We need to perform GRR on operators scanning in defect codes into Manufacturing report system(drop down menu for defects). The subject is PCB with solder joint defects. Concern being, there are muiltiple characteristics for failure on 1 single board. eg. Solder bridge, insufficient, lifted etc. If I perform Attribute GR on 10 boards, how do I translate these into Pass/ Fail category ?.How should I select the samples, standards and ascertain the Pass/fail.eg. Should I group all count of Solder bridges as one and then see whether each operator is able to detect the exact amount of solder bridges (PASS) else FAIL.?. Pls help

Any suggestions, folks?

This is definitely not an easy one. You have recognized (I assume) that this is an attribute study, rather than a variable study.

In the past, I have used a 20 pc x 3 repetitions x 3 operators on a given defect, with an accept score of 100% correlation (all operators detected and recorded the same criteria; pass or fail). The sample group of 20 pcs was randomly 'salted' with defective parts, and the identification of the parts was concealed from the operators. In addition, the 3 repetitions were performed at different times of the day, on 3 different days, to try and reflect 'real world' conditions (operator fatigue, workload, etc).

Don't know if your customer will accept this, but it is at least a starting point.