MSA using tolerances - Page 116 of the AIAG MSA manual

[Timmot]

I am a bit confused over one point in the MSA manual . . .

I have updated our old MSA speadsheets to the newer MSA book issue 3 and have checked and double checked all the results against other peoples results and everything seems fine . . . . until . . .

We performed an MSA test yesterday and got great results, then I saw on page 116 of the manual that when using the tolerance method I have to replace the TV value by the tolerance divide by six in %EV, %AV, %GRR, I hadn't done this division.

SO I made the adjustments:

%EV = 100 [EV/(tolerance/6)]
%AV = 100 [AV/(tolerance/6)]
%GRR = 100 [GRR/(tolerance/6)]
%PV = 100 [PV/(tolerance/6)]

. . . .and ofcourse got much worse results.

I am confused because the examples I have got from these forums have not replaced the TV values with tolerance/6, only the full tolerance, or do people divided the tolerance before entering the value into the spreadsheet.

I am greatful for some clarity on this point.

Thanks

Timmot

ps. does anyone have a spreadsheet example of the Anova method, thanks.

 

[Howard Atkins]

As one of those responsible for posting examples of edition 3 I admit that I had not paid attention to this issue.
I did not study the text properly and in fact used the previous worksheets and just changed the constants.
I know see that in the 2nd edition (page 60) there is no requirement to divide by 6 as there is in the 3rd edition (page 117).
Have I been misleading the public?
Why is there such a big difference?

 

[Timmot]

Thanks for your reply & ofcourse your right,

I checked the version 2 manual and the by six division is not stated, this seems very odd to me . . . . I think I'll go back to issue 2 and save me and my company a lot of headaches and money as I'm unlikely to every get an acceptable MSA with issuse 3 . . . .

. . seriously though, how can such a dramatic change be made, it seems to me a misstake has been made in one of the issues, the question is which one.



Timmot

 

[Miner]

Unfortunately, I do not have ready access to my MSA, 3rd ed., so I cannot cite the correct pages.

If you take the formula that you cited and divide by 5.15 instead of 6, you should get the same answer as in the 2nd edition. The MSA manual made two changes that affect this example.

In the 2nd edition you multiplied the AV, et al, by 5.15 and divideby the total tolerance. In the third edition, you do not multiply by 6, so you must divide the tolerance by 6.

This simplified the %GRR calculations, which they now emphasize over the P/T ratio,because you did not multiply and divide by the same number. It also reflects the push to 6 instead of 5.15 as a multiplier. However, it resulted in confusion to users of the P/T ratio.

 

[Timmot]

I'm not sure from what version I have updated,
We have used an old excel spreadsheet which is about 5 years old, anyway I can not find anywhere which uses the 5.15 calcalation you mentioned, I saw the K1 and the K2 values differ a lot maybe it has something to do with this!.

I changed the division of 6 to 5,15 in the new spreadsheet I have made using issue 3 but it does not give the same results as the old spreadsheet - the values are higher in the newer, maybe the old spreadsheet has some faulty calculations or are from issue 1.

I will just have accept the miss I made with the division of 6 and that I will probably never get an approved MSA - We make large parts of foam and plastics measured on gigs in CMM's . . . . . . well at least MSA keeps me busy and in a job!

 

[martin elliott]

5.15 change to 6

I'm not sure from what version I have updated,
We have used an old excel spreadsheet which is about 5 years old, anyway I can not find anywhere which uses the 5.15 calcalation you mentioned, I saw the K1 and the K2 values differ a lot maybe it has something to do with this!.

I changed the division of 6 to 5,15 in the new spreadsheet I have made using issue 3 but it does not give the same results as the old spreadsheet - the values are higher in the newer, maybe the old spreadsheet has some faulty calculations or are from issue 1.

I will just have accept the miss I made with the division of 6 and that I will probably never get an approved MSA - We make large parts of foam and plastics measured on gigs in CMM's . . . . . . well at least MSA keeps me busy and in a job!

The information regarding the change to a 6 from 5.15 is described within page vi, i.e. after the MSA 3rd edition Quick guide, just before the table of contents.

Obviously this change degrades the value reported but increases the "safety" of the result.

If you are trying to modify MSA 2 speadsheets you are possibly missing some of the changes and maybe even missing the changes in emphasis on chart interpretation. The metric represents only a small part of the study and to get the most out of the study there is a lot more to look at. Fortunatly most customers are only interested the metric.

Martin
Older but no wiser!

 

[Timmot]

I had missed an imput on the old excel spreadseet and now get the same values as issue 2 by dividing the tolerance by 5,15 in the 3rd issue, also I got back a copy of issue 2 and can now see where the calulations differ, it was as I thought in the K values, I can now see in issue 2 the D2 value was divided by 5,15, but not in the third . . . .

Thanks for the help

Timmot

 

[Miner]

I had missed an imput on the old excel spreadseet and now get the same values as issue 2 by dividing the tolerance by 5,15 in the 3rd issue, also I got back a copy of issue 2 and can now see where the calulations differ, it was as I thought in the K values, I can now see in issue 2 the D2 value was divided by 5,15, but not in the third . . . .

Thanks for the help

Timmot

I found my manual. If you take the K1 and K2 values from the 2nd edition, and divide them by 5.15, you will obtain the new K1 and K2 values found in the 3rd edition. That is why you have to divide the tolerance by 5.15 to get the same answer as in the 2nd edition.

 

[Timmot]

From a helpful lady at AIAG:

On page 117, the first line the text reads... “value of tolerance divided by six” in the denominator. This effectively degrades the GR&R score as calculated by using the formulas in MSA2. Is this calculation correct?

Your logic is correct but your conclusion is not since it is based on an incorrect assumption. The problem is not that the 3rd edition inadvertently degrades (i.e. overestimates) the GRR but that the 2nd edition underestimates it. The 2nd edition went to a standardized 5.15 factor even for the total variation (see conversion of historical process variation on pg 60 top MSA2). This same conversion should have been applied to the tolerance (assuming a process with a Cp =1). It was not and consequently the GRR calculated using the MSA2 approach is for a process with a Cp of 0.86 rather than a minimal Cp =1.


Now, if you want to continue to use the underestimated GRR based on the MSA2 approach you can, but you need to get your customer agreement with this approach."

 

[CREED]

Msa Manual Third Edition

DOES ANY ONE KNOW WHAT THE RECOMMENDATION FOR PV% SHOULD BE. ACCORDING TO THE MSA MANUAL, GRR% SHOULD BE UNDER 10% GOOD, 10%- 30% MAY BE ACCEPTABLE, OVER 30% NOT ACCEPTABLE. ndc SHOULD BE GREATER THAN 5. BUT IT DOES NOT GIVE ANY GUIDE LINES FOR PV%.