Gage RR – 1 Distinct Category

Six Sigma – iSixSigma Forums Old Forums General Gage RR – 1 Distinct Category

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
  • #36749


    1) I have a situation where the total tolerance for the measured characteristic is .020″, but the process yields product with very little variation (cold formed product constrained by a die) and the range of readings for 3 operators, 10 samples and 3 trials is only .004″ (using .001″ calipers).  MINITAB calculates the number of distinct categories at “1”.  Do I still interpret this as unacceptable? 
    2) When taking the measurements for analysis, is it OK for each operator to use their own calipers?  I’m thinking that they should be included for the “measurement system”.



    1.)Yes your interpretition is correct. Your measurement system is not able to distinct one part from the other. You can say that the system reads all the parts as same. The number of distinct catagories  should always be greater than 5.
    2.) The instrument being used for MSA should always be common to all the operators. This is necessary for repeatibility. Apart from this Stability,Bias, Linearity of the instrument should be known.



    I would tend to disagree with your point number 2.  If it is standard procedure for the inspectors to use their own instruments to make measurements, then that variation is part of the whole measurements system and should be measured.  If there is some issue with the instruments having different calibrations, then it would show up as having a reproducibility issue and can then be addressed.  Remember GR&R are not just used to measure the instruments being used, but the measurement system as a whole.
    Any comments?



    1) I would suggest that your measurement system is not capable of discriminating one part from another.  Your measurement error takes up just about all of your process variation therefore you shouldn’t use it to control the process.  I would check your R&R % to tolerance.  Your measurement system might be capable of distinguishing good parts from bad.
    2) I agree with you.  I would have each operator use their own calipers if that is the way that it happens in real life.  Remember that variation between calipers is now stuck in with variation between operators.  If you have a lot of reproducibility variation then you might have to do a subsequent study to separate these two factors.

Viewing 4 posts - 1 through 4 (of 4 total)

The forum ‘General’ is closed to new topics and replies.