Page MenuHomeAleph Objects Inc

Watchdog timer will sometimes not reset printer, leading to thermal runaway.
Open, Needs TriagePublic

Description

Marlin sets up a watchdog timer to reset the printer in the event that the software locks up. This watchdog timer seems to be reliable on AVR based printers, but seems less reliable on the Archim.

This problem manifested itself most recently in T8587, where it caused a thermal runaway. It has also manifested itself on other printers, where in similar circumstances it simply causes a reboot (T8544, T7442).

This has previously been reported in upstream Marlin:

https://github.com/MarlinFirmware/Marlin/issues/13558

The cause of the software lockup in T8587, T8544, T7442 has been resolved in .141 (commit 7f869f3d) but further investigation still needs to be done on why the watchdog timer is failing.

Event Timeline

marcio created this task.Wed, Jul 10, 3:55 PM

Additional tickets related to freezing and thermal runaway T8614

I pushed a new FW .142 that includes a stress testing screen which basically triggers the watchdog timer repeatedly. The hope is that this will cause such printers to freeze. To use the test:

  • Click "Menu"
  • Click "About Printer"
  • Click "(c) 2019 Aleph Objects Inc"
  • Click "Stress Test"

The printer should begin moving, drawing circles on the LCD and switching nozzles. Please do not run this test unattended. If the printer locks up with "Watchdog should bark..." on the screen for longer than a few seconds, this will indicate a thermal runaway. Please note whether it said "Test 2" or "Test 3", turn off printer and set aside printer for R&D to examine.

DaniAO added a subscriber: DaniAO.Thu, Jul 11, 8:24 AM

@marcio I have a beta that I ran the test on and it cycled through the test 3 times, and every time went from stress test 1 to stress test 3 and then would reset itself and start over. I can bring this one with me this morning if you would like

@DaniAO: If the test is causing resetting rather than locking up, then it isn't reproducing the problem. So I need to investigate further. I do have a printer from @samantha that had the same problem, so if you need your printer, you do not need to bring it over. On the other hand, if you don't need it, then bring it over and it gives one more printer to investigate.

Hob added a subscriber: Hob.Thu, Jul 11, 9:22 AM

@marcio I have s beta and ran the test with .142. I let it run for a while without selecting STOP, so cycled it 4 times. It cycled through test 1, 2, and 3, including a momentary message of "Watchdog should bark" for each test without thermal runaway. If you need the unit, let me know, and I'll bring it over.

@Hob: We are looking for a printer the repeatedly and consistenly locks up/overheats when you allow the change filament screen to timeout. How many times have you seen the thermal runaway on your printer?

Hob added a comment.Thu, Jul 11, 10:46 AM

Once, with .139 yesterday. was changing filament on extruder 2. want me to try to replicate?