In this context, I'd like to recommend this Keysight channel about metrology and their different facilities:
https://www.youtube.com/user/KeysightCalRepairBob Stern, their chief metrologist, presents a series of videos, including several virtual lab tours.
Especially the video about Loveland Standards Laboratory is very interesting, which had been built in the 1980ties, exclusively for the verification of the 3458A.
They have their JJA inside, and all 3458A in U.S., were calibrated there.
Recently, a web based seminar covered these subjects also, "Accuracy matters: Calibration Options for Lab Standards Webcast":
http://www.keysight.com/main/eventDetail.jspx?ckey=2689915&nid=-11143.0.00&id=2689915The Q&A session afterwards also was very instructive
Frank
unfortunately this is another topic I missed until now.
afaik VPG is using a Keithley 2002 to meassure their tolerances and if you do not ask for data when ordering the resistors you do not get any data from them. I doubt that they are able/willing to give you the data afterwards.
If you do not have a better reference to compare it to, you unfortunately cannot know if your resistor really is in tolerance or not.
I ask Vishay for some data, but they didn't respond until now.
In the measurement report, which I ordered in advance for my 10ppm resistors, they did not reveal the method, but obviously used a 1kOhm reference resistor, specified -2ppm. The designator of this standard is '1500 G 103', maybe something from ESI.
I assume, they used sort of a bridge, but that would be a 11:1 relationship, maybe explaining the extra error.
They measured each resistor three times, adding 2ppm for each value.
Everything inside three times of plus or minus 10ppm was accepted.
For my 5 resistors, they produced a total of 27, and 20 were 'good'.
Maybe they are using a DMM for bigger tolerance resistors.
Frank
PS: I ordered 9k9998 resistors. Vishay measured them to have values of +7 , +8, and three times +6ppm, greater than the default value.
The relative deviations were correct, but it turned out after several comparisons with calibrated resistors (quarks, acbern), that these all deviated in the other direction, i.e. instead they must have had about -14/-13/-15ppm of nominal. Something went completely wrong there.
First, i would not trust vishay here, simply because its a manufactured part, and not a calibration. Any cal report is probably more expensive than this resistor.
Secondly, somewhere i think i remember having read they use a DCC current comparator bridge. One would expect that, by the way. That however would contradict the bad results. Maybe they just own one