Pressure advance calibration



  • Hi, I ran a calibration script (modified Marlin linear advance script) for pressure advance and I'm not really sure how to diagnose the results:
    0_1546721676318_IMG_20190105_224001_HHT.jpg

    Calibration script:
    https://pastebin.com/JE75Ya5Y

    Config.g:
    https://pastebin.com/fPz16j1q

    Using Bondtech BMG with Bowden (about 600mm length), Capricorn tube and e3d v6. Retraction distance is 3mm@45mm/s

    What am I looking for? What parameters should I play with?

    Thanks!



  • That marlin linear advance test script doesn't seem to translate well to pressure advance.

    A better way to tune in my opinion is to start with the recommended values from the wiki and make some adjustments on the fly (you can send the gcode to change it while printing) so that you can see the results. A model with a few sharp corners and some curves and some flat infill areas is a good test.

    https://duet3d.dozuki.com/Wiki/Pressure_advance
    https://duet3d.dozuki.com/Wiki/GCode#Section_M572_Set_or_report_extruder_pressure_advance

    For Bowden extruders, an S value between 0.1 and 0.25 usually gives the best print quality. Direct extruders need smaller values, for example 0.05. Mixing hot ends need larger values.

    You can also find another script here that may be more suitable: https://forum.duet3d.com/topic/6698/pressure-advance-calibration



  • @phaedrux said in Pressure advance calibration:

    That marlin linear advance test script doesn't seem to translate well to pressure advance.

    As far I know this is exactly the same and for my the Marlin script worked fine on Duet.
    My gcode https://pastebin.com/SSVgKmW2

    @Maut isn't you bed a litte bit to far from the nozzle? And I think you test much more to high settings, since the Duet pressure advance works on different range. Maybe try my g-code out.



  • I messed with pressure advance for a week and got nowhere. I tried that generator also, no luck. I noticed more change adjusting accel and jerk settings.
    I noticed when increasing the pressure advance, when the line became more consistent the speed was slowed down. I used this speed to run a test with no pressure advance and got very similar results.. So I'm still at the start of pressure advance tuning..



  • @rs4race It could be that you don't need to use any pressure compensation. There are many variables such as print speed, nozzle diameter, layer height, direct drive or Bowden etc. In any case, it's important to get acceleration and jerk tuned properly before trying pressure advance.



  • @Phaedrux the linear advance script does exactly the same thing as the pressure advance. The only difference is the layout.

    The PA script stacks the layers so you look at the front of the model and look for the gradients to go away.

    The LA script draws them side by side and you look for which line is most consistent in it's thickness.

    That is the same test.



  • @gnydick I think the main thing is that it relies a lot on bed adhesion being that's it's the first layer. Between that and the extra squish it's hard to control for getting just the effects of pressure advance over the other variables.

    The other script that stacks the layers eliminates the first layer effects but its still a bit artificial.

    As I said I like to see the results on an actual print.



  • @phaedrux True. I first got my values from printing 2 wall cubes without infill. Took the value where the corners were sharp without 'overshoot blob' and the inner wall followed the outer wall. Got values that were similar to the wiki. Last week I used this 'wall' script in the other thread and got way higher values (upto 3 times as high). Using these high values results in not sharp corners and top infill that doesn't always reach to other perimeter.
    So I have better experience with printing corners and curves and adjusting the value live and taking images with a macro lens.



  • @deckingman Do you have any resources for tuning acceleration and jerk?

    @gnydick said in Pressure advance calibration:

    @Phaedrux the linear advance script does exactly the same thing as the pressure advance. The only difference is the layout.

    The PA script stacks the layers so you look at the front of the model and look for the gradients to go away.

    The LA script draws them side by side and you look for which line is most consistent in it's thickness.

    That is the same test.

    Is this script here somewhere?





  • @phaedrux I skimmed through that post, but was hoping there was something more user friendly like the linear advance gcode generator.



  • I have a comprehensive gcode generator - a python script that applies various 'for' loops over user specified geometry (polygons).

    It should, hopefully, be useful to others. It includes 20 different routines including many for retraction, jerk and acceleration for XY and E, pressure advance etc.

    0_1547084142867_IMG_1042.jpg

    This image (of translucent PETG 0.6mm nozzle) shows an example with 2 Mcodes incremented in nested 3 x 3 test bands delineated by wider layers, within which a 3rd Mcode is also varied. Each tower can be a different shape and can be printed at different speeds.

    This example: 3 polygons - 2 squares and a pentagon, each with 2 sides having grooves to mimic the XY test cubes. Configurable from triangles to circles with and without grooves, and number of walls.

    I have been using it successfully for a few weeks and now doing a final check. I will post it here shortly when I have cleaned it up a bit for others (and me) to use.

    Now that the structure of the script is working, any combinations of any Mcode can be readily added by request, or the user.

    I guess there is an interest in including linear advance? I might do that and delay the posting to this forum.
    Question, as I don't yet use linear advance: What is the likely range of parameters for me to start with in order to minimise wild first tests?

    Other priorities? maybe wait till you see it. A day or so.



  • @garis Linear advance or pressure advance? https://duet3d.dozuki.com/Wiki/Pressure_advance


 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.