Bed Self-Calibration
Support ▸ Bed Self-Calibration · ◆ Black · fires the laser
What it does
It measures your machine's real positional accuracy across the whole bed, and builds a picture of where it is off and by how much.
ALPS engraves a grid of marks, you tell it where those marks actually ended up, and it fits a distortion model to the difference.
⚠ How this differs from Bed Camera Calibration
They sound alike and answer opposite questions.
| tool | the question it answers |
|---|---|
| Bed Camera Calibration | where is the camera looking, relative to where the laser marks? |
| Bed Self-Calibration | where does the laser actually mark, relative to where it was told to? |
⭐ You can have a perfectly calibrated camera on a machine that marks half a millimetre off at the far corner. The camera calibration would not notice, because it photographs marks the laser made and takes their position as given. This is the tool that questions that assumption.
Step by step
- Generate — choose rows and cols for the grid. More marks measure more of the bed and take longer.
- The grid is placed as a normal job. You send and run it the usual way — the wizard does not fire the laser behind your back.
- Measure — record where each mark actually landed.
- ALPS fits the distortion field and reports the RMS residual and the worst points.
Reading the result
RMS residual is the headline: how far off the machine is on average, once the model has absorbed what it can.
⭐⭐ Read the residual as what the model could NOT explain, and treat a suspiciously good number with more suspicion than a bad one. A very low residual paired with an implausible fitted distortion usually means the measurements were matched to the wrong marks, not that the machine is excellent. If the fit claims a distortion the machine has no physical way to produce, the correspondence is wrong — go back and check which mark is which.
The worst points matter as much as the average. A machine that is excellent everywhere except one corner has a mechanical problem in that corner, and an average hides it.
⛔ Measure carefully or do not measure at all
The entire output is only as good as your measurements of the plate. This is a caliper job, not an eyeball job.
- Measure from a consistent reference on every mark — the same edge or centre each time.
- ⚠ Keep track of which mark is which. A transposed pair produces a confident fit describing a distortion your machine does not have, and nothing in the numbers will look wrong.
- Let the plate reach room temperature before measuring. Metal that came off a laser is not the size it will be in ten minutes.
What to do with the answer
Mostly: know it. A number for your machine's real accuracy is worth having before you quote a tolerance to a customer, and it is the honest basis for saying "we hold ±X" instead of guessing.
⚠ It is a measurement, not a repair. Distortion that shows up as a consistent scale error, a rotation, or one bad corner is usually mechanical — belts, rails, a bed that is not square — and that is where the fix belongs.
Related
- Bed Camera Calibration (
bed-camera-calibration.md) — the other calibration, answering the other question. Run that one first if you have a camera; far more jobs depend on it. - Material Test (
material-test.md) — accuracy is not the same as a good-looking mark.