Guide · Measurement tools
Bed probes and bed mesh: what they fix and what they cannot
A bed mesh approximates the bed's shape but cannot fix mechanical faults. Klipper's 25-micron probe rule, temperature bias, and what invalidates your Z offset.
We may earn a commission if you buy through links on this page. How we make money
Picks in this topic
- VINCA DCLA‑0605 IP54 digital caliper, 6 in / 150 mmCheck price on Amazon
- Clockwise Tools DPIR‑0105 IP65 digital indicator, 0‑1 in / 25.4 mmCheck price on Amazon
- VINCA DMCA‑0105 IP64 digital outside micrometer, 0‑1 in / 0‑25 mmCheck price on Amazon
Short answer
A bed mesh makes the nozzle follow the bed's shape; it does not make the bed flat or fix loose or worn parts. A probe is only as good as its repeatability: Klipper treats a spread over 25 microns as not accurate enough. Recalibrate Z offset after any change to the hotend, probe or motion system.
Picks in this topicAll best picks
- VINCA DCLA‑0605 IP54 digital caliper, 6 in / 150 mmCheck price on Amazon
- Clockwise Tools DPIR‑0105 IP65 digital indicator, 0‑1 in / 25.4 mmCheck price on Amazon
- VINCA DMCA‑0105 IP64 digital outside micrometer, 0‑1 in / 0‑25 mmCheck price on Amazon
What a mesh does, and what it cannot
A bed mesh measures the bed's height at a grid of points, and the firmware adjusts Z as the nozzle moves. Klipper's documentation states that a bed mesh cannot compensate for mechanical and electrical issues and can only approximate the shape of the bed.[1] It also needs at least 3 probe points per axis, and the probe's Z offset must be calibrated before mesh calibration.
So a mesh is not a substitute for bed tramming or for fixing loose wheels, a bent part or a damaged sheet. Fix the mechanics first, then mesh what is left.
Probe accuracy: the 25-micron rule
A bed probe needs to give the same reading each time it touches the same spot. Klipper's probe guide says a PROBE_ACCURACY range above 25 microns (0.025 mm) means the probe lacks the accuracy for typical bed leveling.[2] A spread of one Z step or up to 5 microns is normal, and the same 25-micron limit applies to its location-bias check.
| PROBE_ACCURACY range | What Klipper says |
|---|---|
| Up to 5 microns, or one Z step | Normal |
| Above 25 microns | Not accurate enough for typical bed leveling |
Repeatability is a separate check from Z offset calibration. A repeatable probe can still have the wrong offset.
Picks that fit2 products
For a Creality Ender-3 or Ender-3 Pro, choose the Creality CR Touch kit: Creality lists those models and publishes a CR Touch firmware build for each, and the Ender-3 needs its 32-bit mainboard.
More on choosing
The BIGTREETECH Eddy Duo is for printers running Klipper only, and its maker names no printer model, so check your mount and bed first. Run PROBE_ACCURACY after fitting either. On Amazon, check that 'Ships from / Sold by' shows Amazon.com, Creality or BIGTREETECH.
Creality CR Touch auto leveling kit
BIGTREETECH Eddy Duo eddy-current bed probe (USB or CAN)
Temperature bias
Klipper says many probes show a systematic bias at different temperatures, and recommends leveling at a consistent temperature after waiting several minutes.[2] Its test is to run PROBE_ACCURACY at room temperature and again at printing temperature, without homing in between, and compare.
What invalidates your Z offset
Klipper states that changes to the motion system, hotend position or probe location invalidate the probe's Z offset calibration, which should then be redone before re-running the bed mesh.[2] It also recommends recalibrating after changing probe speed or sample count.
A nozzle change can move the tip too. E3D says you may need to adjust your Z offset after changing a Revo nozzle.[3]
Recheck Z offset after:
- a new hotend, heat break or nozzle
- moving or replacing the probe
- changes to the motion system or carriage
- changing probe speed or sample count
A mesh without a probe
Klipper can build a bed mesh without a probe, using a manual paper test at each point (METHOD=manual).[1] A probe makes it faster; it is not required.
The paper test's limits
The ellis3dp.com guide says a paper test for Z offset is fine as a rough starting point but gives only relative, not absolute, measurements.[4] Use it to get close, then fine-tune Z offset while watching the first layer.
Next
Questions
Does a bed probe fix a warped bed?
No. Klipper says a bed mesh cannot compensate for mechanical and electrical issues and can only approximate the shape of the bed. It makes the nozzle follow the surface; it does not make the bed flat.[1]
How accurate must a bed probe be?
Klipper's probe guide says a PROBE_ACCURACY range above 25 microns (0.025 mm) means the probe lacks the accuracy for typical bed leveling. A spread of one Z step or up to 5 microns is normal.[2]
Do I redo the Z offset after changing the hotend?
Yes. Klipper says changes to the motion system, hotend position or probe location invalidate the probe's Z offset, which should be redone before re-running the bed mesh. E3D also says you may need to adjust Z offset after a Revo nozzle change.[2][3]
Can I make a bed mesh without a probe?
In Klipper, yes. It can build a mesh with a manual paper test at each point (METHOD=manual).[1]
Is the paper test good enough?
As a starting point. The ellis3dp.com guide says a paper test is fine as a rough start for Z offset but gives only relative, not absolute, measurements.[4]