Using MSA instead of calibration

[Jayfaas]

Would it be possible to use an MSA study instead of a calibration to verify machines? We have some machines in house that have 6 individual LVDTs and we run type 1 studies using a master that is verified annually. I believe the variation on the measurement is like ± 0.010mm for all parts because it only measures one type of part. With that being said, I believe I was told that Type 1 studies are used to verify it annually instead of performing calibrations on the LVDTs. If this is allowed, what are the requirements to make that decision?

 

[Miner]

MSA covers a lot of different types of studies (i.e., bias, linearity, stability, and repeatability and reproducibility). When you mention a Type 1 study, you have now specified it to be a Repeatability study, which addresses precision. Calibration addresses accuracy (i.e., bias and linearity), so a Type 1 study cannot replace calibration.

 

[Jayfaas]

We run type one studies that give us a bias number using a certified setting master.

 

[Miner]

That would partially address calibration, but it only covers one point in the instrument's range. Calibration should include a linearity study in at least 3, and ideally 5 points, spread throughout the instrument's range.

 

[Jayfaas]

I think they may have done it this way because the machine only runs one part. Would that make any difference? The linear difference between the parts would only be the tolerance of the feature measured, which is I believe +/- 10 microns

 

[Miner]

Are the environmental conditions appropriate for calibration? The measurement uncertainty?

 

[Jayfaas]

The calibration COULD be performed. It would consist of six LVDTs that would have to be removed from their housing, put in to our calibrator, which has a resolution of 0.0001 mm, have linearity verified, and then put back into their housings, and all be reset this while also grappling with the downtime affect of production. I understand that either way you do it, production is going to be affected. I just wanted to question the validity of this method and whether or not it would be against IATF 16949. When I was first told that this is how we were doing this, I was a little taken back because we have traditionally always performed calibrations or at least linear verifications on gauges like LVDTs. Once I understood that we were just using a verified master that is verified and used to set the machine annually, coupled with the fact that we only run one part in that machine, it seemed to make a little more sense. Even when I explained this to the auditor recently, he seemed to except the method, which was great, however, I’m not 100% convinced that, this is acceptable by the book.