I have just been given the task of becoming 17025 certified and i want to know if there are any good resources that i can look through to help with interpetation and prctical application of the ISO standard.

Good morning,

I did a search using the key words "guidance 17025" and found this thread (http://elsmar.com/Forums/showthread.php?t=21314&highlight=guidance+17025) for you. It has some links for buying the standard and getting guidance documents as well.

After you get these items, study them and decide how your organization meets the "shalls" and where it does not. You can ask us specific questions at that point.

I hope this helps!

Hello Coleman!

First thing would be to get a copy of the current standard. Second, there is a lot of history here on that standard. Using the search on that (if you haven't done it already) will be helpful.

Some questions:

1. What exactly is it that you do and your scope of work?
2. What kind of shape is your current QMS?
3. Have you already determined uncertainties for your measurement systems?
4. Have you identified the appropriate environment, and are you controlling, measuring, etc.?

Once you determine what your scope is going to be, it will give you some idea of how long (and how much resources) it will take. Making sure management understands and is behind that is a good idea.

We have some very knowledgable people here (unlike myself:tg:) in this area. Luckily, we have been fortunate to have a lot of newer Covers who are in metrology labs that contribute some helpful information.

So, hopefully we can help you in your quest.

Coleman,

Welcome…….

First, you are Accredited to 17025:2005.
Are you going to be testing, calibrating or both.

Most ABs have guidance documents posted on their web site.
There are a few interpretative documents out there.
Try searching the threads here. You will find volumes of great information.
You are not alone here. Give us a little more information about your lab and industry.
Many here will be willing to help.

The best advice I was given:
When you read the standard, you need to ask yourself
Not only what does this mean, but what does this mean to US.

17025:2005 is a broad reaching document intended (I believe) to encompass many types of laboratories.

Well we are an automotive supplier of steel componets with the idea that it is in our best interest to maintain a certified lab so we can quote measuremeants for customers of similar products. In addition to this we would like to 'eventualy' begin calibrating our own gages as well as hopefully creating gages in house at some point down the line. We are currently TS certified.

First thing.....get the Standard.....NCSLI has it at a fair price.

http://www.ncsli.org

Next, get training in uncertainty.....you will need it to get your scope ready.

You will need a lab manual. You can likely write it yourself if you have time. Otherwise get a consultant that is experienced with ISO/IEC 17025 specifically

Your internal auditor can do the QMS side of your lab internal audits, but not the technical side unless the person is a metrology professional or at least has training in the discipline(s) your lab will be involved in.

Management review is a big thing.....the lab review can be rolled up into the parent review if necessary, but the 17025 Standard, Clause 4.15.1 has a prescriptive agenda and all the topics must be addressed.

Proficiency testing or inter-laboratory comparison (PT/ILC) is a requirement under international recognition. You have to complete one prior to accreditation, then four years to complete your scope, under ILAC/APLAC rules.

Also, when you seek your accrediting body, do NOT sign with any of them till you talk to ALL of them.

This should get you going.....we look forward to the questions that you almost certainly will have.....just as we all did.....so ask, we will help if we can!

Are there any suggestions for uncertanty training?
I already have a copy of the standard which is what led me to come here and ask a few questions as my head is still spinning from my first read through! Currently I am doing a gap study to determine areas that need improvement. I was kinda hoping for a plain speech check list just to get me heading in the right direction but i can't seem to find anything like that.
The plan is that we will write our own manual, and my boss would like to start moving foward in March :confused: so many more questions will follow I'm sure... Thanks in advance for all of the help I am sure I will find here.

Coleman, my best advice to you right now...

The post previous to yours was by Hershal. Click on his name and follow his posted threads. He will give you a substantial amount of background for uncertainty calculation, appropriate lab environment, etc.

As far as uncertainty, I believe the Certified Calibration Technician program through ASQ covers that (I may be wrong; someone correct if I am). I have a couple of books on it, if you would like the title.

As far as a checklist, Hershal's previous post would be a good start!:D

I have found this to be a good explanation of uncertainty:

http://www.ukas.com/Library/downloads/publications/M3003.pdf

I believe the CCT covers very basic components of calibration although it may touch on the subject of uncertainty you would be much better off with specific uncertainty training by qualified professionals. The links provided will give you a good starter, I'd suggest though, that you might seek out someone who has been in your specific situation, through asq section, this forum, or other business networking. It'd be quite beneficial to get feedback from someone who instituted a 17025 cal lab under a TS certified company.

The gap analysis is a great starter for getting yourself moving along the road to accreditation and planning out your goals.

Take a look at some 17025 compliant quality manuals to give you some baseline examples there are many floating around on the internet, or you can request a copy from your current calibration suppliers for review.

http://www.a2la.org/training/IntroUncert.cfm would be a decent starter for UNC training.

