Laser Engraver Calibration: Focus, Steps-Per-MM & Squaring the Gantry (2026)

Quick Answer: “Calibrating a laser engraver” is actually three unrelated jobs that get lumped under one word — focus height (how close the lens sits to the material), steps-per-mm accuracy (a firmware value telling the controller how far the motors actually move per command), and gantry squareness (a physical, not a software, property of the frame). Thunder Laser’s own documentation notes that manufacturing tolerance alone can shift a lens’s true focal point 2-3mm from its nominal spec, which is why a focus ramp test matters even on a brand-new machine. Each of the three problems has a distinct symptom — faint/fuzzy burns point to focus, wrong-sized designs point to steps-per-mm, and a square that cuts as a parallelogram points to a squareness problem no setting can fix — so the fastest way to calibrate is to match the symptom to the right one of the three, not run every test every time.

If a term search brought you here, you’ve probably already tried “recalibrating” in whatever sense your machine’s software uses the word, and the problem didn’t go away. That’s usually because the actual issue lives in a different one of the three categories below. This guide splits them apart so you can diagnose from the symptom first, rather than guessing which test to run.

Three different problems, three different fixes

Note that camera calibration is a real fourth category, but it only affects machines with a camera overlay add-on, and it’s thoroughly covered in its own guide linked above — the three below are relevant to every laser engraver regardless of accessories.

1. Focus calibration: the ramp test

Every laser lens converges the beam to its smallest, hottest point at one specific distance — focus right, and that point lands exactly on your material’s surface; focus wrong, and the beam is already spreading back out by the time it hits the material, delivering less energy over a wider area. That’s why a focus problem reads as a faint, fuzzy, or oddly wide engrave rather than a positioning error.

The ramp test finds the real focal height instead of trusting the lens’s printed spec:

1. Cut or 3D-print a ramp (or use a simple wedge of scrap wood) that holds a strip of masking tape or cardboard at a shallow, rising angle under the laser.
2. At low power and a fast speed, fire one continuous line along the full length of the ramp. Because the ramp’s height changes continuously, one end sits too close to the lens and the other too far, with the correct focus distance somewhere in between.
3. Find the thinnest, darkest point on the burned line — that’s your true focus height for this lens.
4. Set your machine’s Z-height (or focus-gauge offset) to match that point, not the lens’s printed nominal distance.

Why this matters even on a new machine: per Thunder Laser’s own ramp-test documentation, manufacturing tolerance on a lens can shift its real focal point 2-3mm away from the number printed in the spec sheet — enough to visibly soften an engrave. Re-run the ramp test any time you swap lenses, change material thickness enough to matter, or notice burns getting fuzzier than they used to be. A simple feeler-gauge or focus-height tool makes repeat setting faster than eyeballing it each time — check current prices on Amazon →. Running a shop where downtime waiting on a replacement tool actually costs money? Try Prime free for 30 days for next-day shipping on calibration tools and spare lenses.

2. Steps-per-mm calibration: fixing wrong-sized output

If your focus and burn quality look perfect but a design that should measure 100mm comes out at 97mm or 103mm, the problem isn’t the laser at all — it’s the firmware’s steps-per-mm value, which on GRBL-based controllers lives in $100 (X-axis) and $101 (Y-axis). This value tells the controller how many motor steps equal one real millimeter of travel; if it’s slightly wrong, every move the machine makes is proportionally off, in the same direction, every time.

How to check and correct it:

1. Mark a clear zero point on the axis you’re testing.
2. Send a known move command — for example G91 G1 X100 F200 to move exactly 100mm on X — and measure the actual distance traveled with a ruler or digital calipers.
3. Calculate the correction: new value = current $100 value × (commanded distance ÷ measured distance). If your current $100 is 800 and a commanded 100mm move only traveled 97mm, the corrected value is 800 × (100 ÷ 97) ≈ 824.7.
4. Enter the new value ($100=824.7) and repeat the test until the commanded and measured distances match.
5. Repeat the whole process for Y using $101.

Use the longest move your bed allows when measuring — a small measurement error matters a lot less over a 300mm test move than over a 20mm one. This value is stored in firmware and persists through power-offs, so once it’s dialed in correctly it shouldn’t drift again unless you change pulleys, belts, or microstepping settings.

3. Gantry squareness: the one software can’t fix

This is the calibration problem people most often misdiagnose as a settings issue, because it looks like it should have a settings fix — and it doesn’t. If a cut square comes out as a parallelogram, or a straight line drifts diagonally across the bed, your X-axis gantry isn’t running perfectly perpendicular to your Y-axis rails. That’s a physical property of the frame and belt alignment, not a number in LightBurn or the firmware.

The diagonal test (the same method Thunder Laser documents for its own machines):

1. Cut a square from scrap material — 5 inches is a good size; bigger squares make the error easier to measure accurately.
2. Measure both diagonals with a tape measure or calipers. By the Pythagorean theorem, a true 5-inch square’s diagonals should both measure about 7.07 inches (slightly less once you account for kerf).
3. If the two diagonals differ — Thunder Laser’s own example found 6.999” versus 7.042” on an out-of-square unit — the gantry needs physical correction.
4. The fix is mechanical: on machines with dual Y-axis belts, shift one belt by a tooth or two relative to the other until the gantry squares up; on single-belt or pulley-driven setups, loosen the drive pulley’s grub screw and rotate it slightly before re-tightening. Re-cut the test square after each adjustment — this is a trial-and-error process, not a one-shot fix.

A LightBurn forum thread on this exact symptom confirms there’s no software workaround beyond manually skewing an individual design as a one-off patch — every new file would need the same manual correction until the gantry itself is squared.

The bottom line

“Laser engraver calibration” covers three genuinely different problems: focus height, fixed with a ramp test and worth rechecking whenever you swap lenses or materials since even a new lens can sit 2-3mm off its printed spec; steps-per-mm, a firmware value ($100/$101 on GRBL) corrected with one measured test move and a simple ratio; and gantry squareness, a physical frame/belt alignment issue that a diagonal-cut test diagnoses but only a wrench — not a setting — can fix. Match your symptom to the right category first and you’ll skip wasting time on the other two. If your calibration issue is specifically about a camera overlay drifting from the real bed position, that’s a separate, fourth category covered in our LightBurn camera calibration guide. For the cleaning and belt-tension checks that help calibration stay accurate for longer, see our laser engraver maintenance schedule, and for power/speed numbers once your machine is dialed in, check our laser engraver settings cheat sheet.