Gauge Repeatability and Reproducibility (Gage R&R) on all shop floor gages?

I want to know whether GR&R studies had to be conducted on all gauges used in the shop floor or only for a family of gauges. (Like 0-25mm Mics, 0-150mm Verniers etc..) ie.. one instrument from that family. I think that it make sense to conduct GR&R for one instrument from one family because what we are trying to conclude is whether that instrument is suitable for measuring a product characteristic.

You don't have to *do* gage R&R on anything. You can use other studies.

The point is MSA as a high level system has to be considered. Basically you address all gages, etc., which you take measurements with which 'affect quality'. In the case of inspection and test gages, typically Gage R&R is part of the standard studies.

You start hitting touchy ground when you want to do an R&R on one gage as evidence of family 'compliance'. If you want to get down to the nitty gritty it can be, and often is, argued that a study has to be done on each gage used for inspection and test on the actual parts and with the actual operators - this is quite often the case.

And you have to look at the family of which you speak. Are they all the same brand from the same manufacturer and are they the same model (etc.)? I saw one situation where a 'qualification by similarity' failed.Same gage and model but one was 5 years older than the first. From that there were/are a lot of reasons why they may *not* be alike.

We come back to common sense and an implied understanding of MSA. As you go through design (APQP) youi have to look at the situation as a whole - What precision is necessary, etc. This is where you decide what you have to do to 'prove' your measurement and test equipment as well as the environment it is used in and such.

You have to be ready to explain to the auditor what you're doing and why. Proof (studies) may or may not be in order, but it appears auditors are wanting more and more evidence - so specific, extended studies will probably be more important in the future.

Can you 'qualify by similarity or family'? Yes - if you have good evidence that your hypothesis is true.

If you are a QS-9000 certified company or you have plans to be the answer is simple. All equipment defined in the control plan must have MSA, which includes GRR. Calipers and mics will need GRR. It is not acceptable to perform MSA on one gage, and have it represent a family of gages of the same type. Remember GRR measures the total variation in a specific gage not a family of gages. No two gages are the same, therefore they will have different levels of variation.

Christian - I disagree with you here. Some common sense is involved. You do not have to do a gage R&R on every instrument on the control plan. It was not long ago that all the big 3 wanted was a gage R&R - it was a defacto standard. Now it came to pass that the big 3 found out about Measurement Systems Analysis. Well, they knew it was there (sorta) 'cause there was this great AIAG manual.

When auditors came and checked out the requirement of QS9000, they really (early on) just looked for common sense, a base understanding of MSA (is the rightr equipment chosen, and such), a strong calibration system and gage R&R. Over the last year most of the registrars are stressing MSA more. I'm not sure why. I know the first report to come to me was that UL was acting an ass again (I get lots of 'complaints' about UL).

I just went thru this with LRQA at Motorola in May. We did not graph studies of bias and such. I'd go into detail but I've typed responses to these MSA issues several times already that I refuse to do it again.

If you take a cruise to ***DEAD LINK REMOVED*** you will find this discussed a number of places.

In the 'new' QS9000 MSA is addressed in 4.11.4 where it states: "Appropriate statistical studies shall be performed..." Keyword = APPROPRIATE

This leaves it to you to do what you can justify as 'appropriate'. If an R&R is not approprite, then it's not appropriate.

We come back to this: Can you explain what you are doing and why?

Hell - most of the auditors I've come up against are MSA stupid anyway. At best most understand one part of MSA as a high level system - they understand Calibration Systems Requirements - NOT MSA. One of the most important parts of the whole thing is upfront - determination of appropriate M&TE - how many autitors have you heard ask to be taken thru the selection process?

My 10 cents....

The question was asked, do all gages have to be studied? My question is, do all products have to be studied? We have gages that are used for a bazillion products, many having different specs. Our data collection is also unusually expensive.

Is there a simple approach to this? Going back to Marc's comments, "what percision is necessary," we need a percision that is enough to prevent significant type I and type II errors. I define this as, "RR cant be more than 30% of tolerance." If my definition holds, I should only have to do studies for the products with the tightest specs. As long as they meet my rule (<30%), I should be ok right?

Also, if I find that many gages are above 30%, but I have action plans in place and work being done to get to the below 30% mark, can I still be certified?

Marc-

Actually you do agree with me; I guess I didn't make the distinction between MSA and GRR. What I meant to say is that a compny must use MSA on all IM&TE in the control plan. MSA not necessarily means GRR. As an example of what I mean, the last QS-9000 registration audit I performed I wrote the company up for not following the MSA manual eventhough they had done GRR studies. Their equipment was mostly calipers. Upon further investigation this particular company did not evaluate linearity, bias, stability, resoultion, range of use etc... How could they evaluate total variation (GRR) if they did not even know the variation inherent in their equipment?

When it says in QS 4.11.4 that "this requirement shall apply to the measurement system referenced in the Control Plan" can I submit results of a system that I have used in another submission or does it have to be specific to the part I am submitting?

Typically R&R is done (or other validation of the system or instrument) on the item as is on the control plan. But - you have to look at frequency and such as to what studies you have to do at all.

And - depends upon the measurement being taken and the system as a whole with consideration to the techniques you are using to validate. For example, R&R is on a specific dimension of a specific part (or multiple dimensions in cases such as a with a CMM - which is really an R since there's only one operator) per the control plan line item number.

I think, Howard, we're back to some common sense. As an auditor I typically expect to see a match with the control plan, though.

Why don't you start a new thread with some details and we can bandy it about.

Regarding GRR, we have had success with our QS9000 auditors demonstrating "family of gages" as they apply to "family of parts." Also we have multiple gages of similar type, and we select random gages within the family when GRR are performed. Our control plans are primarily for the "family of parts," so a "family of gages" approach fits.
Re frequency of GRR, our Ford STE was in today and stated we need to "update our GRR every year, as stated in the AIAG."

I wouldn't have a problem with that approach. You can prove statistically that which gage of the family of gages makes no different in the result. I think some auditors would not have this paradigm, however.