Skip to content

Kerf compensation calculator, half the kerf on every cut edge

A laser beam removes a real strip of material as it cuts, and that strip has to be added back into the drawing or the part comes out the wrong size. Pick external, internal or hole geometry, enter a catalogue or a calibrated kerf, and read the CAM edge offset and the CAD drawing delta side by side.

Open the kerf calculator A free account opens this tool. The interface is in English and nothing you enter leaves your browser.

How it works

Three steps. The geometry type decides the sign, the kerf decides the size.

  1. Choose the geometry type

    Pick external, internal or hole. The rule that decides the sign of the compensation has nothing to do with whether the geometry looks like it sits on the outside or the inside of the part. It depends only on which side of the drawn line the kept material sits on. A part's outer silhouette and an island or tab inside a slot both grow by the kerf, because the kept material is inside the line in both cases. Only a hole, where the material inside the line is the scrap, shrinks.

  2. Enter the kerf, catalogue or calibrated

    Start from the catalogue range for your laser and material: roughly 0.1 to 0.3 millimetres for a fibre laser on metal, and 0.25 to 0.5 millimetres for a CO2 laser on acrylic, wood or MDF. That range is a published typical value, not a measurement of your own machine, nozzle, gas and current power setting. When the fit has to be exact, cut a test piece with no compensation at all, measure it with a digital caliper, and enter the result as your calibrated kerf instead.

  3. Read the CAM offset and the CAD delta

    The tool shows two numbers, because CAM software and a CAD drawing ask for the value in different forms. The CAM edge offset is half the kerf, the distance the toolpath is pushed away from the drawn line on a single edge. The CAD drawing delta is the full kerf, the amount the drawn dimension itself needs to change to produce the finished size you actually want on the part.

The same arithmetic, a different kerf per process

Once the geometry rule is clear, only the kerf value itself changes by material and laser type.

Fibre on metal 0.1 to 0.3 mmCO2 on acrylic, wood, MDF 0.25 to 0.5 mmExternal growsInternal growsHoles shrinkISO 9013 taper

Built around the geometry rule, not a single material table

Most kerf mistakes are not arithmetic. They come from applying the wrong sign to the wrong edge, or trusting a catalogue figure a machine has since drifted away from.

  • External and internal are not what intuition suggests

    Many people assume that everything physically inside a part shrinks and everything on the outside grows, because they quietly equate inside with a hole. That assumption is wrong. A silhouette that fully encloses the part grows by the kerf, and so does an island or a tab sitting inside another part's slot, because in both cases the kept material sits inside the drawn line. The finger of a finger joint is the clearest example: it looks internal, sits inside the sheet, and still grows.

  • Two independent tests confirm the same kerf

    An external test, a square or circle cut with no compensation, comes out smaller than the drawn size by exactly the kerf. An internal hole test, cut the same way, comes out larger than the drawn size by exactly the kerf. Both measure the same physical kerf from opposite directions, so running both and comparing the results is a useful sanity check before you commit a calibrated value to a batch of parts.

  • A finger joint needs twice the kerf, not once

    An uncompensated slot and tab pair of equal nominal width ends up too loose, because the slot has widened by the kerf while the tab has not grown to match. The compensated tab width is the slot width plus twice the kerf, plus an optional overlap for a press fit. Skip this and a joint that should hold itself together with friction alone falls apart in your hands.

  • Taper is read from ISO 9013, not guessed

    On thick material the beam does not stay the same width through the full depth of the cut, so the top and bottom faces of the kerf are not equally wide. ISO 9013 quantifies this with a perpendicularity value that grows with thickness, from a few hundredths of a millimetre on a one millimetre sheet to roughly a tenth of a millimetre at twenty millimetres. The tool derives an indicative taper from that published relationship rather than inventing a number, and says plainly when a critical tolerance on thick stock needs a direct top and bottom measurement instead.

Frequently asked questions

What exactly is kerf, and why can't I just draw the exact size?

Kerf is the strip of material the laser actually removes as it cuts, because the beam is a focused spot with a real diameter, not a mathematical line with zero width. If you draw a part at exactly the finished size and cut without compensation, the machine removes material centred on that line, so every edge loses half a kerf and the part comes out undersized. Compensation pushes the toolpath, or the drawn dimension itself, out by the right amount so the finished part matches the size you actually need.

Why isn't the rule simply that everything inside the part shrinks?

Because the direction of compensation depends on which side of the line the kept material sits on, not on whether the geometry looks physically inside or outside the part. A hole shrinks because the material inside its boundary is thrown away. An island, a tab or the part's own outer silhouette all grow, because in every one of those cases the kept material sits inside the line. Treating every internal-looking feature as a hole is the single most common kerf mistake.

How do I calibrate kerf for my own machine instead of trusting the catalogue?

Cut a test piece with no compensation applied at all, so the drawn dimension in the CAD file is the exact nominal size. Measure the result with a digital caliper, never a ruler, because a ruler's own error is comparable to the kerf you are trying to capture. For an external test such as a square or circle, kerf equals the nominal size minus the measured size. For an internal hole test, kerf equals the measured size minus the nominal size. Running both tests on the same setup and comparing the two results is the most reliable way to confirm the figure before it goes into a batch of production parts.

Why does a finger joint need the kerf compensated on both the tab and the slot?

A slot cut at its nominal width comes out wider by the kerf, because a slot behaves like a hole. A tab of the same nominal width comes out narrower by the kerf, because a tab behaves like an island of kept material. Left uncompensated, that combination doubles the gap and the joint is loose on both sides at once. The corrected tab width is the slot width plus twice the kerf, with an optional small overlap added on top for a deliberate press fit.

Should I use a press fit, a sliding fit or a deliberate gap?

A sliding fit, zero overlap, gives a joint that goes together by hand but does not hold itself, a sensible choice for a prototype you intend to glue afterwards. A press fit adds a small positive overlap so the joint grips on its own without adhesive. A deliberate gap, a small negative overlap, suits moving assemblies or parts meant to be taken apart again. How much interference a material tolerates safely differs sharply: wood and plywood are a cellular structure that compresses locally under load, so they typically accept a larger interference than a glassy thermoplastic like acrylic, which deforms very little elastically before it cracks rather than compresses.

Does the kerf stay constant through the thickness of thick material?

Not exactly. On thin material, under about three millimetres, the difference between the top and bottom of the cut is negligible with a caliper. On thick material the beam is focused at one specific depth and widens gradually away from it, and the assist gas behaves differently at the entry and exit of the cut, so the top and bottom faces of the kerf end up slightly different widths and the wall carries a small taper rather than sitting perfectly perpendicular. If a tight tolerance on thick stock genuinely matters, measure the top and bottom edges separately instead of assuming a perfect cylinder.

Other tools

Part of the EduFacturing tool set for laser cutting and CNC.