FAQ's: Measurement Uncertainty

FAQ's on Uncertainty

As a certified calibration technician and technical manager for a calibration lab, I’ve fielded a lot of questions from customers on uncertainty over the years. Here are a few of the most common and useful.

• What is measurement uncertainty?

  Measurement uncertainty is way to quantify the “grey area” that surrounds a measurement. A 1 inch measurement with an uncertainty of ± 0.1 inch means that the “true” value could be as small as 0.9 inch or as large as 1.1 inch, but is most likely very close to 1 inch.

•  

  Does uncertainty need to be on my certificate?

  Yes, a measurement without an uncertainty ends the traceability chain. It is also required by ISO 17025. You may ask to have uncertainty omitted from your certificate, but you should be aware of the consequences.

•  

  Why is my uncertainty more than half the tolerance? What about 4:1?

  It is not uncommon for devices to have a tolerance that is the same as their resolution. When resolution is considered it adds about 0.6 of the resolution to the measurement uncertainty. Test ratios (TAR & TUR) do not consider the unit under test so 4:1 cannot be calculated from the calibration certificate.

•  

  What is coverage factor?

  Coverage factor is how likely the “true” value is to fall within the stated measurement uncertainty. It is typically stated as k=x, where x is some whole number. A k factor of one is approximately 68%, two is about 95% and 3 is 99.7%. Most calibration companies report measurement uncertainty at a coverage factor of k=2.

•  

  Can I just report the uncertainty from the certificate when I calibrate my gages?

  No. The measurement uncertainty from your calibration certificate is just one factor of a proper uncertainty budget which must be created in order to report a proper measurement uncertainty of the device you are calibrating.

•  

  How do I apply temperature uncertainty to my dimensional gages?

  When considering temperature uncertainty in dimensional measurement, you must multiply the coefficient of thermal expansion of the material being measured, by the length being measured, by the temperature uncertainty. When performing the calculation be aware of units and make sure they are the same. For imperial units the units should be (μin/in/°F) *(in)*(°F) and the result will be in μin. When you look up the coefficient of thermal expansion it might be listed with °R (Rankine), these are the same “size” units as °F, just with a different zero.

•  

  What is a scope of accreditation?

  An accredited calibration company will have a published scope of accreditation. This will list the calibration parameters that they have been assessed to perform. In the case of an ISO 17025 accredited calibration company this will also include a CMC (Calibration and Measurement Capability) for each of the parameters listed.

•  

  Why does XYZ company charge extra for uncertainties? Can they do that?

  Because they can, and it is a bit of a “grey area.” If the calibration is not accredited, then there is no requirement. If the calibration is accredited, then the certificate must have “the uncertainty of measurement and/or a statement of compliance with an identified metrological specification or clauses thereof” (ISO/IEC 17025:2005 5.4.1.b). Different accreditation bodies view the second half of this statement differently, and this gave rise to this multitier approach to service offerings. 

   

   

Why is my uncertainty different this year as compared to last year?

I wrote a blog that goes into more detail about this issue a couple of weeks ago, so we will go with a short answer here. Things change. Measurement uncertainty is not written stone. As equipment ages its ability to measure can change. Sometimes it gets better. Sometimes it gets worse. The functionality of the equipment is part of the measurement uncertainty. Calibration labs are also constantly refining their offerings, and accrediting bodies are also changing what they expect to see included in uncertainty budgets. All these factors can cause reported uncertainties to change from year to year. If you see a significant change, then you should ask. There is probably a good reason.

Can you make my uncertainty smaller?

Maybe, probably not, and definitely not without repeating the measurements. If you are looking for a particular measurement uncertainty, then it is usually best to have that conversation with your calibration provider up front, and also spell it out in your purchase order.

Where is my uncertainty?

Measurement uncertainty is typically reported in one of two ways on your certificate. If the device has a fairly lengthy list of measurement parameters, then uncertainties will be reported as a line item. In this method of uncertainty reporting each measurement or group of measurements will have an uncertainty associated with it. If the device has a single or small number of parameters that are measured, then the associated uncertainties will typically be reported in the header or footer of the document. 

                                                        Have more questions?