I checked everything on y axis but at the end it was my y Stepper connector that made the issue. It was not an original so I replaced it with an original and no more issues...

Previously I had ensured proper lubrication, alignment and belt-tension, to no avail. Ordering and installing new linear bearings and rods did not help either.

Placing a fan next to the Y-Stepper motor reduced the number of Y-Crashes by 90%, but still did not entirely solve it. (Besides, I did not consider requiring additional external cooling really a solution.)

Eventually, I bought and installed a new Y-Stepper motor from Prusa, and that seems to have fixed it. So far zero crashes after 48 print hours.

I think one possible cause is that some printers (that ones run into this phantom Y-axis crash detection problem) have a Y-axis belt that is a bit too tight to begin with. Under many conditions, the printer works just fine despite the belt being a bit too tight. But when there are high printer environment temperatures (during hot weather, when using an enclosure without enough ventilation, etc.), the printer Y-axis might get a tiny bit longer due to thermal expansion, or possibly the stepper motor requires more current under higher temps. If the problem is thermal expansion, this extra length when hot then creates extra tension on the Y-axis belt.  In either case, the printer crosses over the threshold to where it starts thinking the Y-axis is crashing due to excessive current consumption on the Y-axis stepper motor.

It's very easy to try this solution on the MK3:

1. Check your Y-axis belt tension value: https://help.prusa3d.com/article/crash-detection_2100#check-the-tension-of-your-belts

     a. Menu: Support > Belt Status

     b. If the Y-axis belt tension value is near or under 240, your belt is too tight.  Move on to step 2.

     c. If the Y-axis belt tension value is good (~275), this solution will likely not solve your phantom Y-axis crashing issue.  Try other solutions.

2. If your belt is too tight, lower the tension:

     a. I accomplished this by backing out the two screws holding the front Y-axis belt idler, to create a small (approximately 1mm) gap between the belt idler and the frame.  See Prusa photo below.  The gap opens between the blue arrows in the photo.

     b. Find the two screws labeled with the purple arrow in the photo.  Back out both screws the same amount, a few turns to start with.  This was easy to do when using a ball-end hex key.  Back out the top screw from above and the bottom screw from below the screen, no need to remove the screen.

3. Check Y-axis belt tension:

     a. Menu: Calibration > Belt self test

     b. Check that the Y-axis belt tension value has increased away from 240 and is ideally somewhere around 275 (as this value increases, the belt tension is decreasing).

     c. If the value is increasing but not yet near the ideal 275, keep backing out both screws and run the self test to recheck the tension until the value is good.

4. Confirm that this solution worked:

     a. Print a model that previously failed for you due to phantom Y-axis crashes under similar temperature conditions. If it prints okay, then reducing your Y-axis belt tension likely solved your problem.

     b. If it worked for you (or you had high belt tension, but lowering the tension did not work), please reply to this post so others know what works to solve this frustrating issue!  Good luck!

Prusa photo: Back out the two screws (purple arrow) a little to increase the gap (between the blue arrows), between the idler (orange component) and the frame.

Posted by: @rob-latour

I'm having this problem too - using PTEG for a cylindrical model.   I tried a variety of the suggestions above, but none worked for me.  However, I then put my gcode thru this OctoPrint addon  https://plugins.octoprint.org/plugins/arc_welder/ and the issue seems to have gone away for the print I am printing now (however, the first layer looks rough).  In any case, in my case, I think it may be slicer related.  I am currently using PRUSA Slicer 2.7.4+ Windows.

That's a nice workaround.

Slicer does not fix nor introduce the issue, since the issue is in the hardware / firmware. However, you can definitely slice the model in a way that reduces or eliminates the conditions that lead to the crash - all you need to do is avoid a rapid acceleration coming out of a 180-degree U-turn on Y-Axis in your path.

When doing circle-like perimeters, the stock slicer likes to abruptly change directions. Say, you are doing an outer perimeter clockwise, reach the seam at full print speed, then tell the printer to jump to the adjacent inner perimeter and do it anti-clockwise, also at full speed. The moment the bed changes the direction and has to decelerate / accelerate causes the Y-Axis stepper to draw a lot of current (to overcome the inertia from the bed). This current exceeds a reporting threshold and gets reported as a possible crash. And the warmer the motor is - the more current it draws - the closer it gets to the reporting threshold in these U-turns.

The plugin seems to be reducing the 180-degree u-turns with arcs. This reduces the jitter (bed jumping back and forth) and, likely, also reduces the load on the motor.

Posted by: @steve457

I just disabled the crash protection in the printer settings, and haven't had any issues since doing that. Ideally, I would like to have crash protection enabled, but it doesn't seem to work correctly.

Two years later, and it still happens to me. Specifically,when doing long-ish prints with PC CF in a temperature controlled enclosure (set to <50°C, actual temperature around 45°C), usually after ~5 hours of printing (while printing the Mk.4/MMU 3 upgrade parts...).

Thanks for the hint with disabling crash protection. I also dialled down the enclosure temp to <40° C, hoping either or both measures will let me finish printing those upgrade parts. And yes - I am already prepared to improve cooling of the Trinamics for the Mk.4, see Walter's printable at https://www.printables.com/model/580484-cover-for-adding-an-80-mm-noctua-fan-to-the-mk4-el .

Like a lot of you my model that was causing issues was mostly cylindrical. I don't use an enclosure, and my Y-axis stepper and stepper drivers weren't getting particularly warm. In the end the problem seems to be my custom slicing profile interacting poorly with the crash detection logic of the printer.

My profile uses the max volumetric speed feature of PrusaSlicer ( https://help.prusa3d.com/article/max-volumetric-speed_127176 ). I've set a max volumetric flow rate of 10mm3/s and set normal perimeter/infill/etc speeds to "0" which lets PrusaSlicer set speeds automatically. Importantly all acceleration and jerk settings are the stock defaults, which should cap any sudden Y movements to acceptable values. I've used this profile without issue for months with really good results, a nice balance of speed and quality.

The problem seems to be that this slicing profile caused the Y-axis to make some very rapid start/stop motions on the particular model I was printing. I have no way of knowing if these moves were actually exceeding the slicer acceleration/jerk settings but they sure sounded terrible, like grinding! I ended up doing a full teardown and repacking of the bearings thinking something had gotten damaged 😭. Once I was sure the Y-axis was butter-smooth, I realized the slicer could be involved I resliced with stock settings and  the print finished with no Y-axis crashes.

Summary: it'd be really useful to know what slicing profiles people who are experiencing this issue are using. If you're using something custom, my recommendation is to try your model with stock settings, even if it's painfully slow, just as a comparison point. And more than anything I hope my pain can be a reminder to consider that slicer settings could be affecting things!