BTW, any reason these replies didn't show up when I click on the "view replies since last visit"?

This is also a problem that I have had in the last few days......

Thanx for the suggestions thus far. I think it will be easy. Easier than the measurement uncertainty that changes EVERY time we get an assessment, even by the SAME auditor 3 different times!

I thought no one loved me......... :nopity:

Jerry

This will sound harsh.....welcome to MU. Believe it or not, it's not easier from the assessor side.

Hershal

Easier than the measurement uncertainty that changes EVERY time we get an assessment, even by the SAME auditor 3 different times!


This will sound harsh.....welcome to MU. Believe it or not, it's not easier from the assessor side.

Hershal

I think ridiculous is a better term. So when he shows up on the 3rd visit, sees what I've been doing since his LAST (2nd) visit and says "No, that's not right either. I think you should do it THIS way", I should just redo everything?

Remember, on the 2nd visit, he said "Well, what I told you the FIRST time isn't right, so do it THIS way". Now he's changed it AGAIN.

Why should I do anything, when next time he'll have "learned something new" and I'll have to change it again??

I may not understand MU completely but I have an idea of what it tries to describe. All the mathematical formulae, "rectangular distributions", etc... get my head spinning though. I thought I had it the first three times we were assessed...

After 3 changes by the same assessor, and still unresolved, I've kind of lost respect for both the assessor and A2LA, and even the concept altogether. This last assessment when he started changing his previous method I just put my head in my hands and went "Ok..." (In Naval Nuclear Power School we had an imaginary button called the "I believe" button. Worked for anti-matter theory and other concepts like this.)

Jerry
PS- Sorry, this probably should be on its own thread by now! :topic:

Jerry,

I hope this is a more appropriate title for your efforts with MU (measurement uncertainty).

Remember, MU is only one thing: It is the mathematical description of errors (systemic and random) associated with any measurement. That's all it is.

Another point: All the existing MU models (save one) come from the Metrology world, so if you are a test lab, there is a REALLY good chance that the MU model does not fit what you do. The one exception I am aware of is a model developed for what I call extreme testing (fire, wood, ultra-large structure), and soon to be handed off for validation.

Given that, it is no surprise to be exceptionally frustrated (I'm being PC here, the true term is likely different).

For those reading who have never dealt with MU, one of the bad indicators is that the "experts" don't even agree.......so expecting total consistency is a bit of a challenge.

Any accreditation assessor for a lab should try to be practical. With MU that is not always possible, but the attempt should be made.

Feel free to vent regarding MU. I'll join you.

Hershal

Feel free to vent regarding MU. I'll join you.


My only complaint (ok, maybe I have more) is the fact that the same assessor changes his plans every time & I have to do the tap dance to figure out what HE wants. Doesn't exactly make me feel like he knows what he's doing, but we have to appease him to keep our little A2LA logo.......

Especially when our Mom&Pop customers have never even heard of MU.....

Jerry
PS- Any way to shift this to a MU-disgust thread for others to join in? :confused:

I second the motion....

Hershal

Hopefully done now....

This looks like this was split off of a different thread. Let me know if you need it moved, merged with the original thread and that thread moved or whatever.

This is exactly what I wanted Marc, thanx.

I was just complaining about Measurement Uncertainty and how in 3 audits from the same assessor, we've had to change our MU 3 times. He "learns something new" every time and we have to jump when told to.

Why should I do anything this time, when it'll change again next time around? I've kind of lost faith/interest in this area and just don't care anymore. I'll put whatever number he wants now, because our customers don't care what it is anyway.

We supply stds for coating/plating thickness measuring equipment, and service the equipment as well. I have a general idea of what MU is, but when the mathematical formulae come out, "rectangular distribution", etc... My head spins...

We've been doing the same thing for 10+ years, accredited 25 / 17025 for 4 years, but somehow our MU keeps changing?

Jerry Stem

This is also a problem that I have had in the last few days......
I don't know if anyone is having a refresh problem still, but if so let me know.

Umm, OK - I retitled this thread to address the MU issue. This isn't my forte so I can't help, but now the thread should be focused.

Comments, from others? What's the scoop on MU? Should this be a continuing issue?

Most certainly this is a continuing issue!