Are there any suggestions for uncertanty training?


I posted a note in one of the sub-areas about some upcoming uncertainty training.....

Here is the link to what I posted.....

http://elsmar.com/Forums/showthread.php?t=24299

a quick visit to nist.gov/nvlap can help you with a checklist. You can access NIST HB 150, which is the NVLAP guide to accreditation. You can also download the HB 150 checklist from this home page. If you check off every requirement on it, you have exceeded requirements of 17025:2005.

Also, Quametec Corporation offers the best uncertainty training out there.

nist.gov/nvlap - This site has been a tremendous help - Thanks.

Does anyone know of any similar online resources concerning SPC or GD&T?

nist.gov/nvlap - This site has been a tremendous help - Thanks.

Does anyone know of any similar online resources concerning SPC or GD&T?

NIST has an excellent site (http://www.itl.nist.gov/div898/handbook/) devoted to engineering statistics, but I'm not aware of anything similar for GD&T. There's a general GD&T glossary here (http://www.etinews.com/gdt_glossary.html), and there are some explanations of some of the most-used symbols here (http://www.actphx.com/gd_and_t/gd_and_t.html).

OK I am trying to develop the procedures we use for measurement strategy - I there any information out there as to how strategies are determined?
Also where can i reserch topics such as the 10:1 calibration rule and why it is used?

In section 4.1.3 of ISO 17025 there is a requirement for the management system in remote locations - if i do not intend to work in remote locations does the policy still need to be incorporated?

Also - Is there any reference matarial out there that can help me to determine what the best mesuremant strategy is and why it is necessary to mesure a certain way?

As for the remote part.....document that you do not have any remote commitments or work and leave it at that.....your AB should accept that.

In section 4.1.3 of ISO 17025 there is a requirement for the management system in remote locations - if i do not intend to work in remote locations does the policy still need to be incorporated?

Also - Is there any reference matarial out there that can help me to determine what the best mesuremant strategy is and why it is necessary to mesure a certain way?

I'm thinking that you need some expert help here, not just on a clause by clause basis. Have you been to any implementation training? That's a great place to share this type of question, rather than trying to answer each item as it presents itself. May I also ask, why is an internal lab going for accreditation? If you don't offer your services to outside customers, and get paid to calibrate or test their equipment/products, I rather doubt that accrediting an internal lab is going to a) be worth the efforts and b) be cost justifiable.
TS 16949 doesn't require you to have an accreditedlab and I've worked with plenty of TS companies which had a lab but didn't go to the cost/work of accrediting it.

Are your management fully informed to make this (costly) decision?
:2cents:

Well we are a TS certified production company but my imediate superior would like to make our department profitable and in order to do this he feels it is necessary to become ISO certified so we can begin quoting layout work for other companies. I do not believe he is aware of the cost that will be associated with this and he gave the project to me to begin the first steps of determining what it will take to reach the standard before we start spending mone on aquiring the standard.
A training class would require money so i am trying to get the individual questions answered here by the wealth of knowledge that seems to be available... Any suggestions would be greatly appreciated though.

First and main suggestion.....get ANS/ISO/IEC 17025:2005 and read it for yourself.....we will help you all we can, but ultimately you need to know for your own satisfaction what it says.....then you can work on setting up the systems that you will need to manage the lab under the Standard.....and then you also get to investigate what accrediting body (AB) you should use.....

well I have already reviewed the management system section (4) of the requirement which is where my questions are coming from - I just want to get an idea for how other people interpet the standard. I also will be looking for the more efficient ways to conform when we encounter an issue.

As I mention.....read the entire Standard.....accreditation to ANS/ISO/IEC 17025:2005 includes the Section 4 Clauses.....the normal QMS stuff.....but ALSO includes and in fact CONCENTRATES on the Section 5 Techincal issues.....that is the heart of the Standard.....please continue the questions, we relish them.....I only seek here to impart an understanding of the accreditation, should you seek that path.....

Is there any reference matarial or standardized documentation that i can get which details measurement stratagies and why they are prefered specificly in reference to section 5.4?

Is there any reference matarial or standardized documentation that i can get which details measurement stratagies and why they are prefered specificly in reference to section 5.4?

Coleman:

If I understand you correctly, your lab is looking to perform layout inspections. Is that correct? The application of ISO/IEC 17025 is for calibration and testing. I'm not exactly certain that layouts are really 'testing' and, therefore, the technical requirements of section 5.4 aren't fully applicable. You could adopt a number of criteria to apply, like the methods used to determine the size/shape of round/cylindrical shapes, but (frankly) I'm still struggling to see why you would want to go for accreditation for layout work.

