A first article inspection is the cheapest argument you will ever have with a supplier
A first article inspection is the formal, documented check of the first part off a production-intent process, measured against every characteristic on the drawing and recorded before any more parts are made. It is not a sample you eyeball, and it is not a favour a supplier does you. It is the moment a drawing stops being a description of intent and becomes a description of a machine, a fixture, a programme and an operator — and it is the document that tells you whether those four things agree with each other.
Most buyers meet this document at the worst possible time: after a container has shipped, or after an assembly line has stopped. The parts looked right. The packaging was right. The certificate in the envelope said "inspection passed". Then the mating part would not go on, or the anodised batch came in two shades, or the thread gauge sat one turn proud. Every one of those failures was visible on paper weeks earlier, in a document nobody asked for.
Three ordering mistakes cause most first article failures, and none of them start in the machine shop. The first is that nobody ordered a report at all — the purchase order says "parts per drawing" and stops there, so the supplier makes parts per drawing and ships them. The second is that one was ordered but nobody said which standard it should follow, so you receive a nineteen-page PDF in a format your quality team cannot file. The third is that the package arrived, someone skimmed the first page, saw a column of numbers inside tolerance, and filed it — without noticing that the report measured 40 of the 61 dimensions on the drawing, or that the material certificate described a different heat.
This is written for whoever has to explain that failure to their own team. It sets out what belongs in the report, how the four main document families differ, what each process family generates, how much calendar time documentation adds, and the eight things to look for when a package lands in your inbox.
What a first article inspection actually proves — and what it does not
A first article inspection proves that one specific configuration of machines, tooling, programme and material produced one part that met the drawing on one day. That is a narrow claim. Most first article disputes start in the gap between what buyers assume it proves and what it actually says.
Location tolerances follow different rules from size tolerances: our true position guide walks through the datum frame, projected zones and what we can realistically hold.
What it does establish
It establishes that the process *can* make the part. It confirms the programme is correct, the fixture locates the part the way the datum structure assumes, the tool offsets produce the intended geometry, and the raw material meets the specified grade. It confirms the inspection method itself is capable of resolving the tolerances the drawing can reasonably call out.
It also forces a conversation that no drawing can force. To balloon a drawing, someone has to decide which of the 61 dimensions is characteristic 1 and which is 61. That ordering is a statement about what matters, and getting it wrong is how a supplier spends two days measuring cosmetic edges while a true-position callout goes unmeasured.
Three things a first article never proves
What it does not establish is narrower than most people assume. Future parts are a separate exercise: capability is measured over a run, which is exactly why PPAP asks for capability studies on top of the dimensional report. Surface finish on batch two is an anodising problem — bath age and alloy lot, not first-part geometry — and surface treatment records are what carry that evidence. Packaging, labelling and count accuracy sit outside its scope. And whether the part can still be made in nine months, after the tooling has been pulled, serviced and re-set, is a question the report was never asked.
One German audio customer learned this the expensive way — in 2024, on a control knob bore, eight months and one repeat order later. Their first article package was clean: full dimensional report, material certificate, correct surface roughness figures. The repeat order came off the same fixture with a 0.04 mm shift on a bore that set the depth of the knob. Nothing in the original report was wrong; the report had simply described a moment, and the moment had passed. The fix was not a better report. It was a rule: any job that has been off the machine long enough for tooling to be re-set gets the affected characteristics re-measured before the next shipment, and that line goes into the quote.
When a new first article inspection is required
A new first article inspection is required when the process changes, not when the clock runs: a new part number or form-fit-function revision, a new manufacturing source or machine cell, a new material supplier or heat lot, a process or tooling change, a lapse in production, or a corrective action. Quantity alone never triggers one.
- A new part number, or a revision that changes form, fit or function.
- A new manufacturing source, including a move of the same part to a different machine cell inside the same factory.
- A change of material supplier or material lot, where the certificate cannot be traced to the same melt.
- A change of process — for example moving a feature from milling to a second-operation fixture, or changing a heat-treatment route.
- A tooling change or major refurbishment that alters how the part is located.
- A lapse in production, commonly treated as two years, after which most schemes require the first article to be repeated.
