gman10192
commented
6 months ago Spamming of M107 was found to be the cause of my printer locking up the extruder, X, and Y axis. The strange part was the Z and tool lock (U) axis continued to function normally while the software/UI showed the print continuing. It was as if the printer thought X and Y were moving but they physically did not.
It required a power cycle or reset of the main board to unlock X and Y axis.
Jubilee build running a Duet 3 Main Board 6HC and Duet 3 Tool Boards 1LC for each of four tools. Running Duet/RRF firmware 3.5.0 rc2.
cdheiser
bobjackman
4res0
DementevAlex
awabom
Description of the bug
When using "dynamic fan speeds" M106 and M107 commands are created very close together, often after every single G1 command when printing circular features, resulting in the printer building up a buffer of hundreds of fan commands. This results in the fan pulsing at a high speed above what was requested by the slicer if fan "kick starting" is enabled in the printer firmware. If "kick starting" is disabled then the fan runs for extended periods after it should have already turned off due to the large buffer of hundreds of fan commands. Originally I thought maybe this could be fixed in firmware. I am using Klipper and it only allows a new fan command to be executed every 100ms, so I thought removing this limitation would fix the issue. However the more I thought about it, I realized that it does not matter what firmware you are using, this will still be an issue. With no fan command execution delay, the physical hardware pin controlling the fan PWM could be turned on and off uncontrollably at whatever speed the gcode file is interpreted. This could potentially destroy fans connected to printers that have no limit on how fast the fan commands are interpreted. I have destroyed fans in the past by using a PWM frequency that was too high (I think the capacitors burn up because of AC currents going through them).
The video shows how I can replicate the issue by simply spamming M106 and M107 at the printer using the console in Klipper. Note that fan "kick starting" is enabled in the video since it exaggerates the RPM fluctuation so you can hear it.
https://github.com/prusa3d/PrusaSlicer/assets/65746045/7925f8f2-c399-45ab-a62d-3dad2d2fbb8a
Project file & How to reproduce
M106 M107 Issue.3mf.zip Open the project, slice the part and view the gcode on layer 4. In areas where dynamic fan speed is working, there is an excessive number of M106 and M107 commands.
Checklist of files included above
Version of PrusaSlicer
2.7.1
Operating system
macOS Sonoma 14.2.1
Printer model
Voron 2.4 running Klipper v0.11.0-304