MSA Using MTB Questions

Six Sigma – iSixSigma Forums Old Forums General MSA Using MTB Questions

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
  • #42706


    1.Β  If you have two categories of variation – accuracy and precision – and the Gage R&R (continuous) provides repeatability and reproducability measures (both which represent precision) where is accuracy represented in the MTB session info?
    2. Where does a test-retest fit into the Gage R&R study?

    Same operator, same part, same device, multiple measures
    Calculate and compare Sample Avg to Std (accuracy) and
    Sample std deviation to tolerance (s < 1/10 of tolerance for precision) into the Gage R&R study?
    3.Β  Can you / Do you prioritize the the %Tol, %Study, %Cont measures, or is it a “fail one, fail all” (based on Rules of Thumb)
    Thanks!Β  Sorry for so many questions.


    Neeraj Katare

    For Accuracy, pl use Minitab Option-Gage Linearity and Bias Study
    Stat > Quality Tools > Gage Study > Gage Linearity and Bias Study
    Gage linearity tells you how accurate your measurements are through the expected range of the measurements. It answers the question, “Does my gage have the same accuracy for all sizes of objects being measured?”
    Gage bias examines the difference between the observed average measurement and a reference or master value. It answers the question, “How biased is my gage when compared to a master value?”



    Thank you so much!




    SG:DonΒ’t worry about the
    questions, MSA is a large and very important part of every project.You will be doing three
    tests. At each test, if you donΒ’t pass, then go to jail, do not collect $200.First test Β– resolution
    checkCan you even hope to read
    the gauge to the precision you require? The readout of the gauge must divide
    the tolerance window into 10 divisions to be acceptable. Would a sundial be
    adequate to measure time while running a production line? You canΒ’t read
    seconds. No, you canΒ’t use it. The motion of the planets is a very well
    understood phenomenon, but you just will never be able to read seconds from a
    sundial.Second test Β–
    test-retestYou can find the pure
    equipment variation by having one operator, one gauge, and one part. Take (say)
    15 measurements, find the standard deviation and compare it with the tolerance
    width. This will give you pure gauge error. By the way, if you are measuring
    one part over and over, why not measure a known standard? You purchase these
    for exactly this purpose. The difference between the mean of the 15
    measurements and the number on the calibrated standard will give you the bias
    of the gauge. This is not found as a separate Minitab function, you use the graphical
    summary to make the calculations. There is no rule-of-thumb for judging whether
    the bias is acceptable or not. This is because it is so easy to correct. The
    device can be re-zeroed or the bias can be corrected by subtracting it from all
    readings.Second test variation Β–
    gauge linearity and biasIf you are going to use the
    gauge in a variety of settings with a large range of values, then you may
    consider using a range of standards and conducting a gauge linearity and bias
    study. This is a bit more involved than the test-retest because it requires a
    range of standards that cover the range of interest. It will also allow you to
    determine the bias as a function of the size of the measurement.Third test Β– Gauge
    R&RThe big meal deal, using
    actual instruments in field conditions by the workers who will be using them.
    It is best to use a sample of parts from real production.Minitab does divide the
    precision into the two categories of operator and equipment. This is the reason
    behind all the different combinations of operator and part numbers. By looking
    at the variation in data from one operator measuring the same part multiple
    times, then you can quantify the pure error. This number will be similar to the
    one from the test-retest, but remember the test-retest was done under ideal
    conditions, the Gauge R&R is done under field conditions. The equipment
    error here is likely to be larger than the one determined during the
    test-retest.To interpret the results
    you can treat the rules-of-thumb more as guidelines.If the total Gauge R&R
    (the combination of equipment error and operator error) is less than 10% of the
    tolerance window, then it is considered OK. The %Study variation always totals
    100% and merely allows you to Pareto the sources of variation. If you only see
    %Study variation without the %Tolerance variation, you forgot to enter a value
    for the Β‘tolerance windowΒ’ found under Β‘optionsΒ’.If the process variation is
    very large, you will have a selection of parts where you have a lot of them
    outside the specification limits. In this case you can have the Β‘number of
    distinct categoriesΒ’ rule-of-thumb indicate the gauge is acceptable, but the
    Gauge R&R may be much larger then 10% of the tolerance window. This is
    because when the process is really bad, then even a poor gauge can tell whether
    a part is way-too-large, too-large, OK, too-small, and way-to-small (5
    categories) while not telling you too much about where the parts are with
    respect to the Upper-OK limit and Lower-OK limit.I do see a lot of people
    neglect the test-retest step, because they get the same information out of the
    Gauge R&R.Cheers, BTDT



    BTDT, thanks so much….Per usual, that was VERY helpful …… Cheers!



    SG:You’re welcomeCheers, BTDT… and I got post 90,0000 too



    Hey, BTDT..
    Was this coincedence or you do it on purpose?Β  I mean you being in 90,000 post.
    Anyway, I’m next to you. 90001 perhaps….

Viewing 7 posts - 1 through 7 (of 7 total)

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