I have the GUM Works Manual, I have excerpts from the Engineering Statistics Handbook, but I would like a step by step (simple)method for calculating and reporting the measurement uncertainty for M&TE calibration.
QS9000 says our calibration results should include this information, but provide no clear method of doing so. Does this stuff really have to be rocket science. I'm a big proponent of the kiss principle, "Keep it simple stupid".

------------------
Dave S.

I have been reading about measurement uncertainty for some time. I have even set up a special sub-directory within the pdf_files directory. I know of no 'easy' answer. I suggest you read through some of the threads in that directory.

I did a forum search and found numerous 'hits' including http://Elsmar.com/ubb/Forum4/HTML/000072.html -- You might want to check out the thread.

That said, if ANYONE feels they can explain uncertainty easily and / or simply, I sure would like them to help us all out.

David,

There is no clear cut method of calculating measurement uncertainty. I highly suggest taking a course to become familiar with the concept. Once you understand the concept, and the statistics behind it, it actually becomes fairly easy. I say easy, as opposed to simple. After you take the classes, it is easiest to come up with a general list of contributing factors to calibration uncertainty, such as temperature growth of UUT and standard/M&TE, deformation, repeatability, master uncertainty, resolution error, etc.

I've been where you are, take the class. The books make great sense to me now, but before it was put into plain english, they were well above my head. To save yourself, and your company, some money I suggest you start with the simple one day "Intro" course. Quite often that is all that you will need.

Ryan

Could someone please suggest source(s) for the training Mr. Wilde suggests?

Thank you.

There is no easy way to calculate uncertainty as each type of measurement has it's own uncertainty contributors.

The easiest way I have found to calculate my uncertainty is to follow an example. You can find a very good example at *** DEAD LINK REMOVED *** click on the orange document.

You can also ask an expert. I called my local registration authority in Canada and talked with a metrologist for about an hour. He laid out what factors may contribute to uncertainty of my specific tests.

Now all I have to do is calculate it. That should be easy right? ;-/

I had great luck with the training provided by Quamtec Corporation. 810-775-5495 ask for James Jenkins. Jenkins@quamtec.com

Originally posted by Benjamin:
Could someone please suggest source(s) for the training Mr. Wilde suggests?

Thank you.

Mitutoyo Technical Seminars may be a good source: One coming up May 17 in Charlotte, NC - "Estimating Measuring Uncertainty" - 1 day for $325/person. Touted as an entry level course covering ANSI/NCSL Z540-2, PUMA method based on ISO/TR14253-2. Check out thier web site: www.mitutoyo.com (http://www.mitutoyo.com)

Try:
www.a2la.org (http://www.a2la.org)

Look under "recent postings". There are $$$ and free documents and findings.

ASD...

Yes, I was like you, looking around for something NOT like the GUM document to figure uncertainty and I think I have found it. Get document ISO/TS 14253-2 This is the GUM PUMA method. This document will make you happy.

There is more to uncertainty calculations then comming up with numbers to pass an audit. We originally used a spread sheet from a consultant to just drop in our numbers and get some type of result. After months of preperation our A2LA assessment for calibration services started. Within the first few hours of a three day assessment, our assessor advised me that he will not accredit a lab that does not have someone on staff that can do detailed uncertainty budgets. I was shocked! After he explained his logic I understood and agreed. The process requires a person that truly understands all the variables, and in our case there are many. Thanks to his insistance to learning all of the details, we are a better lab. The person who recommended taking the Mitutoyo course is correct. I sent a Lab Technician to a course taught by Dr. Hendrick Neilson (Mitutoyo and A2LA instructor) and the technician's understanding is much better. Dr. Neilson also does private consulting, He can be contacted through A2LA.

------------------

Just finished the Mitutoyo class on uncertainty here in Charlotte. A nice basic introductory course that cleared up a lot of issues for me. It will still be hard work but now I know which direction to take.

Thanks for the recommendation.

Benjamin, would you recommend the class? Just got the flier in the mail today for the Milwaukee seminar. The agenda looked very interesting.
Your thoughts?

For Ken K.

The class was well organized, inexpensive,and pragmatic; the instructor knowledgable -- Dr. Henrik S. Nielsen.

All examples came from the dimensional metrology world -- calipers, etc. He gives you a basic intuitive understanding of what uncertainty is and how to go about identifying uncertainty contributors and their distributions. He works through an example in class that is very detailed.


You will not come out an expert, but if you know and understand your instruments and measurement processes, you will get a good basic (notice that word again) understanding that will get you started towards calculating your own uncertainty budgets.

As a result of the class, I have a better understanding of many of the Cove discussions on uncertainty. The more calculations I do, the better I am getting. Hope the auditor agrees.

Also, he does a good job answering individual questions, so come armed. Speak up in class when you do not understand.