Z-Offset Keeps Changing? Your Bed Probe Is the Usual Suspect

Last week your first layer was dialed in. This week the nozzle drags on one side of the plate and leaves gaps on the other — same profile, same spool, same machine. You nudge the Z-offset, it holds for two prints, then it walks off again. Nothing in your slicer broke. Your bed probe did.

Your Z-Offset Is a Relationship, Not a Number

People treat Z-offset as a setting you tune once. It's really the distance between two things: where the probe triggers and where the nozzle tip actually sits. Move either one and every first layer after it moves too. On a carbon fiber machine you move both more often than you think.

Three Things That Move the Offset

Plate swaps. Inductive probes read the steel sheet underneath, not the PEI you print on. Swap to a different magnetic plate and the stack height changes by 0.05–0.2 mm — that is the entire first layer. Eddy-current probes read the sheet surface instead, which is better, but a new plate is still a new number. Re-check the offset with every plate change, and label your plates so you know which ones run identically.

Nozzle wear. Hardened steel wears slower than brass, not never. As the tip erodes, the nozzle sits shorter relative to the probe, so the machine believes it is closer to the bed than it is. That reads as Z-offset drift with no setting changed. Log nozzle hours the way you log spools.

Heat and dust. A probe measured cold was not measured in your print conditions. Aluminum and steel grow, and a 50–60°C chamber is not the machine you probed at room temperature. Carbon fiber dust is the other one: grit in the pin bore of a spring-pin probe makes trigger height wander. Wipe the probe body and pin with a dry lint-free cloth when you clean the machine.

The Ten-Minute Drift Check

Re-run the offset after every plate swap, nozzle change, or any print that lifted on its own. Probe the same three points — center, front-left, rear-right — and write the numbers down. More than 0.05 mm right after a plate swap is the plate, so you re-shim it instead of compensating in the slicer. Drift of 0.1 mm inside one session with no hardware change points at a loose probe mount or a bracket that has taken too much vibration. Check that before you touch a single slicer value. If everything is tight and the number still walks, print a first-layer patch and let the part tell you what's happening.

What We Use

We run the BIQU Cryogrip Pro build plate because the sheet stack is consistent and it peels off clean — a plate swap doesn't lob a random 0.1 mm into your offset. And we keep TINMORRY PA6-CF loaded because diameter holds spool to spool, so when the first layer moves you know it is the machine and not the filament.

Affiliate disclosure: some links are affiliate links — you pay the same price, we earn a small commission that keeps our testing free.