1LC - extruder randomly stopping/reversing - ≤3.5.0-rc1

[Exerqtor]

@dc42 said in 1LC - extruder randomly stopping/reversing - ≤3.5.0-rc1:

@Exerqtor I reproduced the issue on the bench and fixed it. Please try this tool board firmware.

Duet3Firmware_TOOL1LC.bin

Awesome! I'm warming up the bed now before i run a test with the new binaries 🤟

[dc42]

@Exerqtor FYI the latest binaries of everything are now at https://www.dropbox.com/scl/fo/ljnewqssbl9bdo9vex37f/h?rlkey=m11nef0mvuc9pc8wri7rabf1b&dl=0.

[Exerqtor]

@dc42 Ok, I just installed both the Mini (even though it's the MMS version) & 1LC from that drive now and am about to fire up the test, I'll update this post once that's done.

---

It finished without any issues! High five on that fix, it's been annoying me for quite some time now 😂

[dc42]

@Exerqtor thanks, I'm glad to get to the bottom of this. The problem was caused when the GCode called for a very short move that would take less than 2 microseconds to execute. When the move time was truncated to integer to send over CAN, the move ended up as taking zero time yet having nonzero extrusion. This resulted in a floating point divide-by-zero error. I changed the floating point library on the 1LC to a faster but less IEEE-conformant one between 3.4beta4 and 3.4rc1, and I think this caused a change in behaviour after the divide by zero occurred.

Now that we have solved this one, I'd like to revisit the issue with WiFi connection causing crashes. I note that you are doing a lot of things with an LED string connected to the Duet 3 Mini dedicated LED output. Please try changing enough of that LED macro code so that M150 commands to that output don't happen; then try running with DWC connected over WiFi. I am wondering whether the DMA to that port when you use M150 could be interfering with the WiFi DMA.

[droftarts]

@Exerqtor bug fixed, and I still haven't got my printer up and running to test! I've been rather busy on other things, sorry.

Ian

[jay_s_uk]

@droftarts I'll let you "draw" this one to a close

[jens55]

@dc42, could this bug that you just squashed also be causing my huge number of hiccups as well as the bogus Vin min/max numbers?
I noticed that a few of the M122 Bxxx reports posted in this thread also have large number of hiccups (although not as large as the ones I am seeing) as well as the bad voltage readings.
Please advise on what files I should upgrade to implement this fix so I can try out if it maybe fixes my issues.

[droftarts]

@jens55 see @dc42’s message above for link to latest binaries.

Ian

[jens55]

@droftarts
Do I just upgrade the .bin files? What are the .map files for? Do I just upgrade MB6HC.bin and Tool1LC.bin (I only have the 6HC and the 1LC boards) or do I need to upgrade other stuff?

With the .bin files uploaded, this is what I see for the version information:

Nothing seems to have changed on the 1LC firmware and only the MB6HC firmware changed. Is it ok to have three different versions of firmware (between the 6HC, 1LC and DWC)?

[dc42]

@jens55 yes just use the MB6HC and TOOL1LC .bin files. When you have installed them you can retrieve the build dates using M115. Ignore the .map files, they are for me in case there is a problem.

[jens55]

@dc42
Alas, that fix did nothing to my hiccup problem