I have attempted to calibrate the Linear Advance settings twice now, using the often-linked teachingtech instructions and the Marlin gcode generator.
I have been unable to generate results resembling those in the instructions. In particular, there is very little difference between the lines, and expanding the range does not change this (it should). Selecting the most even line and zooming in on that range does not improve matters: the set of lines looks the same. In fact, the lines always look remarkably similar (the 0 or bottom line being the most even, followed by the one immediately above it and the second line from the top) with only the numbering serving to distinguish one set of lines from another.
This has occurred with two different filament, two different nozzles, and most recently with an inline filament drier.
Is there another way to determine the linear advance settings? Is it possible the uneven bed is screwing this up, even with the saved mesh being loaded in the g-code?
There is definitely a change, for example between 0.2 and 0.7 it clearly gets worse. Then from 0.8 to 1.0 it gets better, and from 0.1 down to 0.0 it gets better - almost as if the proximity to the border of the print is the determining factor (which makes zero sense, of course). Expanding the range to 0-2, or limiting it to either 0.75-1.25 or 0.0-0.25, produces remarkably similar sets of lines, rather than progressively more (or less) even lines as illustrated in the instructions.
Thanks. I was just downstairs running two tests, 0.05 to 0.1 in .01 increments, and 0.075 to 0.125 in .01 increments. These indicated 0.085 as a candidate value, though I’ll try a third test with 0.005 increments.
I found that adjusting the Z offset made a big difference. I generally use -0.25 on prints with this particular filament, and had set the gcode generator to use that. When I reduced the Z offset to -0.1, the lines had less of the random thickness changes I was seeing (at no Z offset, the lines did not stick to the bed).
I decided to redo the bed calibration because the need to do a -0.10 to -0.25 Z offset every time was getting old. Used the Luban console, did the 21-point calibration I think, carefully set the Z offset using a 1mm gage bock and then feeler gauges to get the desired 0.1mm nozzle height. Did a first-level calibration print and it looked okay, a little low but serviceable. Fired up the K-value test I had generated and … nothing. The center of the bed was too low for the actual K-value lines to stick to it.
There was some mild cursing and I called it a night.
