This morning I came to work to a perfectly printed set of plates. So I started another instance of the exact same print, went for a swim at the municipal swimming pool nearby, came back to check on it just in case, and it had done the same thing it did on Friday.
This time, nobody was at the office (it’s Sunday) so I know nobody monkeyed with the print. Clearly the plate hasn’t moved. So those hypotheses are out.
The belts look tight - although the teeth feel a bit chewed up, particularly the table belt. But I doubt this is what caused the slip, as it seems to slip randomly in both directions at the same time.
At this point, I’m placing my bet on the print head coming so close to the edges that it hits the limit(s) when it’s unlucky, or it causes the stepper motors’ counters to overflow or something - i.e. it hits a firmware bug. The PrusaSlicer software seems very confident that I can print this close to the edges, and indeed the printer does it, but I wonder if it’s overly optimistic.
So I bunched up all the plates closer to the center of the bed and re-sliced, so the printer never prints less than ¼ inch from any one edge. Let’s see what happens with that.
Okay so I’ve readjusted the belts’ tension this morning. The X belt was a bit loose and the Y belt was a bit too tight, strangely enough. Both in the green though. So I set them to the middle of the range. Good thing I checked anyway because all the screws on the X motor mount were so loose they were on the verge of letting go completely.
And since I was at it, I cleaned the entire enclosure and re-lubed the rails and the linear bearings.
However, I don’t think the printer faulted because of any of this: I checked where the hard limit was in X and Y and found that the nozzle came right at the edge of the previous prints that failed on the left and on the top:
At this point, I’m convinced the printer got lucky twice when it printed that print right, and when it didn’t, it was the result of the carriage hitting the limits when the motor(s) overshot the extreme positions a bit. I re-sliced that print to leave some margin with what PrusaSlicer believes are the limits of the printing area and I came to a perfectly printed set of parts today. And the new set I started 2 hours ago seems fine too.
So I think the lesson here is that PrusaSlicer is a bit too optimistic with how large you can print, both in X and Y. It pays not to believe it too much.
EDIT: 6 hours later, another perfect batch. So I think my theory is confirmed.