Gage R&R Sample Size - Sample Range for gage R&R - What is the requirement?

Subject: Re: MSA /Staples/Templeton/Johnson
Date: Thu, 29 Jul 1999 09:50:44 -0600
From: ISO Standards Discussion

From: Kevin Johnson
Subject: RE: MSA /Staples/Templeton/Johnson

Patricia,

Regarding your message below,

> From: Patricia Staples
> Subject: Q: MSA /Staples
>
> Has anyone heard of a requirement within the MSA that the 10 pieces checked
> for R&R have to be within spec?
> My customer said that my R&R was incorrect because I was checking parts
> that were out of spec.
> I told him that I needed to know that my gage was repeatable throughout
> the spec and below the spec. Any comments?
>

The important range to test across is not spec, but rather the range of product variation the instrument will see. You will want to know the components of variability across all "pieces" you will be measuring with this device. If all you product is always in spec, then maybe spec range is meaningful, but only beacause this is the range of the product, not just because of the spec range itself.

However, having said that, it is possible that, for a particular device and product, that the variability could more (or less) outside the spec than within it. So, the R&R numbers could look better or worse, depending on this range. And, this could concern the customer.

I understand your position clearly. I do not really understand the customer's. Do they understand the funamentals behind MSA?

Kevin Johnson
Sony.com

[Marc]

Any contemporary 'thoughts' on this question?

[wmarhel]

One approach I like to take is that once a process has been proven out (man-machine-method), than any new people hired (or even as part of the interview process) can be tested to discern their ability. I have always kept a few parts as "golden samples" for this very purpose. New gages or methods can also be proven out this way.

It's important to make sure that the entire system can reproduce accurate, repeatable measurements so if a part happens to fall outside the specification then that is okay. The tool should have had the accuracy proven during the calibration process, but it's usually a lot easier to be accurate in a controlled environment versus measuring parts that have flashing, un-finished surfaces, etc.

In my opinion, the biggest focus should be on the repeatability of the individuals as a large difference identifies areas of opportunity for training. If its an accuracy issue with the tool than that can be rectified easier, although possibly at cost, through the repair of the tool or purchase of one that is more appropriate to the task. Even if the method is cumbersome and FUBAR, there should still be some consistency between the individuals.

[Geese]

As far as actual numbers for the sample size go, the MSA minimum number to be used is 15 to ensure confidence according to MSA 3rd edition, even though the forms and K values are made out for 10. We use 10 ourselves and have never had a problem.

Concerning parts out of spec, the customer might not know that its not the purpose of the GRR. The intention of the MSA GR&R is to show how good your measurement technique is conerning your gages and people. However, there is a value calculated called Rp (Average Part Range) that comes into the calculations. This value is based on the range "between parts", not appraisers. Also, the top of the form for the GR&R does specify the part spec or is supposed to (.113 to .115 for instance on a wire diameter). These spec values do not come into the calculations at all from what I've seen though. In the end, the customer is always right or so I've heard and I guess to someone who doesn't know the why, it would be a redlight if they saw a measurement of .116 on the above mentioned wire diameter.

Besides those 2 considerations, the last thing you need to make sure of is that your ndc is greater than or equal to 5, your %GRR is less than or equal to 10, and none of your Ranges on repeatability exceed the calculated UCL (10:1 rule helps here). These are the play it safe numbers. Some customers are less strict, but in my experience its best to meet the maximum standard in trying to cut down on confusion when dealing with different customer needs.

[Caster]

Originally Posted by TheOtherMe

Subject: Re: MSA /Staples/Templeton/Johnson
Date: Thu, 29 Jul 1999 09:50:44 -0600
From: ISO Standards Discussion

> Has anyone heard of a requirement within the MSA that the 10 pieces checked for R&R have to be within spec?
> My customer said that my R&R was incorrect because I was checking parts
> that were out of spec.
> I told him that I needed to know that my gage was repeatable throughout
> the spec and below the spec. Any comments?

I understand your position clearly. I do not really understand the customer's. Do they understand the funamentals behind MSA?

Kevin Johnson
Sony.com

The customer may actually have a really deep MSA understanding.

MSA v3 page 74 says "select part for study that span the entire process operating range".

Process range compared to tolerance is capability.

If you tested parts out of spec...perhaps you are showing your process isn't capable?

Maybe that is what the customer was driving at?

Or more likely, they haven't a clue, and just got lucky.

[Geese]

Yeah,

I agree with you, but the intention of the GRR is the measurement abilities of the gages and people. Capability study deals more with the part itself falling into spec. I try to keep the 2 separate.

We carry many of the same parts and keep on file GRR's we update annually or semi-annually to cut down on time taken to complete one. Whereas the capability study needs to be done on the individual production runs.