Why Your Laser Isn't Cutting All the Way Through: A Test-by-Test Decision Table
Work the causes in cost order, not in the order they occur to you. Focus and material thickness account for the majority of cases and cost nothing to check; a hazy lens or a weak air assist come next; tube age is last because it's the expensive answer and the one people jump to first. Being 1 mm off focus with a 2-inch lens costs about two-thirds of your power density on its own.
Start with focus and material thickness. Between them they explain most of these cases, they cost nothing to check, and they're the two things that change silently between one job and the next.
Raising the power is the wrong first move. It masks a focus error rather than fixing it, drives more heat into the kerf walls, and shortens tube life on a CO2 machine for no gain. The table below runs the causes in the order that a workshop actually eliminates them: free tests first, expensive answers last.
Work the table from the top
Order Suspect The test What confirms it Fix, and rough cost 1 Focus is off Ramp test: tape a strip of card at a shallow angle, cut one line along it at low power The narrowest, darkest point of the burn sits above or below your set Z height Reset Z. Free. 2 Material is thicker or damper than you assumed Calipers at four corners; press a fingernail into the edge "3 mm" ply measures 3.4 mm, or the cut steams and hisses instead of smoking Dry the stock, or re-cut to actual thickness. Free. 3 Lens or protective window fouled Remove it and hold it against a bright light at an angle Visible haze, spatter freckles, or a rainbow film that doesn't wipe evenly Clean with isopropyl and lens tissue, or replace. £20-70. 4 Focus lens fitted the wrong way up Look at the curve: a plano-convex CO2 focus lens goes convex side toward the incoming beam It cuts, but badly, and no Z setting makes the ramp test crisp Flip it. Free. 5 Air assist too weak or misaimed Hold a strip of paper under the nozzle at working height Paper barely moves, or the smoke plume is visibly sitting in the beam path Better pump and a tighter nozzle. £30-90. 6 Air assist too strong on wood Turn it down and re-run the same file Flame flare during the cut, blown-out kerf, scorched top face Fit a regulator. £15. 7 Power, speed or minimum-power mismatch Ten-step power ramp on scrap at fixed speed Cut depth doesn't track the power steps, or plateaus early Correct the min/max power settings for your PWM range. Free. 8 Mirror alignment has drifted Tape shot at all four bed corners with the gantry at each extreme Spot moves more than about 1 mm between corners, or the cut fails at one end of the bed only Realign. Free, but budget an hour. 9 Bed not flat, or the sheet is bowed Feeler gauge at five points; press the middle of the sheet The cut fails only in the middle, or only where the sheet lifts Magnets, hold-down pins, or a flatter bed. £10-40. 10 CO2 tube ageing or power supply sagging Re-run your own baseline file and compare to the tile you kept; check tube current against the tube's own datasheet The same file cuts measurably shallower than it did when the machine was new New tube. £150-400 plus alignment. 11 Diode module window fouled or the module is overheating Inspect the small protective window; touch the heatsink after ten minutes Output falls off part way into a long job; window shows a burnt film Replace the window, £5-15. Improve airflow. 12 Wrong material for the wavelength Check the material against the machine type Clear acrylic or clear glass on a diode; white or pale EVA on a diode; bare metal on either No fix. Change material or change machine. Go top to bottom and stop at the first confirmed cause. If you jump to row 10 first you'll spend hundreds of pounds on a tube that wasn't the problem, which is by some margin the most common expensive mistake in this hobby. Focus explains about half of these
Power density falls with the square of spot diameter, and the spot grows fast either side of the waist. On a 2-inch lens the beam is genuinely sharp over a window less than a millimetre and a half wide, which is roughly the thickness of a sheet of card.
Here's what a focus error actually costs, calculated for a 10.6 µm beam through a 2-inch lens with a 7 mm raw beam, and the same errors through a 4-inch lens for comparison.
Focus error Spot diameter, 2 in lens Power density vs in focus, 2 in What it looks like on the cut Power density vs in focus, 4 in 0.25 mm 52 µm 89% Nothing visible 99% 0.5 mm 60 µm 67% Slightly wider kerf, more char 97% 0.75 mm 71 µm 47% Cut severs but the underside is ragged 94% 1.0 mm 85 µm 34% Last 20% of the thickness holds on 89% 1.5 mm 114 µm 18% Deep score, no cut 78% 2.0 mm 146 µm 11% Wide brown groove 67% 3.0 mm 213 µm 5% Surface mark only 47% Compare the two right-hand columns before blaming your power. A 1 mm focus error on a 2-inch lens throws away two-thirds of your delivered intensity; the same error on a 4-inch lens costs about a tenth. That contrast is also the reason a longer lens rescues cuts on warped or uneven stock. It isn't stronger — it's more tolerant.
