Certain effects on parallel channels will cancel themselves others will not, I wonder why

[manisnotabird]

I was curious about aliasing and upsampling and how much of a difference it would make it certain real world cases. So I started a loop, made a parallel channel, inverted the phase on the parallel channel (which, as I expected produced silence). Then I loaded an instance of Blamsoft DC-1 Distortion on each channel. My plan was to investigate how much difference there was between the various "quality" settings on the back. But with identical settings including "quality" they didn't cancel. Further investigations lead to some interesting results. Some effects I'd expect wouldn't cancel: if they had any kind of non-synced LFO with randomly assigned phase, or "noise" built-in, like say reverb with modulation, or lo-fi effects with "dither" options. But I would expect two distortions on identical settings to cancel. Curiously, a Scream4 on the "overdrive" setting will cancel another inverted Scream4 with identical settings, but not if both are on the "Distortion" setting. I'm just curious why some built-in effects/rack extensions/VSTs that don't obviously feature unsynced-LFOs with randomly assigned phase don't cancel.

[antic604]

Curiously, a Scream4 on the "overdrive" setting will cancel another inverted Scream4 with identical settings, but not if both are on the "Distortion" setting. I'm just curious why some built-in effects/rack extensions/VSTs that don't obviously feature unsynced-LFOs with randomly assigned phase don't cancel.

Perhaps it adds a random element / processing? Or works differently for positive / negative parts of the audio waveform?

[manisnotabird]

Or works differently for positive / negative parts of the audio waveform?

I guess I didn't consider if the inversion button at the top of the mixer works pre-inserts or post-inserts.

[manisnotabird]

Or works differently for positive / negative parts of the audio waveform?

I guess I didn't consider if the inversion button at the top of the mixer works pre-inserts or post-inserts.

This is (at least) part of it. I used a Selig Gain to invert the signal, and placed in it the insert and switched it between pre-distortion and post.

[manisnotabird]

The Scream4 now completely cancels in "distortion" mode with the Selig Gain doing the inversion post-distortion, but the Blamsoft doesn't. It must have a "hidden" random LFO doing something.

[antic604]

Or works differently for positive / negative parts of the audio waveform?

I guess I didn't consider if the inversion button at the top of the mixer works pre-inserts or post-inserts.

That's a good point (it's pre-inserts), but I was thinking of something else - an algorithm may for example push the waveform "up" and then fold it, instead of pushing it both up & down from the center; so that would obviously sound different for reversed phase waveforms and thus not cancel.

[manisnotabird]

I guess I didn't consider if the inversion button at the top of the mixer works pre-inserts or post-inserts.

That's a good point (it's pre-inserts), but I was thinking of something else - an algorithm may for example push the waveform "up" and then fold it, instead of pushing it both up & down from the center; so that would obviously sound different for reversed phase waveforms and thus not cancel.

I don't follow. If the phase reversal is placed *after* the effect in the signal chain it should cancel a parallel channel with the same effect non-inverted assuming there's no "random" element of the effect. I was previously using the invert button on the mixer which placed the phase reversal pre-effect, and thus effects which treat the top half of the waveform different than the button half wouldn't cancel.

[diminished]

First of all, did you try it with a mono signal?
I don't have DC-1 but DC-9 works as expected.
DC-1's shop page states "simulating nonlinear circuits" and I think there's your answer if you have not made any logical errors in your cabling. There might be something embedded which simulates some sort of drifting, think: free running LFO/modulation, a random seed of a paramenter, something like that. Complete cancellation would obviously only work in a linear, time-invariant system.

Edit: that's not it, see below.

[manisnotabird]

My understanding of phrase “non-linear” is doesn’t necessarily imply any randomness. Not sure why it would make any difference if the DC-1s input was mono or stereo. The little signal chart in the back of the rack extension shows “dual mono” for a stereo input which is what I’d expect: the way to get stereo from a BOSS DS-1 is to have two of them.

[diminished]

Not sure why it would make any difference if the DC-1s input was mono or stereo.

That was just my idea for keeping things simple, if it's dual mono then all the better.

My understanding of phrase “non-linear” is doesn’t necessarily imply any randomness.

If it's modeled after an analog device then chances are the time-variance of the modeled circuitry is also somewhere in there. Think of oscillator drift for example. Ofc only Blamsoft knows what's doing what - but I wouldn't be suprised by such behavior from a device that claims to be "simulating nonlinear circuits"

[antic604]

That's a good point (it's pre-inserts), but I was thinking of something else - an algorithm may for example push the waveform "up" and then fold it, instead of pushing it both up & down from the center; so that would obviously sound different for reversed phase waveforms and thus not cancel.

I don't follow. If the phase reversal is placed *after* the effect in the signal chain it should cancel a parallel channel with the same effect non-inverted assuming there's no "random" element of the effect. I was previously using the invert button on the mixer which placed the phase reversal pre-effect, and thus effects which treat the top half of the waveform different than the button half wouldn't cancel.

Yeah, you're right. I was still thinking pre-effect since - that was my understanding - you wanted to see if the added processing still cancels out.

[guitfnky]

maybe try un-inverting the one channel, and bounce both to new audio tracks and compare waveforms? might be interesting to see what’s going on visually, if nothing else.

[diminished]

Okay so I installed the trial to see for myself and there's complete cancellation if you put DC-1 as an insert into both channels. I used the free Itsy RE for post fx phase inversion.
Reason Lite file attached.
So there might be something else going on in your project?

dc1phaseinvert.zip

(33.93 KiB) Downloaded 35 times