Test Gauge Failure

[Instnapp05]

Hi All,

An Ashcroft test gauge model#1082(pressure range 0-1500 PSI) with a 0.25% accuracy FSO was set up for calibration on an Ametek DWT having an accuracy of 0.1% of nominal value and was found to be within 0.25% tolerance, tested okay.

However, same(gauge) was sent to a third party for certification and they had used a DWT having an accuracy of 0.015% of reading .When they performed an as found verification they indicated to me that the gauge was out of tolerance and failed calibration.

Test equipment used with a lower accuracy for calibration will show a significant difference in results but if same instrument(gauge) was calibrated with a 0.1% and passed, are we going to consider the gauge failed if used with a 0.015 % test equipment?

Thanks and appreciate your feedback.

 

[MikeHalloran]

If you didn't personally observe the third party test, they probably screwed it up.

The most common mistake in using a DWT is failing to manually rotate the weights to remove seal friction from the piston packing gland. That's why the weights are supplied as round plates that stack and nest neatly.

Without spilling the beans about that secret, ask for a copy of their test procedure.




Mike Halloran
Pembroke Pines, FL, USA

 

[xnuke]

Regarding the first test with the Ametek DWT, many companies would not accept the results simply because the calibrator uncertainty isn't four or more times less than the uncertainty of the device under test. (Yes, I refuse to say accuracy when uncertainty is meant. I wish everyone else would. End of rant.)

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[IRstuff]

Most of the product pages show Ametek DWTs with much better performance than what's cited

http://www.jmtestsystems.com/pages/productdetail.asp?ID=21

I'm curious how the calculations were made. Is the instrument very close to its uncertainty limit?

TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[Instnapp05]

Hi All,

The results are as follows;

When I tested the same gauge using a 0.05 % accuracy digital test gauge;

Input(PSI) Actual(PSI)

0 0
300 300.25
600 600.32
900 900.41
1200 1200.57
1500 1500.63

All readings were found to be within +/- 0.75 FSO.

Third Party results;

Input(PSI) Actual(PSI)

0 0
300 299.09
600 599.11
900 899.23
1200 1199.42
1500 1499.51


Since they had used a lower accuracy (0.015%),based on their results above, why is the gauge considered failed if found acceptable using a 0.05% test gauge for verification?

Appreciate the feedback.

 

[xnuke]

Good question as to why they said it failed the gauge calibration check. Based on ±0.25% of full-scale output uncertainty for the gauge, each reading must be within ±3 psig of the nominal reading at each of the five points. You can't calibrate out random uncertainty, so the gauge should never be expected to have a smaller uncertainty than ±3 psig of the nominal reading. All test readings from both testers certainly are within that range in the calibration checks. I'd ask about the tolerances they're using.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[IRstuff]

Perhaps your 3rd party didn't understand the requirements. Did either or both places provide the scale and offset errors?

My quick look suggests that the random error is VERY small. The only anomaly is the 0 psi reading from the 3rd party, which, based on scale and offset, should have been closer to -1 psi, which would still have been well within the requirement.

A true "calibration" would measure the scale and offset errors, which are then applied to the resultant data and the residual error is the uncalibratable error.

TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[IRstuff]

Below is a graph, calculations from Mathcad, showing the uncalibratable errors as being quite small, for both cases. You can also see that the second guy's data indicates a scale and offset error in the residual errors, which, if you replace the 0 psi with -1 psi, results in much lower residuals.

Not sure why the text is munged.



TTFN
I can do absolutely anything. I'm an expert!
faq731-376 forum1529

 

[danw2]

I'm really impressed with the out-to-two decimal places values for the calibration measurements.

The 1082 is a bourdon tube gauge with a 270 deg dial scale.

It has either 5 psi or 2 psi minor divisions, depending on the dial diameter.

Getting a reading to the 2nd decimal place is truly an admirable task. Eagle eyes.

The gauge performance is even better than reported/expected using my desk calculator which shows 0.25% of 1500 to be almost 4 psi; 3.75 psi to be exact.

If it were me I'd be looking for another "3rd party" cal lab.

 

[Instnapp]

All,

Thanks for the feedback.