A narrow set of study parts can produce a favorable tolerance-based result and an unfavorable process-based R&R result at the same time. In this case, the reported 3.3% of tolerance indicates useful inspection capability against the 1.0 mm tolerance band, while 31.2% of observed process variation and four distinct categories show that the available parts do not adequately challenge the measurement system for process assessment.
Separate the two measurement-system decisions
| Decision | Reported result | Interpretation |
|---|---|---|
| Can the system distinguish conforming from nonconforming parts? | 3.3% of tolerance | The measurement variation is small relative to the specified tolerance. |
| Can the system resolve current process variation? | 31.2% of process variation; four categories | The result is weak because the sampled parts provide little part-to-part variation. |
These results answer different questions and are not contradictory. The first uses the specification width as its reference. The second compares measurement variation with the variation represented by the selected parts.
Why a capable process can weaken the R&R percentage
The nominal dimension is 21 mm with a tolerance of plus or minus 0.5 mm, giving a total tolerance width of 1.0 mm. The study parts span only about 0.2 mm, while the reported capability index is greater than 4. This means the study represents a much narrower range than the permissible tolerance band.
When part-to-part variation becomes small, measurement variation occupies a larger percentage of the observed study variation even if the measurement system itself has not changed. Producing a wider range of parts in a simulation therefore improves the process-based percentage by changing its denominator; it does not independently improve the camera or software.
Reconcile the reported variance outputs first
| Reported item | Value | Required check |
|---|---|---|
| Total variance | 0.00031 | Confirm whether this is the total variance-component output. |
| Between-part variance | 0.0002825 | This represents most of the reported total variance. |
| R&R variance | 0.00000304 | Its square root is approximately 0.00174 mm. |
| Reported total standard deviation | 0.0177 mm | This is consistent with the square root of 0.00031 after rounding. |
| Reported between-part standard deviation | 0.0168 mm | This is consistent with the square root of 0.0002825 after rounding. |
| Reported total R&R standard deviation | 0.055 | This does not reconcile with the stated R&R variance; verify the decimal point, units, and output column. |
The stated 9.7% R&R figure also requires column verification. Dividing the stated R&R variance by total variance gives about 0.98%, whereas comparing their standard deviations gives about 9.9%. Do not mix variance contribution, standard-deviation ratios, study variation, and percentage of tolerance in one acceptance decision.
Select study parts that support the intended conclusion
- Define whether the required conclusion concerns tolerance classification, resolution of current process variation, or both.
- Confirm and document the prior test showing no material loading or operator influence before applying the selected MSA 3 approach.
- For a process-based assessment, select parts covering the relevant tolerance band instead of using only tightly clustered production parts.
- If producing such parts requires costly tooling, propose measurements of approved products spanning the camera system's applicable absolute measurement range. The evidence describes checks at dimensions 3, 12, 20, 33, 40, and 60 for a target near 35, but this remains an alternative plausibility argument rather than direct evidence for the original characteristic.
- Obtain customer or auditor approval for that substitution and document why the alternative products represent the same measurement principle, setup, and applicable range.
- Repeat the analysis and report tolerance-based and process-based results separately.
Measuring a different dimension merely to increase variation does not automatically validate the specified critical characteristic. Use that route only when its applicability is documented and accepted by the responsible approval authority.
Verify the final measurement-system claim
Confirm that the analysis uses consistent units and that every percentage is tied to its denominator. Verify the suspected 0.055 standard-deviation entry against the raw software output. The final report should state explicitly whether the system is accepted for conformity decisions against the 20.5 mm to 21.5 mm limits, for assessment of the narrow current process distribution, or for both.
Retain the 0.0001 mm displayed resolution as an instrument setting, not proof of measurement capability. Capability must come from the repeatability results, the selected part range, and the approved acceptance basis.
FAQ
Why is R&R 31.2% when percentage of tolerance is only 3.3%?
The denominators differ: 3.3% references the 1.0 mm tolerance band, while 31.2% references the much narrower variation represented by the study parts.
Can a highly capable process cause a poor process-based R&R result?
Yes. If the sampled parts are tightly clustered, measurement variation becomes a larger fraction of observed process variation and the study may produce only four distinct categories.
Must deliberately nonconforming parts be produced for this MSA?
Not necessarily. Parts spanning the relevant tolerance band are preferred; approved products covering the applicable measurement range may support an alternative study when their equivalence is documented and accepted by the customer or auditor.