Good points Dick. A good question to ask yourself is what level of documentation do you really need? And what is the cost per PPM. You may find a local lab with less accreditation charging half then price that is fully capable of of calibrating to the limit of the 732A but cannot document to the level of a primary standards lab. Add to that the fact that the 732A cal'ed locally would not be subject to the abuse of shipping and the calibration on you bench may exceed that of a NIST JJA cal. Yes a 1-2PPM Cals is not as sexy as a .1PPM Cal but in the real world the results when used in you home lab my be the same.
Just look at the effects of temp and humidity on calibration and you will see my point.`to yours.
Thomas Knox

Thomas wrote:

Very, very doubtful.  Very few cal labs have a 732A or equivalent,
much less anything better.  The only labs with better uncertainty
than a properly working 732A are those with JJAs.  If you look at the
NIST NVLAP accreditation list and run down it, looking at each lab's
"Scope of Accreditation," you will only find 4 or 5 labs on the list
with better uncertainty than a properly working 732A (I'm not sure
you will find any on the A2LA accreditation list, but I haven't run
down it lately).  The Fluke cal lab and the Los Alamos and Sandia
standards labs are three of those four or five (plus, of course, NIST
itself).  Boeing (Seattle) is another.  Interestingly, you will find
many labs that are rigorously accredited to only .003% or so (30
ppm), because the best voltage standard they own is an HP 34401A
DMM.  Even the HP Houston cal lab is certified to only 0.0007%, or 7
ppm (using a Fluke 5700A calibrator).

To get a calibration with an uncertainty of 1 or 2 ppm, the lab would
need, at a minimum, a 732A or 732B to compare with (as well as a 720A
Kelvin-Varley bridge, or equivalent, and a null meter that can
reliably be read to 0.1uV, if you want the calibration certified to 1
or 2 ppm at voltages other than 10v).  I don't think there are even
ten labs on the NVLAP list that claim to have a 732A or B (the
equipment used is often listed in the "remarks" column).

It does not take long to run down the whole list -- it's a short list
and the "Scope of Accreditation" documents load fast.  I recommend
the exercise, to get a feel for what's out there.  Same with the A2LA
list, but it is longer and not as well organized and it usually takes
2 or 3 steps (running off to the lab's site) to get to the "Scope of
Accreditation."  (If you look at A2LA labs, pay attention to the lab
class and only look at "open" commercial labs -- the non-commercial
ones do not take in third-party calibration work.)

A list of NVLAP-accredited labs can be found here:

http://ts.nist.gov/standards/scopes/dclow.htm

There seems to be this myth of cal labs that can do just as good a
job as the expensive, accredited labs, but don't bother with
accreditation so they are much cheaper.  First, note that to do a job
as good as an expensive, accredited lab, any lab would have to do the
same documentation as the accredited lab.  If there is no
documentation, there can be no claim as to the calibration's
uncertainty.  Having done the documentation, which is the
time-consuming (thus, expensive) part, no commercial cal lab is going
to do without the accreditation (which is nothing but an audit of the
lab's procedures and documentation).  I stress again -- if there is
no documentation, there can be no claim as to the uncertainty of a
lab's work.  And since the documentation is the part that contributes
most to the cost, there simply are not any commercial labs that can
claim to have uncertainties on par with accredited cal labs, but are
not themselves accredited.

Best regards,

Charles

Hi Charles;
Perhaps you are correct. I work in the research area of metrology. I just assumed a number of labs have multi-cell standards like the Wavetek 4910 or Fluke 7010s or 734A and if recently cal'ed they should be more then capable of calibrating a 732A for home use. True the documentation is not there but does that matter in a majority of Volt-Nut applications. And if you ship VS local calibration you may have documentation stating lower uncertainty, but a local cal may actually be more accurate.  I also think the number of JJA's is larger then most think. Just this week a couple labs added 10 Volt Programmable Josephson voltage standards capable of producing AC and DC voltages. http://www.wmi.badw.de/teaching/Lecturenotes/AS/AS2013_Chapter6_Slides.pdf
Again my point is do you need the documentation, and if not there are alternatives. Especially in a non temperature and humidity controlled environment that will affect both your standards and DUT. Don't get me wrong I am all about documentation and repeatability but not everyone needs or has the $300,000+ for a 10 Volt Programmable Josephson voltage standard. Again I float the idea of Volt-Nuts creating their own network to compare standards locally after a recent calibration.

