Trying to understand LIN_ADVANCE...again :-) #9142
Comments
|
Please give us an indication of which line is first and which one is the last one and also of the print direction of the lines. I'm a little bit confused at the moment. |
|
The lines closest to the camera (e.g. bottoms of the images) were first, and all were printed left-to-right. |
|
I guess your extruder is skipping steps during the high K lines, at least that's the only explanation I have for seeing the opposite compared to what I expect from increasing K.
I recommend printing another test pattern, but this time stick to the default values in the Speed and K-factor sections. |
|
Just for information, I'm working at an improved version 2 of LIN_ADVANCE at the moment. You can follow the status (even if I don't write every progress there) in #9048. |
I doubt it. 47:9 gearing, a brand new hob gear (less than 2 weeks old, not even 1 kg of filament put through it), and 100% certainty that the hotend is clean, and a spring/idler capable of crushing the filament flat (though obviously I don't have it tensioned to do that 😄). At worst, the hob might slip on the filament, but this extruder design makes a distinct sound when it does.
For a genuine J-head, 10mm³/s is pretty low. According to the manufacturer, it's capable of about twice that flow rate; indeed I've pushed it to 200 mm/s (0.4 mm nozzle, 0.304 mm layer height) and while those speeds make my bot's other mechanical inadequacies apparent, the extruder/hotend performed as designed - of course at the time, I was not using LIN_ADVANCE. But I get your point.
OK, I'll give that a try. |
It's not a question of quality but of inertia. The stepper itself has to accelerate to execute the step signal. If it gets too much steps within a too short amount of time, it will just not follow the signal and jump to the next stable full-step position. |
|
Oh right, makes sense. |
|
This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs. |
This is Marlin 1.1.6 (commit 8dc10fb)
So...I decided to give LIN_ADVANCE another go, since I have since upgraded to a newer Marlin, a new hotend (now a genuine J-head v8), a new geared direct-feed extruder of my own design (https://www.thingiverse.com/thing:2744682). Instead of wasting a bunch of filament on random models, I went for the LIN_ADVANCE calibration test pattern (http://marlinfw.org/tools/lin_advance/k-factor.html)... and rather than ask a bunch of questions, maybe I should let some photos do the talking:
K=50 to 150, speeds of 5 mm/s and 100 mm/s (to test a 20:1 flow rate difference, similar to kinds of differences Slic3r may produce during a typical print, say an interior perimeter or sparse infill versus the thinnest bits of gap fill).
K=50 to 150, speeds of 40 mm/s and 80 mm/s (to test a 2:1 flow rate difference)
When the difference is small, the test pattern can get pretty close to perfect. Indeed, previous tests showed K=42 to be the sweet spot for the test pattern, but that setting is quite wrong when I try a real calibration print e.g. a Benchy or Ctrl-v v3.
In Marlin, I've left the E:D ratio at 0/auto.
What do I do now?
Here's the gcode output from the 40/80 test (the 5/100 test differed only by speed, all other parameters unchanged):
Diff showing my settings:
The text was updated successfully, but these errors were encountered: