Laser welding for sheet metal: when it beats MIG

A 1.5 mm CRCA door skin welded to its frame with MIG at 350 A comes out of the fixture with 3–5 mm of bow over a metre, and the painter fills it. Laser welding at 2 kW and 3 m/min puts about 40 J/mm into the same joint instead of 1.36 kJ/mm, and the panel stays inside 0.8 mm. The catch is that the laser only makes that weld if the two edges are within 0.1 mm of each other, which moves the problem from the welding cell to the fixture and the blank.

Laser welding for sheet metal is a fit-up process with a light source attached. The beam is 0.1–0.6 mm wide at focus, the keyhole it drills is narrower than the pool a MIG arc creates, and the filler that lets MIG bridge a sloppy gap does not exist in an autogenous weld. Get the gap, the mismatch and the surface clean and the joint is fast, flat and nearly free of spatter. Miss any of the three and the same machine produces undercut, drop-through or a seam that looks continuous and leaks.

The comparison that matters on a drawing is not laser against MIG in the abstract. It is whether the part can hold 0.1t of gap through handling, coating and fixture wear. If it can, laser wins on distortion and post-weld work. If it cannot, robotic MIG with a pulsed schedule is the cheaper route, and it will still hold ±0.5 mm on a frame.

Laser welding for sheet metal: the limits that decide the joint

Values below are for autogenous welding of mild and CRCA sheet with a fibre laser at 1.07 µm. Aluminium shifts power up and speed down; stainless shifts gas and cleaning.

The gap column is the one to design to. A laser weld on 1.2 mm sheet fails at a 0.3 mm gap by dropping through, and it fails at 0.05 mm on galvanised sheet by trapping zinc. The window is narrow and it sits in a different place for coated stock.

Why fit-up decides the weld before the machine does

A fibre laser couples into steel through a keyhole: the beam vaporises a narrow column, the column absorbs most of the energy, and the melt wall behind it closes as the beam moves. That mechanism needs the two faces of the joint to be in contact at the point where the beam passes. If they are apart, the beam goes through the gap and heats the fixture, or it melts one edge and drops the pool onto the floor.

The arithmetic of the tolerance is simple and unforgiving. A 1.2 mm joint at 0.1t allows 0.12 mm of gap. A laser-cut blank off the 12 kW machine holds ±0.1 mm on the profile, so one blank is inside the window and two stacked blanks are not. In practice a lap joint is set with a 0.1 mm shim and clamped every 150 mm on 1.5 mm sheet; at 1.0 mm the clamp pitch falls to 100 mm because the sheet deflects between clamps by more than the gap allowance.

Mismatch is a separate tolerance. A butt joint between a 1.2 mm and a 1.5 mm sheet has 0.3 mm of step before anyone touches it, and the keyhole will wander to the thicker side. Either match the gauges, or put the step on the non-cosmetic face and accept 0.15 mm of undercut that has to be dressed. Neither is free, and both should be visible on the drawing rather than discovered at first-off.

The arithmetic on a 300 mm seam

Take a 300 mm lap joint in 1.5 mm CRCA, welded both sides of a return flange. At 2 kW and 3 m/min the arc time is 6 seconds per side. MIG at 350 A, 26 V and 400 mm/min takes 45 seconds for the same 300 mm, plus about 15 seconds of spatter removal before paint.

Heat input per millimetre is where the two processes separate:

• Laser: 2,000 W ÷ 50 mm/s = 40 J/mm
• MIG: 60 × 26 × 350 ÷ (1,000 × 400) = 1,365 J/mm

The ratio is 34:1. Distortion does not scale linearly with heat input, because the panel's stiffness and the restraint fight it, but on 1.5 mm sheet the measured result is 0.3–0.8 mm of bow for the laser and 3–5 mm for the MIG. That difference is what decides whether the part goes to the paint line or to a straightening press.

Total time tells a different story. Clamping a 300 mm seam to 0.12 mm takes 25–35 seconds with toggle clamps and a shimmed back-up, and that time does not shrink with volume because every part needs it. Add load and unload and the laser cell is at 40–45 seconds against 60 for manual MIG. The laser wins on distortion and post-weld work; it does not win by a factor of seven on the floor.

That is why the decision is volume-sensitive in a way that the distortion argument is not. A shimmed, toggle-clamped fixture earns its keep by removing straightening, dressing and paint rework from the route, not by cutting weld seconds. At low volume the fixture is a cost with no payback, and a robotic MIG cell running an existing fixture is the cheaper answer even though each weld takes seven times longer.

