False Pass/Fail Information from GR&R (Gage R&R) Minitab

[kornfind]

Hello,

My question is based on interpreting Minitab data with GR&R being done for production process improvement.

We run PCBs on Automated Test Systems with a number of parameters being tested. For instance, for voltage measurements, I have a tolerance value, a measured value and an expected value.

I read in some earlier posts that I should consider the %Tolerance value off the Minitab GRR report? Is that right?

Also, can I extrapolate false Passes and false failures from a GR&R study? If yes, how? If no, is there any other SPC method I should consider to find False Passes and False Failures' percentage?

Thanks,
BE

 

[Miner]

Provided your test equipment is used for inspection, you should use % Tolerance.

The answer to your second question is yes. You can create a gauge performance curve as shown in the AIAG manual. The file attached to my blog will create this curve.

 

[Allattar]

There is an option, under the Gage R&R options that on entering specification limits, you can ask for probabilities of misclassification.

This should be the values you are looking for. (without the curves )

 

[kornfind]

Hi Miner,

Thanks but I couldn't find any direct info that tells me some X% False Pass and X% of False Failures. How do I get that info especially from the Minitab data report generated as in the attached file.

Appreciate your help.

Thanks,
BE

 

[Allattar]

If you are using 15 or 16 it should be in the gage R&R options to display probabilities of misclassification. You can select this on entering spec limits.

 

[Miner]

In Minitab, select Stat > Quality Tools > Gage Study > Gage R&R Study (Crossed) > Options. Select "Enter at least one specification limit", and check the "Display probabilities of misclassification" box.

The session window with display Joint and Conditional probabilities.

• Joint probabilities assume no prior knowledge of whether the product inspected is good or bad.
• Conditional probabilities assume prior knowledge. This means you are either sampling from a pile of rework or previously inspected and accepted product.

 

[kornfind]

All,
Thanks that helped. I have a couple more related questions,
1. What does the * mean under Probablilities of Misclassification?
Example: Conditional Probability
False Accept *
False Reject 0.537
2. What does "Degenerate bivariate detected: The variance-covariance matrix has
determinant = 0. Unable to proceed" mean? I get this at times under Probablilities of Misclassification?

Thanks for all the help
KF