The test grid.

Settings are inherited; work is local. Borrow a row, then make it yours on scrap."

Where do laser settings come from? The honest answer has two parts, and the first-cut guide carries both. First: the makers publish parameter rows for real machines — the app's table stores starting rows for Creality's Falcon2 (12–40 W) and the Falcon series plus A1, and xTool's D1 at 5 W and 10 W, each row read from that maker's own parameter page and saved with a checksum of the page's text, so the provenance travels with the number. Second: every one of those rows stops being exact the moment it lands on your bench — a different lens state, a different altitude, a different day, a different sheet from a different pallet on the same pile. The row is a starting point for a test grid.

The grid, sized honestly

A test grid is a small ladder cut on scrap of the material you're actually cutting — the same thickness, the same brand if the source matters: a range of power against a range of speeds, cut on one sheet, read by eye and finger (did it cut through? did it scorch? did the edge collapse into a melt?). The energy rule from the never-cut post organizes the ladder: each candidate row is an E value — watts × power fraction × passes ÷ speed — and the grid says which rung of your machine's reality cuts clean. Commit the row that reads best, and write it as yours.

The two nuances the table can't carry

Two things change between the borrowed row and your own, and both deserve the grid more than arithmetic. A second pass concentrates heat differently from one slow pass of the same total energy — two 0.8 J/mm passes behave unlike one 1.6 J/mm pass in material the beam touches twice while still glowing. And the material's own lot matters: the guide's plywood note runs through this site's whole bench — the nominal size is a neighbor of the true size, and the scrap tells you.

The rows the app refuses to sell

The honesty rules stay the guide's: when a maker stops publishing numbers (xTool's newer machines carry their settings inside their own software, with nothing on a page to cite), the table leaves them out rather than inventing rows. And cross-type conversion stays refused — a CO₂ row translated to a diode would lie through clear acrylic, which passes diode light where the CO₂ beam cuts. The grid on scrap of the real material answers where the borrowed table can't.

Cut the ladder, read it by eye and finger, and write the row down as yours. — the bench crew