BMC (Best Measurement Capability) exceeding Tolerance!

[rmf180]

I've been told by our calibration house that it is not unusual for BMC to be greater than tolerance due to the nature of the calculations. This doesn't make sense to me. If I am calibrating (certifying) gage blocks to +4/-2 micro inches, why is it acceptable for the BMC to be greater than 5 micro inches? Isn't this similar to measuring parts at ± .0005" with a micrometer with accuracy of .001"? I'm not an expert on uncertainty which leads to my confusion. Any help is greatly appreciated!

[Jennifer Kirley]

I'm with you, it sounds like baloney to me. What is the item the calibration house is talking about?

[Phil Fields]

What is "Best Measurement Capability", I have not heard of this.

Phil

[Stijloor]

Originally Posted by Phil Fields

What is "Best Measurement Capability", I have not heard of this.

Phil

Phil,

As an example, look at this document, page 4, Par C.

That may clarify.

Stijloor.

[Mchurch]

Originally Posted by rmf180

I've been told by our calibration house that it is not unusual for BMC to be greater than tolerance due to the nature of the calculations. This doesn't make sense to me. If I am calibrating (certifying) gage blocks to +4/-2 micro inches, why is it acceptable for the BMC to be greater than 5 micro inches? Isn't this similar to measuring parts at ± .0005" with a micrometer with accuracy of .001"? I'm not an expert on uncertainty which leads to my confusion. Any help is greatly appreciated!

Your calibration house is not misleading you.

This is a common practice when you approach high accuracy calibrations. Gage blocks are often used as an example of this as well, since accross the entire industry virtually nobody except a small few of the manufactorers and National or International Standards labratories can hold an acceptable TUR on a gage block.

Your calibration house is most likely giving you thier reported measurements and stating the uncertainty with those numbers. Since every single measurement will be an undetermined (Uncertainty>Tolerance) they must do this to comply with most calibration standards ex. 17025, 10012, Z540.

Remember that when you see a Best Measurement Capability, it is an expression of what can be proved to an auditor. You may be 10 times better then that or 20 times worse then the number, but you must be able to defend your reasoning and prove that you can accomplish what you claim.

For example if your calibration house is measuring your 4 In gage block;
They would have thier 4 inch gage block that is of tighter tolerance +2/-1 that they send off to more accurate lab that gives them a reported uncertainty of +/- 2.1. They would have a system of comparison between the two blocks resolution .1 micro or better. They would have repeatability errors on this number due to a whole lot of different factors and human skill, probably a +/- .4 error, then you add in enviromental conditions... If they can only measure surface or contact temperature at +/- .5 degrees C accurately they have just added +/- 23 micro of uncertainty. 4 inches * 11.5 uIn/In/Deg C expansion of steel * .5 degrees of uncertainty...

When you do the math and something as simple as not being able to prove they can measure the tempurature more accurately destroys thier BMC.

At then end of it all, the average BMC on gage blocks you will see from third party cal labs is something around 4-5.5 uIn + 1-2 uIn/In. They can adaquetly test your items but have only proven themselves to these numbers for a variety of different reaons.

Just know that you are saving money by sending your product to these compainies, your are risking false failures on a good product or false acceptance on a bad product, determine your own demands for the uncertainty of your calibration, if need be, send any failed unit to a different lab for a second check (of higher accuracy, for gage blocks I would recommend Starrett, Mitutoyo, Nist).

Hope I was of some help.
Mark.

[Hershal]

Originally Posted by rmf180

I've been told by our calibration house that it is not unusual for BMC to be greater than tolerance due to the nature of the calculations. This doesn't make sense to me. If I am calibrating (certifying) gage blocks to +4/-2 micro inches, why is it acceptable for the BMC to be greater than 5 micro inches? Isn't this similar to measuring parts at ± .0005" with a micrometer with accuracy of .001"? I'm not an expert on uncertainty which leads to my confusion. Any help is greatly appreciated!

First question is what size and grade gage blocks are you calibrating? A 1" grade 0 gage block at a high end lab may have an uncertainty of 3-5 micro inches, or a grade H with a laser interferometer may achieve much less uncertainty across 1", say 500 nanoinches.....

But even the best calipers, using a grade 0 gage block will be pressed to get under 250 microinches, and micrometers perhaps 100 microinches.....part of that is resolution error and parralax error.....

Please do not take this the wrong way.....your question is NOT unfounded.....so may I suggest a good course on uncertainty?

There are many available now and coming up, such as the courses from Quametec and Integrated Sciences Group and also International Accreditation Service.....you should check the websites over the next couple of weeks, contact all those and other organizations, and determine which is the best match, based on cost and location.

Hope this helps.

[Hershal]

Originally Posted by Phil Fields

What is "Best Measurement Capability", I have not heard of this.

Phil

In theory, Best Measurement Capability, or BMC, is a means of determining the least uncertainty of a measurement that a calibration laboratory will normally achieve.....

This is different than the current discussion of BMC and CMC that is going on at the National Measurement Institute (NMI) level.....

Now, reality, and a bit of venting by me, so I will no doubt be trouble - again - in the international arena.....

Best Measurement Capability, or BMC was originally defined as the lowest uncertainty a laboratory could achieve using the "most nearly ideal equipment and conditions".....or put another way (I will be a bit elaborate, please do not be offended).....you use your best tech, best environmental conditions, have the Priest swinging incense, sprinkling Holy Water, give burnt offerings, whatever it takes to get your BMCs.....

Put another way, this is a fantasy number with only a few execptions.....one is the Fluke (T-Cal +/-5 deg C, I year absolute uncertainty) - WHERE - the 4:1 TUR can be maintained, in which case, Fluke has convinced me that this is the - EFFECTIVE - delivered uncertainty.....

OTHERWISE.....BMC is a pure fantasy number.....and I suspect I may be the only calibration assessor you ever hear say that.....but reality, you the customer are unlikely at best to ever see your lab's BMC.....

Never, ever, use BMC to select your cal lab.....ignore that number if at all possible.....ACTUAL uncertainties tell you something.....BMCs mean squat.....

Put another way, ignore BMC.....concentrate on real, delivered quality.....that is important, BMC's are immaterial.....

Just my personal opinion, no one else's opinion, this is only me, just Hershal.....

[Mchurch]

Originally Posted by Hershal

OTHERWISE.....BMC is a pure fantasy number.....and I suspect I may be the only calibration assessor you ever hear say that.....but reality, you the customer are unlikely at best to ever see your lab's BMC.....

Never, ever, use BMC to select your cal lab.....ignore that number if at all possible.....ACTUAL uncertainties tell you something.....BMCs mean squat.....

lol at the ideal scenario for achieving your labs BMC.

Hershal makes a good point here though, One thing you should probably look at is the reported uncertainties you are recieving. I find it a big red flag when sending equipment for calibration and recieving a labs BMC as the stated uncertainty.

There are only a couple scenarios in which this would be acceptable, One is that they fluffed thier Type B uncertainties on thier BMC so that it is achievable in virtually all conditions thier labratory might be in. Another scenario is, and its the scary one, you are just recieving the same number for each calibration they preform and they are not actually calculating the uncertainty for your calibration individuallly.

If you have a question about the uncertainty you are recieving on your calibrations, ask them for the uncertainty budget associated with those numbers to see what took place. Most calibration providers (that are accredited) should be able to provide this upon request with little to no problem.