Gauge R&R for Hardness Tester

Six Sigma – iSixSigma Forums General Forums Methodology Gauge R&R for Hardness Tester

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
  • #55959


    Dear All,

    I am doing six sigma project on reduction on variability in martensite %, Presently i am in measure phase, while doing Gauge R&R in Rockwell hardness tester i am not getting Gauge R&R.

    Sample used: Standard Block
    Tolerance in standard block: Upper limit – Lower Limit = 2 HRC

    Machine specification:
    Make : Matsuzawa
    Model : Rockwell
    Machine Error : +/- 1.0 HRC
    Reading Type : Digital
    Decimal Range: 0.1 HRC
    Calibration : Valid
    Calibration type: External

    Gauge R&R study:
    Sample – 1 (Standard Block for all test)
    # of operator: 3
    # of replication: 2

    Gage R&R Results:

    Source VarComp (of VarComp)
    Total Gage R&R 0.0157417 100.00
    Repeatability 0.0148333 94.23
    Reproducibility 0.0009083 5.77
    Part-To-Part 0.0000000 0.00
    Total Variation 0.0157417 100.00

    Process tolerance = 2

    Study Var %Study Var %Tolerance
    Source StdDev (SD) (6 × SD) (%SV) (SV/Toler)
    Total Gage R&R 0.125466 0.752795 100.00 37.64
    Repeatability 0.121792 0.730753 97.07 36.54
    Reproducibility 0.030139 0.180831 24.02 9.04
    Part-To-Part 0.000000 0.000000 0.00 0.00
    Total Variation 0.125466 0.752795 100.00 37.64

    Number of Distinct Categories = 1

    Seeking your direction to proceed further.




    Chris Seider

    did you check the site? Lots of stuff on gage R&R. Promise! :)


    Mike Carnell

    @badrinarayanan I saw this in the iSixSigma news letter and have meant to get back to this all week. When you do a Gage Study the first thing you look at is number of Distinct Categories. If you do not have enough then none of the other numbers mean anything. In your case you have 1. Basically your gage has no resolution.

    This can be caused by a lot of things the first being the gage has no capability. The second thing, and you need to check this, is that if your samples are all the same then the analysis comes out that the gage cannot tell the difference when in fact you chose samples incorrectly.

    You need to resolve this issue before you even worry about the other numbers.


    Ken Feldman

    @cseider, I’m surprised you didn’t take the time to read the original post more closely. Mike is right. With only 1 number of distinct categories, all you can tell is that there is a “part”. Two categories will only let you tell good and bad. And so forth….. Selecting a variety of parts composed of “good”, “bad” and some in-between will give you the chance of getting enough variety that you can actually determine the GR&R. Otherwise the noise of the measurement system overwhelms the signal associated with the parts. So, to reiterate what Mike said, it is either a gage with no resolution or parts with no variation.


    Chris Seider


    Yes, you and Mike were more exact but I was thinking they hadn’t researched much but you and Mike were correct–I just didn’t be as pointed with my help.

    Hope FL treating you well.


    Mike Carnell

    @cseider Are you going to take that? He doesn’t show up here for a couple years acting like a bully after you have been carrying the load all this time.


    Chris Seider

    LOL, @mikecarnell “water off a duck’s back”.

    Hope HOU went well. At least the Astros did well.


    Mike Carnell

    @Cseider Hou went well or at least successful. Thanks for asking.

    Darth is getting old, Actually he was always old so he is just getting older and grouchier.

Viewing 8 posts - 1 through 8 (of 8 total)

You must be logged in to reply to this topic.