I had an odd failure case today. I've been trying to print a successful model with my newly installed MMU2S on a MK3S+ - specifically this one which seems to be a popular first print.
https://www.printables.com/model/2364-sheep-multi-material
Before printing I've re-calibrated the Z axis, did a first layer + belt tests. The printer (when it gets material to the nozzle at the right rate) is printing as well as it ever did. I re-checked and cleaned all the bondtech gears in the MMU2 with a pin, the idler screws are fairly loose as I had a lot of grinding initially. Both filaments load correctly, at least via the printer menu. I've also checked the filament IR sensor and the FINDA, and they seem fine. I'm using fresh filament as I found old PLA was too brittle and would often snap.
I've had two main issues during the print.
Using the gcode supplied, at the first filament swap the print fails. The nozzle pauses over the purge block, there's a lot of back/forth with the MMU and eventually it just dumps a huge blob of filament in-place. It's not clear what's causing this aside from the obvious - the extruder is activated and the nozzle isn't moving. I've tried this twice with the same failure and I've also seen it happen on print start, before the purge line begins.
Using my own gcode from Prusa Slicer from the 3mf, the print got probably halfway. Plenty of filament swaps, so I thought we were onto a winner. However, at some point either an unload or a load failed and rather than halting the print, it just continued... The error message that the MMU2 filament load failed was actually displayed on the LCD while the printer was going. At least part of the problem was that the nozzle was cold (I don't know why it cooled at all, unless it was an E-stop because the load wasn't fixed in time). Not sure if this was an issue with Octoprint still sending commands or something else, but I would expect the printer to refuse gcode until the error on the MMU is cleared. I'll try again direct from the SD card.
Are there any common problems which would cause either of these? Given the successful reviews, I would expect the GCode to work out of the box (indeed the first bit of the model is fine). I re-checked the IR sensor and it reliably triggers with an the 2.5mm hex key. It's pretty stiff forcing the gears apart, but it's the same in Prusa's debugging video.
Even more odd is that during the second print, I was using an old spool which snapped quite early on. I was able to swap in a fresh spool, and the mid-print reload worked without any real problems. So the system clearly can respond and recover from an error state, but maybe I hit some other bug.
I am using a piece of Capricorn tubing to replace the MMU2 -> Extruder section, but I used the provided one as a reference so I don't think that should be an issue (also if the printer is using the sensors to detect extrusion length then it shouldn't matter anyway). Also when I generated the gcode myself, most of the swaps completed fine. Also given how long the print got, it seems like drag isn't likely to be an issue.
