Case file

Laser-Cut Chassis Assembly: One Route to Finish

A welded assembly is a process-sequencing problem wearing a geometry problem's clothes. Each part can be within tolerance and the assembly still come out out-of-square, because welding adds heat and heat adds movement. This file is about controlling that movement rather than measuring it afterwards.

Reviewed by Tom, Senior Process Engineer·Dongguan source factory · 11+ years in precision sheet metal
telecom equipment chassis spcc steel 1 — laser cut chassis assembly
Short answer

A welded chassis assembled from laser-cut and formed parts, where the engineering difficulty was not any single component but holding the assembly square through welding. A jig built around two datum holes removed the distortion that had been accumulating from clamping by hand.

The assembly

A rectangular chassis roughly 600 × 400 × 150 mm built from 1.5 mm SPCC steel: a base pan, two side walls, two end plates and an internal divider. Joined by TIG on the outside corners with the visible faces dressed. Flat pack to the customer, assembled by us.

The functional requirement was that the mounting feet should sit in one plane and that the two side walls should be parallel within 0.5 mm over the full length, because the slide-in module that mounts inside is made to a fixed width.

Where the tolerance went

Individually, every part was inside its drawing tolerance. Clamped by hand and tacked, the assembly came out with walls that were parallel at the ends and pinched in the middle, and a base that had bowed between the feet. The error was small — under a millimetre — and entirely the result of weld shrinkage acting on an unsupported panel.

Two further effects compounded it. Tack-welding in one pass and finishing the weld later meant the assembly had already moved before the final weld locked it in. And because the side walls were clamped to the base rather than to each other, the base flatness error transferred into the wall spacing.

The fixes, in the order they were applied

ChangeReason
Added two datum holes pierced in the base pan at the laser stageGives the jig one unambiguous reference instead of relying on edges
Built a fixture locating on those datums and clamping the walls to each otherStops base flatness error transferring into wall spacing
Changed the sequence to complete the weld in one heat, alternating sidesReduces the total heat input and balances shrinkage across the part
Added a temporary brace across the open top, removed after coolingHolds the critical dimension while the material is hottest and weakest
Moved dressing to after the assembly cooled fullyGrinding a hot weld relieves stress and re-distorts what was straight
Sequence changes that brought the assembly into tolerance

Inspection

The inspection plan checks the datums first, then the features measured from them, then the assembly as a whole. Wall parallelism is measured at three positions along the length rather than one, because a single measurement would have passed the pinched assembly that started this work.

A first-article dimensional report is issued against the ballooned assembly drawing for the first unit and after any change to the sequence or the fixture. That is the point at which a jig problem is still cheap to fix.

What transfers

  • On welded assemblies, design the datums into the parts at the cutting stage; a datum invented at assembly is a datum that moves.
  • Sequence is a specification. Two shops can weld the same parts to the same drawing and produce different assemblies.
  • Distortion control beats distortion correction — a jig costs less than dressing and re-straightening every unit.
  • Name the critical assembly dimension on the drawing. If it is not named, nobody is holding it.

Frequently asked questions

Why does a welded assembly go out of square when the parts are in tolerance?

Because welding adds heat and heat causes movement. Shrinkage acts on the whole assembly, not on individual parts, and an unsupported panel will bow before the weld locks it. Tolerances on components do not guarantee tolerances on the assembly.

Do you build the welding fixture or does the customer?

We build it, because the fixture has to match the sequence we are going to run. It is quoted as a one-off element so the customer can see what it costs, and it stays with the programme rather than being absorbed into a unit price.

Would a heavier gauge have avoided this?

Partly, at the cost of weight and material on every piece. The fixture and sequence changes solved it at 1.5 mm, and they are a one-off cost rather than a recurring one — which is the better trade on a production programme.

Can the assembly be shipped flat and welded on site?

It can be shipped flat and assembled elsewhere, but then the welding quality and the squareness become the customer's problem and the fixture does not travel with the parts. We would only recommend that where freight volume genuinely dominates the total cost.

Send the drawing, get a real answer

An engineer reviews your model for manufacturability and returns a costed quotation — with the DFM observations that would change the price.