First-article inspection and CMM reports on a small EDC batch

Inspection is where small-batch buyers either overspend on paperwork or underspend and find out about a problem when 200 finished parts arrive. The right level for EDC sits in a narrow band, and it is worth knowing where.

What a first article actually is

One part from the beginning of the run, measured against every dimension on the drawing, with the results recorded. Its job is to prove the setup is correct before the rest of the batch is cut.

For EDC work this is the single most valuable inspection you can buy. It is one part’s worth of measuring time, and it catches setup errors, misread drawings and unit mix-ups while the cost of being wrong is one part rather than the whole order.

Ask for it on every new part. Ask for a photograph with it.

What the report should contain

• Every dimension on the drawing, numbered to match a ballooned copy of your drawing
• Nominal, tolerance, actual measured value
• The instrument used for each — caliper, micrometer, CMM, gauge
• Material certificate reference
• Date, and who measured

If the shop returns a report with round numbers matching nominal exactly on every feature, it was not measured. Real data has scatter.

CMM versus hand tools

A coordinate measuring machine is necessary for true position, form and anything under about 0.02 mm. It is not necessary for a thickness or a simple diameter, where a micrometer is faster and just as accurate.

Asking for CMM on everything is a common overspecification. Say which features need it — usually your screw pattern true position and any datum-referenced geometry — and let the rest be measured conventionally. See tolerances and fits for which features those are.

In-process and final inspection

For a 50 to 500 piece EDC run, a sensible structure:

First article: full dimensional, one part, before the run continues.

In-process: the two or three critical dimensions checked at a stated frequency — every tenth part is typical. Enough to catch tool wear drifting a dimension mid-run.

Final: 100 percent visual on every part, because cosmetic defects are the real risk in EDC and no sampling plan catches them. Dimensional sampling on critical features per a stated AQL.

Do not ask for 100 percent dimensional on every feature of every part. On 200 parts with 40 dimensions that is 8,000 measurements, and you will pay for every one.

Visual standards — the hard part

Cosmetic acceptance is where EDC disputes actually happen, and it is the hardest thing to specify. “No scratches” is not a standard; every machined part has something under magnification.

Better approach: agree a physical boundary sample. One part that represents the worst acceptable finish, approved by you, kept at the shop. Everything is judged against it. This is worth more than any written description and it is the thing to establish during the first article stage.

Where that is impractical, specify viewing conditions: “no defect visible to the unaided eye at 300 mm under normal office lighting” is at least testable.

Packaging and the last hundred metres

Half the cosmetic damage on finished EDC parts happens after inspection. Specify individual bagging, interleaving for stacked flat parts, and anti-tarnish for copper alloys. A perfect part that arrives with a rub mark from its neighbour has failed just as completely.

What to write in the RFQ

“First article dimensional report required before production continues, with ballooned drawing. Critical dimensions marked C on drawing to be CMM measured. 100 percent visual against approved boundary sample. In-process check of critical dimensions every 10 parts.”

That is proportionate, it is cheap at EDC volumes, and it puts the conversation about quality before the parts are cut rather than after. Send the drawing and we will mark up what is worth measuring.

Measurement uncertainty, and why two reports disagree

You measure a part at 3.02 mm. The shop’s report says 3.00 mm. Neither of you is lying.

Every measurement carries uncertainty from the instrument, the operator, temperature and the part itself. A hand micrometer in a warm workshop has real uncertainty in the region of several microns. A CMM in a controlled room has far less, but not zero.

Two practical consequences:

A tolerance tighter than roughly four times your measurement uncertainty is not meaningfully verifiable. If you are checking with calipers, a ±0.02 mm callout is beyond what you can confirm.

Agree the measurement method, not just the number. “3.00 ±0.05, measured with a micrometer at the marked location” removes most disputes. Where you measure on a part that is not perfectly parallel matters as much as what you measure with.

Temperature

Metrology standard is 20 degrees Celsius. An aluminium part measured at 32 degrees in a Vietnamese workshop in the afternoon reads measurably larger than the same part at 20 degrees — aluminium expands roughly 23 micrometres per metre per degree.

On a 100 mm part, a 12 degree difference is around 28 micrometres. That is bigger than many EDC tolerances.

For parts with tolerances of ±0.05 mm and looser, this is noise. For anything tighter, specify that critical measurements are taken in a temperature-controlled area after the part has stabilised. A shop with a proper metrology room does this anyway; asking tells you whether it has one.

Ballooned drawings

A ballooned drawing has every dimension numbered in a circle, and the inspection report lists results against those numbers.

Without it, a report is a list of numbers that somebody has to correlate to the drawing by hand, which is slow and error-prone. With it, you can check a first article in five minutes.

Ask for it. Most shops will produce one from your drawing; supplying one yourself guarantees the numbering matches what you expect.

Gauges for production

For repeating production, custom go/no-go gauges are often cheaper than measuring.

A simple plate with pins at your screw pattern positions checks true position in two seconds per part, with no operator judgement. A thickness gauge does the same for a single dimension. For a part you will order repeatedly, the gauge pays for itself quickly and it makes 100 percent checking of a critical feature economically realistic.

This is worth raising at first article stage, once you know the part is right.

Sampling plans

If you specify dimensional sampling, reference a standard rather than inventing one. ISO 2859 (equivalent to the older MIL-STD-105) defines sample sizes and acceptance numbers for a given lot size and AQL.

For EDC hardware, AQL 1.0 or 2.5 on dimensional characteristics and AQL 0.65 on critical cosmetic surfaces is a reasonable starting position. State it; “spot check a few” is not a specification and cannot be enforced.

The boundary sample, in practice

Worth expanding because it is the highest-value thing in this article.

Make two identical boundary samples. One stays with the shop, one stays with you. Both are photographed and dated. Everything produced is judged against it.

When a dispute arises — and on cosmetic parts one eventually will — you are comparing a part to a physical object you both agreed on, not to two different memories of a conversation. It converts an argument into a measurement.

Establish it during the first article stage, before production. Trying to agree one after a batch has been rejected is a much harder conversation.

The first article is where the whole brief meets one real part; the full route is in our guide to custom EDC manufacturing.

FAQ

Should I ask for a first article on a repeat order? Yes, if anything changed: new material lot, new machine, a gap of six months, a revision. For a straight repeat within a few weeks on the same setup, it is usually unnecessary.

What does a first article cost? At EDC scale, typically a modest fixed charge plus the part. It is the cheapest insurance in the whole process.

Can I inspect the parts myself instead? You can, and you should spot check. But inspecting after delivery means a problem is discovered when the whole batch exists, which is exactly the situation first article inspection prevents.

What is a reasonable AQL for cosmetic defects on EDC? Tighter than for dimensional, because cosmetics are what customers see. AQL 0.65 to 1.0 on visible surfaces, with a boundary sample defining what counts as a defect.

Should I pay for a third-party inspection? For a first order with a new supplier at meaningful value, it can be worth it. For a 50-piece batch it usually costs more than the order. A good first article report plus photographs achieves most of the benefit.

What documentation should come with the shipment? Packing list, material certificates if requested, first article report, and any in-process records. Ask for the list up front so it is quoted rather than negotiated.