iSixSigma

Gage R/R with Geometric Tolerances

Six Sigma – iSixSigma Forums General Forums Methodology Gage R/R with Geometric Tolerances

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #54381

    Hello Everyone! I searched through the forums and couldn’t find a straight answer so I figured I just ask. I have a part which uses alot of geomteric tolerances; mainly profiles. The profiles have a 0.006 band (+/-0.003) and we use a vision system to inspect the parts. The output from the vision system is the maximum positive deviation and maximum negative deviation of the profile from nominal. Obviously if either one of those is outside the profile width, the part fails. Should I execute a typical gage R/R on the largest deviation from nominal, on each deviation separately, or something else?

    Thanks for the help!
    Mike

    0
    #195142

    Chris Seider
    Participant

    I am guessing the device can measure the positive and negative deviation in the same setup of the machine? If so, run a gage R&R separately on the positive deviations and then on the same deviations with 2 columns of data in your Minitab worksheet. This may sound conservative but at least you can confirm there’s nothing wrong with measurements both way.

    My recommendation of doing both types of measurement would stay the same even if you have to measure both deviations with 2 different setups.

    0
    #195143

    Are you using a Shadowgraph? Those tolerances seem very tight, without an optical bench and a 3-D coordinate measurement system in a temperature controlled room. But that might just be my ignorance :-)

    0
    #195144

    MBBinWI
    Participant

    Mike: Since you are enquiring about doing a GR&R I’m assuming that you are interested in understanding the ability of the tool to accurately evaluate in this case profiles. In addition to the measurement itself, you will need to evaluate the measurement ability in various areas across which the tool can measure – does the tool have a “sweet spot” or does it drop off in accuracy when you get out to the extremes of the evaluation area? Is the measurement error constant or does it change based on the dimensions of the item being measured? Does the variation in measurement change based on the thickness of the item being measured?
    Depending on the answers to these questions (and there may be more, I’d need to understand the tool and how it measures the parts), will dictate just how to approach the MSA. As you can probably tell from the questions, this could be a rather complex MSA depending on how critical variation in your measurement system is to your decision making.

    0
    #195222

    Chris Seider
    Participant

    To Mike Ross:

    Nice job on follow up and giving us an update.

    0
    #195223

    MBBinWI
    Participant

    Mike: You need to evaluate over the full band width.

    0
Viewing 6 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic.