Uncertainty and Resolution of Standard

[Bryan]

I use the Quametec Uncertainty tool box for official stuff and where the measurement is a calculated result and needs sensitivity coeffecients but for a lot of more common stuff I just use excel.

[Bryan]

Sorry to dig up & kick a dead horse, am working on my budget for Fluke 5520A/5522A, 11 to 20.5 amp range.
The 5520A resolution on the range is 100mA.  For a hypothetical 20 amp output
I get 0.016A type B expanded uncertainty w/k=2.  When I include the influence of that 100mA resolution of the 5520A it kicks it up to 0.060 amps and contributes over 70% of the uncertainty.  In a case like this I acknowledge it on the budget but I won't give it any weighting.   
G110 (pg12)says "*Must consider with documentation of the consideration made."
My position is I did consider it and discounted it because a reasonable person would do the same for a "best" estimate.

[beadwork]

The 5520A resolution on the 20 A range is 100 uA.

[Bryan]

The 5520A resolution on the 20 A range is 100 uA.

Agghhhh, you are right, error in the DCI Spec table (Getting Started, Mar 2003), helps to get off your ass and check in with reality.  The effect is nil.  I withdraw my snide comments.

[RFCAL]

The specification of the UUT should be the driving force in any uncertainty calculation. If it is not, further examination of the contributors is required.--quoted from A2LA

[Pylarinos]

Try the calibration quides of Euramet they are quite helpful or
M3003 uncertainty quide of UKAS.
As A2LA and UKAS are full members of the ILAC according to the Mutual Recognition Agreement you can use M3003 to get some ideas.

[Pylarinos]

An usual uncertaity budget for a voltmeter is the following:

1) Calibration of the standard
2 Difference between previous and last standard calibration of the standard.
3) Resolution of the voltmeter
4) Repeatability of the voltmeter
5)Uncertaity
6)Expanded Uncertainty

[Pylarinos]

Sory for my mistakes

A typical uncertainty budget for a voltmeter is the following:

1) Calibration of the standard (in ppm)
2 Difference between previous and last  calibration of the standard.(in ppm)
3) Resolution of the voltmeter (in ppm)
4) Repeatability of the voltmeter (in ppm)
5) Uncertainty (in ppm)
6) Expanded Uncertainty

[CalLabSolutions]

1) Calibration of the standard
2 Difference between previous and last standard calibration of the standard.
3) Resolution of the voltmeter
4) Repeatability of the voltmeter
5)Uncertaity
6)Expanded Uncertainty

I don't see any need to Item 2.  And they only time I have seen it used is when a lab is trying to say my standard is better that the published specifications (Item 5 I am assuning).

Item 6.. All the contributors (1,3,4,5) should be converted to a K=2.  I have never seen it as an additional contributor.

Mike

[CalLabSolutions]

Here is a link to A2LA's requirements.  And it give a sample budget.

http://www.a2la.org/guidance/A2LA_G110.pdf

Mike

[Pylarinos]

Mr. Michael Schwartz

Item 2 is a required contribution to the uncertainty, indicates the drift between the two last calibrations of your standard in use.
Item 6 is not an extra uncertainty contribution but the total uncertainty (item 5 calculated from items 1 to 4) multiplied by 2.
I can assure you that I have been through A2LA uncertainty procedures which are very helpful, but in Europe the uncertainty guides European Accreditation EA-4/02 or UKAS M3003 follow this method who leads to the same results.

Panagiotis Pylarinos

[USMC kalibrater]

I have written many budgets, sat with auditors while they review said budgets and not once have I been asked why I didn't include #2.  As far as I can tell #2's value should be captured in the standard's uncertainty or minimally so small that its negligible in the final result.
Either way I see no need to be so defensive I'm certain given any items, 10 people might come up with 10 different budgets and roughly the same uncertainty.  Just be certain mine would be the most correct...buwhahahahahaha 

[Bryan]

I think the #2 value would typically be covered by the uncertainty specs of the calibrator if using something like a 5720A but if one were using a 732A/B claiming less that spec then some analysis of drift might be included.  I do that running some Guildline current shunts, with several years of history they have a "manufacturer limit of error" (100 ppm) and several years worth of data so I am applying the standard deviation of the reported values, (much smaller).  That leads me to conflict if I should call this type A as it comes from a statistical analysis of measurements as opposed to type B because it is a part of the uncertainty of my measurement "system". 

[USMC kalibrater]

Bryan I would agree that its a type A.  I would do the same to tighten uncertainties on an old OmniTrak flow calibrator at a lab I use to run.

[RFCAL]

I would have to agree with Mike, item 2 would NOT be included in any of our budgets, nor is it required. It might be nice to have, but not required.