- A corrective action that changes how a characteristic is made, so the old report no longer describes reality.
Quantity is not on that list. Making 500 parts does not trigger a second one. Making 500 parts on a different machine does.
What goes into a first article inspection report
Strip away the standard-specific formatting and every credible first article inspection report — often called a FAIR — contains the same five blocks. If one is missing, the report is incomplete, whatever letterhead it arrives on.
| Block | What it contains | What goes wrong |
|---|---|---|
| Part identification | Part number, revision, drawing number, quantity submitted, date, supplier and sub-tier sources | Revision mismatch — report says rev C, drawing is rev D |
| Ballooned drawing | Every drawing characteristic numbered, cross-referenced to the results | Characteristics ballooned but never measured, or measured but never ballooned |
| Dimensional results | Nominal, tolerance, measured value and pass/fail per characteristic | Rounded values; no actual reading, only "OK"; no measurement uncertainty noted |
| Material and special process evidence | Mill certificate, heat-treat record, plating or anodising record, coating thickness | Certificate for the right grade but the wrong heat lot |
| Sign-off | Name and function of the person accepting the results, and the date | Unsigned report, or signed by the person who made the part |
The ballooned drawing is the index, not a decoration
A ballooned drawing is the drawing with every characteristic assigned a numbered balloon, usually in a sequence that follows the drawing's own dimension order or a numbering convention the customer specifies. Its job is indexation: characteristic 27 in the results table must mean the same thing to the person who measured it and the person reading it three weeks later.
Two failure modes dominate. The first is partial ballooning — the supplier balloons the 30 dimensions that are easy to reach with calipers and quietly omits the geometric tolerances, the thread class callouts and the surface roughness symbols. The numbers will all pass. The part will still fail assembly. The second is balloon drift between revisions: the drawing is revised, the balloons are not renumbered, and the results table now refers to a geometry that no longer exists.
When the ballooned drawing and the results table arrive as separate files, match them yourself: the highest balloon number should equal the number of result rows.
Dimensional results: what a CMM inspection report must contain
A dimensional inspection report — a CMM inspection report, when the values come off a coordinate measuring machine — gives you nominal, upper and lower limit, actual deviation and the pass/fail decision per characteristic, plus the machine, programme reference and operator. That is what you want for geometric tolerances: true position, concentricity, profile. Those cannot be resolved with hand tools.
Holding it once is one thing; holding it across a run is another, and our control plan and capability study guide sets out the four indices and what each one actually tells you.
How each of those features gets measured is set out in our CNC inspection methods comparison, including what ISO 10360-2 actually certifies on a CMM.
The report should also tell you about measurement capability, and this is where a lot of packages quietly fail. The usual rule of thumb is that the measuring system should resolve no more than about a tenth of the tolerance band; strict aerospace reviewers ask for closer to a tenth of that. So if a supplier reports a pass on a ±0.01 mm dimension, ask what measured it — the answer determines whether the pass means anything.
| Tolerance band | Instrument that can legitimately judge it | Instrument that cannot |
|---|---|---|
| ±0.05 mm and wider | Caliper, height gauge | — |
| ±0.02 mm to ±0.05 mm | Micrometer, bore gauge | Caliper |
| ±0.005 mm to ±0.02 mm | CMM, air gauge, comparator | Micrometer alone |
| Geometric: position, profile, concentricity | CMM with a stated programme | Any hand tool |
| Surface texture Ra / Rz | Profilometer, with cut-off and evaluation length stated | Visual comparison blocks |
The same discipline applies on surfaces: a surface roughness figure is meaningless without the cut-off, the evaluation length and the direction of measurement, which is why we treat surface roughness specification as part of the drawing rather than a finishing preference.
Material evidence: EN 10204, and the heat-lot problem
Material traceability is where a technically competent package most often falls apart. EN 10204 (*Metallic products — Types of inspection documents*) defines the types of inspection document a mill can issue: type 3.1 means the mill's own inspection representative — employed by the mill, but required to be independent of the production department — declared the results; type 3.2 adds an independent third party or the customer's representative. A 3.1 certificate is normal and adequate for most commercial work. What is *not* adequate is a certificate for the correct alloy that cannot be tied to the bar stock the parts came from.
