Layer shift
A layer that does not sit square on the one below it. Before anything mechanical, check the first section — on a machine with nozzles this large, the most common “layer shift” is the nozzle dragging through the previous layer, and no belt is involved.
First: is it a shift at all?
Section titled “First: is it a shift at all?”Setting the layer height too high for the nozzle makes the nozzle clip the layer below. It reads as a shift and it is not one.
The shipped process profiles settle it. Layer height sits at roughly half the nozzle diameter, line width a little above it:
| Nozzle | Layer height | Line width |
|---|---|---|
| 1.0 mm | 0.6 mm | 1.2 mm |
| 1.8 mm | 0.9 mm | 1.9 mm |
| 3.0 mm | 1.3 mm | 3.2 mm |
| 5.0 mm | 2.5 mm | 5.5 mm |
| 8.0 mm | 4.0 mm | 9.5 mm |
Pick the process profile that matches your nozzle and these follow automatically. If you have overridden them, that is the first thing to put back.
A clean horizontal offset
Section titled “A clean horizontal offset”Layers offset cleanly sideways, with the geometry otherwise intact, is motion. Two pages own it: the extruder motor skipping if the offset comes with a clicking sound, and belts and pulleys if it does not.
The same offset, at the same height, every print
Section titled “The same offset, at the same height, every print”A consistent offset — say 1 mm — that appears at the same layer height on every print, and not on a second machine running the same G-code, is friction on that axis: the belt or the rail.
- Turn the printer off and move the extruder by hand along Y. It should be smooth, with only the slight even drag of the motor. Any roughness or a sticking point is the fault
- Inspect the white timing belts. A casing worn through shows fine steel wires at the surface — replace the belt
- Lubricate the linear rails — see Lubricating linear rails. Not WD-40
- Until you can, reducing travel acceleration — non-print moves only — often lets prints finish cleanly. It does not change print time
A random layer, on one print only
Section titled “A random layer, on one print only”A shift at a random height that does not repeat is usually the nozzle hitting a bead it left in the wrong place on the layer before — wet material drooling out of the nozzle during travel. Wet pellets flow softer at the same temperature.
Look at the print for bubbles or a rough surface, the other signs of wet material, and dry it: Starting the dryer.
Shifts that line up with the vacuum
Section titled “Shifts that line up with the vacuum”Shifts at random moments that coincide with the vacuum pump starting or stopping are electrical. The pump’s inrush drops the supply for a moment and motor control stumbles.
Put the printer and the vacuum on separate circuits — the power box has two mains inputs for exactly this — and never share a power strip or extension reel between them. See Power supply 24V.
The print stops with stepper too far in past
Section titled “The print stops with stepper too far in past”Not a shift but often mistaken for one. G-code made of thousands of tiny segments — curves exported from CAD rather than sliced — asks the control board for moves faster than it can consume them, and Klipper gives up. The fix is upstream: raise the curve-to-polyline tolerance where the G-code is generated. See Klipper MCU communication lost.
Related notes
Section titled “Related notes”- Print quality — which page
- Extruder motor skipping
- Belt and pulley mechanical issues
- Nozzle set — the layer-height rule, from the component side
- Starting the dryer