Calibration Standards

[mpsmith]

I am looking for a standard calibration deviation tolerance.

i.e. if you have .010 ±.005 given to the machinist to cut. What will the + percentage be for some one to calibrate to, and is there such a thing?

 

[Andy981]

NASL and ISO standards recomend the manufacturer to establish a measurement uncertainty factor for dimension calibrations.

Years ago a 10:1 factor was used. But it was expensive. Now-a-days a 4:1 ratio is prefered. Per your example of .010" +/- .005" The tolerance of .005" is divided by 4. Therefore your calibrating instrument myst be accurate to .00125 or better.

 

[cobb]

Andy981 where is this 4:1 standered, I havent seen anything on this, I still ues the 10:1

Thanks Scott Cobb
CSM Manufacturing, Inc
csm-mfg.com

 

[IRstuff]

MIL-STD-45662 Calibration Systems Requirements, paragraph 5.2 Adequacy of measurement standards, states:

"...the collective uncertainty of the measurement standards shall not exceed 25 percent of the acceptable tolerance for each characteristic being calibrated. ..."

Note that this is somewhat tighter than what one might reasonably do with RSS'ing of the component tolerances, since the words imply a simple arithmetic sum of the tolerances. Usually, though, most customers accept the argument for RSS'ing of the tolerances.

TTFN

 

[IRstuff]

Sorry, missed a couple of bigger ones due to too much haste.

MIL-HDBK-52B Evaluation of Contractor's Calibration System, paragraph 5.1 Calibration system requirements, states:

"...This is commonly called a 4:1 test accuracy ratio..."

Likewise MIL-HDBK-1839A Handbook Calibration and Measurement Requirements, paragraph 5.4.3 Test Uncertainty Ratio, states:

"The recommended TMDE shall be capable of measuring or generating to a higher accuracy than the measurement parameters being supported. Unless otherwise specified, a minimum TUR of 4 to 1 is desired. The actual TUR shall be documented." TTFN

 

[mpsmith]

Thanks for all the info it has been a great help!

 

[IRstuff]

Should point out that in a normal environment, you don't use the calibration standard for measurement purposes, such as checking finished parts. You'd use calipers or someting similar. These are at best secondary standards, so they're a 4:1 ratio from the primary or secondary standard, which in turn is 4:1 ratio from a NIST standard.

Therefore, the NIST standard could be as much as 4x4x4=64 times smaller tolerance than anything you're actually measuring.

For linear distances, the above is quite easily obtainable, since distance measurement is reasonably accurate down to the submicrometer level. In other areas, 1% relative accuracy is pretty darn good. TTFN

 

[hldhstg]

I was reading this thread and it's very interesting. I was looking around in the NIST website and trying to find this 4:1 standard, but was unable to locate it. Do you have a URL or some sort of code/regulation number that references NIST as recognizing 4:1 as the standard? I'd really like more information on this issue if possible. Thank you.

 

[IRstuff]

the 4:1 ratio is actually a military thing, per MIL-HDBK-52

TTFN

 

[hldhstg]

I understand that. I thought it was adopted by NIST as insdustry standard as well?

 

[IRstuff]

ISO 10012-1 paragraph 4.3 page 5 Guidance indicates:

"The error attributable to calibration should be as small as possible. In most areas of measurement, it should be no more than one third and preferrably one tenth of the permissible error of the confirmed equipment."

TTFN

 

[grvldrvr]

Back on March 7, Andy981 states that in the case of +/-.005 tolerance, the 4:1 ratio would apply to .005 for an accuracy requirement of .00125. I have always thought the 4:1 ratio applies to the total tolerance, or .010 in this case, giving an accuracy requirement of .0025. Comments?
Thanks
SMR

 

[IRstuff]

+/-.00125, what's the issue?

TTFN