MCU 'EBB' shutdown: Rescheduled timer in the past despite low MCU usage?

General Discussion
1

Encountering the issue when printing at fairly high speed (Gyroid infill at 300mm/s).

The MCU in question is a BTT EBB36 CAN Toolhead board, connected via CAN at 250k.
The error puzzles me, as the MCU is a fairly fast (50Mhz) STM32 and I see very low usage rates throughout, both in Mainsail and the log file. Although I might be reading the later wrong.

Log attached:
klippy(16).log (3.4 MB)

2

Your log contains multiple (serious) errors:
Resetting prediction variance 12162.662: freq=64000062 diff=507688 stddev=4911.189

I could imagine that this error in consequence leads to the rescheduled error.
I’m lacking the experience to help with this error, maybe @koconnor could chime in.

1 Like
3

Seemed to be cause by a too high step rate on the extruder motor, maybe in some conjunction with CAN communication?

The extruder has a fairly low rotation distance. I lowered microsteps and the problem seems to be gone.

4

IIRC you had a microstepping of 16. Even if you print at 300 mm/s, this should neither be causing an issue with this prediction variance nor a rescheduled error.

5

True, should be quite a bit below 100k steps per second in any case. The canbus utilisation on the hostside was in 10-20% range at those speeds, so seemingly fine?

I tried to experiment a bit with the motion capture and data logging API, which resulted in the error almost immediately. But cabus utlisation was also way higher in this case.

I also never encountered the issue on the second CAN toolhead, which uses a similarly clocked STM32F0 instead of the G0 on the EBB toolhead board.

6

A Resetting prediction variance error is a severe error - it indicates the host is unable to make accurate predictions of the micro-controller clock.

It’s unclear what the root source of your errors is.

I did notice that the canbus speed is set to 250000 - that’s rather low and I would not recommend that speed. If you’ve reduced the speed due to errors you were getting at higher/normal speeds then I would track down the root cause of those communication errors and return to 500000 or higher.

-Kevin