Body
In part 1 and part 2 of this series, we looked at the numbers of parts, operators, and replicates used in a gauge repeatability and reproducibility (GR&R) study and how accurately we could estimate %Contribution based on the choice for each. In doing so, I wanted to provide you with valuable and interesting information, but mostly I hoped to make you like me. I mean like me so much that if I said you were doing something flat-out wrong and had been for years, you’d hear me out before reverting back to being just indifferent toward me.
…
Want to continue?
Log in or create a FREE account.
By logging in you agree to receive communication from Quality Digest.
Privacy Policy.
Comments
Non-Dedicated gages
What is your suggestion for gages that are not dedicated to a particular part? i.e. CMM's, Optical Comparators. These items could be used for very small as well as very large products. At least around here they could be used on several hundred different items.
Various measurements
My suggestion would be to use the rules for measurement uncertainty (ISO GUM) in stead of Gage R&R and always report the measurement results with the measurement uncertainty.
Number of distinct catagories
Frequently I don't know the underlying distribution and if I sample randomly I often get NDC < 5, forcing me to find parts that have higher part to part variation than the random samples. What to do?
Sampling parts and NDC
The problem is that when you organize a Gage R&R you use the parts that are available at that time and these are often produced in a small time frame so show low part variation and as a result you get a bad NDC even if the absolute measurement errror is very small. I never sample, I take 10 parts and use either tolerance width or historical variation as the basis to compare to. In fact most often tolerance width.
Add new comment