Ultimaker 2+ unloads filament during the middle of a print?!?! Gcode problem?

[Stringman 2]

This is a strange one, never seen this in years of using a couple of Ultimaker 2 + extended systems.     (firmware ver 2.6.2ex).    About an hour and a couple if millimeters in height into a print, I checked the progress and realized that while the head was still darting around, all the filament had unloaded from the head and the bowden tube.  It was as the result you'd get if you paused it to change filament - all the filament reversed out of the extrusion path - BUT THE PRINT WAS STILL EXECUTING.    The filament had not broken and the end was rounded like it had been pulled out while hot.   Well, huh.   Fortunately not that far into the print, so clean the plate, start it over.       ..... And it happened AGAIN.    Was printing great, about 2mm in height on the model, extruding fine.    Go away, come back in a few minutes and again, it had retracted and dumped all the filament out of the tube, but was still cheerfully pretending it was printing.

 

I'm going to try it again, but first I'm going to re-slice the model.  I mean, I don't know of anything that you can put into gcode that tells the printer to dump the filament, but  it seems to have happened at the same point in the print, maybe there was some data error somehow.   Anybody else ever seen that happen?

[Stringman 2]

Welp,  re-slicing (with all the same settings) seems to have fixed this, so I speculate that maybe data corruption in the process of writing or reading the print to/from the SD card, or a damaged SD card (it's been heavily used) maybe made the UM "think" it saw a gcode to retract/unload the filament completely.  I wouldn't have thought such a code could be a thing but that was before a quick web search turned up several results similar to this: https://www.bondtech.se/2018/05/03/load-unload-script/

[GregValiant 1,408]

So you have this bad sector on the memory card.  The operating system doesn't know it's there and Cura has no way to know it's there.

You tell Cura to save the file to the SD, or maybe save it to the hard drive and then you will copy it to the SD.  Nothing checks the SD for errors, the OpSys happily gets the file to the SD card.

But the middle of the file has a gap in it because layers 13 to 17 were written to the bad sector.  The printer is going along and jumps from layer 12 to layer 18 and in so doing went from E4017.12345 past a G92 E0 (which was missed because it was in the bad sector) and the next E-value it sees is E17.12345.  That will back up the extruder 4000mm which in anybody's book is a hell of a retraction.

 

A really good scenario (and one that has been reported) is when the end of FileA.gcode is in the bad sector and terminates there, and the beginning of FileB.gcode is in the bad sector but continues into good sectors.  In that case the printer can be printing FileA but it doesn't have an End-of-File marker so it finds FileB which doesn't have a Beginning-of-File marker and it merrily keeps on going but it's now printing the wrong gcode file.

 

3D printing.  Ya gotta love it or you'd go nuts in a heartbeat.

 

 

[Stringman 2]

Thanks for the detailed scenario!  That is consistent with what I saw (and with the fact that it was my oldest SD card, which I will now toss.)

 

[GregValiant 1,408]

Unless there is some gross malfunction within the card, re-formatting the card usually fixes the issue.

[ahoeben 2,010]

But do you really want to risk the failure of a big print over the cost of a new sd card?