Gage R&R for Automated Measurement Equipment Systems

I am resurrecting this thread, due to study requirement which was dropped in my lap. I have had very little experience with anything other than the 'normal' GRR studies, and would appreciate any assistance or guidance that could be provided.

The defined study parameters are as follows:

1) 2 different but identical gauges
2) 4 different but identical nests for each gauge
3) 4 known value 'master' samples (1 gross pass, 1 marginal pass, 1 marginal fail, 1 gross fail)
4) Each part to be measured in each nest 10 times

The study parameters were agreed on some time ago, and it is not feasible to have them altered at this time.

I would appreciate any input in how best to do this analysis, both for testing each of the 8 nests individually (i.e. as a stand-alone gauge) as well as each set of 4 nests together.

Thanks in advance.

From the title of the thread and the mention of multiple nests, when you say gauges, do you mean automated test machines, not your typical handheld or tabletop gauge?

Yes, they are automated test machines - robotically loaded and unloaded. There is no human interaction at all for this measurement process.

Is the output categorical (pass/fail) or continuous. If categorical can you get continuous data?

If continuous data you can perform the test as outlined and plot results in a multi-vari of multi-variate chart. I ave done a this exact sam Egypt of study before (the details are different but not important; the analysis is what matters). See my presentation MSA - Not your Father’s Gauge R&R
Page 56, Count of Swabs in Bag example…

You can do the same thing with categorical data - the idea is that your test method can repeatably distinguish between, god, marginal good, marginal bad and bad units. The key area of course is the marginal units. If using continuous data add the spec limit to distinguish between the marginal good and bad units

Thanks Bev. The output is continuous data.

This is what Bev's article described. You can add another layer for the good/bad/marginal parts as well.