CMM Inspection Reports and First Article Inspection: What You Should Receive

You asked for a CMM report with the first delivery and received a PDF listing thirty dimensions, none of which are the four that matter, plus a note saying "all dimensions within tolerance." Two rows are marked red with no explanation. The document looks official and tells you almost nothing.

A coordinate measuring machine produces whatever the programmer asked it to measure. That makes the report a direct reflection of the questions the drawing posed. If the drawing does not identify the important characteristics, the report will measure the convenient ones, because those are the ones a programmer can pick up from a model in an afternoon.

What a CMM inspection report contains

A CMM report is a set of measured values compared against nominal values and limits, produced under stated conditions. A complete one has six parts:

A ballooned drawing. Each measured characteristic carries a number matching a balloon on the drawing. Without ballooning, the report is a list floating free of the part.

The measurement plan. Which machine, probe tip, alignment and datum reference frame were used. If the alignment is not stated, positional results cannot be interpreted.

Nominal, actual, deviation and limits. A report showing only pass or fail removes the information you need to see a trend.

Measurement uncertainty or equipment capability, typically stated as MPE, for example (1.8 + L/300) µm on a mid-range bridge CMM. This matters when a result sits within 10% of a limit.

Environmental conditions. Standard practice is 20 °C ± 1 °C for machine and part, with soak time stated. Aluminum and steel expand differently, and a part measured straight off a warm pallet will not match one measured the next morning.

Traceability and sign-off. Who measured it, when, with which calibrated equipment, against which drawing revision.

Evaluation method decides the answer

For each feature, the inspector chooses how many points to touch, how to fit the geometry, how to reference it and whether to filter the data. A bore can be evaluated as a least-squares circle, a maximum inscribed cylinder or a minimum circumscribed cylinder, and on a press-fit those three differ by several micrometres.

This is how a report can be accurate and misleading at once. A Ø40 H7 bore reported at 40.012 with a least-squares fit is probably fine, but a maximum-inscribed evaluation tells you whether the mating shaft will actually enter.

For functional fits, name the evaluation method in the drawing notes. One line saves a week of correspondence.

First article inspection: the flow

A first article inspection is a full verification of the first part from a new process, program, fixture or drawing revision. It is not a sample check.

1. Drawing and model review. Confirm the revision, resolve ambiguous callouts, identify characteristics that cannot be measured as drawn.
2. Ballooning. Every dimension, GD&T frame, note and finish callout gets a sequential number on a marked-up drawing.
3. Measurement plan and program. Sequencing the alignments and features, with probe angles that can physically reach each surface.
4. Machining the article, ideally on the production fixture rather than a one-off setup.
5. Measurement, split between CMM for geometry, roughness tester for finish, spectrometer or hardness tester for material, and hand tools for features the CMM cannot reach.
6. Report and disposition. Every characteristic marked conforming or not, with a root cause and corrective action for anything out.
7. Customer approval before volume release.

The step buyers underestimate is the fourth. An article made on a carefully supervised one-off setup proves the drawing can be met, not that the production process will meet it every time. Ask whether it came off the production fixture.

Critical dimensions versus a full layout

These are not quality levels, they are different questions. A full layout answers "can this part be made to print?" A capability study answers "will it stay there for the next 2,000 pieces?" A sampling report answers "did this lot drift?" Buying the wrong one wastes money in both directions.

How to specify inspection on the drawing

The drawing is the only document that reliably reaches the inspector, so the requirement belongs on it rather than in an email that gets lost between sales and the shop floor.

• Mark important characteristics with a symbol or a note such as "CTQ" and a short list, so ballooning scope is defined before the program is written.
• State the report type in the title block: "First article inspection report with ballooned drawing, full layout, required with first delivery."
• Give a datum reference frame a CMM can physically establish. Three mutually perpendicular planes on accessible surfaces are ideal; a datum on a feature the probe cannot reach forces an inferred alignment that adds uncertainty.
• Attach a measurement location to finish callouts. "Ra 0.8 µm on bore ID, measured axially at mid-depth" is inspectable. "Smooth finish" is not.

The recurring buyer mistakes are predictable. Requesting a report without saying what to measure produces a report of convenient features. Over-tolerancing forty dimensions forces a full layout on every lot without improving function; if your drawing uses ISO 2768 general tolerances, state the class, because a US inch drawing sent to a metric shop is a reliable source of avoidable non-conformances.

Reading a marginal result

A dimension at 60% of its band is fine. A dimension at 95% of its band on a part you intend to order 2,000 of is a warning: it conforms, but the process is running near the edge and normal variation will start producing failures.

Read the pattern, not the rows. If five bores are all biased 0.008 mm high, the program or machine has a systematic offset that will eventually cross the limit. If deviations scatter randomly, the process is centred and only spread is a concern.

At JSHS, first article reports go out with the ballooned drawing attached, because the drawing is what makes the numbers readable. If your quality system needs a specific template, include it in the RFQ and we will quote against it.

Frequently asked questions

Do I need a CMM report for every order?

Usually not. A first article plus per-lot sampling covers most production needs at a fraction of a full layout. Reserve layouts for new parts, new processes and drawing changes.

How accurate is a CMM?

A typical job-shop bridge CMM is specified around (1.8 + L/300) µm, so roughly 2 µm on small features and a few micrometres on a 500 mm part. That is well inside a ±0.005 mm tolerance, but only at 20 °C with proper fixturing.

Can a CMM measure everything?

No. Deep bores, undercuts, sharp internal corners, surface roughness and material properties need other instruments. A good report states which characteristics were measured otherwise and with what.

What if a first article fails?

The shop should identify the root cause, correct the process and produce a new article from the corrected process. Re-measuring the same part until it passes is not a corrective action, and a report showing the same serial number twice is worth questioning.