Thomas Knox

Thomas wrote:

If the documentation has not been done, there is no basis upon which
to make ANY claim about the accuracy of a calibration.  Really, the
documentation IS the calibration.  Having a voltage standard that is
exactly 10.000000v but not knowing for sure what the voltage or the
calibration uncertainty is, is no better than having a voltage
standard that is tens of uV off and not knowing for sure what the
voltage is.  The documentation is the only way the adjustment becomes
useful -- now you know exactly how far you can have confidence in the
voltage.  Indeed, a standard that is tens of uV off is a perfectly
fine standard as long as you know the uncertainty -- which you would
only if the documentation has been done.  So yes, the documentation
is absolutely critical to any calibration, accredited or not.  And no
commercial lab that puts in the extreme effort to document its
results to the 1 ppm level or better is going to pass on
accreditation, because it would seriously limit their business.

There are mountains of data on 732As that get shipped to Fluke every
year, and there is no problem getting uncertainties at the 0.1 ppm
level back at the customer's facility.  This is just a non-issue with
732As.  (The same is true of humidity, and temperature as long as you
stay within 5C of the cal temperature.)

The number of JJAs deployed is not the issue -- lots of research
facilities have them, but they do not calibrate third-party
instruments.  For someone wanting to get an instrument calibrated,
the question is, how many commercial calibration labs have them?  And
the answer is, only a very, very few.  As I said before, take a look
at the NVLAP lab certifications.

Best regards,

Charles

Charles is difficult to disagree you are simply explaining the system that has evolved over decades that insures the accuracy of our standards labs, and I have found I am barely smart enough to get coffee for some of the Senior Time and Volt Nuts members. You have made your point very eloquently and I don't necessarily disagree with anything you say. But I feel reading your response I have failed to make my point.
I will make one last effort using a example. If I send a 732A to Fluke and have it calibrated, and you stop by with yours shortly after mine returns, has a day or so to settle, and appear correct with my 3458A, you could then calibrate your 732A using of mine and feel fairly confident you standard is also a few PPM from 10 volts. Now lets stay you repeat this with another Volt-Nut when his standard returns. You do not have documents proving the accuracy of your standard because none of the standards used are parts of a accredited lab, but you can be confident yours is correct and You have save nearly $1000 dollars. There is what you can prove, but there is also what you know using common sense. Many Volt-Nuts also work in Metrology or know someone who does. In my work these days solutions at the cutting edge of Metrology are often as much art as science. That intuition often leads to advances that are both document-able and repeatable. So as I brought up earlier if Volt-Nuts compared standards and shared data locally they could build there own model and uncertainty algorithm that although not recognized by accrediting bodies could achieve similar results.  If you can afford to fully document the instruments in you lab I encourage you to do so and support the industry, but that can be beyond the budget of even some serious researchers and exploring alternatives seems to be what the Volt-Nuts are all about. Also I have been collecting data for a number of years hoping I can find time to do a serious study showing the effects of shipping on calibration. And I can tell you my early data points toward the fact that a substantial part of the yearly uncertainty happens during shipping. I am also convinvced that in that next decade instruments will include an environmental sensor package, and cal will be based more on environmental expose then simple time. As Physicists have pushed the limits of Metrology in recent years there has been little research in insuring these improved accuracies reach the end user. Expect changes in coming years.

