Linear advance calibration: What "Lin Advance Version" for UM2+?

[gdog 4]

I am dealing with some issues on the infill in a PETG print.  I suspect there is over-extrusion at the end of short lines that is getting caught on the nozzle and making a mess. It only seem to happen on the short infill lines for a narrow section of the print (see pic for details).  Everything else prints beautifully, so I'm loath to adjust fan / nozzle temps / retraction /etc.  anymore.  The print looks beautiful on the outside, and only concern is that the integral strength is not what it should be.  Anyway, Linear Advance Constant k-value seems like a promising solution... so added the linear advance plugin from the Cura marketplace, but I am looking to do that "k-value test print" to figure out what settings to use.  (https://marlinfw.org/tools/lin_advance/k-factor.html)

 

So the question is this: for an Ultimaker 2+, should I select "Line Advance Version" 1.0 (says for Marlin 1.1) or 1.5 (says its for Marlin 1.9/2.0+).  I understand that Ultimaker firmware is a fork of Marlin, but nowhere could I find whether it's implementation of linear advance - which appears late to the party - is done according to the older or newer Marlin protocol. 

 

Anyone who ran this k-value test on a UM2 / UM2+, what version did you use, and has it given you the information you needed in terms of the Cura plugin to get more uniform lines with less initial under-extrusion as well as less over-extrusion at each line end?

Edited March 19, 2022 by gdog

[Smithy 1,146]

As far as I know UM printers don't support linear advance.

But I am sure @ahoeben knows more details...

[gdog 4]

4 minutes ago, Smithy said:

As far as I know UM printers don't support linear advance.

But I am sure @ahoeben knows more details...

 

 Well that is disappointing to hear.  It seems like something is very helpful.  Going back to look at what I thought were search results suggesting it was supported, it turns out they only talk about "Ultimaker Cura" and not the actual Ultimaker printers.  I guess that makes sense why it was a 3rd party plug-in.  

 

Which brings up the question: what to do to fix this problem?  I can try lowering the infill print speed, but since the longer infill lines on the other side of the model print fine, I'm not so sure.  If I had to speculate about what's going on, the issue could be that the end-of-line over-extrusion is happening everywhere but with lower fan speeds, on longer lines, when the nozzle passes over the blob, it's bonded enough to the layer below not to stick and drag away with the nozzle.  But on the short infill line, the mass of goo is too sticky and gets mopped around by the nozzle, making a mess of clumps. 

 

I suppose maybe I will try a different infill pattern before doing anything else, or maybe us the "avoid printed areas when traveling" setting.

[gdog 4]

Got to looking through Cura very carefully on this issue.   Seems like the "Retraction Extra Prime Amount" might be something related to the linear advance stuff, except for the fact that it's only applicable to the points where the filament retracts.  Which isn't at the end of an infill line that then travels over the wall to the next infill line.  But then I also saw a setting "Infill Wipe Distance", which also looks promising.  In my settings it is .0875mm, which seems like it might be on the low side is there is overextrusion at the end of these infill lines.  Anyone have thoughts on these ideas?

[gr5 2,271]

Linear advance works great for printers where acceleration is slow and the feeder is on the print head.  The UM2 is a bowden system.  The values you would need for linear advance are so huge that the stepper motor would fail.  It just can't accelerate enough to do the linear advance algorithm. It's like asking a car to do 0 to 100mph in 1 millisecond.  The stepper just can't do it.  Partly because the XY acceleration is so damn high on the UM2 (I think it defaults to 5 meters/sec (meters! not mm).

 

The accerlation is so high on X and Y because the head is so light.  Because there is no feeder there.

 

Having the fast acceleration means the print head doesn't slow down much on corners.  Which also means you don't really need linear advance as much (although it would help - if the feeder could handle the acclerations).

 

So it's a tradeoff.  One that results (for most people) in higher quality prints on the UM2 than many other printers and also faster printing than many other printers.

[gdog 4]

3 hours ago, gr5 said:

...

 

Having the fast acceleration means the print head doesn't slow down much on corners.  Which also means you don't really need linear advance as much (although it would help - if the feeder could handle the acclerations).

 

 

A key observation that would help explain why a choice of a different infill could change the amount of "blobs" as some other sources suggest... because if it is making triangles or hex, it doesn't need to stop at the end of the grid "lines" and travel to the next line, it just keep printing a complex shape. 

[ahoeben 2,016]

The Ultimaker 2 and 2+ firmware do not support linear advance. There is a very old implementation of ADVANCE in the code, but it is disabled in the official firmware releases:

https://github.com/Ultimaker/UM2.1-Firmware/blob/f6e69344c00d7f300dace730990652ba614a2105/Marlin/Configuration_adv.h#L229

You could compile your own firmware from the source and try to enable it, but I would not expect much from it.

[gdog 4]

I had a reasonably good improvement changing the infill from "line" to "gyroid", which basically adds some linear distance to each infill extrusion.  Not perfect, but acceptable.   I also reversed the orientation based on the fact that all the problem seemed to be on one side of the print.  In the reversed orientation, the issues were mostly on the opposite side of the print this time, meaning it might have something to do with the fan being over one side of the print more than the other....  and as I paid a bit more attention, is seems like maybe the "globs" weren't from end-of-line over extrusion, but in fact from a bit of layer that wasn't getting good adhesion to the layer below it and stuck to the nozzle instead, smudged around, and deposited somewhere else, then repeated again the next time the print head passed over.  If anything it seemed like some of the over-deposits were at the start of each line, possibly due to the prime-after-retract setting being too high?

[gr5 2,271]

Prime after retract - I've never set that to anything but zero.  I guess I don't get the purpose.  It implies the head was leaking at some point.  Yeah having more fans is always good.