Formulas
Cg = (K/100 · T) / (6 · s)Cgk = (K/200 · T − |x̄ − x_ref|) / (3 · s)- Bias t-test:
t = (x̄ − x_ref) / (s / √n)with n − 1 degrees of freedom.
T is the tolerance (USL − LSL), s the standard deviation of the repeated measurements, K the share of the tolerance (20 % by default).
Running the study
- Use a master or a production part with a calibrated reference value close to the middle of the tolerance.
- Measure it at least 25 times, removing and re-positioning the part each time as in production.
- Check the gauge resolution: it should be 5 % of the tolerance or finer.
When the gauge passes, continue with a Gage R&R study to include operator and part variation, and a linearity and bias study over the operating range.
Frequently asked questions
›What is a Type 1 gauge study?
One operator measures one reference part with a known value 25 or 50 times under repeatable conditions. It shows whether the gauge itself is precise enough (Cg) and whether it measures correctly (Cgk includes the bias) before running a full Gage R&R.
›What are good Cg and Cgk values?
The common requirement is Cg ≥ 1.33 and Cgk ≥ 1.33 with K = 20 % of the tolerance, as in Bosch booklet 10 and VDA volume 5. Some companies use 1.67 for critical characteristics.
›What does K = 20 % mean?
Cg compares 6 standard deviations of the gauge with 20 % of the tolerance. In other words, the gauge spread may use at most a fifth of the tolerance. Cgk compares 3 standard deviations plus the bias with 10 % of the tolerance.
›What if the bias is significant?
The t-test shows whether the average differs from the reference value more than chance would explain. A significant bias usually means the gauge needs adjusting or the master value is wrong. Cgk is reduced by the bias.