Thomas Knox

I have to side with Tom on this one. Accreditation only provides independant verifcation of processess and proceedures. It provides increased confidence (in both the human and statistical senses) in the service provided. It does not prove that your standard is correct. The typical time nut does not need accredited calibration. True, if a commecial cal lab is any good it will have accreditation, but if they only do work in-house they may not need it. If I could get access to a JJ in a local non accredited research lab to compare my 10V  reference to, I'd have very high confidence that it is correct. If  I can make repeated visits over time and document those, even better. I personally think that this would be better than hot shipping it to a remote accredited lab. It might also release limited funds for a second or third reference for even greater confidence. I've had to recall aircraft equipment that was mis-adjusted due to a bench test set being
calibrated incorrectly. All the links in the chain were accredited and documented. This was finally traced to the manufacturers reference standard being off following calibration (at a high level). The conclusion was a shift during transport. They had calibrated several bench test sets before we caught it. It was only because the unit failed badly on this minor parameter during an aircraft check at the same base as the workshop and then failed again after repair that it was spotted. I continually complain about over calibration at work. For example 12 monthly calibration of variable bench DC power supplies. just stick an "indication only" sticker on them and use an external meter if the voltage is critical.

Robert G8RPI.

From: Tom Knox actast@hotmail.com
To: Discussion of precise voltage measurement volt-nuts@febo.com
Sent: Sunday, 9 March 2014, 21:07
Subject: Re: [volt-nuts] A Fluke 732A

Charles is difficult to disagree you are simply explaining the system that has evolved over decades that insures the accuracy of our standards labs, and I have found I am barely smart enough to get coffee for some of the Senior Time and Volt Nuts members. You have made your point very eloquently and I don't necessarily disagree with anything you say. But I feel reading your response I have failed to make my point.
I will make one last effort using a example. If I send a 732A to Fluke and have it calibrated, and you stop by with yours shortly after mine returns, has a day or so to settle, and appear correct with my 3458A, you could then calibrate your 732A using of mine and feel fairly confident you standard is also a few PPM from 10 volts. Now lets stay you repeat this with another Volt-Nut when his standard returns. You do not have documents proving the accuracy of your standard because none of the standards used are parts of a accredited lab, but you can be confident yours is correct and You have save nearly $1000 dollars. There is what you can prove, but there is also what you know using common sense. Many Volt-Nuts also work in Metrology or know someone who does. In my work these days solutions at the cutting edge of Metrology are often as much art as science. That intuition often leads to advances that are both document-able and repeatable. So as I brought up
earlier if Volt-Nuts compared
  standards and shared data locally they could build there own model and uncertainty algorithm that although not recognized by accrediting bodies could achieve similar results.  If you can afford to fully document the instruments in you lab I encourage you to do so and support the industry, but that can be beyond the budget of even some serious researchers and exploring alternatives seems to be what the Volt-Nuts are all about. Also I have been collecting data for a number of years hoping I can find time to do a serious study showing the effects of shipping on calibration. And I can tell you my early data points toward the fact that a substantial part of the yearly uncertainty happens during shipping. I am also convinvced that in that next decade instruments will include an environmental sensor package, and cal will be based more on environmental expose then simple time. As Physicists have pushed the limits of Metrology in recent years there has been
little research in insuring these
  improved accuracies reach the end user. Expect changes in coming years.

Thomas Knox

volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.

Thomas wrote:

Documentation and accreditation are two entirely different
things.  In this case, the documentation would be the informal, "on
[date] I compared and adjusted this standard to Thomas's 732A that
was just back from calibration and sanity-checked with a 3458A, using
a [Fluke 845A null meter, or whatever]."  Knowing the published
uncertainties of the 732A, I can even calculate the uncertainty of my
732A as of the moment of calibration, and at intervals thereafter
based on the published uncertainties.

Checking against another recently-calibrated 732A at a later date can
confirm or deny that mine remains within the published uncertainty at
that time.  If it does, that fact becomes part of my (informal)
documentation.  If it doesn't, the presumption is that one of the
732As is broken.  We can suspect that it is mine, but we cannot know
that without further testing.

Best regards,

Charles