Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
This is the second one of these I have used (one belonged to my school, this one is my own) and they have both suffered the same problem: layer shifts and ruined jobs. I am not sure I fixed the other one, I'll know if they use it, but mine is having similar issues. It has about 13 days of use on it, 9 days when I bought it. It was sold as having layer shift issues and a bad filament sensor. I adjusted the X belt tension and bought a filament sensor I don't seem to need, and all has been well. But now it seems jobs over 25-30mm in height will crash unless they are in vase mode. I printed something about 15mm tall without issue last night.
X is at 252, Y is at 272. I have not fooled with the Y-belt. It was recalibrated last night as I am also seeing what looks like an uneven print bed, where one corner is notable lower than the others (left front). Running the stress free first layer printable (I keep on every SD card) shows that and it's not great.
This machine makes a bit of squeaking and grumbling as it goes through its motions. Not sure what can be done about that. X is almost all the way loosened…the tension bolt is backed out almost all the way. I am not even sure how to adjust the Y-belt. Mystified that there isn't a tensioner like many belt-driven devices have.
I know there are apps to measure the sound of the belt and there is a PETG only printable (I have some PETG on the way, but its also a puzzle that something couldn't be made in PLA. These should be precise enough for that. What I would like is some kind of measurement that isn't based on a sound or a tool I can't currently make.
I currently have 3 3D printers and would like to get down to one reliable one. The Prusa should be the winner here (against an Ender 3v2 and an AnyCubic mega S) but none of them are 100% reliable or capable of delivering quality work all the time.
RE: Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
I own a MK2.5 (very simular to the MK3) and had a couple of the same issues.
Of all things you mention, I think the not even bed, would worry me most.
1. The sheet is laying on a heatbed and the heatbed on the carriage (6 mm thick aluminium). I guess you need to stand on it or hit it with a hammer to get that carriage out of shape. So search for loosened screws, nuts in the construction there.
2. The carriage lays on 3 bearings and hangs on 2 axis what should be parrallel. I measured mine (6 years old or so) and the axis where mounted about 0,1 mm more apart then the carriage. So only in the middle it was running OK, as soon it go to the end of the axis (front + back) there was tension, causing layer shifts and also caused the bearings to wear out to the sides. I adjusted all (keep measuring as you tighten the nuts, difference in tension on inside and outside gives a difference of about 0,1 to 0,2 mm). When you loosen the belt (step motor screws, one out and loosen the other) it is not hold by the stepmotor and you feel exactly if there are obstructions. (this repair (adjustment + drylin bearings) solved my layer shifting problems).
3. For the sheet to get it leveled out, I also mounted springs instead of spacers. so I can adjust the bed (and now it is within 0,1mm all around). Don't forget to do z-calibration after these actions.
4. Layer shifting in X-direction, could be the tension on the belt (because the axis are in printed blocks and not movable up-and downwards.). You could shorten the belt one or two links. Here also loosen the stepmotor is helpfull.
Clean the axis and maybe a bit of lubrication. For tension of the belts I printed an instrument from printables in PETG, but was not very usefull, the sound is most easy to check. Just use the app. And better a bit too hight then too low. (those belts don't wear out very fast.
We will do what we have always done. We will find hope in the impossible.
RE:
this is great info, thanks. I'll withhold my skepticism on the sound check, then. I have used but it seems hard to accept that the belt tension can be verified that way. I'll dig into the rest. The bed unevenness is new, it was level when the machine arrived so maybe something has happened underneath.
The bed visualizer plugin used to create this map seems to have gone away or been renamed…but you can see the unevenness.
RE: Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
I just ran it again on a different Rasp π unit (it looks like the bed visualizer is bundled win the OctoPrint distribution now?) and this is the result now.
This is the unit that will be permanently attached to the Prusa.
RE: Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
The variance you measure is 0.3 to 0.4 mm. And it is a regular angle (my machine was more like a waved bed).
The mesh leveling will compensate for that. So printing should not be affected (too much, I think).
I would check the nuts in the middle where the Y-axis are mounted to the vertical frame. Might be that there is a small difference.
We are talking about 0.4 mm, so you will not see that without calipers and a flat workbench, also turning the machine 180 degrees and remeasuring (to compensate for irregularities in your workbench). If loosening and tightening a nut, only one at the time (see my explanation on the Y-axis above), to avoid moving the setting in longitudinal direction. Work precisely, because it is easy to make the difference bigger then smaller 😉
NB: But best is to start with a X-calibration (not XYZ!), on my machine this was essential. Seems this calibration is stored in memory and the mesh leveling goes on top of that.
We will do what we have always done. We will find hope in the impossible.
RE: Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
I spent some time adjusting belts today, another area where are some differences of opinion…this video suggests different values than Prusa does. I was unable to get the adjustments to match the video but they aligned pretty well with the Prusa ones. I then ran a full calibration sequence (I tried running just a Z calibration but the nozzle plunged into the heatbed and I had to reset the machine). My first layer calibration (this one, not the one in firmware) is looking good, no unevenness on the bed, an offset of -1.5. If this passes without any issues, I'll try to actually make something. Sometimes it feels like more time goes into care and feeding/calibration and fiddling than actual productive work.
RE: Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
Sometimes it feels like more time goes into care and feeding/calibration and fiddling than actual productive work.
Join the club :-). And maybe it helps to know that:
- Making a meal takes about two times as much time then eating it.
- NASA worked decades with thousands of people to set one man on the moon (for a couple of hours).
- A musican needs about 10.000 hours of concentrated practice to play one piece well.
We will do what we have always done. We will find hope in the impossible.
RE: Wanted: definitive how to on diagnosing and fixing layer shifting on the MK35S
That explains my lack of musical success but a meal often feeds many people or can be made to last one person for many meals. 😏
It's the repetitive nature of it, the calibrate and re-calibrate that gets wearing. Prusa is not the worst for this, at least from my experience: I have one machine that requires bed leveling between jobs and that gets old very quickly. It's the lack of quantifiable metrics, like feeler gauges or any other measurement that I find puzzling. Surely a Z-offset could be set that rather than having to run a job each time.
Trying out a couple of jobs, trying to figure out the limits and tolerances.