Extruder Crashes down into print then fails
I have a problem with my Prusa Mk3, operating under the most recent firmware. I have owned it for 6 months now, and it was working like a dream up until a couple weeks ago. I have been trying to print a large print that I have successfully printed before with this same printer. However, my printer is not cooperating. The total print time is over 100 hours at normal speed and .1mm layer height. At some point, usually 89-92% complete, the extruder will, for some reason, push down into the print, melting a nozzle shaped depression into it, then the extruder will raise on the z axis to a really high height and continue to try to print way too high and just create a coiled ball of material. I have now tried different settings, speeds, and filaments and it has happened in the past six attempts to print this piece.
I can print smaller items with no issue, but longer prints will have this happen. There is no consistency as to the height this occurs at, how long into a print it happens, or even the XYZ coordinate where it occurs. Each time it has failed this way, it was at a different height and at different XY location. The piece I want to print is not complicated and does not even need supports. I have double checked my layer height and everything appears to be fine.
I have tried searching for a solution, ran a self test, recalibrated XYZ, and everything appears to be just fine. I have no idea what the issue is. ANY help would be greatly appreciated.
Re: Extruder Crashes down into print then fails
A photo is really helpful to understand the issue. Otherwise people like me will guess and may tell you it was the recent full moon that caused it. 
Re: Extruder Crashes down into print then fails
I appreciate your response. Here are a few pictures of my most recent failed attempt. The first image shows what I found. The print had gone perfectly for the first 35 or so hours. Then, as you can see in the second picture, the nozzle pressed down into the print, melting a nozzle shaped depression into it. The extruder then, having met with resistance while lowering, raised to the height that you can see in the third picture and tried to print at that height, leaving a coiled mess. The past two times, but none before that, the printer stopped due to an extruder fan error when a piece of stringed filament stopped the fan.
Re: Extruder Crashes down into print then fails
Nothing odd stands out except warpage. But it might be something weird in the slice, too. Or firmware.
Large sized prints have a tendency to warp and lift off the plate while printing. The warp can be slow, an edge lifting a tiny bit per hour, or fast where an edge pops free and jams into the nozzle. Looks like you might have gotten the latter -- but round prints are less prone -- so I'm not 100% convinced.
Here's an example of a warp in progress. Square corners are the worst.
If you remove the dead part, check the base with a good straight edge. If you see any warp whatsoever, it's partly adhesion issues. Increasing bed temp to 65c or even 70c will help. Others will tell you an enclosure can help more.
If you see any layers with overextruded bulges, like plastic has been smushed sideways, they are a precursor to a full snap off the bed.
Also - there have been a few reports of filament sensor weirdness. Wouldn't hurt during your next attempt to turn the filament sensor OFF at the LCD. Just in case.
Re: Extruder Crashes down into print then fails
Thanks again for the response. This last attempt did have the filament sensor disabled. Also, there is no warping and the base is fully adhered to the steel sheet. I've looked across the topmost printed layer and it does not appear to have any raised areas that would have caught on the nozzle. Instead, the nozzle seems to have depressed too far at one point. Could it be an intermittent PINDA error, where it can't tell the correct height? I'm at a loss. Thanks.
Re: Extruder Crashes down into print then fails
And just for clarification, the misshapen part of the print on the second picture was just the result of the extruder pushing down into the wall too far. It melted the depression into the print and displaced plastic around it. prior to the failure, there was no raised or uneven portion of the print for the nozzle to collide with.
Re: Extruder Crashes down into print then fails
Are you printing via Octopi or SD?
If Octopi, I'm wondering about bad USB cable, USB cable routing running into extruder bundle, or power supply going bad on Raspberry.
Also if Octopi and the underpowered Raspberry Pi Zero W, I would check if any new plugins are using up too much CPU.
Re: Extruder Crashes down into print then fails
good question, I'm just printing through SD. It's weird that it doesn't mess up at the same time or place each time. I've tried recreating the gcode file and it still messes up. I've used both the Prusa slicer and Cura.
Re: Extruder Crashes down into print then fails
The printer is getting a bad g-code from somewhere. The only other way I can remotely conceive of a spontaneous move happening is if an axis gets sticky, and missed step recovery re-home moves the nozzle into the printed object. That might fit with it starting to happen after the printer ages. Are the axes and bearings still free moving through their entire range?
Re: Extruder Crashes down into print then fails
Thanks for the thoughts. The axes all work great. After the failed prints, I checked the height on both Z axis sides, in case a motor was going bad on one side, and it was still completely level after the failures. The X and Y axes are without issue as well.
Re: Extruder Crashes down into print then fails
And the PINDA is only used to sense the bed for Layer 1 operations. Everything that happens above layer one is software based. The only way the nozzle would move below the current printing level is a software glitch forgetting where it is in space, or the part moving.
Since the part hasn't moved, you're left with a glitch of some sort. USB error - like a PC going into sleep mode - a bad cable as Guy has suggested; something odd about the gcode itself; but it could also be a printer hardware problem - a broken wire in one of the bundles could be flexing just right at height and causing a short - though there's nothing on the Z-Axis that is flexing - but you could manually run Z up and down a few times and watch for anything odd.
Another possible source - especially on an older printer - is the SD socket. There are known to go intermittent with use. Insert the SD card and then wiggle it around. See if you get signs the contacts aren't stable. There is a ThingyVerse part that you can install that presses on the SD card connector to help prevent lost SD connections.
The other possibility is a bad gcode file. Is it possible the card is old enough it's having read / write issues? They do go bad - flash memory is only good for so many erasures and then starts to fail: and some technologies don't use wear leveling, so 25,000 overwrites puts you into this territory. The only way to know if this is happening is to do a binary compare of the file after writing it to the SD card.