Three solder defects on one assembly are not the same observation as one defect on each of three assemblies. Both matter, but they describe different outcomes. Define whether a quality metric counts defects, nonconforming boards, test attempts, or opportunities before comparing suppliers or production periods.
NIST’s attributes-control-chart overview distinguishes charts for the fraction of defective product from charts for defects per unit. The appropriate summary depends on what is counted and on the statistical assumptions of the chosen method.
Write the counting rule
Define a defect category with enough detail that inspectors can apply it consistently. State whether repeated observations of the same physical defect are merged, whether reinspection findings are new events, and how multiple defects on one board affect the board-level status.
Choose the denominator explicitly: inspected boards, assembled boards, inspected joints, or another defined exposure. Keep the measured population separate from the total shipment when only a sample was inspected. Report missing inspection records separately rather than assigning those boards an assumed passing status.
Use a small worked example
Suppose 20 assemblies are inspected and two assemblies contain five recorded defects in total. The observed nonconforming-board fraction is 2/20, while the observed defects per inspected assembly is 5/20. Neither number should be labeled simply as defect rate without explaining its meaning.
| Metric | Required identity |
|---|---|
| Nonconforming board fraction | Each board counted once in the numerator |
| Defects per board | Each defined defect counted with its board link |
| First-pass test yield | Initial valid attempt for each eligible unit |
| Retest activity | Additional attempts and their reasons |
Check comparability before trending
- Confirm the same defect definitions were used.
- Review changes in inspection coverage or detection method.
- Separate variants with materially different inspection opportunities.
- Reconcile board counts with the production population.
- Preserve the raw defect-to-board relationships.
A more sensitive inspection process may report more defects even when the underlying process has not worsened. A simpler board may have fewer opportunities for a defect than a dense assembly. Explain these differences before treating a lower aggregate number as evidence of better manufacturing.
Do not add percentages from different populations without an appropriate combined numerator and denominator. Retain the counts behind each percentage so reviewers can reconstruct the result and understand small-sample variation.
Close the report with plainly named metrics and their counting rules. The retest-history guide explains how repeated executions can otherwise distort unit-level yield, while the sampling-plan guide defines the population an inspection decision actually covers.
Move from reading to a build-ready project
Use the fabrication and assembly checklists to prepare drawings, production files and test requirements.
Open the DFM checklist