
Check the manual as i wrote. That part has nothing to do with PW just like all the reat of my comments did'nt.

Please read my posts again more closely. I'm not at any point claming your testing has to do with PW.

[Tausendberg]

Sorry, maybe there was a misunderstanding.

All I'm trying to do is to suggest that this out of sync behavior of duplicated signals shouldn't be happening.

For instance, in Reaktor Blocks, I can duplicate signals from an oscillator and run it through filters or any other modules and the signals remain perfectly in sync.

I'm bringing this up because, I mean, I never noticed this exact issue until I examined it closely, Arturia might not even be aware of this bug.