A first article inspection is not a box-ticking ritual and it is not a favour the supplier does for you. It is the moment a manufacturing process is proven, on production tooling, with production people, to be capable of making your part. Everything after it is just monitoring. Everything before it is hope.
Key takeaways
- FAI approves a process, not a part. A sample made on a different machine is not a valid first article.
- Two frameworks dominate: AS9102 (three forms, aerospace and defence) and AIAG PPAP (elements and five submission levels, automotive).
- Most enclosure buyers need neither in full. A right-sized eight-document package does the job.
- Balloon the drawing once. That single artefact feeds dimensional results in either framework.
- Re-FAI triggers are where programs get caught: design change, process change, source change, long production lapse.
What a first article inspection actually proves
The first article is the first part — or a defined small number of parts — produced on the production process: same machine, same tooling, same fixture, same operators, same inspection method. That is the whole point. If a supplier sends you a sample made on a prototype route, you have learned that somebody can make the part, not that the line can.
This is why the two most common buyer mistakes are both about scope: accepting a prototype as a first article, and accepting a single part as proof. One part tells you one part was good. It tells you nothing about scatter.
AS9102 and PPAP: the same idea, different paperwork
| Aspect | AS9102 (aerospace, defence) | AIAG PPAP (automotive) |
|---|---|---|
| Structure | Three forms: part accountability, product/material accountability, characteristic accountability | A defined element set, submitted at one of five levels |
| Core artefact | Ballooned drawing plus the three forms | Ballooned drawing plus process evidence |
| Statistical capability | Not part of the three-form package; variation control of key characteristics normally arrives as a separate specification or customer flow-down | Required by the customer, with thresholds set by the customer rather than by the framework |
| FMEA and control plan | Not the focus of the package | Central to the submission |
| Approval document | The completed first-article report, signed by the supplier's authorised quality representative | Part submission warrant signed by the supplier and approved by the customer |
| Trigger to redo | Design change, source change, process change, extended production lapse | Change notification per the customer's requirement |
Both frameworks punish carelessness, just different kinds. AS9102 fails you for a missing characteristic on the ballooned drawing. PPAP fails you for missing process evidence.
What to measure on a sheet metal enclosure
Enclosures have a specific set of characteristics that matter, and they are not all dimensional.
| Characteristic group | Examples | Typical method | Note |
|---|---|---|---|
| Overall envelope | Height, width, depth, diagonal | Tape, gauge, CMM | Diagonal catches racking better than the sides do |
| Feature location | Hole patterns, PEM positions, rail pitch | CMM or gauge | Measure after welding and after coating |
| Formed geometry | Bend angles, flange lengths, hem tightness | Protractor, gauge | Post-weld angles differ from press-brake angles |
| Flatness and gap | Door gap uniformity, panel flatness | Feeler gauge, straightedge | See weld distortion control |
| Coating | Dry film thickness, adhesion, colour | DFT gauge, cross-cut | Specify the test and the acceptance number |
| Sealing | IP rating by actual test | Spray or dust test per the declared rating | Claimed and tested are different things |
| Function | Door swing, latch operation, gland plate fit | Fit check | Often skipped, often the field complaint |
| Material identity | Mill certificate to the ordered grade | Certificate review | Lot traceability matters later |
Two things belong on that list that buyers routinely leave off: coating thickness and an actual IP test. Both are cheap at first article and expensive at field failure.
The document package: what to ask for
A right-sized first-article package for an enclosure program:
- Ballooned drawing — every design characteristic numbered, tied to the released revision.
- Dimensional results — each balloon number with nominal, tolerance and actual measured value.
- Material certificates — mill cert for the sheet, plus certificates for purchased hardware.
- Special process certificates — coating, welding qualification, heat treatment if applicable.
- Inspection method list — which instrument measured which characteristic, with calibration status.
- Process flow and control plan — the operations in order and where each characteristic is checked.
- Capability data on the characteristics that matter — on the few the customer names, at the sample size and threshold the customer specifies.
- Sample parts — retained, labelled, and traceable to the report.
If the program is automotive, expand items 6 and 7 into the full element set the customer requires. If it is aerospace or defence, use the three-form structure and keep process documentation out of the package — auditors want the forms clean.
Five things that trigger a re-run
- Design change affecting form, fit or function — including a drawing revision you consider minor.
- Manufacturing source change — new supplier, new site, new sub-tier processor.
- Process or tooling change — new machine, new fixture, new coating line, new welding procedure.
