Published 8 October 2026 · 7 minute read
In many AS9100 plants the first article lives in a binder in the quality office, the calibration register lives in a spreadsheet on one person's laptop, and the production line knows about neither. Each is done properly on its own. The gap is between them: a line can start a new revision with no approved first article, and a lab result can be taken with a micrometer that went out of calibration last week. Nobody notices until an audit, or a customer escape.
AS9100D clause 8.5.1.3 calls for production process verification, typically a first article inspection to AS9102, before production of a new part or revision. Clause 7.1.5 covers monitoring and measuring resources, including calibration. This article is about connecting both to the line so they act as gates rather than as filing.
AS9102 defines three forms:
A full first article is performed for a new part, and a partial or delta first article after a change to the design, the process, the tooling, the location or the source of material. The trigger list matters: most plants are disciplined about the first one and loose about the rest.
On a composite line the "part" is often a material product rather than a machined drawing, so the characteristics are things like areal weight, resin content and volatile content, made to a recipe and revision. In machining it is a part number and revision with dimensional characteristics. The structure is the same.
On paper, Form 3 has a column for the measured value and a column for "conforms: yes/no", and the same person fills in both. That is where errors hide. A value transcribed as 0.512 against a maximum of 0.510 gets a tick because the inspector was reading the next line.
The better arrangement is that the inspector enters the requirement once, from the drawing or specification, and the measured value from the instrument; pass or fail is computed from the numbers. The approver then reviews numbers, not ticks, and any non-conforming characteristic is impossible to miss because the system marks it, not the inspector.
A first article is a verification. Verifying your own work is not a verification. The engineer or inspector who prepared the first article submits it; a different, authorised person (usually Quality) reviews and approves it with an electronic signature. The software should simply refuse to let the preparer approve, rather than rely on a procedure that says they should not. The same separation of duties runs through the rest of a release, as covered in who can release a nonconforming roll.
An approved first article is only useful if production depends on it. Two connections make that happen:
With both in place, "did we do the first article for revision C?" stops being a question for the binder. Product made to revision C simply cannot leave without it.
Clause 7.1.5 expects instruments used to verify product to be calibrated or verified at intervals, identified so their status is known, and protected from adjustments that would invalidate the result. When an instrument is found out of tolerance, the plant must assess the validity of earlier results taken with it.
A calibration register should hold, for each instrument:
The as-found reading is the one that matters for the past. If a gauge was found out of tolerance, the as-found value tells you how wrong the last interval's results could have been, and which product needs reassessment. A register that only records "calibrated, passed" loses that.
This is where most plants stop short. The register exists, the due list is accurate, and still nothing prevents an inspector from picking up an overdue gauge, because the result sheet does not know which gauge was used.
The fix is to make the instrument part of every result. When a lab result or inspection value is entered, the instrument is chosen from the register, and the register decides what can be chosen: an instrument that is overdue, or whose latest calibration has not been approved, does not appear as a choice. The result stores the instrument and its calibration status at the time of the test, and the certificate of conformance can list it.
Two benefits follow. An uncalibrated instrument cannot be recorded against product in the first place. And because every result names its instrument, when an instrument is later found out of tolerance the results it touched can be found by instrument, which turns the clause 7.1.5 reassessment from an archaeology project into a query.
For a composite line, the first article is per recipe revision and the instruments are the lab's balances and test equipment. For machining, it is per part number and revision, with CMMs, micrometers and gauges in the register. For special processes such as heat treatment, plating and NDT, calibrated sensors and test equipment feed a record per furnace or bath load. For assembly, torque tools and test stands are typical. What changes is the instrument list and the characteristics; the gates are the same. The AS9100 manufacturing overview shows how they fit into one flow.
In PulseMQ, first article inspections sit on the line's Quality tab. Forms 1 to 3 are filled in on screen; pass or fail is computed from the requirement and the measured values; the preparer submits and a different Quality user approves with a signature (password, authenticator code and reason). A recipe can be flagged as first article required, and then a unit made to that revision cannot be released without an approved first article for it.
The calibration screen holds the instrument register and the due list. A calibration is recorded with as-found and as-left readings and approved by a second person. Lab result entry reads the register: an instrument not in calibration cannot be chosen, and every result stores the instrument and its status at the time. Those results land on the unit's record, described in roll-level traceability for composite lines, and from there on the certificate. For broader context on lot genealogy see batch tracking and traceability; for comparing process data against a known good run, see what is a golden run.
PulseMQ keeps the evidence for clauses 8.5.1.3 and 7.1.5. It does not replace the AS9102 standard, certify a plant or make it compliant; how first articles and calibration are set up for your parts and instruments is agreed and qualified on your line during a pilot.
Design partner program. Pilot PulseMQ on one AS9100 line, with your first articles and your instrument register connected to real production.
See the design partner programWhat are AS9102 forms 1, 2 and 3?
Form 1 is part number accountability (the part, its revision and any sub-assemblies). Form 2 is product accountability for raw materials, specifications and special processes. Form 3 is characteristic accountability: every design characteristic, its requirement and the measured result.
Why should FAI pass or fail be computed rather than typed?
When the inspector types both the measurement and the verdict, a transcription slip or optimism can produce a pass on an out-of-tolerance value. Computing the verdict from the requirement and the measured value removes that step, and the approver reviews numbers instead of ticks.
What does AS9100D 7.1.5 require for measuring instruments?
Instruments used to verify product must be suitable, calibrated or verified at specified intervals, identified so their status can be determined, and safeguarded. If an instrument is found out of tolerance, the manufacturer must assess the validity of earlier results taken with it.
How do you stop an uncalibrated gauge from being used for an inspection result?
Make the instrument part of the result. If the result entry reads the calibration register and refuses an instrument that is overdue or has no approved calibration, the uncalibrated gauge cannot be recorded against product in the first place.