Stuttering axes

Hi there,

my prusa 2 developed a problem i cannot get my head around:
-after about 1 minute of printing all axes start stuttering, as if planner or stepper are stuck in some routine that is sucking too much cpu power. if i pause the machine and home the axes they move smoothly, but seem to disrespect the max_feedrate (they move waaaay to fast) but home. if i unpause after that the stuttering continues. no gcode-resends (because of crc or something else) are happening. this happends always at the same time, no matter what i'm printing.

at first i thought of this as a problem with marlin, but after configuring repetier and experiencing the same problem (now after about 3 minutes) i came to the conclusion that it cannot be the firmware.

this is what i did so far:
-checked 12v and 5v with alternative power supplies
-replaced sparfun quadstep (btw: horrible) with 4x A4988
-replaced mosfet bank with a completely new one
-checked and temp. replaced 12v power cabling
-replaced arduino mega2560
-generated new gcode with different slicer (slic3r vs cura)
-sent gcode with pronterface instead of octoprint
-temp. disabled endstops

nothing changed. if i modify the gcode to only move 1 or 2 axes, the error does not appear. when moving 3 axes the problem appears. i already swapped pins of most components (endstops, steppers) involved. this all sounds very much like a firmware error, but why is it also happening with repetier which i used a completely different and more conservative settings for?

I'm out of debugging ideas, Help

Try to print via SD-card.
If that works, it´s your PC who´s slowing data transfer.
Disable screensaver, internet connection and any other background prog. that may block the PC.
-Olaf

As this is happening with pronterface (on a pc) and octoprint (on an raspberry pi) i consider that to be already ruled out. Plus 250000 baudrate.

This would also not explain why the stuttering continues instantly after unpause (which should allow marlin to fill it's command buffer) or the homing with a too high speed ignoring max_feed.

edit: soldered sdcard to the mega, no change.

Edited 1 time(s). Last edit at 08/05/2015 08:21AM by pette.

are you actively cooling your electronics?

yes.

overheating electronics also cannot explain all of the symptoms

update, day 20 of debugging:

Installing an old marlin version (0.98) i still head laying around in my svn proved that the hardware is working.

Managed to build a test-gcode that reproduces the problem after seconds. Found the cause of the massive homing speeds, and after adding tons of debug code to the planner the stuttering is gone, still have to find the exact cause.

How much microstepping are you using and do you have a display?

I've read somewhere that with too much going on (full graphic display, 1/32 Microstepping) deltas already get into trouble. So maybe you also have too much going on?

Cause: Ground cables too long which resulted in too high resistance and 1-2 volt difference between ground connections, some logic levels to climb above 2 volts, causing stepper driver havoc. Rewired ground to a star config (instead of bus), everything works again. Different behavior per firmware was caused by the new late-enable option.

Thanks for sharing!
The errors you had, indicated more than one problem. That's always hard to find...
-Olaf