Where a UDI mark is involved, the first article also has to confirm the code grade on the specified surface: UDI direct part marking on machined medical parts.
Keep the two documents separate when you file them. The mill certificate certifies the material; a certificate of conformity certifies that the part was made to the drawing. A CoC that cites the mill certificate number links the two. A CoC with no certificate reference behind it is a statement, not evidence.
Practical check: the certificate should carry a heat or lot number that also appears on the supplier's own traveller, and ideally on the part itself — see traceability marking. The quantity on the certificate should plausibly cover the quantity of parts: a certificate covering 4 tonnes of 6061 does not demonstrate that your 200 housings came from it. Material choice drives all of this, and our machining materials guide sets out which grades are normally stocked with certification.
Choosing between part marking methods comes down to depth, substrate and coating. Our part marking methods comparison sets out all six with production numbers.
Special processes, coatings and substance declarations
Anodising, plating, heat treatment and painting are special processes: you cannot verify the result by looking at the finished part, so the evidence has to be the process record. For an anodised aluminium part that means the bath record, a dye-uptake or seal result where specified, and a coating thickness reading; for a heat-treated component it means the furnace chart and hardness results, not just a hardness number.
Alongside it sits the substance paperwork: RoHS and REACH declarations, and where relevant conflict-minerals statements. These are not part of the dimensional first article inspection, but they arrive in the same envelope and they are checked by a different team, so we send them as a separate, clearly named section.
FAI vs PPAP: what actually changes
This is the question that produces the most confusion, because the two are not alternatives. FAI is an inspection activity; PPAP is an approval package that contains an inspection activity. A first article inspection asks "did this part meet the drawing?" A PPAP submission asks "can this process keep meeting the drawing, at rate, under a controlled plan, and here is the evidence for every element you asked for?"
The comparison that matters
| First article inspection (FAI) | PPAP | |
|---|---|---|
| Owner of the standard | Aerospace origin (AS9102), widely borrowed | Automotive, AIAG, within IATF 16949 supply chains |
| Question it answers | Does this one part meet every characteristic? | Is this production process approved to run at rate? |
| Typical trigger | First part, new revision, new source | New or changed part before series production release |
| Core artefact | Ballooned drawing plus dimensional results | Dimensional results plus process evidence and a signed warrant |
| Process evidence | Not required | Control plan, PFMEA, capability study, measurement system analysis |
| Sign-off | Inspector or quality representative | Part Submission Warrant, signed by the supplier and received by the customer |
| Used outside its home industry | Very often, as a generic term | Almost always inside automotive and its Tier 2/3 chain |
The practical consequence: if you are buying a machined bracket for a non-automotive product and you ask for "PPAP", you will either receive an automotive package you do not need, or you will be told it does not apply. If you are buying for a carmaker's Tier 1, a bare FAI will be rejected.
The five PPAP submission levels
What is PPAP Level 3? PPAP Level 3 is the warrant plus samples plus complete supporting data, and it is the default requirement for a new part entering a series programme.
Ninety readings settle it. Our repeatability and reproducibility guide shows what goes into the study and what we hand back with it.
A submission level does not describe how much work the supplier does. It describes how much of that work has to be sent to the customer. The supplier still has to own all of it.
| Level | What is submitted | What it is used for |
|---|---|---|
| Level 1 | Part Submission Warrant only, plus an appearance approval report where applicable | Low-risk repeat parts; customer already holds the data |
| Level 2 | Warrant, samples, and limited supporting data | Common for minor changes to an approved part |
| Level 3 | Warrant, samples, and complete supporting data | The default for a new part into a series programme |
| Level 4 | Warrant plus whatever else the customer specifies | Customer-specific; often a subset with named additions |
| Level 5 | Warrant, samples and complete supporting data, reviewed on site at the supplier's plant | New sources, new technology, or after a serious quality failure |
PPAP Level 3 is the one buyers are usually thinking of when they say "send PPAP". It is also the one that takes calendar time, because complete supporting data means the control plan, the process flow, the PFMEA, the measurement system study and the capability study have to exist in writing before anything is submitted — and in an IATF 16949 environment those documents were being written during process design, not after the first part came off. For automotive work, this is the milestone documentation map most programmes follow.
