iSixSigma

Using GR&R to Determine if a Product/Device Is Acceptable

Six Sigma – iSixSigma Forums General Forums Methodology Using GR&R to Determine if a Product/Device Is Acceptable

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #55809

    Kevin Yee
    Participant

    Hello all experts, masters, gurus, sifu. I would like like to seek for your enlightenment in GR&R. I have come across someone using the GR&R as described below.

    Sample : 60 meters
    2 sets of test jigs.

    Mounted the 60 meters on the first test jigs. Feed the current via the test jig into the meter and record the meters output. So there is 60 readings. Repeat the current feeding and output recording. There are 120 readings, 60 from the first and second test respectively.

    Using the same 60 meters and repeat the above using the second test jigs.
    Therefore, there are 240 readings collected.

    In the minitab,

    Part numbers = meters (60 of them)
    Operators = Jigs (2 sets)
    Measurement data = output reading (240) of them.

    so from the GR&R, what can we conclude? can the meters be accepted? are the test jigs reliable to produce trustworthy output reading?

    Another question, does anyone has the experience in using GR&R to judge if a testing device is good enough to be accepted during new equipment buy-off and commissioning?

    Sorry for such a long question.

    0
    #201801

    Chris Seider
    Participant

    If you’re doing only 2 “operators”, it’s highly recommended to get 3 measurements of each part so you have a good estimation of the s.d. of the gage measurement system. One caveat–people think that machines have no interaction with people placing the parts into the system and that can be a mistake–sample prep can be a big deal. Just an fyi.

    Hopefully your 60 parts covers a large portion of typical process variation.

    Read on this site for ideas on how to interpret your results.

    0
    #201809

    Kevin Yee
    Participant

    HI Chris,

    Thanks for the reply.

    How do i ensure or verify if the 60 parts covers a large portion of typical process variation?

    0
    #201811

    Mike Carnell
    Participant

    @kevinyee This is just my thought based on what I am picturing your setup as. It sounds like aautomated/semi-automated test. You need to consider randomizing between runs.

    Just my opinion.

    0
    #201814

    Chris Seider
    Participant

    @kevinyee I’m assuming you’ve done a process capability in the past?


    @mike-carnell
    Great way to force process variation quickly if that was your hint.

    0
Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.