If you are already ISO/TS 16949 compliant and registered, you had to meet the requirements for an internal lab, so why not use that 'qualification'? Are you being asked by customers to be accredited to ISO/IEC 17025? IS your boss fully on the same page as you about the application of the lab. standard?

We are able to do layout work for parts that are company produces however we have found that in order to do work for other companies there customers require us to be 17025 certified otherwise they will not recognize the results as 3rd party compliant asper section 7.6.3.2 of the ISO/TS 16949.

Coleman:
If you have customers asking for you to be meet the requirements of ISO/IEC 17025, they might just be wishing! I would respond that, as an internal lab. under the TS standard, you meet the relevant requirements of ISO/IEC 17025. You might provide your lab scope etc. to support this.

I'm seriously doubting that your boss will want to go to the expense of a full accreditation. Mostly, those labs which do layout also do some kind of functional testing of the products they layout/inspect - tensile, melt flow, abrasion, salt spray etc., for which it would be appropriate to call 'testing'.

If you need to I'd be happy to conference with you and him - even though I'm officially not a consultant from last Friday - so my advice is free............!!

The reasoning behind the want to move foward with 17025 acreditation is due to the fact that it will empower us as an organization to move foward with contracting layout work because it meets or exceeds any requirment that a potential customer may have, without the need for them to go through any effort to get approval from there customer. I agree that it is not a "Requirement" to do layout work however it is the only 3rd party acreditation that will exceed the requirements of a customer without any effort on the part of the customer.

Coleman:
I've taken a look on the web and found a couple of labs who do layouts. I had it in the back of my mind that someone was accredited, and there are some which don't (to my surprise) also offer testing. One does offer calibration.

So, I agree that it can be done and reverse my original comments. I do feel that it is a major (cost) undertaking, but it's your boss's call, ultimately. please feel free to ask more questions here and we'll do our best to answer.

Can anyone give me specific information on contributing factors toward uncertainty effecting dimensional layout results?

Typically, you have to estimate uncertainty. You should do R & R studies on types/families of equipment used to perform the layouts. Then there's the effects of the equipment's calibration results. There may be some aspects relating to the properties of the materials, like plastics vs metals and so on. A copy of the MSA will help you determine what can affect the results and whether they are a significant source of variation/uncertainty.

Can anyone give me specific information on contributing factors toward uncertainty effecting dimensional layout results?

I'm not necessarily recommending A2LA.

Having said that, they have a guidance document available on their web site. It was free, last time I looked.

Does your layout conform to any established (ASTM, ISO, etc.) method?

Per the A2LA guidance document, if you use established test methods it is not necessary for you to go through the rigamarole of establishing uncertainty. For example, if you use an ASTM test method to determine the physical properties of some plastic AND condition the specimens according to the ASTM method called out in the test method AND take the samples according to the ASTM sampling method called out in the test method, you are exempt from establishing uncertainty.

AIAG publishes a Measurement Systems Analysis Manual. It will give you a good grounding in the basics as well as some specific methods for gage R&R. Also there is an old IBM publication called _Process Control, Capability and Improvement_, which contains two methods for gage R&R. I'm not sure how you could get your hands on the IBM pub. I do know you can't have mine.:D

Are you strong on stats?

Typically, you have to estimate uncertainty.
<snip>There may be some aspects relating to the properties of the materials, like plastics vs metals and so on.

Good points, there.

Don't forget environmental factors: temp, humidity, vibration, phase of the moon, etc.

Many, many, many years ago I did some work for a calibration lab that sat on an 18-foot-thick, spring-suspended concrete slab. The purpose was to isolate the items calibrated from vibration. Item calibrated? Missile warheads. Desired uncertainty of measurement: zero.

Are you strong on stats?
I'm not strong in anything but i am getting stronger... I'm kinda learning as i go.

I'm not strong in anything but i am getting stronger... I'm kinda learning as i go.

Gage R&R has a statistical basis.

The _For Dummies_ series has a beginning stats book and workbook. It might be a good idea for you to spend some time with them.

Then there are some SPC concepts of which you should have a decent grasp. Don't know of an _SPC for Dummies_ book, though. Maybe some other Cover can suggest a decent SPC tutorial.

And beware of software! Statistical software is great for doing the math, but not so good at making sure you apply it correctly.

Folks, C'Mon....................

Uncertainty is not really all that difficult...........

The heart of uncertainty is this...........you have been dealing with uncertainty since you first let go the coffee table to walk across to where moma was at.....the only REAL difference in that and measurement uncertainty is that it is applied to a specific concept, and that you describe uncertainty in a formal mathematical model.....

GR&R and other studies can help establish your uncertainty.....but get the concept first.....

Now, it is an intregal part of 17025, but understand its application also.....

It simply tells you the errors associated with a measurement or series of measrements.....expressed at approximately 95% confidence.....or put another way, 95 times out of a hundred, the true value will be the stated value plus or minus the uncertainty.....but you will never know if you hit true value unfortunately.....

