Buyers love quoting "±0.1mm plate, ±0.05mm tube". It sounds precise. In assembly, those numbers fight each other.
A 4mm symmetric 3K twill plate holds ±0.1mm flatness. A 12mm OD carbon tube holds ±0.05mm OD. You bond the tube into a milled pocket in the plate. Now three tolerances stack:
Plate pocket: ±0.1mm
Tube OD: ±0.05mm
Bond line: 0.1–0.2mm (resin squeeze-out)
Worst case clearance = 0.1 + 0.05 + 0.2 = 0.35mm. Best case interference = -0.1 - 0.05 - 0.1 = -0.25mm.
That means your "tight joint" can rattle or your "slip fit" can crack the plate.
The fix is not tighter tolerance. It is assigned datum.
Good RFQ says:
Datum A = tube centerline
Datum B = plate top face
Pocket tolerance referenced to tube, not to plate edge
Bond gap specified as 0.15mm nominal, not "as little as possible"
Common reject patterns we see
Buyer dims tube to its own axis, plate to its own corner → CMM passes, assembly drifts
Matte both sides, but no Ra call-out → bond line varies 0.08–0.22mm
Holes in plate ±0.1mm, holes in tube ±0.05mm, no coaxiality call-out → fastener binds
RFQ line we want to read
"Plate pocket and tube OD share Datum A. Coaxiality Ø0.15mm. Bond gap 0.15mm nominal, allowable 0.10–0.20mm."
That one line removes 80% of clarification loops.
Where it matters
UAV gimbal yokes
Telescoping mast joints
Exoskeleton knee brackets
Robot wrist plates
Carbon fiber is precise. The RFQ has to be precise first.
FAQ
Q: Should I just ask for ±0.02mm on everything?
No. Tighter tolerance on carbon plate raises cost 30–60% and still fails if datums differ.
Q: Who should redraw when plate and tube don't match?
The buyer's mechanical engineer, not the carbon factory. Supplier can propose, but datum ownership stays with design side.
Q: Is bond gap really that important?
Yes. Carbon-to-carbon epoxy wants a controlled film. Too thin = starved joint. Too thick = shear creep under vibration.
