Measurement comparison for CMMs?

[Graeme]

I have been contacted by someone looking for something that is not in my area of expertise ... so I hope some one can help. since I am not familiar with the work area or equipment, I hope I am interpreting this correctly -- and that this forum is appropriate.

A company has a CMM that is calibrated by the manufacturer. There are some questions about measurements. I believe they are looking for some kind of measurement comparison program, an arrangement is send to different participating organizations to be measured. The measurements are reported and analyzed anonymously. the ressults are an overall set of values and uncertainty for the dimensions of the artifact, and a report to each organization about how well they perform compared to the others.

Is there such a program that anyone knows of?

TIA,

[RosieA]

I've never heard of this program, but we just bought a new CMM and after it was set up and calibrated by the manufacturer, we did a Gage R&R study to further validate it.

I'm not sure why it would be necessary to validate it outside your own operation, unless there is a concern about CMM measurements in general. I know we have non-believers in my company, despite doing a gage R&R comparison study to manual gaging that laid to rest any questions.

Darn Luddites!

[Claes Gefvenberg]

In Reply to Parent Post by Graeme

Is there such a program that anyone knows of?

We did something similar with the tensile testers in our group. I don't know the details, but doing the same thing for CMM's could have merit. It feels good to know that results match... or not, in which case it's important to find out why.

/Claes

[gszekely]

Hello Graeme !
My interpretation may be wrong. Here it is. We use video CMM's for plastic part measurement.
I have met the situation, when a tool was approved, based on product dimensions, and when we got it, my measurements were out of tolerance. We use in different factories, several types of CMM's, fixtures, are manufactured in different places, and the measuring prgrams are build by different persons. So shortly, when I have a global project, when parts are manufactured in different plants, and dimensions are critical for assembly, I do the following:
Prepare one set of samples
Run gage R&R in my factory, and I'm participating in it.
My mesurement is the standard
Then send samples, to first second and third factory, and they perform the measurement on the same samples.
For me the most important, is how far are they, from my measurement, and not the R&R results.
Just my approach, to the problem described above, if I did not missunderstud.
I suppose that you know, how a video CMM is calibrated. There is a standard program, run on a calibrated etalon.
György

[Miner]

I remember seeing an ASQ publication years ago that dealt with inter-laboratory correlations. I don't know if it is still available.

There a several other options that come to mind. One would be to perform an MSA using the different CMMs as different operators. The Reproducibility would be the measurement variation between CMMs.

Another would be to scatter plot, box plot or perform regression analysis between labs.

[Jim Wynne]

In Reply to Parent Post by Graeme

I have been contacted by someone looking for something that is not in my area of expertise ... so I hope some one can help. since I am not familiar with the work area or equipment, I hope I am interpreting this correctly -- and that this forum is appropriate.

A company has a CMM that is calibrated by the manufacturer. There are some questions about measurements. I believe they are looking for some kind of measurement comparison program, an arrangement is send to different participating organizations to be measured. The measurements are reported and analyzed anonymously. the ressults are an overall set of values and uncertainty for the dimensions of the artifact, and a report to each organization about how well they perform compared to the others.

Is there such a program that anyone knows of?

TIA,

This is known as "round robin" testing. A group of companies measure the same part--and the results are analyzed. I've attached a research paper from NIST on the subject, which might shed some light. I think that more information may be obtained from CMM manufacturers, who are likely to have similar information.

[Miner]

In Reply to Parent Post by JSW05

This is known as "round robin" testing. A group of companies measure the same part--and the results are analyzed. I've attached a research paper from NIST on the subject, which might shed some light. I think that more information may be obtained from CMM manufacturers, who are likely to have similar information.

This would be a good model to base your study on.

However, I would expect much higher variation between CMMs when using actual product and fixtures versus ball-plates. NIST is using a rather idealized situation here with an extremely uniform product and a modular product and fixture.

[Jim Wynne]

In Reply to Parent Post by Miner

This would be a good model to base your study on.

However, I would expect much higher variation between CMMs when using actual product and fixtures versus ball-plates. NIST is using a rather idealized situation here with an extremely uniform product and a modular product and fixture.

Absolutely. When you add the vagaries of drawing interpretation and datum construction the chances for variation increase significantly. This sort of testing must be very carefully designed and controlled if the results are going to mean anything. Part of the design and control might be no control--give someone a part and a drawing and a CMM and see what happens--or it might be controlled for fixturing, integrity of specifications, etc. Like all other experiments, the design depends on what you're trying to find out.