-
Notifications
You must be signed in to change notification settings - Fork 0
Understanding Your Report
Equipment, performer, date, study design, and regulatory reference. This is the audit trail header — it must be accurate for 21 CFR, ISO 13485, or PPAP records.
| Column | What it means |
|---|---|
| Study Variation | 5.15-sigma value — band covering 99% of measurements |
| % of TV | Percentage of Total Variation from this component |
| Assessment | GR&R row: Acceptable/Marginal/Unacceptable. EV/AV rows: Low/Moderate/High |
GR&R row: This is the pass/fail line. Green = Acceptable, amber = Marginal, red = Unacceptable.
Reading EV vs AV:
- EV dominant → gage hardware problem (worn parts, poor resolution, environmental noise)
- AV dominant → operator technique problem (inconsistent force, fixture setup, training)
- AV > EV → almost always fixable without replacing the gage
%Tolerance rows (when --tolerance provided): same thresholds but relative to the engineering spec.
Shows R-bar-bar, UCL_R, X-diff, Rp, and all K-constants. Allows an auditor to verify the math from first principles without re-running the tool.
Grand mean and average range per operator. Identifies which operator has highest within-part variation or is the outlier driving AV.
Color-coded summary: %GRR, ndc, and %Tolerance(GRR) if applicable.
Variance components bar chart — EV / AV / GRR / PV with 10% and 30% threshold lines.
R-Chart — Range by part and operator with UCL_R line. Points above UCL_R flag erratic technique or part-handling issues.
X-bar Chart — Means by part and operator.
- Parallel lines → good: part variation is real
- One line consistently above/below → operator bias (AV source)
- Lines cross or diverge → operator-by-part interaction — needs fixture standardisation
Plain-language corrective action guidance specific to the study verdict.
Self-contained single HTML file — open in any browser, no server needed.
Click to hide/show individual operators on R-chart and X-bar chart. Useful for isolating the outlier operator when AV is the dominant source.
Horizontal track showing result position across green (<=10%), amber (10-30%), and red (>30%) zones.
-
95% CI on GR&R — e.g.
[0.00553, 0.03159]— statistical uncertainty on the GRR estimate -
ndc (tolerance-based) — when
--tolerance/--usl/--lslprovided
Both PDF and console warn when any individual range exceeds UCL_R. Investigate before accepting the study for regulatory use.
If AV^2 goes negative (finite-sample correction exceeds raw spread), AV is clamped to 0. This is mathematically correct — it means operator variation is indistinguishable from gage noise. It is a good outcome, not an error. The note is included for transparency.