What PPAP adds beyond a dimensional report
Four things, roughly in order of how often they are missing:
- A control plan that names each characteristic, the control method, the sample size and the frequency. Without it, "inspection" is whatever the operator did that morning.
- A capability study on the characteristics the customer designates, usually expressed as Ppk or Cpk. The AIAG PPAP manual does state acceptance criteria — commonly 1.67 for special characteristics and 1.33 more generally — but customers override them routinely. Write your number into your own specification rather than accepting a supplier's default.
- Measurement system analysis, typically a gauge repeatability and reproducibility study. Under 10% passes, 10–30% is conditional depending on the importance of the characteristic, and above 30% is a rejection; the count of distinct categories should be at least five. Whether the percentage is taken against the tolerance or against total study variation depends on the method, so state which one you mean.
- A signed warrant in which the supplier states that the submission is representative of the production process. That sentence is the legal heart of the package.
For a medical device buyer the equivalent question is not PPAP but whether the supplier's quality system is appropriate to the device class, which is a different conversation — see our notes on ISO 13485 machining and on medical device machining.
ISIR and EMPB: the German VDA route
Germany runs a third system again. If your drawing comes from a German or Austrian OEM, the document you are asked for is probably not a FAI report and not a PPAP package. It is an ISIR — an Initial Sample Inspection Report — known in German as an EMPB, *Erstmusterprüfbericht*.
What an ISIR contains
Structurally it is close to a first article inspection report: the part is identified, every drawing characteristic is numbered and ballooned, each is measured and reported with nominal, tolerance and actual value, and material and special-process evidence is attached. The differences are in the German automotive convention. The report form is the one prescribed by the customer or by VDA Volume 2, *Sicherung der Qualität von Lieferungen*, which is the German industry's procedure for assuring the quality of supplier deliveries. The characteristic numbering typically follows a strict drawing-order convention. And the cover sheet carries a formal sample-status decision rather than a simple pass/fail.
The EMPB is also stricter about two things the English-language world is relaxed about. It expects the sample to come from series-intent tooling and series-intent process settings — a part made on a prototype fixture is not a valid sample. And it expects measurement results to be reported as actual values, with the measuring equipment named, not as "within tolerance".
FAI, PPAP and ISIR side by side
| FAI / AS9102 | PPAP | ISIR / EMPB | |
|---|---|---|---|
| Origin | Aerospace | AIAG / automotive, North America and its supply chain | VDA / German automotive |
| Document shape | Three forms: part accountability, product accountability, characteristic accountability | Warrant plus around eighteen named elements | Cover sheet plus ballooned drawing, results and evidence |
| Process evidence | Minimal | Central — control plan, PFMEA, capability, MSA | Present, but lighter than PPAP |
| Typical buyer | Aerospace primes, and anyone borrowing the format | Carmakers and their Tier 1/2 chain | German and Central European OEMs and Tier 1s |
| What it is called | First article inspection report | Production part approval process submission | Initial sample inspection report / Erstmusterprüfbericht |
How to tell which one you are being asked for
Read the purchase order and the supplier quality requirements document together. Three words usually settle it. If the document names AS9102 or mentions forms, it is FAI. If it names a submission level or a warrant, it is PPAP. If it names *Erstmuster*, *Erstmusterprüfbericht*, EMPB or *Bemusterung*, it is ISIR. Where a customer operates in both worlds they will often accept one format and ask for named additions, which is fine provided the additions are written down before the parts are made.
This happens in both directions. A buyer requests "full PPAP" for a German programme and receives a file the receiving quality team cannot process; or a buyer asks for a FAI where the customer's portal expects an EMPB. One sentence in the PO — "initial sample report to [named form], revision [x], including material certificate EN 10204 3.1" — removes the ambiguity entirely.
The AS9102 form structure, in plain terms
Even outside aerospace, AS9102 (currently Revision C) is the most widely copied first article inspection format, so it is worth knowing how it is built. It uses three forms, and the split is logical.
Form 1 — Part Number Accountability
Who made what: part number, revision, the drawing it was made to, the serial number of the actual sample part — usually a laser-marked serial number rather than a handwritten tag — the supplier, and any sub-tier sources that performed work on it. This is the form that catches the "the report describes a different revision" failure.