If you take five "Experts" in MU, even well-known ones, give them all the same scenario (mutually blind), with little exact specific data, you will get five different answers, and the answers may well vary significantly. :confused:

This is bad enough for calibration labs. It gets WAY-Y-Y-Y worse for test labs. :frust:

I am looking forward to this thread. There is a lot of well-deserved frustration that needs to be vented :mad: , and hopefully some realistic outlooks that can be put forth.

Hershal

I have an even more fundamental problem with Uncertainty.

After I am done all the tests and calculations and can express the uncertainty...no one cares.

Ever tried to explain this to the plant manager?

We're going to use the gage anyway. It is what the industry uses. It is what our customer uses. It is the state of the art.

People will claim improvement if the gage reading improves, and look for heads if gage readings get worse.

It is virtually impossible to convince people that the measurement they got has some variation associated with it. They know what the gage reads, after all anyone can see what it reads.

So...now that we have to deal with 17025 cal labs (automotive requirement) and know the uncertainty of our plug gages...how much better are we as a supplier?

I can't find ANY payback. But I sure see real costs. Cal certs now cost 2 to 5 times more.

Show me the money indeed. All I see is cost.

Does anyone have a good news story, where understanding of uncertainty prevented a problem/saved money?

Most certainly this is a continuing issue!lYou 'know your stuff'. What per cent of total tolerance should / can measurement uncertainty 'take'? Let's say you have a dimension toleranced at +/- 1.0 mm. You do a gage R&R and come up with a low number. How does MU come into play here? MU is supposed to be too low to affect measurement of the dimension with consideration to gage R&R 'error', yes???

** NOTE ** There is rarely, if ever, a time when measurement uncertainty (MU) would acceptably exceed by 0.1 x - yes or no???

Most certainly this is a continuing issue!

If you take five "Experts" in MU, even well-known ones, give them all the same scenario (mutually blind), with little exact specific data, you will get five different answers, and the answers may well vary significantly.

Hershal

How about my case? Take ONE "Expert", give him the same scenario 3 different times, and I get 3 different answers??? This is where this thread came from. I mentioned at the end of a post on cal. intervals and customer agreement that it would be easier to fix than the MU issue from an assessor that keeps changing his mind every time he shows up.

I understand that the passage of time will give new insight, more training, and a better overall grasp of a concept, but why should I have to pay for that by jumping thru hoops every year to come up with a clever new way of expressing MU?

Our industry uses methods of measuring plating thickness that don't even come in contact with the part or standard being measured. We measure our calibration standard by XRayFluorescence and if it reads back w/i 5% of the marked value, we then measure the customer's standard looking for the same thing. That's it. Temp, humidity, etc... are no factor.

Jerry
PS- There are two other co's accredited by A2LA for the same field, we were first, and all three scopes of accreditation list the same values! When our assessor changed ours the last time, they eventually changed accordingly.

Not a bad way to put it Marc.

True, the MU should have essentially no real effect, until one is measuring way down into the dirt in that sceanario, especially if it is right at the edge of tolerance. Then the MU could push the measurement beyond the tolerance. Then the question is: did the item actually pass?

Having said that, it depends a bit. MU is really only the mathematical description of errors (systemic and random) associated with a given measurement. that is the first easy part. The actual calculator portion is the next easy part. It is all the effort between those two points that is hard.

A micrometer, measuring in a resoanbly controlled environment, can easily go to 3-4 places right of decimal with no significant effect from MU....MOST of the time. Calipers are a bit more limited. Other types of measurements will vary. If you see MU on the cert from your accredited cal lab, it typically is well under 1%. Most metrologists (especially the "white coat" Metrologists) go into cardiac arrest if they go above 1%.

A test lab has different test parameters, and the MU will be correspondingly different. Typically the MU will be much higher. The MU for a test lab can easily go over 25%. Some tests may have MU over 100%. Fire test is a prime example there.

Caster brings up a good point. Customers often don't care about it. So then, the actual reason to have it is partly historical, and partly process related.

Historically, only the NMIs (National Measurement Institutes) did MU, ad everyone else used the 4:1 rule. Now that low cost instruments are easily obtainable that have the ability to rival what NIST had in the 1970s, MU is often necessary to address the erros in a measurement.

MU should be used as a process improvement tool. The exercise of developing the MU should be viewed as a means of defining and exploring the measurement process. Identify improvements and shortcomings. Then, you get some actual value from it.

