Hi

The R & R studies, should be made to all equipment's refereed in the control plane, or only the ones that measure critical or significant characterises



------------------

Only measurement systems used on the "key / sinifigant characteristics" must have Gage R&R studies done on them. However, some auditors would LIKE to see them done on all, but it is not required.

I am sorry to dissapoint you but 4.11.4
Appropriate statistical studies shall be conducted to analyze the variation.........This requirement shall apply to measurement syatems referenced in the control plan.

Ok

All measurement systems referenced in the control plan, should have appropriate statistics studies, but if a study is made for a measurer system, for instance a calliper, this study is valid to that measurement system although it is use in a lot of control plans?


------------------

Hi Luis:

We have used the same results from a measurement system study for other areas, as long as we determined that the methodology was the same. Likewise, we have used one 0-6" caliper and applied the results to other areas that use 0-6" calipers, wven if they are not the exact same calipers.

You should be careful when determining when to do this. If you have 8 stations and 8 sets of identical measurement devices, measuring like features, you could decide to only use a couple of these for the R&R determination.

Also, the actual R&R result - the actual error, not the % of tolerance result - could be applied to another tolerance on another part, again if the measurement ssytem is similar enough.

I agree with this for EV not AV. Let's say you have a caliper with an R&R on a part dimension from part X measuring 10.0 mm. It is a flat, easy to measure dimension on a piece. Then, with the same instrument, you measure a 'hard to reach' dimension on another part of 0.5 mm. While the EV should remain approximately the same (though different part of the equipment range), AV can vary significantly.

What I am interested in, and heard a bit about at a training class for MSA...

How much stock in the Gage R&R studies can their be? I mean, for instance, if I am measuring a part that is 108mm X 25mm, and it is a pressure-sensitive fabric... it isn't REALLY possible to expect that three people using vernier calipers are going to measure each of 10 parts in EXACTLY the same spot as each has previously measured... let alone one person to the next.

Think about it... that type of "exactness" is a requirement of an effective measurement system study... or was I misinformed? If I measure the 25mm dimension on the left-third of the part on one pass, the middle on the second, and the right on the third... how does that truly impact the result?

Thoughts? Comments?

ALM

What does it means EV and AV?

EV = Equipment Variation

AV = Appraiser Variation

ALM

In the case of the pressure sensitive cloth, measurement equipment and techniques should have been looked at during the early design phase. If the measurement is important you would pick a measurement instrument appropriate for the material and conditions. A calipers probably wouldn't be the best tool for material. The point is to find out whether it's even worth taking a measurement. Let's say the gage R&R is 25%AV and 30%EV - that's 50% of your tolerance. Now the question becomes whether your measurement means anything or not.

Gage R&Rs are a tool - no more and no less. There is also the aspect of measurement uncertainty to consider.

Don't read into this any more than is here. For example, there are alturnatives to Gage R&R - but you have to be ready to justify why you use what ever tool you use. Gage R&Rs do not make sense for every measurement condition. Automotive just happens to not know much else.

What other kind of studies that are refered in your answer?

Could you send me some examples so a I can decide if it is worth.

Were can I find information about that?

Thanks in advance.

Luis Alves

------------------




[This message has been edited by Luis Alves (edited 08 July 1999).]

Analysis of Variance (ANOVA) is one. Doing analysis plots is another. Automatic gages you do an R on - not and R&R.

Do you have a copy of the MSA manual?

The bottom line in the automotive world is that R&Rs are very, very acceptable and 'understood'. If you decide to do something else, you had best be ready to explain how your methodology fulfills the intent of MSA. And remember that to fulfill the intent you should be able to show that measurement is considered and planned from early in the APQP process.

To get back to: The R & R studies, should be made to all equipment's refereed in the control plane, or only the ones that measure critical or significant characterisesI have seen companies with 2 control plans: One with customer critical characteristics and one with their own 'special' characteristics because they were told "If it's on the control plan it's subject to R&R (or other methodology)". This is, to me, streching things (not to mention complicating things) a bit. There is a column on the control plan format recommended in the APQP manual labeled 'Special Characteristic Class' - use it.

For more info, do a search here in the forum for "Critical Characteristic".

[This message has been edited by Marc Smith (edited 08 July 1999).]

So in your opinion, only R&R studies should be conducted to special characteristic, but as you can read in Howard Atkins reply 3rd Edition of QS9000 obliges that all measurement systems refereed in control plan should have appropriate statistical studies.

In our plant we have two kinds of characteristics, the one that are refereed by our customer as special characteristics and the internal ones, which are decide in the APQP process and give us the idea how the process is been conducted.

Which ones should have R&R studies or others, and how can we accomplish QS9000.

Thanks in advance.

Luis Alves

This is why some companies have 2 control plans or include some characteristics in 'work instructions'. The key here is the definition of critical vs special or other characteristics. Your customer may or may not have critical characteristics - I have seen situations where no critical (or special) characteristics defined.

I have in front of me a control plan with probably 20 entries, but only 3 are designated critical and thus only R&R studies are done on the three critical characteristics.

This is another 'registrar interpretation issue'. LRQA cited a company for no control plan where there were no critical or special characteristics defined by customer or internally. The auditor said "Well, you have to have SOMETHING!" Zat so?

In addition, 4.11.4 says: This requirement shall apply to measurement systems referenced in the control plan It does not say to every line item on the control plan. I have seen companies leave out defined measurement systems on items that are not 'critical' and thus there are line items in the control plan where, for example, the Control Method is 'First Off Report' - not a measurement system per se.

On the other hand think about a critical characteristic such as presence of a hole. Control is by light beam. Here, as you can see, you have a critical characteristic for which no measurement device is used at all. How does 4.11.4 fit here? OK - it's an attribute check, but think about it.

Have I confused the issue for you or has this helped?

There is a way of doing R&R for attribute.
I believe that 2 control palns is wrong as you then can get confused as which is correct.
If your customer comes to do Run & Rate where he checks that all the documentation is in place and you are working according to the PPAP, will you then use your 1st or 2nd control plan?
The problem with R&R is that to get good results you must have a wide spread of measurements, that means you must have out of tolerance parts, in the MSA book it says that if you make a control chart at least half of the points must be out of control.
The mathematics of statistics cannot deal with non-variance.
It is possible to work with family of equipment, one type of caliber for all measurements, as I see it the more important factor is the variance between operators. The point as to where the part is measured is a good one but the test is that the measurement point is the same otherwise the measurement has no value. This problem also applies when measuring non metalic parts that are "soft" and the amount of pressure that the operator uses is a critical part of the system.
The demand is reasonable IMHO the practice is impossible.

Howard,

Thanx for the info. I agree that 2 control plans is a bad idea, but I have seen it done. They were exclusive so it was not a matter of 'one or the other'.

oh, i am confused with the answer.
who can show me the direct result?