Cg, Cgk and GRR of a CMM

[ARevenga]

Hello everyone,
I am working for Bosch in Automotive Brand and for every component made by an external supplier, Bosch is requesting them to perform an individual Cg, Cgk and GRR of all Functional/Safety dimensions which are measured by CMM (e.g. position tolerance of a plane/ parallelism,...). From your perspective, does it make sense? Once known that CMM is calibrated, what should I request to Supplier to guarantee suitability of their CMM (measurement method already agreed)?

My manager is pushing me to get these values of each ICL dimension which is driving me crazy... if CMM is calibrated (I reffer to tolerances +/- 0,1 or higher).

I don't know how could supplier do a Cg and Cgk of a Pos.tolerance between planes and anyway It would not make sense to me.

Thanks in advance for your feedback!

 

[Miner]

The first thing to understand is that there is no relationship between calibration and GRR/Cg/Cgk. There are two aspects of measurement variation:

• Accuracy (Bias) - This is measured by bias/linearity/stability studies, and is controlled by calibration.
• Precision (repeatability and reproducibility) - This is not controlled by calibration. It is measured by GRR studies and is controlled by the measurement system design, procedures and training.

A GRR study can be performed on any continuous measurement. In very specific cases (fully automated test, including fixture loading), a repeatability study only may be more appropriate.

 

[Welshwizard]

Hello ARevenga,

I have come across this with an aerospace company where they were actually being asked to perform 20 to 30 GRR studies on one part with multiple features! They quickly rationalised these features down into families with respect to the features and then performed repeatability studies.

Over time they proved that the measurements were consistent and that the standard deviation was commensurate with their specifications. They also used fixtures which they demonstrated added no significant extra variation upon loading and unloading, thus reproducibility was not a factor.

Of course it would also be normal to produce a certificate to demonstrate that the CMM had been calibrated and even better was been regularly monitored for the consistency of this bias.

Hope this helps

 

[Miner]

This link ( Intro to MSA of Continuous Data - Part 10: Creating gauge families ) will provide some guidance in creating gage families as suggested by @Welshwizard .

 

[Matt's Quality Handle]

A GRR study can be performed on any continuous measurement. In very specific cases (fully automated test, including fixture loading), a repeatability study only may be more appropriate.

A repeatability study is probably sufficient, depending on their CMM operations. I have seen operator differences in fixturing come into play.

It can also be a good test of the honesty of your supplier. I had to call BS on a supplier that did a 10x3x3 over 20 features. And every CMM measurement point agreed to within .0001 inches!

 

[Miner]

A repeatability study is probably sufficient, depending on their CMM operations. I have seen operator differences in fixturing come into play.

It can also be a good test of the honesty of your supplier. I had to call BS on a supplier that did a 10x3x3 over 20 features. And every CMM measurement point agreed to within .0001 inches!

It would also depend on whether the CMM is running from a program or is a manual CMM.