Skewed print from loose pulley

Over the weekend, I was tracking down an oddity. The delta had been previous printing dead straight and square but suddenly developed a skewed print for the lower 1.5cm layers. It would continue after that point straight up to completion without any further disruptions. The skew consistently shifted -3mm on "X" and +1mm on "Y". I first thought it had to do with Marlin's bed levelling so I ran auto calibrates, remeasured everything and used the latest Marlin and repeated the calibrations. The skewing persisted. It appeared it might be something to do with a levelling fade height since the skewing always stopped at the same height.

I rechecked the belts and tightened things up a bit - they weren't loose to begin with but it was worth a shot. The skewing remained but was now stretched up to about 3.5cm, with the same X/Y displacement from centre.

After another round of checking everything, I discovered one of the motor pulleys was loose. It wasn't exactly sliding everywhere sort of loose, but was loose nevertheless. With that tightened up, everything is now printing just fine once again.

Can someone explain why this skewing might occur in such a consistent way? I had done over 20 test prints, countless auto calibrations, moved the effector around in all directions, and the skewing area was isolated to such a small range.

Errors due to a loose pulley or belt will really depend on where the carriage reverses its up/down direction. During the reversal, the carriage will remain stationary (or even get pushed around by the other motors?) whilst the motor moves to take up the slack. How this manifests itself will probably depend on what object you're printing and how your slicer generates G-code.

If you're printing a thin-walled vertical cylinder in spiral vase mode on a perfectly level bed, then I'd expect errors to be the same on every layer. This assumes that the nozzle is always supposed to travel in the same circle in the same direction, so that the movements of the carriages are simple, with only one reversal of direction for each carriage per layer. If you're printing a cylinder that has multiple inner and outer walls, then you might get consistent but opposite errors on the inner and outer walls if they're printed in opposite directions. If you're printing in layers, then each layer change will involve a reversal for one or more of the carriages. If your object is a conic solid, then there will be more or less amounts of infill at various layers, which may or may not involve changes to the motion of the nozzle, as less infill is needed.

Your slicer will try to optimise things. e.g if you have several parallel traces, it will do them in opposite directions. In which case the errors would be at both ends.

On top of all of this, the auto-levelling will rotate the object by a few degrees in 3 dimensions, so that what the slicer created as a horizontal motion will also involve motion in the Z axis, which in turn will change the movements of all 3 carriages.

Thanks for the detailed explanation. The problem was driving me crazy as it was seemingly a software issue. For alignment, or to show the skewing, I was printing a 20x20x150mm square U shape, walls are four lines thick, with vertical alignment lines down each side. I'm just glad it turned out to be something simple to fix.

I had something similar with a belt that had got stretched, Object skewed but only the lower 60mm after that it was straight. It was the part of the belt that wraps around the pulleys when the effector is near the bed. Steel/PU belts, 16t pulleys, not a viable long term combination.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Did the steel in those belts actually break? I thought they were designed to take a lot more tension than the usual reinforced belts. I usually don't tighten them too much but I do make sure they are all at the same tension and printing has been just fine.

If only it was easy to get (or punch your own) steel timing belts, then we would be all set!

Yes the steel wires broke gradually and as they did the belt stretched. Yet is was the small radius pulley that did it. I'm quite certain with much larger pulleys those belts would still be working.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions