1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

How long is a Gage RnR Valid

Discussion in 'Gage R&R and MSA - Measurement Systems Analysis' started by Larry Ricketts, Jan 8, 2016.

  1. Larry Ricketts

    Larry Ricketts Member

    Joined:
    Jan 8, 2016
    Messages:
    7
    Likes Received:
    0
    Trophy Points:
    1
    The question came up with our auditor on how long a gage r n r is good for? Is there a time limit, such as should they be conducted annually? I see nothing in the MSA 4th edition that covers this. I would assume that as long as the measuring device is the same, and calibrated once it is qualified to measure a specific dimension I would believe it would always be good with no need to repeat the test annually?
     
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    576
    Likes Received:
    492
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    An R&R has no expiration timeframe. As long as the measurement system (e.g., device, procedure, feature. etc.) remains unchanged the R&R is valid. If the system changes, then the need for a new R&R should be evaluated.
     
    Vthouta and Bev D like this.
  3. Golfman25

    Golfman25 Well-Known Member

    Joined:
    Nov 6, 2015
    Messages:
    816
    Likes Received:
    402
    Trophy Points:
    62
    You must have my auditor. You'll be doing gage studies 24/7. Good luck. :)
     
  4. Eric Twiname

    Eric Twiname Well-Known Member

    Joined:
    Jul 31, 2015
    Messages:
    329
    Likes Received:
    232
    Trophy Points:
    42
    Location:
    Northeast USA
    ...or until a customer requires you to redo it...

    In my experience, validity is in the timepiece of the beholder, regardless of how right on point Miner is (and he is right).
     
  5. Emmyd

    Emmyd Member

    Joined:
    Jul 31, 2015
    Messages:
    38
    Likes Received:
    38
    Trophy Points:
    17
    Location:
    Tennessee
    I agree - usually your customer dictates this timeframe. Our largest customer requires the gage R&R be less than a year old at the time of the PPAP submittal. So, we redo ours once a year.
     
    WCHorn likes this.
  6. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    261
    Likes Received:
    168
    Trophy Points:
    42
    Location:
    North Carolina
    Miner is correct, but the question of "when should it be conducted again" has not been answered. And that's where the master parts come in. You should identify at least 2 master parts when you do your Gage R&R. One of the results of your Gage R&R, if it is done correctly, should be the gage uncertainty. Example: you have a CMM that is measuring a part and there is an ID on the part. The uncertainty is the +/-0.008 mm type number. Which means, any given time you measure a part, you "know" the ID to within +/-0.008 mm. When all this fancy studying is going on, you set aside master parts, usually painted a different color, and you measure them and record the values. So I pick myself two master parts, one on the low end and one on the high end. Let's say LOW has ID = 78.035 mm and HIGH has ID = 78.515 mm.

    Fast forward to "now" and you are wondering if your gage is still valid. Or maybe maintenance was done on it. Or maybe a fork truck bumped into it. Whatever. You know what you master parts were when it was good. You know what your uncertainty was when it was good. So you remeasure your master parts. If your new measurements are within the uncertainty from the old measurements, you are OK. If they aren't, you need to diagnose the gage and find out why. This can even be automated into the controller in the case of an automated machine. We run leak testers that every 100 parts, they automatically grab one of the two masters and check it. If the master is very different than last time, the system stops. And we fiddle with the leak tester until it measure the master correctly again.

    Word of warning - the master can "wear" depending on the measurement. So they may change over time with frequent checking.

    An exception - if the gage is being used in a root cause analysis, part of DMAIC, you should redo the Gage R&R as part of "M". That's just good practice. There's a lot of improperly done Gage R&Rs out there..... Unless for some reason you are CERTAIN it is OK. (Like you did one last week for a different study).

    And I have to state my continual annoyance with CMMs. You take a CMM with Program X and measure part X+ and pass a Gage R&R, you cannot just assume that on the same CMM Program Y measuring part Y+ is OK. The program itself is the largest contributor to accuracy.
     
  7. Candi1024

    Candi1024 Well-Known Member

    Joined:
    Jul 30, 2015
    Messages:
    129
    Likes Received:
    83
    Trophy Points:
    27
    Location:
    Pennsylvania
    I'd like to pause on this issue for a moment.

    Are you suggesting that the Gage R&R needs to be repeated for each part?

    Or perhaps you are stating that the program, as entered, needs verification? In other words, a "double check" that the program is entered per drawing specs.
     
  8. Eric Twiname

    Eric Twiname Well-Known Member

    Joined:
    Jul 31, 2015
    Messages:
    329
    Likes Received:
    232
    Trophy Points:
    42
    Location:
    Northeast USA
    Not going to speak for ncwalker...but toward the topic...

    GRR tests a test method...not a piece of equipment, not a part

    If I make 14 different parts and use three different programs on a CMM to measure aspects of them, I need:

    1. a GRR for program 1 measurements on a subset of parts that can be considered similar for that test method.
    2. same for program 1 for the other subsets...
    NOTE: If five of my parts have width between 0.5cm and 1.0 cm...I'll put them all in the same GRR. ...but not the 14cm part.

    3. same for program 2
    4. same for program 3

    CMM accuracy and repeatability is different across 10cm than it is across 1cm.
    It is different for circles than it is for lines.

    It isn't about the tool, it's about what you use the tool for, and how you use the tool, and the tool...all combined. ...The test method.

    I can write an awesome program #1.
    That doesn't mean I didn't totally goof on program #2.

    I remember beating my head against GRR with a CMM trying to locate the distance between circle centers.
    Simply using the closed spline object recognition instead of the circle object recognition shifted my GRR from 18 to 0.2....on the same CMM with the same parts.
    It wasn't the CMM, it was how I was using it for that part.
     
  9. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    605
    Likes Received:
    663
    Trophy Points:
    92
    Location:
    Maine
    I would add that it isn't just the dimension range that creates a unique GRR, but the feature type. a wall thickness that is curved and has very little clearance between adjacent features is much more difficult to measure than a straight wall thickness with plenty of clearance between adjacent features.

    And Candi1024, the issue here is a Customer or auditor that is requiring a GRR to be redone on some regular basis. This is a clear misinterpretation of the intent of GRR. Probably someone who is confusing calibration with GRR or is just a 'check the box' kind of person...
     
  10. QUALI

    QUALI New Member

    Joined:
    Aug 19, 2015
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Besides the requirements from Customer (usually to be done yearly, as we need to submitt PSW...), on our internal procedure we stated that the GRR should be repeated whenever there is a considerable maintenance, where gage componentes were disassembled, or , if there was a big change on the design of the Gage, so we need to confirm it is repeating properly.
     
  11. Eric Twiname

    Eric Twiname Well-Known Member

    Joined:
    Jul 31, 2015
    Messages:
    329
    Likes Received:
    232
    Trophy Points:
    42
    Location:
    Northeast USA
    That approach works, but I would push a bit on some of the areas...

    Considerable maintenance...I would recalibrate, not re-GRR
    Disassembled...I would recalibrate, not re-GRR
    Design change...well that's really a new gage now, thus a new test method, thus GRR is a good idea.

    JMO based on cost/benefit
     
    Bev D likes this.
  12. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    261
    Likes Received:
    168
    Trophy Points:
    42
    Location:
    North Carolina
    Eric Twiname is authorized to speak for me on the subject of CMM and Gage R&R. :)

    Exactly that. Just because the Brown and Sharpe guy comes in and says you have a volumetric accuracy of 0.002 mm on your CMM does not mean it is a good as a system. Because you could then turn right around and use a Sharpie as a probe tip in some stupid programming choice. What would the volumetric accuracy then mean?

    Most of us have been in the seat where we didn't like, say, a diameter a CMM reported and say the infamous "we need more probe hits." Why? If we mean to imply that the PROGRAM can affect the ACCURACY (which it does, and the only reason adding more probe hits gives us a better diameter) then that means the CMM/Program/Fixture combination stands alone as a Gage system and must be treated as one.

    We can definitely make some simplifying assumptions. If I am CMM measuring a men's shoe and I work up a good program that passes a Gage R&R, I am pretty safe in saying the same scheme (which includes number of probe hits, tip selection, acceleration of tip on approach, order of construction of alignment system) will also work for a childs shoe, because that change of scale is not unreasonable. But I would not say I can then assume I have a passing Gage R&R for a hat or a glove, which would require a different measurement scheme.

    But here's the rub....

    Let's say I have this program for a men's shoe, and I wish to convince myself that the child's shoe program is ALSO good. By the time I step through all the code ensuring this is the case, and tying up my valuable CMM programmer (most places have a small number of programmers and a large number of operators) I may as well go ahead and just do the darn Gage R&R on the child's shoe. It will probably take less time AND avoid the "human error" mistake of going through the code line by line. (Have you ever done such a comparison? Tedious doesn't begin to describe it.)

    Further, I may be a shoe factory, but I am not the customer. What happens if Sperry gives me an account for men's shoes. But the children's shoes are going to Buster Brown (yes, I am that old)? I do my Gage R&R all nice and neat for my adult shoe going to Sperry. Then when Buster Brown says "convince me you are controlling the child's shoe we buy" what will please them as a customer?

    I could say:
    a) Oh. Well, we did this Gage R&R on your competitor's similar part and it was OK, so it will be good enough for you....
    b) Here you are, Mr. Buster Brown, your very own Gage R&R all nice and neat.

    All these things need to be factored in when selecting your "simplifying assumptions."

    Final comment: We do an ANOVA Gage R&R where we measure 14 parts 3 times. Much better than the Xbar - R method where it is 3x3x10. It is OK because we hold the part by an inconsequential feature and measure the datums every time. If you use the other scheme of hold the part by the datums and dial in the fixture to the CMM, then you MUST include the operator effects because different people will clamp the part differently. But that's another soap box.
     
    Eric Twiname likes this.
  13. Eric Twiname

    Eric Twiname Well-Known Member

    Joined:
    Jul 31, 2015
    Messages:
    329
    Likes Received:
    232
    Trophy Points:
    42
    Location:
    Northeast USA
    LOL...thanks!
     
  14. Candi1024

    Candi1024 Well-Known Member

    Joined:
    Jul 30, 2015
    Messages:
    129
    Likes Received:
    83
    Trophy Points:
    27
    Location:
    Pennsylvania
    Thank you, both of you!

    I need to look at doing verification or R&R of some sort our CMM. However my situation is a bit different than yours. Ours isn't coded, but instead uses software so its easier to review. (which most likely isn't validated) We also only measure about 3-6 parameters. I am in medical device not automotive, and this measurement is done early on as sort of a double check. For this reason we don't consider it a "critical" measurement, as errors will also be found further down the production line.

    However I do need to verify that parameters are entered correctly, and that the readings are as accurate and precise and required.

    Side note, I hope to be able to move into a R&D/ New products quality position in the future. If only the designers had any idea what kind of restrictions they put on us when specs are tighter than need to be, and how hard it is to change once it has all been qualified.
     
  15. Eric Twiname

    Eric Twiname Well-Known Member

    Joined:
    Jul 31, 2015
    Messages:
    329
    Likes Received:
    232
    Trophy Points:
    42
    Location:
    Northeast USA
    Best of luck to you Candi,

    Something to consider:
    You mention that you don't consider it a "critical" measurement.
    Consider "Do I make any decisions based on these results?".
    That's usually the first trigger for me to be sure the inputs to my decision are accurate and reliable.
    Non-critical decisions still benefit from accurate data inputs.
     
  16. Candi1024

    Candi1024 Well-Known Member

    Joined:
    Jul 30, 2015
    Messages:
    129
    Likes Received:
    83
    Trophy Points:
    27
    Location:
    Pennsylvania
    I agree.
    It's not 'critical' because then this test would be put under more scrutiny per CFR 820. But that doesn't exempt us from making an accurate and reliable measurement. Really, any measurement that isn't accurate and reliable is useless.
     
    Eric Twiname likes this.
  17. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    261
    Likes Received:
    168
    Trophy Points:
    42
    Location:
    North Carolina
    Not trying to be combative, please rethink this statement. It has nothing to do with being in automotive. If you have a gage that is measuring something and giving you continuous output ... meaning a value on some scale, not OK/NOK or Pass/Fail ... then a Gage R&R (done properly) will tell you if the measurement system is effective or not. There are some nuances, such as if the gage is destructive to what is being measured, but there are also Gage R&Rs that meet those needs as well. I just want to be clear, it is NOT an automotive thing. I also want to point out something about Pass/Fail - if you have some medical thingy that is using a transducer or pickup, but the device itself is coded to give you a Pass/Fail with some red light or green light, you can still do a Gage R&R on the voltage of the transducer. Because an analog signal, that has different values, is getting converted to a digitial output (pass fail) with some programmatic or circuitry behind the scenes. My last comment in my Gage R&R tutorial - a common mistake I see EVERYWHERE - when you buy a gage, put into the contract that you don't pay until the gage PASSES A GAGE R&R. Many times people are performing some thing that they are the experts in, and they need a gage, but they are NOT the experts in the gaging itself. You're an expert baker, and you know you need to have a well controlled oven for your souffle (holy crap! I spelled souffle right the first time!!!), but you have no idea how a thermocouple works. So you buy a controller from an oven expert and ASSUME he's a thermocouple expert. 6 months into your business, you can't get souffles right and find out the controller for your oven can't control the temperature nearly well enough. But that guy has been paid in full, and you can't get his attention. Connect his final payment milestone to a Gage R&R. If he balks, he may not be the expert controller supplier you thought. I see this bite people in the butt over and over again. You do not have to be an expert on the gage at all to use a Gage R&R to evaluate it as a measurement system. You DO have to understand the proper procedures of conducting a Gage R&R.

    I hope you realize, the further down the line you catch the defect, the costlier the defect is. It may be that the cost of measuring it upstream far exceeds the added cost of catching it downstream. It is both a quality and a business decision as to how much gaging one does.

    Like most quality folks, I salivate at the thought of being able to beat up a designer over something. :) But let's be fair. In THEIR world, loosening a tolerance typically initiates costly and monotonous testing. Tightening one typically means reissuing a print. The fault lies higher up in the chain of command, where someone should be evaluating the savings by loosening a tolerance against the testing cost of proving it out. It's a decision that should be made by facts, but frequently relegated to opinion.
     
    Bev D likes this.