Evenly spaced horizontal bands running up the whole part, light and dark or slightly proud and slightly recessed, repeating at a constant interval regardless of the model's shape.
The constancy is the diagnostic. A defect that appears at the same interval from the plate to the top of the print is being driven by something periodic, and there are only three periodic things in a printer that can do it.
Start with calipers, not settings
Measure the distance between band centres over as many bands as you can span, then divide. Averaging over ten bands gives a far better figure than measuring one gap.
Now compare that number against your Z leadscrew's lead — the vertical distance the axis moves per full revolution of the screw, which is 8 mm on the common four-start screws fitted to most desktop machines and 2 mm on single-start ones.
- Spacing matches the lead: the screw or its coupling is the cause. Read on to the mechanical section.
- Spacing matches an exact number of layers: something is happening once per layer, not once per revolution. Look at flow and at layer-change behaviour.
- Spacing matches neither: thermal cycling, which has a period in time rather than in distance and therefore drifts with print speed.
That one measurement removes two thirds of the possible causes for the cost of thirty seconds.
The mechanical cause
A leadscrew that is bent, or one that is rigidly coupled to a motor whose shaft is not perfectly concentric with it, moves the gantry very slightly up and down once per revolution. The amplitude needed to be visible is small — well under a tenth of a millimetre — and this is why a screw that looks straight on the bench still produces bands.
The usual remedies, in order:
- Loosen the Z motor mounts, run the axis up and down its full travel, and retighten with the motor free to find its own alignment. This costs nothing and fixes a surprising number of machines, because it removes a screw being forced sideways by its own mounting.
- Fit a flexible or Oldham coupler in place of a rigid one, so motor eccentricity is not transmitted into the screw.
- Check that the anti-backlash nut is not binding against a screw that is slightly out of true.
- Replace a genuinely bent screw. Roll it on a flat surface to check; the wobble is visible.
The thermal cause, which people miss
A heated bed does not hold one temperature. It overshoots, cools, and reheats on a cycle whose period depends on the controller's tuning. The plate expands and contracts a little each time it does that, and on a machine where the print sits on the plate, the whole part moves relative to the nozzle.
The signature is a band spacing that does not match anything mechanical and that changes when print speed changes, because the bands are being laid down at a rate set by the clock rather than by geometry. The fix is a PID tune of the bed, not a mechanical repair — and a cheap way to test the theory is to print the same object with the bed off after the first few layers, if the material permits it.
The flow cause
If the spacing corresponds to a whole number of layers, the Z axis is probably innocent. Something is varying once per layer instead: a retraction and restart at each layer change laying a fractionally different amount, or an extruder gear whose own eccentricity produces a slow variation in delivered volume.
The tell for the extruder version is that the pattern persists when the model's cross-section changes, and that it does not scale with layer height the way a mechanical fault does.
The three patterns this gets confused with
Bands that only follow corners and fade away are ringing. A fine shimmer with no measurable spacing is salmon skin. Blotchy definition on gentle curves with no regular pitch is uneven layer lines on curves, which is a geometry effect rather than a periodic one — although a machine with mild Z banding shows it there first, because a shallow curve magnifies any vertical error.
Where opinion splits
Ask about Z banding and you will get "your leadscrews are bent" almost immediately, because that is the most photogenic cause. In practice a fair share of reported cases on well-built machines turn out to be bed temperature swing or flow variation, and the people arguing for each have usually only ever seen their own machine's version.
The measurement above is what makes the argument unnecessary. Take the spacing first, and only then decide whose theory you are testing.
Materials worth noting
Any glossy filament — silk grades especially — makes small height variations dramatically visible by reflecting light off each ridge. Matte and fibre-filled grades hide the same error almost completely. If bands appeared the day you switched to a metallic PLA, the machine did not change; its ability to hide the defect did.