So, how do you handle this?

Get training from someone reputable and good.....then as you work your uncertainties, follow the KISS rule.......Keep it Short and Simple.........

Now, since this question was posted by someone seeking accreditation, it is fair to let you know that - WHERE - a given standard (e.g., ASTM) has a value for uncertainty that actually gives you information, you can likely use that.

You will take on several important questions in this forum as you prepare for accreditation.....please do keep asking, it is healthy and shows active participation.....and more importanly, you may ask the very questions that someone else might be a bit shy to ask.....

Your manual, selection of an accrediting body, how to document authorized signatories, proficiency testing and inter-laboratory comparison (PT/ILC) are just a few of the questions to address.....

Hope this helps.

Good points, there.

Don't forget environmental factors: temp, humidity, vibration, phase of the moon, etc.

Many, many, many years ago I did some work for a calibration lab that sat on an 18-foot-thick, spring-suspended concrete slab. The purpose was to isolate the items calibrated from vibration. Item calibrated? Missile warheads. Desired uncertainty of measurement: zero.

This is all good information however I am trying to establish the guidelines for uncertainty specificy regaurding Layout work (Dimensional Testing) Is there any resource information on this topic that you can direct me to... Aside from taking a general class for measuring uncertainty?

This is all good information however I am trying to establish the guidelines for uncertainty specificy regaurding Layout work (Dimensional Testing) Is there any resource information on this topic that you can direct me to... Aside from taking a general class for measuring uncertainty?

I am not recommending A2LA over other CB's...

That said, try this url:
http://www.a2la.org/documents/docfinder.cfm

Select "Guidance Documents" in the pulldown menu. Look for _G103 - A2LA Guide for Estimation of Uncertainty for Dimensional Calibration and Testing Results (http://www.a2la.org/guidance/est_mu_dimen.pdf)_

Also, are you using ASTM methods?

I am not recommending A2LA as your 17025 certification body.

Cliff,

I do not know A2LA too well. Can you clarify why you would not recommend A2LA?

Stijloor.

Cliff,

I do not know A2LA too well. Can you clarify why you would not recommend A2LA?

Stijloor.

Ooops, maybe I didn't word that right.:o

I'm not pushing A2LA in preference to other bodies, is what I meant to say.

Ooops, maybe I didn't word that right.:o

I'm not pushing A2LA in preference to other bodies, is what I meant to say.

Thanks Cliff, that makes sense.

Stijloor.

I am not recommending A2LA over other CB's...

That said, try this url:
http://www.a2la.org/documents/docfinder.cfm

Select "Guidance Documents" in the pulldown menu. Look for _G103 - A2LA Guide for Estimation of Uncertainty for Dimensional Calibration and Testing Results (http://www.a2la.org/guidance/est_mu_dimen.pdf)_

Also, are you using ASTM methods?


Couple of questions regaurding the A2LA guide...
1)In the examples i.e. Table 1 Uncertainty budget for 0.5-inch XX Plain Plug Gage They show the formuls used for measuring uncertainty when using a CMM probe to measure a round steel surface. How are the standard uncertainty measurements calculated?
2)can someone please expand on the significne and diferences between standard uncertainty and expanded uncertainty?
3)How can i research the type of uncertainty and how it is applied (A vs. B)

I was given the same task 2 years ago.
If you have never done this sort of work before, I would highly recommend a pre-certification audit to ascertain where you neeed the most focus.
Are you doing this alone?
What kind of lab?
What kind of test?

:D

Couple of questions regaurding the A2LA guide...
1)In the examples i.e. Table 1 Uncertainty budget for 0.5-inch XX Plain Plug Gage They show the formuls used for measuring uncertainty when using a CMM probe to measure a round steel surface. How are the standard uncertainty measurements calculated?
2)can someone please expand on the significne and diferences between standard uncertainty and expanded uncertainty?
3)How can i research the type of uncertainty and how it is applied (A vs. B)

Coleman:
I think you're headed in the wrong direction. If you're looking at uncertainty using that example, you're looking at a calibration, when your organization's lab is doing layout work. Therefore you run the risk of doing uncertainty work which is not required. In addition, although it's an initially higher cost, why not find a class? In the long run it will be cheaper and less frustrating that fishing for answers to all your questions here. Once you get some basic training, the specifics could be addressed by our fellow Covers.

:2cents:

Coleman:
I think you're headed in the wrong direction. If you're looking at uncertainty using that example, you're looking at a calibration, when your organization's lab is doing layout work. Therefore you run the risk of doing uncertainty work which is not required. In addition, although it's an initially higher cost, why not find a class? In the long run it will be cheaper and less frustrating that fishing for answers to all your questions here. Once you get some basic training, the specifics could be addressed by our fellow Covers.

