Very interesting article concerning MSA (Measurement System Analysis)

[Black arrow]

Hi there,

When i read this article by Q-DAS director Dr. Ing. Edgar Dietrich i really changed my mind when it becomes to selecting reference figures and how they affects the result in the end.Interseting what Q-DAS director recommend and what arguments he has for that choice. This can be read from "Reference Figures affecting the result" and forward.

http://www.q-das.de/en/current/piq-journal-32010/msa-4th-edition/

 

[Black arrow]

Hi again,

I would be glad if someone comments the article. I agree with him that selecting parts is critical and i think in the end he preferres tolerance as to be compared to. Seems to be resonable. Tolerance is constant, easier to compare measuring equipment with measuring equipment, easier to compare new ones with old,....
What do you say about this?

But i do not know what he mean and disagree with him what he wrote under "Part variation".

" The proposed and preferred part variation is a resasonable reference figure, for sure. However, in order to use this figure, the taken parts must be evenly spread within tolerance and some parts have to lie outside tolerance limits"

This way is not correct way according to manual reference MSA 4 th ed (AIAG) or?

 

[Miner]

I disagree with him on two points. The section on Parts Variation is dead wrong. You should never hand select parts in such a manner. Secondly, you do need to compare measurement variation to the actual process variation. A gage may be suitable for inspection, but be totally useless for SPC because it cannot "see" the part variation.

 

[WCHorn]

The section regarding part variation makes more sense if applied to attribute gage studies. For those studies, you need product spread throughout the tolerance and even outside the tolerance to make sure the gage can detect good from bad.

If the gage is a variable gage used only for inspection, then comparing the R&R to the tolerance is fine. The location of the samples within the tolerance is not important.

If the gage is a variable gage used for statistical process control, then I agree with Miner. You must know how the gage performs versus product variation.

 

[Black arrow]

For sure, It is important to select parts that are representing the process variation. However I find i it quite difficult to select parts in practice and we have lately been using standard deviation from earlier made cap.study. Is this correct approach according to AIAG guidelines?

 

[Miner]

Yes. You may use the standard deviation or the Cp index from a prior capability study. This is specifically mentioned in the 4th edition manual.

I strongly recommend using the former approach specifically because it is difficult to find 10 representative samples.

 

[Black arrow]

Ok so if i understand you correct, using % Process option in Minitab do not conflict to MSA reference manual.

Thanks!

 

[Miner]

That is correct.

 

[Bev D]

i would also be leery of this article simply because it's written by someone at the company that wants to sell your their software.

my recommendation is to stick with independent sources such as donald wheeler...