A wide histogram of PCB test results does not reveal how much spread comes from the boards and how much comes from the measurement process. Contact variation, instrument noise, setup differences, and time-dependent drift can broaden the observed distribution. Investigate those contributors before concluding that manufacturing has become less consistent.
NIST’s measurement variability discussion distinguishes short-term behavior from variation across longer intervals. Repeating one measurement several times in immediate succession does not characterize every source of variation that appears during normal production.
Ask two different questions
First, how much does the measurement change when the same stable unit is measured repeatedly under a defined procedure? Second, how much do results differ between units sampled from the population of interest? Keep those data sets identifiable. Combining them without labels makes later interpretation difficult.
Choose units that remain sufficiently stable during the study. If the measured property changes with warmup, charging, firmware state, or handling, define preparation and stabilization conditions. Otherwise, real changes within the unit may be incorrectly assigned to station repeatability.
Vary the factors that matter
| Study element | What it can reveal |
|---|---|
| Repeated readings without reloading | Short-term acquisition variation |
| Unload and reload the same unit | Contact and setup contribution |
| Different operators or stations | Reproducibility under the chosen conditions |
| Measurements over time | Drift or longer-term changes |
Use an analysis plan suitable for the measurement and decision. Do not subtract one standard deviation from another or apply a generic percentage threshold without checking the statistical assumptions. Involve the responsible measurement specialist when estimating separate variance or uncertainty components.
Use findings to choose the next action
- Preserve unit, attempt, station, and time identity in the data.
- Investigate obvious measurement instability first.
- Repeat the relevant study after correcting the measurement process.
- Then assess the remaining product-to-product distribution.
- State which conditions the conclusion covers.
A narrow repeatability result also does not prove accuracy. A station can repeat a biased value consistently. Preserve sample-selection details as well: a deliberately diverse engineering set should not be presented as a random estimate of the production population. Calibration evidence and method validation address different questions and should remain part of the overall measurement review.
Close the investigation with a clear account of what was observed and which cause remains uncertain. This prevents a fixture problem from driving unnecessary board changes, or a real product variation from being dismissed as noise. For comparing separate stations, use the station correlation plan alongside this population study.
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