:2cents:
Coleman:

Andy is right.

Coleman:
I think you're headed in the wrong direction. If you're looking at uncertainty using that example, you're looking at a calibration, when your organization's lab is doing layout work. Therefore you run the risk of doing uncertainty work which is not required. In addition, although it's an initially higher cost, why not find a class? In the long run it will be cheaper and less frustrating that fishing for answers to all your questions here. Once you get some basic training, the specifics could be addressed by our fellow Covers.

:2cents:

I don't dissagree with this statement in any way however taking a class NOW is not an option and making progress regaurdless of how much progress is a requirement so I am patient and content to move at a snails pace for the time being. Further more I personaly have found that i get MUCH more out of a class when i know what i want to get out of it before hand. By asking my questions here i am gaining a better understanding of what i need to get from the class I do intend to take at some point in the not TOO distant future.

At any rate my reason for asking about the example that i choe to ask about was mostly because i wanted to see what they were using to get there data - not because i wanted to apply the specific model to my situation. The more knowledge i have about how uncertainty works the more confident i will be in applying its practice to my needs.

That being said my question still stands... :confused:

WOHA WOHA WOHA..........Let's get real.....layout is NOT that huge a deal for uncertainty it seems to me.....let's not get so worked up we go to war.....yes there is an uncertainty.....but let's put it in perspective.....

This is a layout.....in an - INTERPRETIVE - sense, the same as TESTING, not calibration.....yes, you need to develop your uncertainties.....but test uncertainties are different than cal uncertainties.....

You need to work your uncertainties.....but whatever the numbers are, they are.....I would not sweat the numbers, regardless of what they are..........

Hope this helps.

Hershal,
Thanks! That was the point I was making; that by simply looking at an example of an uncertainty calculation isn't going to be helpful! In fact, it will lead to more work, frustration and possibly problems at the accreditation audit, when no-one can fully explain how they got to their figures!

As you say, layout uncertainty is uncomplicated. Uncertainty simply needs to be estimated. The considerations have already been posted in this thread.

Coleman, if a class isn't an option, then at least find someone who can visit you lab for a day and hand hold you through the analysis to put you on the right track. It'll be better for you in the long run!

WOHA WOHA WOHA..........Let's get real.....layout is NOT that huge a deal for uncertainty it seems to me.....let's not get so worked up we go to war.....

Sorry i was not getting worked up - I appreciate all the feedback/input - it helps ... Really! I was just trying to clarify my situation not trying to start a war.

Thanks again - I'm supposed to be getting some on site assistance today so that will probably get me going on some new questions.

Although i am still unable to understand the questions that i posted above. And the fact that it is a calibration and not a test brings forth some more confusion. The uncertainty budger is for calibration but how can you be said to calibrate something if you are not making any modifications? At this point wouldn't you just be verifying or testing the gage pin? Would there really be any practical difference between measuring the diameter of a gage pin with micrometers vs. measuring a dowel pin with micrometers (Aside from accuracy requirements)?
I understand that you are said to be "Calibrating" something when the item in question is intended to be used for measurement purposes however it seems like uncertainty should be determined by the practical application of the actual proccess that is being done.

Although i am still unable to understand the questions that i posted above. And the fact that it is a calibration and not a test brings forth some more confusion. The uncertainty budger is for calibration but how can you be said to calibrate something if you are not making any modifications?

Perhaps I can dive in here. :bigwave: Hope I don't cause more confusion.

It is normal and OK to use the word calibration when comparing your artefact (pin or whatever) to eg. a reference artefact with a traceable calibration certificate, even though no user modification is possible. However, the found difference between your pin and the reference (ie. the bias) is not part of the uncertainty budget (or at least does not have to be) because you know it and will presumably allow for it in some way.

What is part of the budget is the uncertainty of the bias, for which a best estimate would be a combination of the uncertainty of either your two micrometer measurements, or one + the uncertainty of the reference quoted by some external source.

Your question 2 is easier to answer but it is in the A2LA guide and others. Standard uncertainty always in modern practice refers to +/- 1 standard deviation (~67% coverage) and expanded uncertainty refers to +/- 2 standard deviations (~95% coverage). There is no significance other than that.

The A2LA guide cites the UKAS document M3003 as a reference and it has been revised recently as in this link. A slightly different approach sometimes helps. http://www.ukas.com/Library/downloads/publications/M3003.pdf

And the fact that it is a calibration and not a test brings forth some more confusion. The uncertainty budger is for calibration but how can you be said to calibrate something if you are not making any modifications? At this point wouldn't you just be verifying or testing the gage pin? Would there really be any practical difference between measuring the diameter of a gage pin with micrometers vs. measuring a dowel pin with micrometers (Aside from accuracy requirements)?
I understand that you are said to be "Calibrating" something when the item in question is intended to be used for measurement purposes however it seems like uncertainty should be determined by the practical application of the actual proccess that is being done.

