MSA (measurement system analysis) procedure on CMM machines

I use two CMM machines and sitting here wondering about the correct use of our MSA procedure.

I will outline a typical procedure and maybe someone can give me some feedback:

We measure car interior components, a lot of "soft" parts which makes the MSA quite difficult.

1. We are two operators who measure ten times each on a single part, why would we use more parts ? is it not the R&R of the fixtures & operators we want to test not the test part itself, are there any demands to use more than one part per test ?.

2. We measure 6 points, two in each direction - X, Y and Z, usually on a flat surface with quite high tolerances - if there are any such surfaces. Is there anywhere that states that difficult areas with small tolerances should be measured ?

Thanks for any feedback

We have two too!

Timmot said:

I use two CMM machines and sitting here wondering about the correct use of our MSA procedure.

Click to expand...

Timmot said:

I will outline a typical procedure and maybe someone can give me some feedback:

We measure car interior components, a lot of "soft" parts which makes the MSA quite difficult.

1. We are two operators who measure ten times each on a single part, why would we use more parts ? is it not the R&R of the fixtures & operators we want to test not the test part itself, are there any demands to use more than one part per test ?.

2. We measure 6 points, two in each direction - X, Y and Z, usually on a flat surface with quite high tolerances - if there are any such surfaces. Is there anywhere that states that difficult areas with small tolerances should be measured ?

Thanks for any feedback

Click to expand...

Timmot

I'm surprised no one has answered yet. I'm no expert but here goes.

As you say the operator has no influence, they just set and forget.

Do you have the AIAG MSA version 3 book? It covers this, I'm at home so I can't quote the pages.

Basically we run 10 parts using the CMMs as "operators". Run them against each other. Then call your CMM builder to find out why they are so different. We've been at it about a year now.

1) You need more than 1 part. The parts should span the range of typical process variation (or specification), even if you have to make them at great cost. Search this thread for reasons why.

2) Soft parts with tight tolerances...you are likely to get "bad" MSA results. Again, a search here or the AIAG MSA book will tell why. State of the art measurement systems on parts with little variation will show "bad" MSA. Of course it will, you're trying to measure close to nothing at the limits of your gage. The CMM is not "bad" however!

The MSA book offers this helpful advice when this happens "develop better systems". If you do, I'll buy the first one from you.

Good luck!

Timmot said:

I use two CMM machines and sitting here wondering about the correct use of our MSA procedure.

I will outline a typical procedure and maybe someone can give me some feedback:

We measure car interior components, a lot of "soft" parts which makes the MSA quite difficult.

1. We are two operators who measure ten times each on a single part, why would we use more parts ? is it not the R&R of the fixtures & operators we want to test not the test part itself, are there any demands to use more than one part per test ?.

2. We measure 6 points, two in each direction - X, Y and Z, usually on a flat surface with quite high tolerances - if there are any such surfaces. Is there anywhere that states that difficult areas with small tolerances should be measured ?

Thanks for any feedback

Click to expand...

Are the CMMs DCC (Direct Computer Controlled) or manual? If they are DCC, the two CMMs themselves can be treated as operators. If they are manual, the two operators must be treated as operators. You could throw in the second CMM with one operator as the third.

You want to measure multiple parts for at least two good reasons. One, you want to see part variation in the study. Second, part of the repeatability variation is from placement of the parts into the fixture. Multiple measurement of the same part would not accomplish this.

You may want to perform the MSA on each of the 6 points and coordinates. You may find that the quality of the measurement varies significantly depending on the location and coordinate direction. This would require more analysis, but it sounds like you are already collecting the data on all six points.

Since you measure soft parts, you may consider performing a DOE with speed of hit, ball diameter, probe length, trip pressure, etc. as control variables, and MSA 6*StdDev as the response.

Thanks for the feedback,

We do have two automated and one manual CMM and a faro arm with a laser.
All MSAs are done on the automated CMMs (LK).

Miners suggestion about doing a DOE on the CMM is a good idea, when (or if), I have the time this would be an interesting test.

I got hold of a copy of the MSA handbook, version 3 and realized we were using calcalations from an earlier version.

I have read the book a couple of time which has helped me understand why we have to use more than one parts and we have now set a minimum critera of 2 operators, 5 parts and two measurments on each part. I had no previous information on MSA what so ever, just an excel spreadsheet which i filled in.

We will continue to use the tolerance method as standard I am still wondering about what tolerances to set, I can't find any information about this in the manual, I must be blind (or fell asleep while reading), anyone ?

If, for example I was measuring a rear seat which has a tolerance of +/- 10mm over 98% of its surface and +/- 1mm for 2%, does the MSA manual say should I then use +/- 1mm for the entire test.

Thanks again for any help

I'd be real interested to hear if you have any success with that stratagee.

....We certainly did not.

There are a number of related threads here. Please check them out.

Timmot said:

I am still wondering about what tolerances to set, I can't find any information about this in the manual, I must be blind (or fell asleep while reading), anyone ?

If, for example I was measuring a rear seat which has a tolerance of +/- 10mm over 98% of its surface and +/- 1mm for 2%, does the MSA manual say should I then use +/- 1mm for the entire test.

Click to expand...

Use both. You may find that the CMM is adequate for 98% of the surface and inadequate for the 2%. Remember, GRR compared to tolerance only tells you if the CMM is adequate for Pass/Fail decisions.

The GRR as %TV tells you if it is adequate for process control. MSA 3rd Ed. places more emphasis on the process control aspect.