Re: How do I get the sample for GRR study
Beiquan said:
Then my question is:
1.Which gage should I use in this measurement system, Vernier Caliper or micrometer ?
2.Why the GRR results were different with the same gage , sampling method or specification tolerance?
Use the micrometer. It meets all of the MSA criteria.
The vernier is the gage that is providing inconsistent results, not the micrometer. The reason for this is very simple, and I recommend that you add this as another criteria to your MSA report. The vernier appears to have a resolution of 0.020 versus the micrometer resolution of 0.001. This shows up in the Range chart of the two vernier studies. You should have at least 5 possible measurement values (including zero) under the UCL of the range chart. The two vernier studies only have 2 and 3 possible values under the upper control limit.
When your resolution is inadequate, you will obtain inconsistent reults in an MSA because the resolution can work for you as in the 3rd study, or against you as in the first study. The test for resolution that I described should be the very first screen because if it fails this screen, the other results are invalid.
I recommend using the following tests in this order prior to interpreting the %GRR, ndc, P/T Ratio, etc.:
Step 1: Measurement Stability: Range chart must be in control.
Step 2: Resolution/Discrimination: Need five levels of discrimination (including zero) below the UCL of the Range chart. Less than 25% may be zero. Should not see obvious stratification.
Step 3: Bias: There should not be obvious differences between operators in the Average Chart.
Step 4: Capability: Greater than 50% of the points should be outside the control limits on the Average Chart.
If the MSA study fails any of these tests, the numerical results will not be valid, and will probably change each time you repeat the study.