Hope this helps.

Hershal.

Jerry,

I did not mean to suggest your example is less important. I suspect it may be more the rule than the exception.

Understand I cannot comment on a colleague organization, only for the one I am part of.

Personal opinion, this is Hershal, not ISO or anyone else (do I have enough caveats?), changes in scope due to MU calculations should be evolutionary, not revolutionary, unless there is really good reason. Also, MU, even the BMCs, will change over time. Minor adjustments in the BMCs may well be required. That said, what you described does not appear to fit that model. Asking for an explanation at the next level up from the auditor may be a good move.

BMCs are the way MU is expressed on the scope of accreditation. The definition of BMCs, from ILAC G4 is:

“Best Measurement Capability” is the smallest uncertainty of measurement that a laboratory can achieve within its scope of accreditation when performing more or less routine calibrations of nearly ideal measurement standards or of nearly ideal measuring instruments. Best Measurement Capabilities are expressed as uncertainties at approximately the 95% level of confidence, usually using a coverage factor of k = 2. The measurement uncertainty of a specific calibration performed by the laboratory may be greater than the least uncertainty due to the behavior of the customer’s device, to the environment (if the calibration is performed in the field), and to influences from the circumstances of the specific calibration.

In the U.S. another note is often added:

NOTE: Uncertainty of measurement for calibrations performed on-site may be higher than those reported above.

Hershal

Interesting discussion.


It gets WAY-Y-Y-Y worse for test labs.

We are an internal test lab. Been accredited since early 2003. Never been asked to see any uncertainty estimates for any of the testing we do. In fact, at our sureveillance audit, our assessor wouldn't even look at some estimates I did.

So I sit here waiting for our upcoming surveillance II audit, with a new auditor, wondering if I should keep what I have or expand on it. Most of our material requirements only list a minimum value the material must meet. There is no upper value. I think it would be a total waste of time and energy to even begin to calculate MU for these tests.

That could change is February... :mad:

Almost forgot...the only guide I ever read that made any sense regarding MU for testing comes from A2LA.

Guide for the Estimation of Measurement Uncertainty in Testing.
July 2002 by Thomas M. Adams.

Downloaded from their websight.

What kinds of tests? Also what standards (if external such as ASTM)?

Hershal

Jerry,

I did not mean to suggest your example is less important. I suspect it may be more the rule than the exception.



I didn't take it that way either. Just showing my frustration with "the system"... Tried changing ours AGAIN, sent it in last week. Let's see what A2LA says this time...

Jerry Stem

This is exactly what I wanted Marc, thanx.

I was just complaining about Measurement Uncertainty and how in 3 audits from the same assessor, we've had to change our MU 3 times. He "learns something new" every time and we have to jump when told to.

Why should I do anything this time, when it'll change again next time around? I've kind of lost faith/interest in this area and just don't care anymore. I'll put whatever number he wants now, because our customers don't care what it is anyway.

We supply stds for coating/plating thickness measuring equipment, and service the equipment as well. I have a general idea of what MU is, but when the mathematical formulae come out, "rectangular distribution", etc... My head spins...

We've been doing the same thing for 10+ years, accredited 25 / 17025 for 4 years, but somehow our MU keeps changing?

Jerry Stem


Well, as kind of an update, I just talked to our assessor last week. After the 4th reply to A2LA with our various deficiencies, our A2LA rep basically said I had MU at 1 sigma and needed it at 2 sigma. I said "So all I have to do is double it and we're done?" When she said yes, I was done. I told her I would put 20% on our scope if that ended this mess because none of our customers care.

She forwarded my new info to the assesor. He called last week and the first thing out of his mouth was "I have a problem with MU, if you hadn't noticed". Gee, 5 months since the assessment and still going round-n-round, no I hadn't.

I told him what I did, when he said he didn't understand my procedure (which he had given me the first time he was here). He finally said he was going to ok it but flag it for the next assessor to check our methods and see if it should be changed. I already know it will change the next time, it has the last 3 times!

Jerry
Not a big fan of A2LA/17025/MU..... :mad:

Jerry,

I understand the frustration.....will you be at NCSLI this year? if so, I'll buy you a beer and let you vent...... :argue:

Hershal