Pressure advance causing reverse extrusion

dc42

@jim546 what firmware version are you running?

jim546

@dc42
Board: Duet 2 Maestro (2Maestro)
Firmware: RepRapFirmware for Duet 2 Maestro 3.4.0 (2022-03-15)

dc42

@jim546 can you provide a short section (a few 10s of lines) of a GCode file that demonstrates the problem?

jim546

@dc42
This gcode recreates the problem.
new 1.gcode

jim546

I've revised the testing g code to get rid of the extruder priming moves and run it multiple times. Every time I run it with a pressure advance of 1 I get about a net negative 1mm of extruder movement. With no pressure advance I get a positive movement.
The results are repeatable and slowing down the acceleration or jerk does not seem to change them.
both extruders behave exactly the same.
testing 2.gcode

deckingman

@jim546 A pressure advance setting of 1 is an extremely high value so I'm not surprised that you see reverse extruder action. What happens is that, at the start of a move, PA will advance the extruder beyond the amount that it would normally move, but at the end of a move, the extruder will be retarded by the same amount. So the overall amount of extrusion for any given move (or series of small moves) will be unchanged - you just get more at the start but less at the end. If you have a great deal of advance at the start, then you need a great deal of retard at the end and this can mean that the extruder has to go into reverse otherwise you will get overall too much extrusion. I have to use 0.5 PA with one of my mixing hot ends, and I see the same reverse extruder action. It's normal. If you use a lower value PA, then you won't get the reverse action but print quality will suffer if you genuinely need to use a high value.

Phaedrux

@jim546 said in Pressure advance causing reverse extrusion:

the pressure advance is set to 0.72

How did you determine that you need such a high PA value?

jim546

@deckingman The total / net extruder movement is negative 1mm of filament when printing testing 2.gcode.
A negative movement is expected when the printer slows down but it's also expected that when it speeds up there would be a positive extruder movement and the net movement of filament would be the same as without pressure advance. This is NOT what is happening.
The extruder is moving back as expected but is not moving back forwards when the extruder movement speeds up. This results in no filament being extruded or even net reverse extuder movement in this case.
The print works normally and has net positive filament feed when not using pressure advance.
The pressure advance of 1 was just random high value for testing to recreate the problem.
@Phaedrux I used the method here : (https://forum.duet3d.com/topic/6698/pressure-advance-calibration) to determine that .72 was the best pressure advance value.

jim546

Is it possible that something is wrong with my board? wiring? configuration?
is this problem repeatable for anyone else?

Phaedrux

Well there have been some changes to extrusion behavior in 3.4 and there are some other reports of pressure advance not behaving as expected. Can you roll back to 3.3 and see if you experience the same effects?

jim546

@phaedrux Just tested with firmware 3.3 with no change in how it prints this file.
Thanks for the suggestion
I think I'm going to try a full board reset to see if that fixes anything. If that doesn't work maybe I'll have to buy a new board.

jim546

It seems that increasing (yes, increasing) the acceleration and jerk values reduces the problem for some reason.

Phaedrux

What did you change?

jim546

@phaedrux i had updated to firmware 3.4.1 and left my jerk at 120 and acceleration at 500
After having trouble removing support with from prints with pressure advance turned off I turned it on and the seams the print were extremely wide.
After that print took 13 hours I decided to set both jerk and acceleration to 1000 and the seam issue seemed to mostly disappear.
Unfortunately, It didn't seem to have any other effect on the problem.
I tried to further increase jerk and acceleration with no change in the print.
This image has one print with pressure advance off and the other one with it enabled. Both the exact same gcode.