A 6-Meter Conveyor Frame: From Drawings to Machined Pads
This is the story of a welded conveyor frame — what we drew, what the shop welded, what warped, and what we changed for the next one. The frame carried linear rails and a belt drive. It was 6 meters long, made from 80×80×5 mm rectangular steel tube. By the time the rails went on, the frame had twisted 3 mm across the width. We shimmed. Then we redesigned.
Week 1: The drawing
We drew the frame as a simple rectangle of rectangular tube, cross-braced at three stations. The rails mounted directly to the top of the tube. The flatness tolerance on the drawing was “±2 mm” — a number we’d inherited from a previous project that didn’t actually have rails on it. We didn’t question it.
The tube wall was 5 mm. We chose it because 3 mm felt too thin for a 6 m span. At this stage, everything looked fine on screen.
Week 3: The welding
The shop welded it on a jig. They tacked all four corners, checked diagonals, then ran continuous welds along the bottom flange. By the time they finished the long seams, the frame had pulled — the top edge was 1.5 mm shorter than the bottom. The whole frame bowed upward in the middle by 4 mm.
This is predictable. Welds shrink as they cool. A continuous bead along one side pulls the material toward that side. The shop clamped it down during welding, which helped, but they didn’t alternate sides — the long seam on one side was completed before the other side was started. Balanced welding would have distributed the shrinkage.
Week 4: The rails go on
We brought the frame into assembly, set it on floor stands, and tried to bolt the linear rails down. The rail mounting holes didn’t line up. The frame had twisted — not just bowed — because the cross-braces had been welded last, after the main rectangle had already distorted. The twist accumulated.
We measured with a laser: 3.2 mm variation across the width over 6 m. The rails needed 0.05 mm flatness. We shimmed every rail support point. It took two days. The machine ran, but the rails were sitting on a bed of shims that no one would service again.
What changed for the next frame
Three things, in order of impact:
1. Machined mounting pads. We added 12 mm thick steel plates where the rails mount. After welding the frame, we sent it to the CNC mill and faced those plates flat, parallel to each other within 0.05 mm. The rails bolt to the pads, not to the raw tube. The pads are a datum — everything else references them.
2. Balanced welding sequence. We specified in the drawing: weld in 150 mm segments, alternating sides, back-stepping. Not continuous long seams. The shop already knew how; we just wrote it on the print.
3. Bolt-on sub-frame for precision axes. Where we really need straightness, we bolt a machined aluminum extrusion onto the welded frame. The extrusion is straight by extrusion process. The welded frame carries the load; the extrusion carries the precision. No machining needed on the weldment for that axis.
Flatness: what tolerance do you actually need?
| What mounts to the frame | Treatment | Flatness spec |
|---|---|---|
| Guards, supports, non-critical brackets | Weld, square by eye | ±5 mm |
| Conveyor bed, non-critical rails | Weld, clamp, add pads | 0.5 mm |
| Linear guide rails (profile rails) | Machine pads after welding | 0.05 mm |
| Measuring station, vision stand | Anneal, machine all surfaces | 0.02 mm |
The takeaway that stuck
Welded frames carry load. They don’t carry precision. If you’re bolting a linear rail to a 5 mm welded tube, you’re asking the weldment to hold a tolerance it was never going to hold. Add the machined pads, or bolt on a straight sub-frame. The second frame we built with these changes went together without a shim. The rails landed on the pads within spec. No one measured the frame’s twist because it didn’t matter anymore.
The line I put in every frame drawing now: “Precision components bolt to machined pads, not directly to welded surfaces.” It costs 2 hours on the mill and saves 2 days on the floor.
