Hi everyone,
I'm looking for some advice because I've been troubleshooting this for several days without finding the root cause.
Printer
Voron 2.4 R2 (350mm)
FYSETC Spider V3
FYSETC SB Combo V2 (CAN)
Spider configured as USB-to-CAN bridge
CAN bitrate: 500000
Mini PC (Intel Celeron J1800, Debian 13)
Klipper flashed directly via DFU (no Katapult/CanBoot)
Problem
The printer works normally for:
G28
Quad Gantry Level
Bed Mesh Calibration
Probe Accuracy
The probe is also very repeatable.
However, during PROBE_CALIBRATE, if I press TESTZ quickly several times, Klipper eventually crashes with:
MCU 'mcu' shutdown: Timer too close
If I wait about half a second between each TESTZ command, it never crashes.
The issue only appears when many movement commands are sent rapidly.
What I've already checked
CPU load is very low
Plenty of free RAM
No swap usage
SSD is healthy
Replaced the USB cable with a high-quality data cable
Spider is connected directly to the Mini PC (no USB hub)
CAN wiring checked
TMC drivers configured for 16 microsteps
Reflashed Klipper
SB Combo CAN MCU never disconnects
Interesting observations
Earlier, Linux reported several USB disconnect/reconnect events for the Spider.
The Klipper logs show very high retransmission counts before the crash, for example:
bytes_retransmit=5276
and another crash showed:
bytes_retransmit=55512
When everything is working normally, retransmissions stay at zero.
I also occasionally see this during startup:
Got error -1 in can write: (105) No buffer space available
My question
Has anyone experienced something similar with the Spider V3 acting as a USB-to-CAN bridge?
Does this look like:
USB communication issue?
Spider hardware problem?
Firmware/configuration issue?
Something specific to USB-to-CAN bridge mode?
I'm happy to upload my full klippy.log, printer.cfg, or any other information that might help.
Thank you!