Fully agree on the need for a range of parts with a sizes distributed accross the tolerance band and beyond. I would like to add the following points:1) If possible always use 3 operators that have knowledge of the process. 2) You will need appraisers to run an R&R, App1-Notes Down thew readings. App2-Observes the operators and notes down how they use the gauge. This is very usefully for understanding any operator to operator variance shown by the XmR charts.
Parts can span some portion of the spec range; they just can't all be the same or there won't be enough part variation to get a good NDC. Too many QI Macros customers call to complain that they can't get a good GR&R, but it's because their parts are identical. You have to have an NDC of at least 5 to get a good GR&R. You can use the Spec Range to calculate %R&R or the average and range method. If you use the Spec Range, it should be reported. Otherwise, it's not essential to the GR&R.
The whole "requiring bad parts" is misleading. The purpose of the part to part variation metric is to ensure your measurement system is capable of detecting variation. If all parts are the same in the Gage R&R study you have not proven your measurement system is capable of this feat. If your process is statistically stable then you can substitute low part to part variation for a "surrogate process variation" refer to AIAG
Fully agree on the need for a range of parts with a sizes distributed accross the tolerance band and beyond. I would like to add the following points:1) If possible always use 3 operators that have knowledge of the process. 2) You will need appraisers to run an R&R, App1-Notes Down thew readings. App2-Observes the operators and notes down how they use the gauge. This is very usefully for understanding any operator to operator variance shown by the XmR charts.
Also is it correct to say that one can not report a %R&R value without also reporting the corresponding Spec Range?
Parts can span some portion of the spec range; they just can't all be the same or there won't be enough part variation to get a good NDC. Too many QI Macros customers call to complain that they can't get a good GR&R, but it's because their parts are identical. You have to have an NDC of at least 5 to get a good GR&R.
You can use the Spec Range to calculate %R&R or the average and range method.
If you use the Spec Range, it should be reported. Otherwise, it's not essential to the GR&R.
The whole "requiring bad parts" is misleading. The purpose of the part to part variation metric is to ensure your measurement system is capable of detecting variation. If all parts are the same in the Gage R&R study you have not proven your measurement system is capable of this feat. If your process is statistically stable then you can substitute low part to part variation for a "surrogate process variation" refer to AIAG