Random print shifting by big distances

Hi!
It's me again. Previous issues (axes crawling and comms hanging) magically disappeared but new problem shown up. My prints randomly shift by really big distances like 10-30mm so it's not a friction, loose belt, etc. Nothing helped yet. Moving ATX PSU further from electronics, rearranging wiring, motors and chips cooling. I've got toroidal core transformer for heated bed but it's optoisolated. This weird behaviour happens regardless of heated bed state.
It's unusual to me that quick reset and resume works just fine like nothing happened but print is ruined as I don't have endstops yet to automatically recover print. Observations revealed that it happens to just one axis at a time. Sometimes it's a bigger pause and blob from extruder but mostly shifts. Once Y axis almost bumped into idler (15+ cm movement). I've replaced ATmega with a new one and flashed latest (downloaded today) Gen7 branch Teacup.

Please help in troubleshooting this annoying problem, I'm running out of ideas.

Best wishes,
Mark

What type of printer do you operate, can you attach your config.h?

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Hi!
Unfortunately my printer is a custom one, WolfStrap derivative running my version of Gen7 v1.2 (just converted to EAGLE and added some connectors). Config in attachment.

Best wishes,
Mark

There's no reason to excuse for building a custom printer, actually that's what RepRap is about!

Is it possible your stepper drivers overheat? They turn off then (repeatedly) for half a second and ignore the incoming step signals.

Also, reading your description I'm unsure wether these shifts happen due to the controller hanging, the motors missing steps or the motors doing additional movements.

Can't find unusual things in your config.h.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I‛ve got 120mm fan pointed at them. Overheating can cause skipped steps not direction change or smooth movement as it happened with Y axis. It's so random that I think it may be caused by EMI from motor cables. It looks sometimes like mirrored movement. I'm running out od filament from these errors.

Best wishes,
Mark

It looks like it's doing additional move on one axis as almost everytime print resumes just fine. As wrote before, sometimes moves a bit and then resumes like nothing happened (from shifted but on other axes correct point) or just goes to idler at constant speed. I think that 10+cm or even 20+cm movement can't be caused by overheated drivers. It's hard to debug randomness. I may rearrange wiring, more separating stepper wires. To be exact, everything is connected using 2x1.5mm^2 doubly insulated (one insulation per wire and one common shell), no shielding.
Can somene point what to additionally check to get rid of this problem?

Best wishes,
Mark

Edited 1 time(s). Last edit at 06/07/2013 08:04PM by neonek911.

That's right. Overheating can lead to step losses, but not to additional movements.

One thing to check would be what the firmware thinks the position is compared to the actual position. M114 gives you a position report. If pysical and locical positions always match, it's a problem between G-code generation and G-code interpretation. If they don't match, motors move on their own somehow.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I think these errors are gone now. Rewired motors, shielded electronics in an old PSU case and grounded, moved fan a bit away from PCB and it seems that they're gone. Also upgraded board, inductor change to suggested 10uH. Why are checksumming and line numbers checking off by default?
Now I've got another problem, heating resistor burned out, but that's an easy part.

Best wishes,
Mark

Quote

Why are checksumming and line numbers checking off by default?

Because communications are usually reliable enough to make this pointless. Worse, they cost precious bandwidth.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY