TL;DR: A COA that lists material grade without test method and acceptance criteria is not a qualification document — it’s a shipping label with extra columns.
TL;DR: In our incoming inspection protocol, wall thickness deviation above ±0.05mm on aluminium cases triggers automatic lot hold, regardless of supplier COA status.
Wall Thickness Uniformity: The Specification Parameter That Drives Everything Else #
When brand buyers brief us on metal tin or aluminium case requirements, the first document they send usually specifies dimensions, print colours, and finish type. Wall thickness uniformity rarely appears — and that omission causes more downstream problems than any other single spec gap.
Here is why it matters more than the obvious parameters. Tinplate gauge directly controls lid seating torque, emboss depth retention, and lacquer adhesion. A nominal 0.23mm tinplate panel with actual variation of ±0.04mm will produce inconsistent emboss depth across a print run — emboss dies are set to a fixed tonnage, so thinner panels over-emboss and thicker panels under-emboss. For aluminium cases, wall thickness uniformity drives hinge pin fit, lid-to-body closure gap, and anodising layer consistency. Anodising builds at roughly 1–1.5µm per minute in a sulphuric acid bath; if the base aluminium wall varies by more than 0.06mm across a batch, the anodised layer thickness will vary proportionally, producing visible colour difference under directional lighting.
The applicable measurement standard for sheet metal thickness in packaging is ASTM B209 for aluminium alloy sheet and GB/T 2520 for cold-rolled tinplate — GB/T 2520 Clause 5.2 sets the permitted thickness deviation for T-2.5 to T-4 temper tinplate between ±0.01mm and ±0.03mm depending on nominal gauge. Any COA that does not reference one of these standards in the thickness column should be treated as unverified.
Supplier Qualification — What to Request and What the Response Tells You #
When we onboard a new metal tin or aluminium case supplier, our QC team runs what we call the Form QM-14 incoming qualification checklist. The document requests seven specific data fields, and the way a supplier responds to that request tells us more than the data itself.
Ask for the COA with the following fields specified explicitly: nominal wall thickness and measured range (not just nominal), temper designation per GB/T 2520 or ASTM B209, lacquer coating weight in g/m², lacquer adhesion test result per ISO 2409 crosshatch method, weld seam tensile pull force, base metal alloy grade, and batch production date with furnace/coil number traceability.
A qualified supplier responds within 48 hours with a COA that populates every field with measured values and test dates. A marginal supplier sends a template COA with several fields blank or filled with “per standard.” A disqualified supplier sends a product brochure and calls it a COA.
The lacquer coating weight field is particularly diagnostic. For food-contact tins, internal lacquer weight should fall between 8–12 g/m² for epoxy-phenolic systems and 5–8 g/m² for polyester systems. Both ranges are traceable to FDA 21 CFR 175.300 for resinous coatings. A supplier who cannot provide coating weight measured values — not estimated values, measured — has not run a proper incoming or outgoing QC process on their lacquer application line.
One additional request we always make: ask for the supplier’s internal AQL sampling plan for outgoing inspection. The response time alone is informative. Suppliers with a real quality system respond with a document reference within one business day. Suppliers without one respond with a phone call.
Cost-Performance Trade-offs in Metal Packaging Specification #
The most common cost lever buyers try to pull on tinplate tins is gauge reduction. Dropping from 0.23mm to 0.20mm tinplate on a round slip-lid tin reduces material cost by roughly 12–15% per unit at volume, and in some applications that is the correct decision.
Where gauge reduction works: simple cylindrical tins with no emboss, no deep-draw lid, and contents under 300g. The structural demand on the sidewall is low enough that 0.20mm T-3 temper tinplate performs adequately, and the cost saving is real.
Where it fails: rectangular tins with embossed panel artwork, or any tin with a friction-fit lid carrying contents above 500g. On rectangular panels, sidewall flex under handling stress is proportional to the cube of panel width divided by thickness cubed — thin gauge on a wide panel produces visible oil-canning that looks like a defect. We’ve held incoming lots on exactly this basis: nominally compliant gauge, structurally inadequate for the panel geometry.
For aluminium cases, the cost-performance argument is different. Alloy grade matters far more than wall thickness within the typical 1.0–1.5mm range. Switching from 6061-T6 to 6063-T5 saves roughly 8–10% on extrusion cost, but 6063-T5 has a yield strength of approximately 145 MPa versus 6061-T6 at 276 MPa. For a hinge-lid case carrying electronic components or medical devices, that strength differential is not recoverable with finishing. For a cosmetic brush roll case with light-duty closure, 6063-T5 is entirely adequate.
The counterargument on lacquer specification: buyers sometimes request upgraded epoxy-phenolic lacquer on tins that will hold dry, non-reactive products like confectionery or candles. The cost premium over standard polyester lacquer is 6–9% per unit, and for those product categories it buys nothing. Standard polyester internal lacquer at 5–7 g/m² is sufficient — the epoxy-phenolic upgrade is justified for acidic food products, carbonated beverages, or solvent-based contents where pH or chemical compatibility is a real variable.
Incoming Inspection Protocol: Pass/Fail Thresholds We Actually Use #
Our incoming inspection for metal tins and aluminium cases follows a three-tier structure that we developed after tracking 31 incoming lot failures over a 24-month period ending in Q4 2024. Roughly two-thirds of those failures were attributable to dimensional non-conformance that a supplier COA had passed — which told us the COA pass criteria were set too loosely at the supplier end.
Dimensional checks — critical tier:
| Parameter | Nominal | Our Pass Tolerance | Automatic Lot Hold Trigger |
|---|---|---|---|
| Wall thickness (aluminium case) | Per drawing | ±0.05mm | Any single measurement >±0.07mm |
| Tinplate gauge (tinplate tin) | Per grade | ±0.02mm | Deviation from GB/T 2520 Clause 5.2 limit |
| Lid-to-body closure gap | 0.1–0.3mm | ±0.05mm | Gap >0.4mm or <0.05mm |
| Hinge pin diameter (aluminium case) | Per drawing | ±0.03mm | Interference fit failure on 3+ units in 32-piece AQL sample |
| Internal lacquer coating weight | 8–12 g/m² (epoxy) | ±1 g/m² | <7 g/m² on any measured panel |
Sampling follows ISO 2859-1 (equivalent to ASTM E2234) at AQL 1.0 for critical dimensional parameters and AQL 2.5 for visual surface defects. For a standard lot of 5,000 units, that means a 200-piece sample for dimensional checks at normal inspection level II.
Red flags that escalate a lot from standard inspection to 100% inspection: weld seam visible through internal lacquer on food-contact tins, hinge pin corrosion on aluminium cases stored below 60% RH (signals a protective coating failure, not a storage issue), and coating adhesion failures on more than 2 units in the initial crosshatch sample.
One area where practice genuinely varies across the industry: requalification frequency for approved suppliers. Some factories requalify annually on a fixed schedule. Others only re-run full qualification after a supplier reports a formulation change or factory relocation. Our approach is annual full qualification for suppliers providing food-contact or cosmetic-contact tins, and biannual for suppliers providing non-contact decorative tins with stable order history. Neither approach is universally correct — it depends on the regulatory exposure of your end product and the length of your supplier relationship.
Specification Notes for Brand Partners #
When you brief us on a metal tin or aluminium case project, the information that most directly affects sample accuracy and quote precision is: nominal dimensions with toleranced drawing (not sketch), intended product contact category (food, cosmetic, non-contact), required alloy or tinplate grade if you have a preference, surface finish specification for aluminium cases (anodised, powder coat, bare), and any regulatory market requirements (FDA, EU food contact, REACH).
The most common gap in incoming briefs is the absence of a dimensional drawing with tolerances. Brands frequently send a physical reference sample with a note to “match this” — but without a toleranced drawing, we cannot confirm whether the supplier’s COA passes or fails your specification, because the specification does not exist in writing. This causes the first sample iteration almost every time. Providing a dimensioned drawing, even a basic one, removes that cycle.
Our standard sample lead time for metal tins is 18–22 working days from brief approval. Aluminium cases with custom hinge mechanisms run 25–30 working days. Both timelines assume artwork is supplied in AI or PDF format at 300 DPI minimum with Pantone references confirmed. Unresolved colour references are the single most common cause of timeline extension at the sampling stage.
Does wall thickness on a COA always reflect actual production material?
No. COA values are drawn from the supplier’s outgoing QC sample, which may be measured on a different panel location than where failures occur. Our protocol requires us to re-measure wall thickness on a 200-piece sample at incoming, using a calibrated micrometer per ISO 3611. COA values and incoming measurements disagree in roughly one out of every eight lots we receive from new suppliers.
What AQL level should I specify for food-contact metal tins?
AQL 1.0 for critical defects (lacquer failure, weld seam exposure, dimensional non-conformance affecting seal integrity) is the standard we apply, per ISO 2859-1 normal inspection level II. For cosmetic surface defects, AQL 2.5 is typically accepted by brand QA teams unless the product is luxury-tier, in which case AQL 1.5 is more appropriate.
How do I know if an aluminium alloy grade is appropriate for my product weight?
It depends on the closure mechanism and product load. For cases carrying items under 1kg with a simple hinge, 6063-T5 (yield strength ~145 MPa) is adequate. Above 1.5kg or where the hinge bears repeated open-close cycles, 6061-T6 (yield strength ~276 MPa) is the specification we recommend. Between those weights, we review the hinge pin diameter and frequency of use together before specifying.
If a supplier’s COA shows “per standard” for lacquer coating weight, is that acceptable?
No. “Per standard” is not a measured value — it is a declaration that the process was run within a parameter range. For food-contact applications regulated under FDA 21 CFR 175.300 or EU 10/2011, the coating weight must be a measured value from the actual production batch, not a reference to process parameters. We reject COAs that use this language for food or cosmetic contact categories.
What lead time should I budget for a first production run of custom aluminium cases?
From approved sample and confirmed artwork, our standard production lead time is 35–45 working days for aluminium cases. This covers extrusion, machining, anodising, and assembly. Tins with standard constructions run 25–30 working days. Both assume tooling is already made — if new tooling is required, add 15–18 working days to the front end.
Planning a packaging project? Contact our team to request a complimentary specification review and sample quote.
Ran into exactly this with an aluminium case supplier in Foshan — their COA listed “wall thickness: 0.8mm per drawing” with no measurement method, no sampling frequency, nothing. When we did incoming CMM checks on the first production lot, we had units ranging from 0.74mm to 0.89mm in the same batch, which killed the anodising colour consistency across the run. They didn’t think wall variation was a qualification parameter at all, genuinely surprised we held the lot.
The anodising colour shift point hit close to home. We had a 20,000-unit run of hinged aluminium cases for a praline gifting line, and about 8% came back from the retailer with a visible banding effect under the store lighting — a warmer gold stripe running roughly through the middle third of the lid. Took us two weeks to figure out it wasn’t a bath chemistry issue at all; wall thickness on those lids was varying 0.08mm side-to-centre because the supplier had switched aluminium coil source mid-run without flagging it. COA listed alloy grade, looked fine, said nothing about thickness tolerance or measurement method.
The anodising point is relevant to something we ran into switching from virgin 1050-H14 to a 60% post-consumer recycled alloy on a lip balm case line — wall thickness variance across the recycled billets was consistently wider (we were seeing ±0.08–0.09mm batch-to-batch) and the anodised colour shift under directional lighting was exactly the kind of defect that gets caught at retail, not incoming QC.
The anodising colour shift point is real — we actually saw this on a 50ml hip flask case run out of Guangdong, where a 0.08mm wall variance across the batch produced a visibly warmer gold tone on thinner sections after Type II anodising. Took us two production cycles to isolate it because the COA was showing the batch as conforming.
Our incoming AQL cycle on aluminium cases runs fortnightly against a fixed sample size of 32 units per lot, and we’ve had to hold three consecutive lots from a Shenzhen supplier last Q3 because wall thickness kept drifting past ±0.07mm around week 6 of a 12-week run — not at intake, which is what made it hard to catch. The COA was clean every time, which is exactly the problem the article’s describing.
Pushed back on a Ningbo tinplate tin supplier last year over emboss consistency on a 35,000-unit run of 100ml spirit miniature tins — their emboss depth was wandering by nearly 0.3mm across the batch and they kept attributing it to die wear. Took us pulling their incoming coil gauge logs to show it was actually panel thickness variance, running about ±0.035mm against a T-3 temper spec that should’ve held tighter than that. Once they adjusted coil acceptance criteria at their blanking stage the emboss stabilised, but we lost six weeks of production schedule getting there.
The ±0.05mm hold trigger on aluminium cases works cleanly for extrusion-formed walls, but on deep-drawn cases — we run a 58mm diameter candle tin case out of a Dongguan supplier — the springback variation post-draw means you’ll routinely see localised deviation at the shoulder radius that doesn’t reflect actual functional wall integrity. We had to add a zone exclusion to our inspection protocol so that shoulder measurements within 3mm of the draw radius don’t trigger the hold logic, otherwise we were rejecting good stock.
We ran into the tinplate gauge issue specifically on closure gap — 0.23mm nominal T-3 temper panels from a Tianjin supplier drifted to 0.27mm actual on about 15% of units in a 28,000-unit 50ml spirit tin run, and that 0.04mm gain was enough to push lid-to-body gap past 0.4mm on the shoulder radius where tolerances stack. Took three months to get the supplier to accept that their gauge variance was the root cause rather than our tooling.
On the GB/T 2520 T-temper range — do the ±0.01mm limits at T-2.5 actually hold in practice for embossed lid panels, or does the work hardening from the emboss operation itself push effective gauge variation outside spec before you even get to the seating torque issue?
One thing we added to our supplier onboarding pack after a bad 40,000-unit run of aluminium lip cases out of Zhuhai: we now require measurement method callout on the COA — specifically whether wall thickness was taken by micrometer or ultrasonic gauge, because we’ve had suppliers switch methods mid-production and the delta between readings on the same unit can be up to 0.04mm depending on surface finish and operator technique.
Tinplate emboss consistency being gauge-dependent is something we didn’t fully account for until a 0.22mm nominal panel from a Wuxi supplier came in consistently at 0.25mm and our emboss dies were cutting too shallow across the entire 18,000-unit run — had to reset tonnage mid-production.