Calibrating Your Printer
Until you calibrate, every clearance in your models comes from a table of typical values for a typical 0.4 mm machine. Your machine is not typical: it lays a bead a little wide or a little narrow, and every hole in every part inherits that error. Two coupons fix it. Print one small plate, tell the app how twenty holes feel, and the constants your formulas read become measurements. Start from the Calibrate button on each category of the workbench dashboard, which is behind a sign-in: signed out, that link goes to sign-in first.
Before you start
Both wizards need a complete active set. With a slot empty you get Pick an active printer, material, nozzle and quality on the dashboard before calibrating. instead of the flow.
A calibration belongs to one exact combination of printer, nozzle, quality and material, because all four move the numbers. Grade the plate with it in your hand, in one sitting: from memory a week later throws the print away.
The clearance plate
Calibrate clearances offers to Find the exact fit tolerance for this set — how much gap a hole needs so a matching pin slides, grips, or press-fits. The download is not a stock file: the STL is generated for your setup, from a real parametric model rebuilt on the spot, which is why the button sits on Generating… first.
You get twenty graded holes, ten round and ten square, on a plate of 76.8 × 62.4 × 3 mm, plus three loose bars: a round pin, a square pin and a rectangular bar. The bars are nominally 9.6 mm, snapped to whole layers so they print true.
Each set of ten holes is graded on a 14.4 mm pitch in steps of 0.07 mm of diametral clearance, from +0.33 at the loosest down to −0.30 at the tightest: nine steps, 0.63 mm end to end.
Slicing it
One instruction on this screen matters more than the rest: In your slicer, set the first and last layer height to {layer} mm — the same as this set — so the holes come out true to size.
Most slicers ship with a thicker, squashed first layer. In practice that squash spreads the bottom of every hole and makes the plate measure differently from the parts you design with it. Print flat, unscaled and otherwise untouched.
Turn off any elephant-foot or hole compensation your slicer offers. On any FDM machine those corrections are the error this coupon exists to measure, and they hide it.
The four passes
| Pass | Bar you use | What it isolates |
|---|---|---|
Test the round holes |
The round pin | How a curved wall prints, faceting included |
Test the square holes |
The square pin | How a flat wall prints |
Test the rectangular bar — flat |
The rectangular bar, lying flat | The reference measurement |
Test the rectangular bar — on edge |
The same bar, turned on edge | Compared with the flat pass, which axis your printer runs large |
The last two passes are the same bar in the same hole, rotated 90°, and on any FDM machine a slack belt or an axis out of square shows up there as a bar that fits one way and not the other. Follow the order the wizard prescribes: start bottom-right and work left, then the top row right to left.
Grading a hole
For each hole you answer How does the bar fit hole {n}? with one of four verdicts: Slides freely, Friction fit, Press fit, Won't fit. Grade by hand feel, not by caliper: "goes in and stays put" is a physical fact no measurement predicts reliably.
A Won't fit cascades. Holes run loosest to tightest, so once the bar refuses one, every tighter hole is filled in for you.
What comes out
Three panels. Your clearance tolerances is the direct reading: the tightest clearance that still gave each fit. Estimated deviations breaks that down, taking the rectangular bar as the accurate reference, into how far your printer moves a wall. Your calibration constants is what your models read.
Everything on the plate is measured across a diameter and every constant is per side, so the values are halved on the way out. A 0.07 mm step is 0.035 mm per side, the real resolution of this test. Anything finer is noise, and the panel says so: values are quantised to the 0.07 mm test step — treat small or negative numbers as noise.
The three bias terms come out by subtraction across the four passes, read against Friction as the closest thing to line-to-line contact the coupon offers. The three FIT_ constants are the square holes at each of the three fits. A bias term is one-signed, because holes shrink, pins grow and curves need more room, so anything under 0.02 mm is published as zero. A fit level that never occurred is interpolated from the holes either side and flagged estimated, which is what happens to Press fit when a printer's holes all come out loose.
The printability coupon
The second wizard reads off your bridge, overhang, wall, gap and detail limits from one coupon. Its one instruction is not optional: Print with your set's normal profile and no supports, so every feature is tested as it would print in a real part. A coupon printed with supports measures your supports, not your printer.
Five banks of ten features, lettered A for the easiest through J for the hardest:
| Bank | Range, easiest to hardest | Reads off |
|---|---|---|
| Bridges | 2 mm to 20 mm span | MAX_BRIDGE_SPAN |
| Overhangs | 20° to 65° from vertical | MAX_OVERHANG_ANGLE |
| Thin walls | 1.1 mm down to 0.2 mm | MIN_WALL |
| Gaps | 0.55 mm down to 0.1 mm | MIN_GAP |
| Pins | 1.1 mm down to 0.2 mm | MIN_DETAIL |
Read the bottom of the wall and pin banks carefully: it is where this coupon gets mis-graded. The ten steps are 1.1, 1.0, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3 and 0.2 mm, and on a 0.4 mm nozzle only the last two are genuinely below one bead. A 0.4 mm wall is one bead, and 0.5 and 0.6 mm print as one slightly wide bead on any FDM machine, routinely. Expect 0.3 and 0.2 to fail; everything from 0.4 mm up is a wall your printer should be able to make.
Marking the banks
Each bank asks for one judgement about the printed plastic: the longest bridge that printed without sagging, the steepest overhang wedge that came out clean, the thinnest wall that came out solid, the smallest slot that stayed open (did not fuse), and the smallest pin that survived printing.
Per feature you answer Feature {label} — {value} {unit}: did it print well? with Prints well or Fails: fifty marks in total. Be strict. In practice a bridge that sagged and then recovered should be marked a fail, and so should a wall that is there but crumbles under a fingernail. The number you get out is what your models will treat as safe.
Reading the limits
The limit sits at the boundary between your last pass and your first fail, and when both exist the app takes their midpoint and flags the result estimated.
Bridges and overhangs get harder as the number grows, so those yield a maximum; walls, gaps and pins get harder as it shrinks, so those yield a minimum. Millimetre values round to 0.01, angles to whole degrees. From here on, a rib written as MIN_WALL * 3 is grounded in something you printed.
Re-running a calibration
Running a category again replaces the current result and keeps the old one in the history. Re-run after a change that moves the numbers: a new nozzle, a different filament type, a rebuilt hot end, a re-tensioned belt. On any FDM machine those numbers drift with wear and belt tension, so the same plate will not read the same a year from now.
When it fails
Couldn't generate the STL. Please try again. The geometry is built live in your browser and occasionally does not come back. Press it again.
The bar will not go in any hole. Check the first layer: on any FDM machine a squashed one closes the bottom of every hole on the plate. Reslice with the first and last layer at the set layer height.
The bar drops through every hole, including the tightest. Your printer runs small enough to be off the end of this plate. On any FDM machine the extrusion multiplier is the first thing to check before you calibrate against that.
The whole bottom half of the wall bank failed. Only 0.3 and 0.2 mm sit below a single bead, so a 0.6 mm wall that failed is about the machine or the slice, not about the bank. Check the print before you accept the number.
See also
- Printers and Materials: the setup a run is recorded against
- Your Active Print Setup: the constants this run publishes, and how to use them
- Test coupons and calibration: the print-side view of what a coupon can prove
- Choosing the fit: clearance, transition, interference: choosing which fit a joint needs