Form 2 — Product Accountability
What the part is made of and what was done to it: material and specification, special processes and their specifications, functional testing, and the sub-tier suppliers responsible for each. This is where a missing plating record or an untraceable heat lot shows up.
Form 3 — Characteristic Accountability
The measurements, formally the Characteristic Accountability, Verification and Compatibility Evaluation: every ballooned characteristic with its design requirement, the measured result and the pass/fail decision. This is the long one, and it is the one people actually read.
A word on certification, because it causes real damage: AS9102 is a reporting standard, not a certifiable management system. No organisation is "AS9102 certified" in the way it can be ISO 9001 certified. A supplier can hold AS9100 registration for its quality system and *issue an AS9102-format report*; those are two different claims, and a supplier that advertises the first as the second either misunderstands its own certificates or is hoping you do.
Our own position is narrow and specific. We hold IATF 16949 — see also our IATF 16949 guide — together with ISO 9001 and ISO 13485, and we issue dimensional first article reports and PPAP documentation, built to whichever AS9102, ISIR or customer-specific template the programme names. We run 200+ CNC machines across a plant that has been doing this for more than 23 years, hold ±0.01 mm as our standard machining tolerance, and turn prototypes in 3–7 days — which is why the documentation, not the machining, is usually what sets the date. We do not hold AS9100 registration. Where a programme requires it, we tell you at quotation, so you can qualify another source rather than discover it at audit.
How to read an FAI package before you sign it off
Eight checks, in this order. Most bad packages fail at least two of them.
- Revision match — part number and revision on the report match the drawing you released.
- Balloon count — the highest balloon number equals the number of result rows.
- GD&T coverage — every geometric tolerance on the drawing appears as a result.
- Actual values — real numbers, not "OK" or "pass".
- Heat-lot traceability — the certificate lot number ties back to the traveller.
- Special process records — anodising, plating or heat-treat records attached, with parameters.
- Measuring equipment named — CMM, gauge or instrument identified, with resolution.
- Signature and date — a named person, their function, and the date.
| Check | What you are looking for | Why it matters |
|---|---|---|
| 1. Revision match | Part number and revision on the report match the drawing you released | The single most common defect; invalidates everything below it |
| 2. Balloon count | Highest balloon number equals the number of result rows | Catches characteristics that were ballooned but never measured |
| 3. Coverage of GD&T | Every geometric tolerance on the drawing appears as a result | Hand-tool reports routinely skip true position and profile |
| 4. Actual values | Real numbers, not "OK" or "pass" | Without values you cannot judge how close to the limit you are |
| 5. Heat-lot traceability | Certificate lot number ties to the traveller | Catches a right-alloy, wrong-batch certificate |
| 6. Special process records | Anodising, plating or heat-treat record attached, with parameters | These results cannot be verified on the finished part |
| 7. Measuring equipment named | CMM, gauge or instrument identified, with resolution | A pass means nothing if the gauge cannot resolve the tolerance |
| 8. Signature and date | Named person, function, date | An unsigned report has no owner |
Both of those checks are really watching for the same two patterns. The first is selective measurement: everything reported passes, but the set of things reported is smaller than the set of things specified. The second is substitution: the report describes a part that is similar to yours but not identical — an earlier revision, a different finish, a prototype route. Both are invisible if you only read the summary page.
It is also worth reading the out-of-tolerance rows with interest rather than alarm. A characteristic that misses by a micron and is accompanied by a written concession request tells you the supplier found it, understood it and told you. On a complex part, a package with zero deviations deserves a second look: check the coverage before you celebrate.
What the package looks like for each process
The document set differs by process because the failure modes differ. Stamping packages fail on material traceability. Sheet metal packages fail on accumulated dimensions. Machined packages fail on geometry. Surface-treated packages fail on colour.
CNC turning and milling, including five-axis work
The dominant risk is geometric: true position of a hole pattern, concentricity between a bore and an outside diameter, profile of a machined face, and features that only exist after a second setup. A useful package balloons the GD&T explicitly, reports the CMM programme reference, and states which setup each group of characteristics came from.