Running the ramp test properly
- Cut a strip of 3 mm card or thin ply about 200 mm long and prop one end on a 20 mm block so it sits at a shallow angle across the bed.
- Set the machine to your usual focus height at the low end of the ramp. Draw a single line along the length of the strip at low power and moderate speed — enough to mark, not enough to cut.
- Look for the point where the burnt line is narrowest and darkest. Mark it with a pencil.
- Measure the vertical height of the strip at that point. The difference between that and your set focus height is your correction.
- Apply the correction and repeat once. The narrow point should now sit at the height you set. Note the number for that lens and that nozzle.
- Re-run this after any lens change, nozzle change, or bed reassembly. It takes four minutes and it eliminates row 1 of the table permanently.
What a hazy lens really costs
A ZnSe lens doesn't fail dramatically. It picks up a fine film of resin and vaporised material, gets a faint amber cast, and loses transmission gradually enough that you adjust your settings without noticing you're doing it.
Hold it edge-on against a lamp. A clean lens looks like nothing at all. A dirty one shows a haze, and if you can see individual spatter marks it's already costing you serious power. Clean it with lens-grade isopropyl on lens tissue, one wipe per tissue, no circles, no pressure. If a spatter mark has pitted the coating, cleaning won't bring it back and the lens is done.
Mirrors deserve the same look. Three mirrors at 97% each cost you nearly 9% before the lens even sees the beam, and a fingerprint on mirror one costs a great deal more than that.
Air assist, in both directions
Too little air is the classic version. The plume of hot smoke sits in the kerf, absorbs and scatters the beam, and the bottom of the cut never receives the power your settings promised. Symptoms are a cut that severs at the start of a line and fails at the end, and charred kerf walls that smell worse than usual.
Too much air is the version people don't expect. On plywood and MDF, a strong jet feeds oxygen to the burn and turns a cut into a flame that scorches the top face and widens the kerf. Acrylic wants a gentler stream still — hard air frosts the edge that CO2 would otherwise leave clear.
The workable rule: firm and focused for cutting, gentle for engraving, and gentler still for acrylic where edge finish matters. A cheap inline regulator is the best twenty pounds in the workshop.
Cuts fine at one end of the bed, fails at the other
That's an optical path problem, not a power problem, and no setting change will fix it.
On a CO2 machine, the distance from tube to workpiece changes as the head travels, and if the alignment is off the beam clips an aperture or arrives off-centre through the lens at the far corners. Tape shots at all four extremes will show it: the burn mark moves, or turns from a clean dot into a crescent.
On a gantry machine of any type, check bed level too. A bed that's 2 mm low at the back is a 2 mm focus error at the back, and the table above already told you what that costs.
One easily missed variant: the honeycomb bed itself sagging in the middle after a few hundred hours. It looks flat and isn't.
Keep a baseline file
Cut one small test tile the week you commission the machine — a ramp of powers and speeds on a material you'll use often — and keep it in a drawer with the date and settings marked on it.
Twelve months later, run the identical file on identical stock and put the two tiles side by side. That comparison answers the tube-age question in five minutes, for free, with no meter and no guesswork. It's the only test on this page that tells you what your machine has actually lost rather than what it should theoretically deliver.
Nobody does this on day one, and everybody wishes they had on day four hundred.
When the machine really is the limit
Sometimes the honest answer is that the material is beyond the machine. A 10W optical diode will not cut 9 mm ply, whatever you do to the settings, and a 40W CO2 will not give you a clean flame-polished edge on 15 mm acrylic.
The tell is consistency. If the machine cuts your reference material exactly as it always has, your optics pass inspection and your focus test is crisp, then the machine is fine and the job is too big for it. That's a purchasing decision, not a fault.
Before accepting that, though, check one last thing: the lens. Fitting a longer focal length costs a fraction of a new machine and moves the thick-material ceiling more than most people expect.