Coleman:
Your confusion is understandable. Calibration is simply the process of determining how far off the 'true value' you are. It doesn't imply adjustment, just the determination of the value. The uncertainty is a factor in determining some of the contirbutors to why you cannot report the 'exact' value you are measuring. Since calibration is normally performed on measuring equipment, and that usually requires equipment of higher capabilities, then uncertainty becomes a bigger deal, hence the complicated math and so forth.

If you imagine trying to calibrate a device which gauges/measures to 0.0001", then the device used as the calibration master would (generally) be capable of measuring to 0.00001". At this level of measurement, a number of factor can influence the result. Hence the need for calculating uncertainty.

With layouts, it isn't as complex, since you are reporting an observed result, based on an understanding of the normal influences on that type of measurement, derived from studies like R & R, etc. Your calibration results for the CMM used will give you the equipment uncertainty. Compared to the normal values of tolerance on the engineering spec, my guess is it's very capable, with only minor influences of R & R, temp/humidity etc.

:agree1:

Ok ... so this falls back to the idea of determining negligable factors of uncertainty. The idea is i need to create an "uncertainty budget" when i am doing layout work that identifies the items that are not negligable and these items then need to be monitored and calculated to give an expanded uncertainty estimtion. My question here is what is the determining factor that makes a factor negligable - i.e. if i am doing layout work on steel sub componets does the expansion coefficient really matter if i am only reporting the test results to 0.000"? - and why/why not? Do apply the 10to1 rule (or perhapse 100to1 rule...) meaning that any effect that would be smaller than 0.00005" would not be significant and therfore not need to be measured, or is this where i use the rectangular distribution method when calculating my standard deviation? Or am i just confused here again... I think i might be catching on or then again...!
:thanks:

Ok I have a mth question now (bare with me)... When you plot standard deviation charts there are 3 types of charts that i am aware of - Standard bell curve, Triangular distribution and rectangular distribution - the formula is then devided by the square root of 2, 3 or 6 accordingly - what does this devisor represent and how is the number determined?

Ok I have a mth question now (bare with me)... When you plot standard deviation charts there are 3 types of charts that i am aware of - Standard bell curve, Triangular distribution and rectangular distribution - the formula is then devided by the square root of 2, 3 or 6 accordingly - what does this devisor represent and how is the number determined?

Coleman, are you doing a gage R&R study?

Maybe I'm displaying my own ignorance here, but I know of no case where a triangular or rectangular distribution would have anything to do with gage R&R.

What is it again that's stopping you from taking a class? You (and your company) will be better off if you do so.

Ok I have a mth question now (bare with me)... When you plot standard deviation charts there are 3 types of charts that i am aware of - Standard bell curve, Triangular distribution and rectangular distribution - the formula is then devided by the square root of 2, 3 or 6 accordingly - what does this devisor represent and how is the number determined?

and you could encounter a 'bi-modal' distribution when you're considering sources of uncertainty, which has two 'humps'. So, I'm thinking that, like Cliff, to help you out, you'd be really, really much better off finding someone to give you some hands on assistance.......

Ok ... so this falls back to the idea of determining negligable factors of uncertainty. The idea is i need to create an "uncertainty budget" when i am doing layout work that identifies the items that are not negligible and these items then need to be monitored and calculated to give an expanded uncertainty estimation. My question here is what is the determining factor that makes a factor negligible - i.e. if i am doing layout work on steel sub componets does the expansion coefficient really matter if i am only reporting the test results to 0.000"? - and why/why not? Do apply the 10to1 rule (or perhapse 100to1 rule...) meaning that any effect that would be smaller than 0.00005" would not be significant and therefore not need to be measured, or is this where i use the rectangular distribution method when calculating my standard deviation? Or am i just confused here again... I think i might be catching on or then again...!
:thanks:


Developing the uncertainty budget is accounting for the relevant factors that will affect the readings/tests that you are to perform.

The advancements in metrology have shown that there are multiple variables that affect calibrating one device with another. While one variable alone may seem insignificant, with other variables add up to make a difference.

Utilizing the 10 to 1 ratio is needed when you are comparing A to B. Developing uncertainty budget is needed to ascertain the confident value that A represents, and B represents.

Say I have a pressure gauge that needs calibrating. What accuracy do you need the pressure gauge to be calibrated? Well, once you have that, then you need a standard with an accuracy at least four (or ten) times better than that gauge. However, you need to perform the uncertainty analysis on the standard to determine all the relevant factors that will make up that tolerance.