On a five-axis part the report should say whether the part was measured in the as-machined orientation or re-fixtured, because a five-axis part measured after removal can show a datum shift that is an artefact of measurement rather than a machining error. For turned parts, thread class deserves its own line: a thread gauge pass is not the same as a reported pitch diameter. Practical notes are on our CNC turning and CNC milling pages.
Progressive and transfer die stamping
Here the report is about the die, not the part. What you want: the die identification, the station the sample came from, the strip layout reference, burr height and direction on cut edges, and — critically — a note on where in the coil or strip the sample sat. A first-piece sample cut from the lead end of a coil is not representative. Our progressive die design notes explain why station order matters.
Material evidence dominates. Stamping runs on coil, and coil certification is per heat and per coil, so a progressive die programme backed by a single certificate covering six months of production is not traceable. Ask for plating or coating thickness readings on the actual sample parts, not on a test panel, because rack position changes deposit thickness. Our stamping documentation as standard page sets out what ships with a new die, and the same logic applies to transfer die parts.
Sheet metal fabrication
The characteristic that matters is almost never a single dimension; it is an accumulated one. A folded chassis with six bends accumulates error across the whole stack, so the report should give the overall envelope dimension as its own characteristic, not just the individual leg lengths. Hole-to-hole across a bend is the second one to insist on, and bend allowance is where that error originates.
Finish is the other half. Powder coat thickness, colour reference and gloss reading should appear as results, since sheet metal work is usually a cosmetic part and the eye is the final inspector. Tolerance expectations for this process are set out separately in our sheet metal tolerance guide.
Surface treatment
Anodising, plating and coating produce a record, not a measurement. Expect bath or bath-line identification, thickness readings at named locations, and for anodising a seal-quality or dye-uptake result where the specification calls for it; process detail is on our surface treatment page. Colour is the hard one: it is judged against a reference chip under stated lighting, and calling it a number would be dishonest. A colour match is an agreement — a signed reference chip, judged under stated lighting. Ask for the chip with the first article, not after it.
Four processes, four different failure modes, and four different document sets.
How much calendar time documentation adds
Documentation adds calendar time, not machining time, and the two are routinely confused in programme planning. The parts may be ready on Thursday and the package on the following Wednesday.
The sequence, once the first part exists: balloon the drawing, run the measurement, compile results, obtain and attach material certificates, obtain sub-tier special process records, internal review, sign-off, and issue. Internal review is the step that gets squeezed and should not be, because it is where a revision mismatch gets caught.
Three ways to compress it, all of them yours:
- Name the format in the PO. A named form removes a round of clarification.
- Say which characteristics matter. If you designate the twenty critical characteristics up front, the ballooning is faster and the report is shorter. Those same twenty are cheapest to argue about during a DFM review before the drawing is frozen.
- Ask for the material certificate with the order, not with the parts. The mill is not the delay. Realising on Thursday that nobody asked for the certificate is the delay.
Where a programme has a hard launch date, the cheapest accommodation is a partial first article: full dimensional results on the critical characteristics, with the remainder to follow. That is a legitimate arrangement provided it is written down as a partial, with a date for the balance.
First article inspection checklist
Use this first article inspection checklist as the basis of a PO clause or an internal receiving procedure. Copy it into your own template.
| # | Item | Done |
|---|---|---|
| 1 | Part number and revision on the report match the released drawing | ☐ |
| 2 | Report format named in the PO (FAI / AS9102 / PPAP level / ISIR) | ☐ |
| 3 | Ballooned drawing supplied and matched to the results table | ☐ |
| 4 | Every GD&T callout appears as a measured characteristic | ☐ |
| 5 | Actual values reported, not pass/fail only | ☐ |
| 6 | Measuring equipment and resolution stated | ☐ |
| 7 | Material certificate EN 10204 3.1 (or 3.2 where required), heat lot traced | ☐ |
| 8 | Special process records attached — heat treat, plating, anodising, coating thickness | ☐ |
| 9 | Sub-tier suppliers identified where work was subcontracted | ☐ |
| 10 | Sample came from series-intent tooling and process settings | ☐ |
| 11 | RoHS / REACH declarations included where applicable | ☐ |
| 12 | Deviations, if any, listed with a written concession request | ☐ |
| 13 | Named signatory, function and date | ☐ |
| 14 | Report retained as the baseline for future drift comparison | ☐ |
Write items 2, 7 and 11 into the purchase order rather than asking for them afterwards. After the parts are packed, all fourteen become requests instead of terms, and a request can be refused.
