Channel EQ. Is it non-linear?

[romariozen]

Hi,
It doesn't say in the Live manual that it is analog, yet the phases of two identical sounds (one with default channel EQ) don't cancel out.
Is that kind of non-linearity, because if it is, it acts strangely because it seems like the distortion doesn't depend on signal level.

P.S. the most audible results are with the snare.


Thanks in advance

[[jur]]

yet the phases of two identical sounds (one with default channel EQ) don't cancel out.

What's your test setup exactly?
Are you sure about what non-linearity means in the context of an EQ?

[fishmonkey]

from the description in the manual i guess it's modelling an analog console:

https://www.ableton.com/en/manual/live- ... channel-eq

[jlgrimes]

I believe this device is linear. It only models the curves of an Analog EQ. I thought I remember something stating it doesnt Saturate or Compress anything.

[pottering]

It has bands, the bands' crossovers are not completely flat, probably Butterworth crossovers with a +3 dB gain peak at the crossovers.

It is just a small gain change, it is not distortion, you can play a sine wave and see that no noise, harmonics or inharmonics are added.

EDIT: The way a curve appears when High is set to negative values makes me think that curve is there all the time, just out of view of the display, around 20k.

[Angstrom]

It would be very odd if the channelEQ was linear phase as the big daddy EQ8 got a lot of work from Andy Cytomic and others to make it linear drive, linear drive means if you put a sine in you get a sine out. Its not turning sines into saturated driven waves.
But linear phase is about temporal correlation. That's very handy in many circumstances such as a mastering context but not necessarily "better" in all cases, as there are tradeoffs there too.

If EQ8 is linear drive & nonlinear phase then I'd suspect that Channel EQ is too.

[subparuser]

It would be very odd if the channelEQ was linear phase as the big daddy EQ8 got a lot of work from Andy Cytomic and others to make it linear drive, linear drive means if you put a sine in you get a sine out. Its not turning sines into saturated driven waves.
But linear phase is about temporal correlation. That's very handy in many circumstances such as a mastering context but not necessarily "better" in all cases, as there are tradeoffs there too.

If EQ8 is linear drive & nonlinear phase then I'd suspect that Channel EQ is too.

I'm confused - are you saying linear-phase and linear-drive are separate qualities?

It's my understanding you have to engage the oversampling-mode on EQ8 to get it to behave in a linear fashion, it's non-linear by default - perhaps I've misunderstood something along the way?

[Angstrom]

I'm confused - are you saying linear-phase and linear-drive are separate qualities?

It's my understanding you have to engage the oversampling-mode on EQ8 to get it to behave in a linear fashion, it's non-linear by default - perhaps I've misunderstood something along the way?

thats right.

linear drive means ... well, think about Saturator device, its a non linear amplitude device.
The shape you see on the interface (for example the curve for analog clip) is the "non linear " transfer curve for increasing signal amplitude.
So it can turn a sine into a sort of rounded square if you drive the input. That transfer curve output when viewed as an FFT of frequencies in Spectrum shows the initial sine (lets say at 100hz) and some odd number harmonics (300hz, 500hz, etc).

EQ8 used to have filters which used to produce some subtle harmonics when driven, but not since it got upgraded by Cytomic.

Linear PHASE means that the phase of the sine wave will remain time synchronous. It's very different. Most filters act on the time domain as an equal consequence of filtration. A linear phase filter tends to maintain time syncrony, BUT it usually requires a buffer so while it wont time shift relative to its source it WILL incur latency overall - which would then need to be compensated for. So if you had a 2 minute recording of drums and EQed it with linearphase the transients would remain in position - BUT the whole drum track would be offset by the buffer required.
Linear phase filters usually incur quite a bit of latency and use more CPU resources

Now, you mention the HiQ mode. this is different again. It runs "oversampling" to run the EQ at 2x (?) the samplerate to push any aliasing audible artefacts far above the human hearing range.
To oversample requires a process of a sample buffer of just a few sample lengths. When you use the HiQ EQ 8 is adds about 4 samples of latency to that stream. But thats nothing too do with the drive linearity, just the oversampling.
All numbers from memory. All facts may be badly remembered

[subparuser]

Sure, I understand any and all saturation / distortion is a nonlinear process etc - it's just I hadn't ever seen the terms broken out like this before in regards to EQ. It was the terminology that threw me slightly - insofar as "linear-drive" is a term I hadn't encountered before.

I was, however, under the impression that the oversampling mode on EQ8 didn't just decramp, but also ran the device as linear-phase. Looks like I need to find a different EQ for that task then. Shouldn't be a problem, I have plenty to choose from . . . though they might run the CPU a bit harder than EQ8.


Appreciate the more detailed reply, cheers!

[pottering]

I think OP was asking about Channel EQ not being 100% transparent with the settings zeroed.

I think OP assumed the null broke due to distortion, so that's why he was using terms related to distortion.

Just happens "linear" is also used with EQs ("linear phase" as others explained).

I think Channel EQ is linear in the context of distortion (no distortion, a smooth sine wave will stay a smooth sine wave), but "minimum-phase" (not "linear phase") in the context of EQs.

[[jur]]

EQ8 and Channel Eq are not linear phase, and they're not adding saturation.