Why Your CO2 Laser Cuts Fine in One Corner and Fails in Another

(And why corner-by-corner alignment never fixes it)

This is the companion write-up to the interactive guide — eight short animated chapters that show everything below actually happening.

// The Full Write-Up

1. The Symptom Everyone Recognizes

You can cut 3mm ply at the front-left of the bed all day. Move the same job to the back-right and it barely scores. You clean the mirrors. You check the lens. You “align the laser.” Someone on Facebook says check your bed leveling. Someone else says your tube is dying. The problem persists.

What’s actually happening: the beam is not arriving at the lens the same way at every position on the bed. Somewhere along the mirror chain, the beam is either clipping an aperture, missing part of a mirror, or entering the lens off-axis — but only at some positions. Power delivered to the material drops in those regions.

2. One Rule That Explains Everything

A tilt screw rotates the beam around the point where it strikes the mirror. It changes direction only. It can never move that point.

The outgoing beam is a line in space: a starting point (where the beam hits the mirror) plus a direction (set by tilt). The starting point is determined by where the previous component put the beam — the tube’s position, the previous mirror’s position, the rail the carriage rides on. Tilt screws own the direction. Physical hardware position owns the starting point.

Every alignment mistake in every Facebook thread comes from using the tool that controls direction to try to fix a problem of position, or vice versa.

3. The Three Error Types

Type 1 — Angle error (tilt-screw fixable).
The beam leaving Mirror 1 is not parallel to the Y rails, or the beam leaving Mirror 2 is not parallel to the X gantry. As the axis travels, the landing spot on the next mirror walks — offset grows linearly with distance. A tiny angle becomes a big miss at the far end of the bed. This is the error tilt screws exist to fix, and the near/far method (Section 5) is how.

Type 2 — Constant translation offset (hardware fixable, tilt-screw worsenable).
The tube sits a little high. Or a mirror mount is offset. The beam travels perfectly parallel to the rail but displaced — it hits Mirror 2 (say) 2mm below center at every Y position. Same spot near and far, just not centered.

This is often not a problem at all. Mirrors reflect from wherever they’re hit. Plenty of well-cutting machines have nothing dead-center and everything consistent. It only matters if the offset is large enough to clip a mirror edge, an aperture, or the nozzle bore, or to enter the lens badly off-axis.

The classic self-inflicted wound: the spot is low on Mirror 2, so the owner tilts Mirror 1 to “aim” it at center. That works — at one position. Now the beam is angled relative to the rail: centered up close, low when far. They’ve traded a harmless constant offset for a deadly walking one, and the corner-chasing begins. The standard tape method’s instruction to “center the dot on each mirror” quietly assumes all hardware is positioned correctly — when it isn’t, that instruction actively creates the failure.

If the offset genuinely must be fixed: shim or reposition the hardware (tube mounts, mirror mount, carriage). Never the screws.

Type 3 — Rail/gantry geometry error (the hidden one).
The gantry is not level, is twisted, or a rail sags. Now the Mirror 2 carriage itself physically rises, falls, or shifts as the gantry travels along Y. The beam can be perfectly parallel to where the rail should be — the mirror is the thing moving out of the beam’s path.

No tilt screw can fix this. To compensate with angle you’d need the beam crooked in exactly the way the rail is crooked — which breaks the other axis, or breaks mid-travel if the error isn’t linear. This is the error that survives every alignment attempt, produces the endless corner-chasing loop, and almost never gets named in the comments.

Fixes: level and square the gantry itself (shim, adjust rail mounts) — or, on many machines, use the Mirror 2 carriage adjustment: the whole carriage can be moved up/down and left/right so the mirror tracks the beam correctly across the full travel. This adjustment exists on most Chinese gantry machines and is almost never documented or mentioned.

4. Why Corner-by-Corner Alignment Cannot Converge

The intuitive procedure — pulse at corner 1, adjust, pulse at corner 2, adjust, and so on — fails for a structural reason, not a skill reason.

When angle error is present, every adjustment made at one corner changes the beam direction, which changes where it lands at every other corner. Fix corner 3 and you’ve silently broken corners 1 and 2. Most people never re-check the earlier corners; the few who do just keep chasing, because with the error present there is no screw position that satisfies all four corners simultaneously. It’s a loop with no exit.

And when the error is Type 3 (gantry geometry), even a perfect angular alignment leaves residual walk — the target itself is moving. Corner-by-corner can’t even see this, let alone fix it.

5. The Near/Far Method — The Only Alignment That Works

The correct target is not “center of the mirror.” It is: the beam hits the same spot at the nearest and farthest travel positions. Same spot at both extremes = beam parallel to the rail = frozen spot everywhere in between.

Per leg (M1→M2 shown; repeat for M2→M3 and for the nozzle):

1. Stick masking tape over Mirror 2’s face or aperture ring. Leave it there for the whole test — the tape rides with the carriage and is your position record.
2. Gantry at the near end: fire one low-power pulse. Burn mark #1.
3. Gantry at the far end: fire again. Burn mark #2 on the same tape.
4. Compare. Marks coincide → this leg is parallel, done. Marks apart → the gap is your error, made visible.
5. Marks apart: adjust the upstream mirror’s tilt (M1 for this leg), fresh tape, repeat near/far. Iterate until the marks merge.
6. Marks merged but off-center: that’s a constant offset — leave the screws alone. Only act if it threatens to clip something, and act by moving hardware.
7. Marks that refuse to merge after honest iteration = Type 3. Stop turning screws. Check gantry level with the tube off — measure rail height at both ends, check for twist. Fix mechanically or use the Mirror 2 carriage adjustment.

Final checkpoint: tape over the nozzle opening, pulse at multiple bed positions. Beam exits centered everywhere → done, regardless of what any upstream spot looks like.

6. The Diagnostic Flowchart

Tape on next mirror → pulse NEAR → pulse FAR → compare marks
│
├─ Marks coincide, centered ........... leg is good, next leg
├─ Marks coincide, off-center ......... constant offset: leave it
│                                       (unless clipping → move hardware,
│                                        NEVER tilt screws)
├─ Marks apart, converge with
│  upstream tilt + re-test ............ angle error: fixed, next leg
└─ Marks apart, will NOT converge ..... gantry/rail geometry:
                                        level the gantry or adjust
                                        the M2 carriage position