NAPT Proficiency Testing

Started by RFCAL, 08-29-2014 -- 12:07:53

Previous topic - Next topic
Print
Go Down Pages1
User actions

RFCAL

08-29-2014 -- 12:07:53
Has anyone had problems with the way NAPT conducts their tests? I have an Attenuator dB loss PT from last year and this year. Our data passed last year, but this year it failed in 8 different spots. After checking closely, I found the pivot labs data had changed 0.2dB lower from last year to this year. Meanwhile, our data remained close to the same. I.E failure.

Hawaii596

#1
08-29-2014 -- 13:14:13
I have an issue I am waiting for the right moment to bring up regarding Oscilloscope bandwidth PT's.  Using the Fluke 5520A with 1100 MHz scope option, the manual requires you to use the matched cable.  I don't remember the exact numbers, but I did some experimenting using a regular cable versus the matched cable, and compared using the Wavetek 9500 w/9530.

What I found was that (example based on memory) when you do the measurement at 3 Vpp, I believe at 300 MHz, I got a bandwidth reading of around 2.92 Vpp using either the 9500/9530 (no cable involved - direct connect), and about 2.92 Vpp using the matched cable with the 5520A/1100.  But a reading of about 2.74 Vpp using the 5520A and other than the matched cable.  I even charry picked good cables and high quality terminators (including the precision Tektronix one, the Huber-Suhner Type N one, etc.).  I pretty consistently got about 2.74 Vpp that way.  The PASS on the PT is based on the 2.74 Vpp value, which the provider (NAPT) got.  I am pretty convinced that the PT is wrong.  I haven't brought it up yet.  We passed because of increasing uncertainties to account for operator to operator variability, etc.  But I am troubled that the nominal appears to be based on doing it wrong.  Meaning you could fail for doing it right, and be much more likely to pass for doing it wrong.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

Bryan

#2
08-29-2014 -- 16:50:40
Had the same issue with the oscilloscope and ended up increasing our uncertainty after a lot of examination of the effect of different cable/terminations.  Had a second issue on it as well, submitted our findings and reasoning-nothing yet.  I prefer to test myself against calibrations performed by our suppliers, works better for me as we are typically testing the same points.

CalLabSolutions

#3
08-29-2014 -- 18:04:28
I am not an expert here.. By no means..
But I think the problems is the PASS/FAIL part of the whole process.  I don't know, because I have never set down and researched the standard.  But I don't think the PASS/FAIL results from any organization is required or even wise.

To be, the goal is to see how your lab compare to other labs.  Every lab in the loop should have to report the standards the method they used and the standards.  Then review the uncertainties provide by all the labs in the loop and make your own PASS/FAIL evaluation.

I see the problem when labs are comparing apples to figi apples (a far better apple in my opinion).  Less accurate lab standards can skew the results.  Different measurement techniques can skew the results. 

Well that is my 2 cents.
Mike
Michael L. Schwartz
Automation Engineer
Cal Lab Solutions
  Web -  http://www.callabsolutions.com
Phone - 303.317.6670

Hawaii596

#4
09-01-2014 -- 19:15:42
In the case of this one proficiency test measurand, the "correct" answer is the result from the PT provider.  So pass or fail by a participant lab is based on how far their measurand is from the control lab's measurand (plus combined uncertainties).  If there were for example, sixteen participant labs who (for fictitious example for the sake of my issue) all got pretty closely matching results, but the control lab got a different result, outside of the combined uncertainty, all of the participant labs could theoretically fail with good measurands, if the control lab / PT provider, measured incorrectly.

This is what I am concerned about.  In the case of my non-fictitious issue, I believe the PT provider may have had an incorrect value.

For the record, I am not here to denegrate the eithical basis of any PT provider.  I think there are good, conscientious people who may be doing something wrong, and as this is such a widespread type of measurand (i.e.: bandwidth of oscilloscopes), I believe it important enough that I plan to take it up with them.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

RFCAL

#5
09-02-2014 -- 09:39:51
Today will be the first chance I get to speak with a technical person from NAPT, supposedly. Let's see what excuses they come up with. I am very close to switching PT providers.

Hawaii596

#6
09-02-2014 -- 10:33:23
Again, what I believe the issue is, as I mentioned in an earlier post, is that the Fluke 5500A and 5520A scope options come with a matched cable which, according to the manual is to be used for bandwidth measurements, along with the matched feed through termination.  I noticed a lot of the numbers were a little low, which leads me to believe the PT provider and many participants are NOT using that cable and termination.  And so, the PT is skewed to those using the incorrect method.  This is something that generally doesn't make a big difference in false pass/fails, but it does make some difference.  Since the spec is normally greater than 4.2 div versus 6 division reference, the difference is small.  But large enough that in a proficiency test, it invalidates measurements.

I would be interested in what ever you may find on this.  Maybe I'm wrong.  But I'm concerned that I may not be.
"I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind."
Lord Kelvin (1824-1907)
from lecture to the Institute of Civil Engineers, 3 May 1883

RFCAL

#7
09-04-2014 -- 09:58:50
After sending additional data, pin depth readins on the attenuators, I still have not heard back from NAPT Technical.
Print
Go Up Pages1
User actions