A CNC subcontractor's quality evidence is its inspection report, far more than its words. The coordinate measuring machine (CMM) report is the most comprehensive of these, but when read incorrectly it leads either to unnecessary complaints or to real problems being missed. This article deepens our quality assurance article on the report-reading side and explains the core document of the FAI process.
What Is a CMM and What Does It Measure?
A CMM moves a touch-trigger or scanning probe along the X-Y-Z axes to collect points from the part surface, derives geometric features (plane, cylinder, circle, cone) from those points and calculates the dimensions and geometric tolerances between features. Typical bridge-type CMMs have a volumetric accuracy (MPEE) in the order of 1.5–3 µm + L/350–L/250, i.e. roughly 2–3 µm uncertainty on a 100 mm dimension. Its difference from a caliper or micrometer is that it can reliably measure geometric tolerances (position, perpendicularity, flatness, profile) against a single reference system.
Sections of the Report
1. Header block
Part number and name, drawing revision, order/batch number, number of samples measured and serial numbers, measurement program name, date and time, operator, CMM model and last calibration date, probe configuration, ambient temperature. An incomplete header block (especially revision and temperature) lowers the credibility of the report.
2. Datum alignment
The reference system the measurement was taken against: for example "Datum A: bottom face plane (12 points), Datum B: Ø20 hole axis, Datum C: left edge". It must match the datums on the drawing exactly; a different alignment gives different results (see below).
3. Feature table
The body of the report; each row corresponds to a balloon number on the drawing. Typical columns:
| Column | Meaning |
|---|---|
| Balloon / Feature | Balloon number on the drawing and description (e.g. "12 – Ø20 H7 hole") |
| Nominal | Target value on the drawing (20.000) |
| Upper / Lower tolerance | +0.021 / 0 (for H7) |
| Actual | Measured value (20.012) |
| Deviation | Actual − nominal (+0.012) |
| Out of tol. | Amount exceeding the tolerance; blank or 0 means conforming |
| Bar graph | Position of the value within the tolerance band; green inside, yellow close to the limit (75%+), red outside |
| Number of points | Probe points collected for the feature (at least 4–8 for a circle, 6–12 for a plane) |
4. Geometric tolerances
- Position: Twice the deviation of the actual hole axis from its theoretical location (diametral tolerance). If "MMC" or "Ⓜ" is present, bonus tolerance applies: the larger the hole is than its maximum material condition, the wider the position tolerance becomes; the report may show a "bonus" column.
- Flatness: The distance between two parallel planes enclosing the surface points; requires no datum.
- Perpendicularity / Parallelism: Relative to a datum; the row must state which datum.
- Cylindricity, circularity: Require sufficient points or scanning; circularity from 4 points is meaningless.
- Profile of a surface: Deviation from the CAD model on free-form surfaces; usually given as a colour deviation map.
5. Summary and approval
Total number of features, number conforming/non-conforming, result ("Conforming" / "Deviation approval required"), signatures of the inspector and approver.
Why Does Datum Alignment Change the Results?
Same part, same CMM, different datum order: position tolerance results can change by 0.02–0.05 mm. If the drawing says "A|B|C", alignment follows the principle of 3 points on plane A (plane), then 2 on B (orientation), then 1 on C (location): the 3-2-1 rule. If the supplier's report and your own measurement differ, alignment is the first thing to check; second is temperature, third is the number of points and the probe diameter.
Measurement Uncertainty and the Decision Rule
No measurement is exact. According to ISO 14253-1, measurement uncertainty is taken into account in the conformity decision: the uncertainty should not exceed 10% of the tolerance (the 10:1 rule), with 25% accepted at worst. For a dimension toleranced at ±0.01 mm, a CMM uncertainty of 2–3 µm gives a ratio of roughly 1:7–1:10, which is acceptable; this tolerance cannot be measured with a caliper (±0.02–0.03 mm uncertainty). The report may flag values very close to the limit (the last 10% of the tolerance band) as "conditional" because of uncertainty; the decision rule (simple acceptance, guard band) should be agreed with the customer in the quality agreement.
7 Signs of a Fake or Careless Report
- All values exactly on nominal: A deviation of 0.000 on 40 features is not realistic; real measurements show scatter.
- No datum alignment information: It is unclear which reference the geometric tolerances were measured against.
- Number of points not stated or too few: Cylindricity from 3 points or flatness from 4 points.
- No temperature, or far from 20 °C: In aluminium, a 5 °C difference means 12 µm over 100 mm.
- Drawing revision does not match the report.
- Calibration date missing or expired.
- Only a "conforming" stamp, no values: That is a declaration, not a report.
Additional check: measure one sample yourself. If there is no systematic difference on 2–3 critical dimensions, you can trust the rest of the report.
CMM Reports in Series Production
Measuring every part on the CMM is not economical for most parts; typical practice:
- First article (FAI): 100% of features, 1–5 parts.
- Series: Statistical sampling for critical dimensions (AQL or n=5 every X parts), CMM or dedicated gauges; results reported as min/max/mean and, where required, Cp/Cpk.
- Final inspection: Sampling report per batch; under PPAP, according to the control plan.
Report frequency and content should be defined at the quotation stage and included in the price; the framework is described in our PPAP and ISO 9001 articles.
When Comparing with Your Own Measurements
- Use the same datum alignment; if it differs, the comparison is meaningless.
- Let the part rest at 20 °C for at least a few hours.
- Use a similar number of points and probe diameter for the same feature.
- Compare the difference with the uncertainty: a 3 µm difference is not a "difference" if it lies within the combined uncertainty of the two CMMs.
- If the difference is systematic (all values in the same direction), the cause is alignment or temperature; if it is random, the cause is the number of points or surface quality.
Frequently Asked Questions
Should I request a CMM report with every batch?
For critical parts, yes (with sampling); for simple parts, a first article report plus sampling according to the control plan for series is sufficient. The scope should be included in the price.
Should the report be a PDF or a data file?
PDF is standard; for statistical evaluation, CSV/Excel or QDAS format can also be requested.
Is the report valid if the supplier has no CMM?
An inspection report prepared with calipers, micrometers and dial indicators is valid for tolerances of ±0.02 mm and wider; geometric tolerances and tolerances below ±0.01 require a CMM or equivalent equipment.
When is a deviation approval accepted?
The customer can give written approval for small tolerance overruns that do not affect function; the report marks these separately. Acceptance should be converted into a drawing change for subsequent batches.
Conclusion
Read together with its header block, datum alignment, feature table and uncertainty information, the CMM report is a subcontractor's strongest quality evidence. At Tornelle, first articles are delivered with a full inspection report against a ballooned drawing, and series batches with a sampling report according to the control plan. If you would like to define the appropriate inspection scope for your part at the quotation stage, send your technical drawings via our quote form; the inspection and documentation scope comes with the quote.
