When Does Resolution of Measurement Standard Apply in Uncertainty Calculations

Jerry Eldred

Forum Moderator
Super Moderator
I've had my own views on this for a while, but haven't had to deal with it. But it is coming up again, and needs to be properly addressed....

For example, a Fluke 5720A is used to verify 10 Volts DC on an Agilent 34401A DMM. There are various contributors to measurement uncertainty, which is not my question/focus for this question. But regarding resolution....

It makes perfect sense that the resolution of the DMM is a contributor. But where I have a problem is if I simply enter an output of 10 Volts DC from a Fluke 5720A calibrator, there are opinions out there that the resolution of the calibrator is also a contributor. I could even see the least significant non-zero digit of the calibrator being considered the resolution perhaps. But at a setting for a cardinal value, or even non-cardinal value, the resolution is accounted for in its specifications already.

Any knowledgable inputs as to exactly why resolution must be considered a contributor? Or, are there only certain AB's that think that, or is this subject to adjudication and engineering judgment.

I'd appreciate any knowledgable feedback on this anyone could give. Have a great evening all.
 
D

dv8shane

I personally account for the resolution of the calibrator too. The contribution is small but does exist.
 
D

dv8shane

Jerry

Attached is a copy of a budget I did 7 years ago for a Fluke 8506A calibrated with a Datron 4000A. Both units had 7.5 digits of resolution and had accredited calibration with adjustment to nominal just before I took the readings. The contribution is really small but it is there.
 

Attachments

  • 10 VDC Uncertainty Analysis.pdf
    61.9 KB · Views: 345

dwperron

Trusted Information Resource
Here is a link to an A2LA guidance document which covers the rationale for including resolution of the standard in uncertainty calculations. It's written by Klaus Jaeger, and as usual is an excellent presentation.

Go to the A2LA website
Click on Publications
Click on Search A2LA Site
Search for document G110
 

Hershal

Metrologist-Auditor
Trusted Information Resource
The real question is not, is there a difference? There is a difference.

The real question is, does that difference truly make a difference? That I would argue is not always true, because the difference may be so small that the reporting of the uncertainty is not affected. Given Jerry's example, if the uncertainty absent resolution is 10mV, and adding the resolution results in a change of say 0.05mV, then the difference is insignificant, and can simply be accounted for the uncertainty budget as a comment.

Typically, resolution is a more significant factor in mechanical and dimensional calibration, rather than electronic calibration.
 
S

Spleen

In theory, your standard (Fluke 5720A) resolution contribution is supposed to be included in your imported calibration uncertainty (The Fluke 5720A calibration certificate). In your case, the Short-term repeatability will be a bigger contributor. It should be already taken into account during the calibration.
 
B

brian33a

Yes, you are absolutely correct. Unfortunately, you have to prove it. The easiest way to do that, for any given model, is to include the resolution of the calibrator in the list of contributors.

Also, it should be noted that A2LA states (R205 4.2.1):
Every measurement uncertainty shall take into consideration the following standard
contributors, even in cases where they are determined to be insignificant, and documentation of the consideration shall be made:
a) Repeatability (Type A)
b) Resolution
c) Reproducibility
d) Reference Standard Uncertainty
e) Reference Standard Stability
f) Environmental Factors

So, even though the resolution of the 5720A is insignificant, they still require it. I can only imagine they require that because people kept leaving out the resolution of the standard when it was significant (even if barely).
 

dwperron

Trusted Information Resource
Yes, you are absolutely correct. Unfortunately, you have to prove it. The easiest way to do that, for any given model, is to include the resolution of the calibrator in the list of contributors.

Also, it should be noted that A2LA states (R205 4.2.1):
Every measurement uncertainty shall take into consideration the following standard
contributors, even in cases where they are determined to be insignificant, and documentation of the consideration shall be made:
a) Repeatability (Type A)
b) Resolution
c) Reproducibility
d) Reference Standard Uncertainty
e) Reference Standard Stability
f) Environmental Factors

So, even though the resolution of the 5720A is insignificant, they still require it. I can only imagine they require that because people kept leaving out the resolution of the standard when it was significant (even if barely).
You are correct, A2LA does require resolution in the uncertainty budget. Period.
But then, they also get inconsistent, such as this from the same R205 document:

4.0 Requirements for Uncertainty Calculations that support the Calibration and
Measurement Capability (CMC) on the Scope of Accreditation
4.1 Terms and Definitions:
b) Significant: “significant” further means a contributor whose contribution increases the CMC by five percent (5%) or greater

But even though you know that a 5720's resolution will not change the CMC by more than 5% you still have to include it - the reason we were given is because "you can't tell if it is significant unless you actually calculate it". So A2LA says you just have to do it, even if it isn't a significant contributor. But if you have different assessors...
 

Hershal

Metrologist-Auditor
Trusted Information Resource
I would say where the resolution makes a difference in the uncertainty that can be observed, then yes it must be considered. This is typical in calipers and micrometers for example.

A 5720 calibrating a 34401 likely no noticable impact.
 
Top Bottom