Kerf: why your laser-cut part is always a bit too small
A cutter removes material. If the tool runs down the middle of your line, the part finishes one tool width undersize and the hole finishes one tool width oversize. Here is the arithmetic.
The beam has a width
A laser beam is not a mathematical line. It is typically 0.1mm to 0.3mm across on a diode or CO2 machine, and a router bit is whatever the bit says, 3mm and 6mm being common. Whatever the number, the machine removes that much material as it travels.
If the toolpath is exactly your outline, half the beam is inside the part and half is outside. So the part comes out smaller than drawn, by half a beam on each side: one full beam width across.
| Tool | Width | Part finishes | Hole finishes |
|---|---|---|---|
| Diode laser | 0.10mm | 0.10mm undersize | 0.10mm oversize |
| CO2 laser | 0.20mm | 0.20mm undersize | 0.20mm oversize |
| 3mm router bit | 3.00mm | 3.00mm undersize | 3.00mm oversize |
| 6mm router bit | 6.00mm | 6.00mm undersize | 6.00mm oversize |
Holes go the other way
This is the part people get wrong. For an outside cut, the tool should run outside the line. For a hole, it should run inside the line, or the hole opens up by a beam width and the peg you meant to fit into it rattles.
That means a converter cannot just offset everything outward. It has to know which contours are outsides and which are holes, and offset them in opposite directions. Contour nesting decides that: a contour inside another contour is a hole.
Finger joints are where it shows
A box with finger joints has the error twice per joint, once on each mating face. At 0.2mm kerf, a five-finger joint is 1mm of slop, and the box falls apart in your hands. Compensating the kerf turns a rattling box into one you have to tap together, which is what you wanted.
Measure your own
Cut a 20mm square at your usual settings and measure it. The difference is your kerf at that power and speed on that material. It changes with all three, so measure again when any of them change.