Oh, and exactly what Andy said on calibration and the use of the term. If you will review some of our threads here, we have struggled with making sure we are exactly on board with what calibration, verification, and such means when we use the term. It's not that any of us are intellectually snooty:lol:. We just want to assure that we are all talking about the same thing. So we may clarify the use of a term/phrase every so often.:)

Ok I have a mth question now (bare with me)... When you plot standard deviation charts there are 3 types of charts that i am aware of - Standard bell curve, Triangular distribution and rectangular distribution - the formula is then divided by the square root of 2, 3 or 6 accordingly - what does this divisor represent and how is the number determined?

Not to pile on with the others, but this question is quite intriguing.

1. What formula are you referring to?
2. Any/all data will represent some distribution of some kind. Depending upon a various set of variables, the data may be normal, light/ medium/ heavy skewed, Bimodal, Poisson, Binomial, random etc.

You may want to post your data in a spreadsheet so we can see what you're doing.

Lets move in a different direction for the time being...
The first clause within 17025 states that you must be an organization that can be held legaly responsable - How can this be verified?
Also what is required for proficiency testing - How rigorous does the testing need to be, must it be documented? Does it require third party approval?

Not to pile on with the others, but this question is quite intriguing.

1. What formula are you referring to?
2. Any/all data will represent some distribution of some kind. Depending upon a various set of variables, the data may be normal, light/ medium/ heavy skewed, Bimodal, Poisson, Binomial, random etc.

You may want to post your data in a spreadsheet so we can see what you're doing.

I am reviewing the G103 - A2LA Guide for Estimation of Uncertainty of Dimensional Calibration and Testing Results. I am trying to gain a better understanding of the examples provided in the document.

I am reviewing the G103 - A2LA Guide for Estimation of Uncertainty of Dimensional Calibration and Testing Results. I am trying to gain a better understanding of the examples provided in the document.

Coleman, what others have suggested is that understanding this subject doesn't lend itself to being delivered in a forum like this. For those with a basic understanding of calibration and uncertainty calculations we have experts here who can provide guidance, but it's a subject fraught with complexity and without some sort of training you're unlikely to get the knowledge you need. That sort of training is beyond the scope of this forum, or any other I know of.

Well i have broken down and decided to attend a class on measuremant uncertainty. Is there anywhere that i can go to view a free copy of the ISO GUM so i am more comfortable with the material before i attend the class?

Well i have broken down and decided to attend a class on measuremant uncertainty. Is there anywhere that i can go to view a free copy of the ISO GUM so i am more comfortable with the material before i attend the class?

Good luck finding a free copy of the ISO G.U.M.

Here is a link to NIST Technical Note 1297 "Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results"

http://physics.nist.gov/Pubs/guidelines/TN1297/tn1297s.pdf

Coleman, Perhaps the best option for an uncertainty course is to attend Measurement Science Conference (MSC). There are a couple of good uncertainty courses. One to consider there is the one taught by Dilip Shaw, Dilip is a good instructor, and the course has a good reputation.

Also, you can go around after and see vendors of similar devices that you may be using and ask more direct questions.

If you do get to go, drop by our booth and say hi.

http://www.msc-conf.com

Hershal

Hershal's suggestion of taking a class from Dilip is a good one. Dilip lives in Cleveland (or Akron??), Ohio. Since you are also in Cleveland, you may be able to contact him and get some consulting.

You can try to contact the American Society for Qulaity for more information. His e-mail address is Emc3solu*********

His last name is Shah.

I wish someone would have told me about the Measurement Science Conference before i signed up for my class... The class is through Quametec -does anyone know anything about this company?

Quametec is good. I know them well.

I am having a hard time locating the ISO GUM would someone please point me in the right direction - I was told that this would explaing the 2 types of measuremeant uncertainty but i am getting documents that are unrelated when i try to search for the obvious descriptions. :confused:

I am having a hard time locating the ISO GUM would someone please point me in the right direction - I was told that this would explaing the 2 types of measuremeant uncertainty but i am getting documents that are unrelated when i try to search for the obvious descriptions. :confused:

http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detail.htm?csnumber=45315

when I click on the link there is no purchase option? It does however allow me to purchase a different document.

when I click on the link there is no purchase option? It does however allow me to purchase a different document.

Try this link (http://webstore.ansi.org/RecordDetail.aspx?sku=Guide+to+the+Expression+of+Uncertainty+in+Measurem),where the GUM is offered by ANSI.

I am having a hard time locating the ISO GUM would someone please point me in the right direction - I was told that this would explaing the 2 types of measuremeant uncertainty but i am getting documents that are unrelated when i try to search for the obvious descriptions. :confused:

By all means buy the ISO GUM, then put it away and forget it ! is my advice. Yes it does sort of explain type A and Type B uncertainties, but it was written by metrologists for metrologists. I would really recommend re-reading one of the free on-line Guides to the GUM that are cited elsewhere in this thread. I also think it is good advice to read more than one Guide when one version is unclear on a particular topic.

