Sample preparation for Gage R&R study - Heterogeneous data or Homogeneous data

[sinned]

If we have to use "homogeneous" data in GRR study, that means for each model of product having an individual/unique nominal dimension, does it mean that we need to conduct GRR once for every of them even there is little difference btn them ?

Or is it appropriate to use one GRR result to demonstrate equipment effectiveness for different nominal dimensions but having the same tolerance span (as long as the "measurement system" is considered identical) ? In other words, should one GRR be done for one "measurement system" ? i found sometimes it is difficult to convince customer to accept this idea especially when they question the linearity performance of the equipment which is not verified in GRR study.

 

[Miner]

If we have to use "homogeneous" data in GRR study, that means for each model of product having an individual/unique nominal dimension, does it mean that we need to conduct GRR once for every of them even there is little difference btn them ?

Or is it appropriate to use one GRR result to demonstrate equipment effectiveness for different nominal dimensions but having the same tolerance span (as long as the "measurement system" is considered identical) ? In other words, should one GRR be done for one "measurement system" ? i found sometimes it is difficult to convince customer to accept this idea especially when they question the linearity performance of the equipment which is not verified in GRR study.

You can use a single gage study to cover this situation if the following conditions are true:

• The same gage is used for the different nominal dimensions
• A Linearity study has been performed for the entire range of nominal dimensions with acceptable results
• The process variation at each of the nominal dimensions is the same if %GRR is used, or the tolerance is the same if P/T Ratio is used.

You customer has a valid concern if you have not verified the linearity, so you should satisfy them and yourself that this is not an issue.