Quality Assurance and Testing
Quality assurance and test validity
Quality Assurance and Testing tests whether you can plan checks, produce trustworthy measurements, and act on nonconformity. It is less about inspecting everything and more about controlling the process and proving that results are valid.
Core rules
- QA is process-focused and preventive; QC is product-focused and detective. A test result is useful only if the method, sample, instrument, and operator are controlled.
- Calibration must be traceable to a national or international metrology standard, with known uncertainty and a validity period.
- Report measurement uncertainty and apply a decision rule. A simple guard band accepts only when |measured deviation| + U ≤ tolerance; otherwise the conformity decision is indeterminate or requires better measurement.
- Use defined sampling plans; random or stratified samples must represent the lot, and acceptance sampling does not excuse poor process control.
- For nonconformity, contain the effect, determine root cause, implement corrective action, and verify effectiveness.
Worked example
A 10.00 mm feature has a tolerance of ±0.02 mm. A calibrated micrometer reads 10.02 mm with expanded uncertainty U = 0.01 mm (k = 2). The deviation is 0.02 mm, exactly at the tolerance limit. Since 0.02 + 0.01 = 0.03 mm exceeds the 0.02 mm tolerance, the result cannot demonstrate conformity at the stated confidence. The correct response is to remeasure with a lower-uncertainty method or apply an agreed decision rule, not to record the part as passing because the reading is “within” tolerance.
Traps
Confusing precision with accuracy; ignoring calibration status; using a sample that is convenient rather than representative; and closing a corrective action without checking recurrence.