Aluminium, zinc and the metals that argue back

Aluminium reflects about 95% of a 1.07 µm beam when it is cold and polished, and absorbs most of it once the keyhole opens. That transition makes the start of the weld unstable: power has to be ramped, the surface has to be clean, and the oxide that sits on it melts above 2,000 °C. Expect 1.5–2× the power used on steel at the same thickness, faster travel, and helium in the shielding mix to stabilise the plasma.

Cracking is the second issue. 6xxx alloys with a silicon-rich filler weld cleanly; the same alloy autogenously cracks along the centreline because the magnesium silicide leaves a low-melting film at the grain boundary. A 4043 or 5356 filler wire solves it and turns the process into laser hot-wire, which is a different machine and a different quotation.

Galvanised sheet is the third. Zinc boils at 907 °C, well below the steel pool, and the vapour needs a path out. In a zero-gap lap it expands between the sheets and blows through the pool, leaving porosity that shows as a pinhole line after powder coating. A 0.1–0.2 mm gap vents it. So the coated part needs a looser joint than the bare part, which is the opposite of what most drawings assume.

What Ganesh runs, and what we do not

We do not weld with a laser, and we will not quote a drawing as laser welded. The 12 kW fibre laser with the 40 ft × 16 ft bed cuts the blanks the process depends on: flanges at ±0.1 mm, edges inside the ISO 9013 range 4 tolerance band, and cut faces clean enough that a weld prep needs no machining. That edge quality is half of what a laser weld needs; the clamping is the other half.

Welding runs on three CLOOS robotic MIG cells and more than 30 manual MIG/MAG, spot and special-purpose stations, with fixtures designed and built in the in-house tool room. When a drawing specifies a laser seam, the engineering review converts it to one of three routes. The first is robotic MIG with a pulsed transfer at 0.25–0.4 kJ/mm, a backstep sequence and a fixture that holds the flange to 0.5 mm; that is a quarter of the heat of a conventional spray-arc schedule and it holds a 1.5 mm panel inside 1.5 mm of bow. The second is a spot row where the joint is cosmetic and the load is shear. The third is a mechanical joint where the section cannot take heat at all.

The conversion has consequences that belong in the quotation. Robotic MIG needs a weld access window that a laser keyhole does not, so a 6 mm fillet in a 12 mm return is reachable with the torch at 45° and not with a beam at 90°. It also leaves a visible toe that has to be dressed on a painted face. Where both of those are unacceptable, the joint is a design problem, not a welding problem, and the answer is usually a different section.

Elevator frame sections and canopy panels are the work shapes where this question comes up, and the welded assemblies capability sets out the cell and station envelope. The laser cutting capability covers the blank tolerances that any thin-sheet weld route starts from.

What to put on the drawing

1. Process by name, not “weld all round”. A laser symbol and a MIG symbol are different joints with different access and different distortion.
2. Joint type and gap allowance at the interface, with the fixture note that holds it: 0.12 mm maximum on 1.2 mm, 0.15 mm mismatch.
3. Thickness on both sides of the joint, and the gauge match, because a 0.3 mm step is outside any laser window.
4. Coating and whether the weld is made through it, since a galvanised lap needs a venting gap and a zinc repair afterwards.
5. Cosmetic faces, so the weld toe lands on the side that gets no paint or gets dressed.
6. Acceptance: ISO 5817 quality level, the imperfection limits that matter for the joint, and whether the seam has to pass a leak test at a stated pressure.
7. Distortion limit after welding, in millimetres over the panel length, because a flatness callout on the finished part is not the same instruction.

The robotic MIG comparison covers the schedule and cost side of the conversion, and the repeat assembly guide covers fixture and sequence design for the parts that follow it.

When MIG or a mechanical joint is the right answer

MIG is the right answer when the gap cannot be held. A hand-fitted canopy frame, a part that has been through a brake without a checking fixture, and any assembly where the flanges are formed from a sheared blank will sit at 0.4–0.8 mm of gap, and a laser will burn through or miss. Robotic MIG bridges that with wire and a wider arc, and the distortion it adds is manageable on 2 mm and above.

MIG is also right above 4 mm, in deep fillets, and wherever the joint carries load in a direction that needs a leg length rather than a penetration depth. A laser keyhole gives deep penetration in a narrow band; it does not give a 6 mm leg in a corner. Aluminium enclosures usually go to MIG or to a rivet-and-adhesive joint for the same reason, and the aluminium welding guide covers the schedule.

A mechanical joint is right when there is no second side to clamp against, when the panel is pre-coated, or when the assembly has to come apart. A closed box cannot be laser welded from outside without a back purge and a backing bar, and a powder-coated door cannot be welded at all without destroying the film within 20 mm of the seam.

Send the joint detail with the thickness, the gap you can actually hold and the distortion limit. The bending and welded assemblies hub carries the rest of the cluster, and the RFQ route is where a laser-specified joint gets reviewed against the fixture rather than against the process brochure.