Is a low calibration TUR acceptable in the eyes of ISO17025 and TS16949? 4:1 TUR Rule

[ScottyWM]

Hi all, new to the forum, hope in the future I will be able to make some good contributions. For now, I've a question...

My company is TS 16949 certified, and we outsource most of our calibrations to ISO17025 labs. We recently have been arguing/discussing with a lab their TUR. On more than one occasion they have performed their calibration (within their scope) with TUR's below 2:1, most recently they did a dial indicator (micron resolution) using a TUR of 0.5:1 !!

Obviously, we need to reevaluate using them as a calibration source. Our contract comes due in one month and I will address minimum TUR's then.

But my question is whether these low TUR's are acceptable in the eyes of ISO17025 and TS16949? All I see is that the uncertainties must be known and stated. So, I could be ISO17025 compliant if I calibrate CMM's with a yardstick, as long as I state the uncertainties? Obviously, some calibrations are going to have poor TUR's (gage blocks for example), but an indicator??

Actually I guess it's more of an observation than a question. Anyone have any experience dealing with this problem?

 

[Jerry Eldred]

That is a good question, actually one I need to deal with as well. I'm in a similar context (TS16949, internal lab, need to have external labs comply with 17025).

Although I struggle with finding the old "4 to 1" Uncertainty Ratio in the new ISO17025 (maybe someone can help with this one), I'll try to throw in a few usable comments.

TS16949 prescribes essentially the use of accredited labs, or when there are none available, use of the OEM. TS16949 also notes that ISO10012-1 may be used for guidance. So there may be some usable information there as well.

If we forget completely about the old 4 to 1 rule, ISO17025 does state that they should be using recognized procedures (ANSI, EN, etc.) or the original manufacturer's procedure. So if they are using their own inhouse procedure, I would at least ask them to justify that their method complies with one of them.

Additionally, an uncertainty ratio of less than 1:1 (0.5:1 in your example) means that the total uncertainty of their method is only verifying your instrument to their nucertainty. Therefore, a TUR of less than 1:1 is at the least, calibrating your instrument to a new, reduced accuracy. If they are calibrating to less than the published specs on your instrument, they need to get your permission. By the way, I moved to TX from Durham a few years ago, so I am familiar with the various third party labs in your area. I won't comment about any of them here.

The next issue (one of the reasons the elusive 4:1 TUR has been around for so long) is measurement confidence. Specifications are derived based on a statistical confidence that a measurand will be within a given set of limits. 2 Sigma is pretty good, 3 or 4 Sigma is even better. So with the stated accuracy of their standard, I question the confidence of that new "spec." If you apply confidence factor to the equation, I would speculate it "could be" even worse than 0.5:1 (I don't know without reviewing their data).

My recommendation is to ask them a few questions (just to begin with):

1. What procedure are they using? Is it internal? If it is internal, does it comply with the manufacturer's or other recognized methods (ANSI, etc.)?

2. What measurement standards does the OEM or recognized procedure recommend? What is the applicable uncertainty of those standards?

3. Review their scope of accreditation (on your own). Verify what their stated measurement uncertainty is for the calibration in question. Compare the stated uncertainty with the uncertainty they provided you for the calibration in question. If it is not the same, ask them to explain the difference.

There is a significant exception (in my view) to the old 4:1 TUR rule. That is "State of The Art". There are types of calibrations for which 4:1 TUR does not exist. In such circumstances, State of the Art limits apply. One example is some Relative Humidity instruments. I'll avoid brand names, but some of the best hygrometers are only capable of around 2:1.

Hope that was of help. Please feel free to email me if you have any offline questions about the above.

And as always, I would love if some of the ISO17025 experts could tell me where to find the 4:1 TUR rule in current standards, or if it is no longer a requirement.

 

[Hershal]

Actually, ISO/IEC 17025 does not address the TUR concept in the least. It requires measurement uncertainty.

ANSI/NCSL Z540-1-1994 specifies a 4:1 in absence of uncertainties.

The best bet is if your calibration provider can support a 4:1 ratio, then that is best. You should be getting uncerainties if the 4:1 can not be supported. If you have a 0.5:1 calibration of calipers......give some thought to a competitive bid process for a new calibration provider.

I have run quite a few of them.

Hershal