A clause you can paste:
> Supplier shall submit a first article inspection report in [named format, revision x], built to drawing revision [y], covering all dimensions, geometric tolerances and notes. Submission shall include a ballooned drawing matched to the results table, actual measured values, the measuring equipment used, an EN 10204 3.1 material certificate traceable to the heat lot, records of all special processes including heat treatment and coating thickness, and a dated sign-off by a named person. Deviations shall be listed with a written concession request.
Frequently asked questions
What is a first article inspection?
The check of the first part off a production-intent process against every characteristic on the drawing, recorded before more parts are made. The two words that do the work are *first* and *production-intent*: a prototype part or a hand-finished sample does not qualify.
What documents are included in a first article inspection report?
Five blocks: part identification with revision, a ballooned drawing, dimensional results with actual values, material and special-process evidence, and a dated sign-off. RoHS and REACH declarations travel in the same envelope but belong to a different review — see the table above.
What is the difference between FAI and PPAP?
FAI asks whether one part meets the drawing. PPAP asks whether the production process is approved to run at rate, and it contains an FAI-style dimensional report plus process evidence: a control plan, a process flow, a PFMEA, a measurement system study and a capability study, closed out with a signed Part Submission Warrant. PPAP applies inside IATF 16949 automotive supply chains; FAI, in its AS9102 form, originated in aerospace and is now used generically.
What are the five PPAP submission levels?
Level 1 submits the warrant only. Level 2 adds samples and limited supporting data. Level 3 adds samples and complete supporting data, and is the usual requirement for a new part. Level 4 submits the warrant plus whatever else the customer specifies. Level 5 is Level 3 content reviewed on site at the supplier's plant. The level governs what is sent to the customer, not how much work the supplier performs.
Is ISIR the same as PPAP?
No. ISIR — the Initial Sample Inspection Report, or *Erstmusterprüfbericht* in German — is the VDA-format initial sample document used by German and Central European OEMs. It is structurally closer to a first article inspection report than to a PPAP package: it centres on the ballooned drawing and measured results, with material and process evidence attached, and it carries a formal sample-status decision rather than a process approval warrant. A customer operating in both systems will usually accept one format with named additions.
Do I need a ballooned drawing for a first article inspection?
Yes, in practice. The ballooned drawing is the index that makes the results table readable and auditable: without it, a list of 60 measurements cannot be tied back to specific drawing features. Check that the highest balloon number matches the number of result rows; a mismatch means something was ballooned and not measured.
How many parts should be measured in a first article inspection?
One part, fully measured — every characteristic, and every instance of a grouped feature such as "4× holes". Where a customer requires statistical evidence the number rises, commonly three to five pieces for an initial study and more where a formal capability study is specified. For a capability study the parts should be taken across the run rather than consecutively, so that tool wear and setup drift are represented.
Who pays for a first article inspection — the buyer or the supplier?
That is a commercial term, not a technical one, and it should be settled in writing before the parts are made. A full PPAP Level 3 submission with capability studies involves real engineering hours, and third-party inspection or any testing performed outside the supplier's own laboratory is normally billed separately.
What to do on Monday
Pick one live part number and write four sentences into the next purchase order: the report format you expect, the revision it must be built to, the critical characteristics you want ballooned first, and the evidence you require alongside the dimensional results — material certificate type, coating or heat-treat records, substance declarations. Do that and most of the failure modes in this guide never reach you.
If you are at the stage of choosing a supplier rather than qualifying one, the document question belongs in the first conversation, not the last; our notes on choosing a CNC supplier and on supplier evaluation at prototype stage cover that earlier step. And if you would rather hand the whole thing over, send us the drawing, the revision and the format you need — we will quote within 24 hours, and we will tell you in writing which documents come with the parts. Our certifications and quality system and full service list are linked for reference.