If you must buy the GUM - in the USA it was adopted by ANSI as ANSI/NCSL Z540-2-1997.

Good grief folks.....why the fuss?

If you are in the U.S., then go to http://www.ncsli.org and look around till you find ANSI/NCSL Z540-2-1997 and buy it.....then go download TN 1297 (I believe I got the right number).....

Now, we are past that I hope, so let's get real.....uncertainty tells you what the error is in a given measurement....nothing more, nothing less.....it is NOT a be all to end all or a Holy Grail or any of that.....

The GUM, including the US GUM is the most boring document I know.....I have read Government technical procedures that are more exciting.....

You don't use the gum for its excitement.....you use it for its technical content.....for that it is great!

I have a standing challenge BTW.....if you start at the first paragraph of Chapter One and get through the first ten pages without either going to sleep or otherwise loosing focus, I got the first round.....I know exactly one person who I truly believe has done that and he does not drink.....

So let's get real about 17025.....

This is an integrated Standard.....take any Clause by itself including uncertainty and you gt a distorted picture.....look at the whole, where the Clauses describe parts of the whole.....don't take ANY Clause as the whole.....

Hope this helps.....

these are my two favourite reference when it comes to MU :

1. EA reference http://www.european-accreditation.org/n1/doc/EA-4-02.pdf

2. NPL, UK http://publications.npl.co.uk/npl_web/pdf/mgpg11.pdf

I only read GUM AFTER 2 years practicing MU and have enough bravery to not feel stupid against the statisticians. Its really good in making you sleep. Better options against insomnia rather than sleeping pills *wink*

As a part of clause 4.3.1 don't i have to have a controlled copy of the ISO Gum? Also is there anywhere that I can go to find a comprehensive list of the different scopes of acreditation?

As a part of clause 4.3.1 don't i have to have a controlled copy of the ISO Gum? Also is there anywhere that I can go to find a comprehensive list of the different scopes of acreditation?

I can't answer the GUM question, but as far as scope is concerned, it's going to be different for each lab, so there is no "comprehensive list." If a lab is doing calibration of electronic test equipment but nothing else, the scope will cover the types of equipment the lab is qualifed to calibrate and relevant standards. Likewise, if a lab is doing calibration of mechanical measuring devices (calipers, micrometers, etc.) the scope will be limited to those types of devices. The scope reflects what the lab has demonstrated it's qualified to calibrate during the accreditation process.

As a part of clause 4.3.1 don't i have to have a controlled copy of the ISO Gum?

No definitely not. I happen to be looking at my controlled copy right now, the first time for 2 years. Perhaps the only time I would want it is for a citation of some definition of terms and that's only as a participating member of ISO working groups. It's not even on my list of controlled documents reviewed by my 17025 Accreditation Body and they have never asked for it. It's a waste of space. That's why there are so many Guides to the "Guide".

I understand the concept of expansion coeficients and how they contribute to measuremeant uncertainty - can someone please explain to me the effect that humidity has on my uncertainty?
:confused:

What is the calibration in question regarding RH? Some calibrations are in fact affected by RH, some are not.

the uncertainty budget that i am producing is for dimensional testing/layout work - not calibration

Humidity may not play a part in uncertainty. It depends on the material of the part you're laying out. If it's a plastic, humidity might be a factor - some plastics are hygroscopic (mostly nylon, I think). But if you have metals, it shouldn't be significant.

I appreciate the response and as always it seems to bring forth additional questions...

My understanding is that at the time of auditing I will define the measuremeant uncertainty of our Lab - but what you are saying leads me to believe that we will have a different uncertainty value for every type of material that is measured. Typicaly i will be working with steel - but different chemical properties within the steel will cause my uncertainty for things like COE to vary.

Does this mean that i will recalculate my measuremeant uncertainty for each type of material that is being measured?

Also is there any way to document that humidity will be neglagible when measuring steel?

I appreciate the response and as always it seems to bring forth additional questions...

My understanding is that at the time of auditing I will define the measuremeant uncertainty of our Lab - but what you are saying leads me to believe that we will have a different uncertainty value for every type of material that is measured. Typicaly i will be working with steel - but different chemical properties within the steel will cause my uncertainty for things like COE to vary.

Does this mean that i will recalculate my measuremeant uncertainty for each type of material that is being measured?

Also is there any way to document that humidity will be neglagible when measuring steel?

It may be true to say that various steel alloys will give a variety of C o E's. It may be easier, in considering U of M that you treat them as a range within which they fall.

The whole purpose of establishing your U of M is to document their effects - in this case to show they are negligible (humidity) or rest within this range (C of E)