- Extended lapse in production — a part that has not run for a long period needs the process re-proven.
- Corrective action that touched the process — a fix is itself a change.
The lapse trigger is the one that catches people. The tooling may be hanging on the rack unchanged, but the people, the material lot and the machine condition are not what they were.
Seven reasons first-article packages get rejected
| Rejection reason | Why it happens | Fix |
|---|---|---|
| Characteristics not all ballooned | Someone numbered "the important ones" | Balloon every callout. A radius is a characteristic. |
| Wrong drawing revision | The package was built against a revision that has since moved | Freeze the revision and record it on every page |
| Material certificate mismatch | Cert lot does not match the material actually used | Reconcile lot numbers before inspection, not after |
| Expired processor certification | Coating or special-process cert lapsed | Check expiry dates as part of package assembly |
| No statistical evidence on critical characteristics | One measurement recorded where capability was required | Decide up front which characteristics need capability data |
| Unsigned or undated documents | Missing approval block | A missing signature is grounds for full rejection |
| Out-of-spec value with no disposition | A reading outside tolerance recorded honestly but not addressed | Every nonconformance needs a documented disposition |
Notice that most of these are administrative. They are also the cheapest to fix, which is exactly why they are inexcusable.
If you are not automotive and not aerospace
Most enclosure buyers are neither. You still want the discipline, but you do not want to pay for a submission nobody will read. Our recommendation:
- Balloon the drawing. This is the artefact that makes everything else possible.
- Require measured results for every balloon, on a small number of parts from the first production run — not one part, and not a prototype.
- Require material and coating certificates tied to lot numbers.
- Require one functional check: door operation, gland plate fit, latch engagement.
- Require one environmental check: an actual IP test at the declared rating, or a leak test where the rating is not the point.
- Require coating thickness readings at defined locations.
- Ask for a short control plan: which operations, which checks, which instruments.
- Ask for capability data on the two or three characteristics that actually drive assembly.
That is a package a competent fabricator can produce in days, and it closes the failure modes that generate real field cost.
How this fits the rest of the process
FAI sits at the handover between prototype and production, and it depends on the work done before it. If the design has not been through DFM, the first article will pass and the production run will still hurt — see our prototyping to production guide and DFM design guide. If the inspection plan is unclear, read the quality control and inspection guide. If the parts are welded, the dimensional expectations have to be written for a welded assembly, not a folded one — tolerance guide.
The structure you choose also changes what the first article has to prove: a welded shell is judged on post-weld dimensions and seam quality, a modular frame on hole patterns and panel fit. See welded vs modular enclosure construction.
Practically: the dimensions you mark critical on the drawing are the dimensions the first article will be judged on. Choose them deliberately, before you send the RFQ.
Need a first-article package you can actually use?
Tell us the application, the critical characteristics and the markets you ship to. We will build the inspection plan and the documentation package around that, not around a template.
Frequently Asked Questions
What is the difference between FAI and PPAP?
FAI is the practice — prove the production process can make a conforming part. PPAP is one formal framework for doing it, used in automotive, built around a defined element set and submission levels. AS9102 is the other common framework, used in aerospace and defence, built around three forms tied to a ballooned drawing.
How many parts should a first article inspection cover?
More than one. A single part proves one part was good. For general industrial enclosure work, three to five parts from the first production run is a practical starting point. Where the customer requires capability data, the sample size is whatever the customer specifies — that number is a customer requirement, not a framework constant. All parts must come from the production process, not a prototype route.
Can a prototype be used as the first article?
No. If the sample was made on a different machine, with different tooling or by a different process than planned production, it does not prove the production process. Require parts made on the production equipment with the production fixture before approving release.
Do I need the full PPAP or AS9102 package for an enclosure?
Usually not. Most enclosure buyers are neither automotive nor aerospace, and a full submission is disproportionate. A right-sized package — ballooned drawing, dimensional results, material and coating certificates, functional and environmental checks, a short control plan — closes the real failure modes.
What triggers a new first article inspection?
A design change affecting form, fit or function; a change of manufacturing source or site; a change of process, tooling or machine; a long lapse in production; and any corrective action that altered the process.
What is the most common reason a first-article package is rejected?
Administrative gaps. Missing or incomplete ballooning, a drawing revision mismatch, unsigned or undated forms, and material certificates whose lot numbers do not match the parts. These are cheap to prevent and are treated as grounds for full rejection.