@doctrucker said in Inverse non-linearity of extruder:

the documentation notes currently say that the M592 corrections aren't applied to extruder advance and retract moves.

Yes, but i'm not saying they are either.
It's that the corrections are often very violent (on long bowdens) by necessity - for example during infills.

So speeds like the ones tested are very much in the domain of normal printing - even though you won't get speeds like that on average, or on most of the print, you will in peaks, due to forementioned corrections.

So, i'm testing extreme speeds on an uncoupled extruder, doing extruder only moves, where, yes, linear advance is not applied, but it doesn't have to be - as i'm commanding extrusion rates that are (hopefully) similar to what you'd see during violent retractions OR violent linear correction moves.

Hope that makes it clearer.

@doctrucker said in Inverse non-linearity of extruder:

Are those linear filament speeds you are quoting there? If so that's way out side the normal print range.

Correct.
They're outside the normal print range for continuous extrusion, but not for fast retracts and extreme linear advance moves.

How many samples taken?

Not many. 2-3.

How are you carrying out your test, extrude through the hot end or just drive through the extruder drive?

Just the extruder.
I'm currently debugging issues with uneven extrusion, and i did this not so much in order to set up the linear correction as to find what the cause might be. I thought this might reveal any random slippages during retracts or sudden linear advance moves. Much to my surprise, it revealed a weird non-linearity that's irrelevant to the problem, but bizarre enough to warrant investigating further

I have.
In fact, i thought it might be due to pinch roller being too tight.

Then the hobbed wheel would dig into the filament more (lower radius) at low speeds, and possibly less (greater radius) at high speeds.

Maybe i didn't test enough of a range, Titans do grip the filament pretty strongly all the way to the point where the idler is very loose.

Has anyone ever experienced Titan extruder extruding with an inverse non-linearity? As in, it extrudes more the faster the requested feedrate, instead of less? This occurs with the unloaded (no hot end attached) extruder - i still have to test with the hotend and see what happens. The load (from the pressure) will likely kick the curve into its inverse.

Here are my results:

2mm/s - 50mm requested, 50.575mm pulled in
5mm/s - 50mm requested, 50.95mm pulled in
10mm/s - 50mm requested, 51.25mm pulled in
20mm/s - 50mm requested, 51.45mm pulled in
50mm/s - 50mm requested, 51.00mm pulled in

Any ideas on how something like this might be possible from a purely physical/mechanical point of view?