EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: SArepairman on August 10, 2013, 06:19:15 pm
-
So, How are 8.5 digit meters calibrated?
Is there some kind of 50,000$ voltage calibrator that can be set for 9.5 digits?
Or is a regular standard and a external divider used?, I.E. a 6.5 digit EDC unit with a KV divider.
-
Yes, using a Josephson Junction, JJ. There are ~ 3 ways to calibrate the top-line meters and a lot depends on the precise accuracy of the actual meter; for e.g. 3458a has 2 levels of accuracy in ppm depending on what standard option you choose.
If you have the more lax standard, you could opt for a method that uses statistics on a secondary transfer reference. This is the cheapest and most error prone, or more uncertain. The reference is typically not a JJ. ~$600.
www.fluke.ae/comx/applications/3458.pdf (http://www.fluke.ae/comx/applications/3458.pdf)
For the higher accuracy mode, you must send it to a primary lab; they typically maintain a JJ as a reference. Given costs and uncertainty, the best bet is to send it back to Agilent for the 'gold calibration.' $~1600.
For the higher accuracy mode but with the best data and thus lowest possible uncertainty, you send it to NIST. Even if all primary labs have JJs, NIST is the national standard; what differs are the quality of their uncertainty data generated by years of data NIST accumulates on its standards. Agilent would have similar, but NIST is the lab of labs and has had a JJ longer than most labs in the USA. ~$4,000.
http://physics.nist.gov/cuu/Uncertainty/coverage.html (http://physics.nist.gov/cuu/Uncertainty/coverage.html)
-
Many thank to you saturation, I will enjoy reading. :clap:
-
is there a good website showing a JJ off?
like a tear down or at least some pictures
-
I don't think NIST would calibrate something like an Agilent 3458a, though I've never tried or asked. They're not in the business of competing with commercial cal labs. I think the Agilent standards lab in Loveland (which does posses a JJ standard) is the best you can do. Costs more than $600, however, more like twice that.
-
Hi alm, NIST will calibrate anything on their schedules, but as you see, their prices are very non-competitive with market prices in the USA, so indirectly they don't compete with commercial labs.
http://www.nist.gov/calibrations/voltage.cfm#53200S (http://www.nist.gov/calibrations/voltage.cfm#53200S)
53202S Special 25-Point Test of Digital Multimeters (DMMs), by Prearrangement $3937
All the fees are public and listed on their cal web site. Also their relative expanded uncertainty data is public on that URL for their DMM calibration.
Compare with Agilent's schedule:
https://service.home.agilent.com/infoline/public/product-service.aspx?pn=3458A&lc=eng&cc=US (https://service.home.agilent.com/infoline/public/product-service.aspx?pn=3458A&lc=eng&cc=US)
The JJ on a chip, for sale:
http://www.hypres.com/products/voltage-standard/ (http://www.hypres.com/products/voltage-standard/)
An early attempt at a turnkey JJ system, for sale, c 2002:
http://www.prema.com/D/JVSe.htm (http://www.prema.com/D/JVSe.htm)
-
The uncertainty for most of those points is worse or similar to the 24h specs of the 3458a for DCV, DCI or R, and barely better than the 90 days spec, especially for option 002 meters. I would much rather pay for the much cheaper Agilent Standards Lab cal.
I think NIST should have a bunch of documents on JJ standards, as should Fluke, but I don't have time to dig these up right now.
-
Hello,
8 1/2 digits DMMs / calibrators are the longest available scales. Higher resolution can be achieved by other methods, as measurement of differences . More important, the actually required uncertainties for the calibration of electrical units are just below 1ppm, so 7 1/2 digits of resolution is sufficient.
NIST will not accept any 8 1/2 digit DMMs for calibration as standards, only secondary standards as (heated) Weston Standard Cells and Solid State Zener references, which are proven and accepted to be stable enough.
They explicitly disregard DMMs as being standards for DCV or Ohm: http://www.nist.gov/calibrations/voltage.cfm. (http://www.nist.gov/calibrations/voltage.cfm.)
"Voltmeter calibrators, multirange instruments with up to eight decimal digits of adjustability, are not considered by NIST to be standards and are not to be submitted routinely for calibration under this test category."
And that can be easily explained: Although all of those DMM/calibrators have high resolution and high short term + transfer stability, they all have relatively mediocre 1 year stability (uncertainty), compared to dedicated Zener references, as Fluke 732B, Datron 4910, or even compared to the old Weston cells.
Single Zener references have basic stabilities of 1.5-3ppm/yr., and a periodically monitored group of references can have mid term (3mo.) uncertainties of 0.3ppm, and with additional statistical methods (prediction), still well below 1ppm/yr.
In comparison to that, the 3458A has the worst DCV reference, 8ppm/yr, (poor, mistreated LTZ1000A inside :-[), and the best is the Fluke 8508A, 2.7ppm/yr, always relative to the calibration standards used. That's 10-25 times worse than the Zener standards.
In reverse, this comparison reveals, that all 8.5 digit DMM can simply be calibrated by such a Zener reference group.
Fluke also owns commercially available JJ primary standards in its main cal facilities.
So there's no need to calibrate @ NIST.
At NIST or the other primary labs, the Ohm is based on the von-Klitzing-Hall standard
This standard requires the Hall effect probe, and also a cryostat with liquid Helium and a superconducting magnet.
Whether companies like Fluke, Vishay or iet labs own such an apparatus, I don't know.
So very stable secondary standard, as the SR104, are normally used for calibration of DMMs Ohm, as they also are at least 10 times more stable and uncertain (~0.2ppm) than the Ohm references inside 8.5 digits DMMs.
How is the calibration done in practice?
The first category of instruments need different external so called Cardinal Points for each range and for each function.
So for DCV, exact voltages of 100mV, 1V, 10V, 100V, 1000V are required, for Ohm, 10, 100, 1k, 10k, and so forth, and for DCI, ACV and ACI equivalently.
For that task, you need a calibrator (e.g. a Fluke 5720A) as a source for all those calibration points.
That leads directly to the question, how this calibrator is calibrated, as the problem is obviously shifted one instance further.
As an answer, there exists a second category of instruments, where the 5720A, and also the HP3458A belong to:
Both are AutoCal instruments, that means they are calibrated from two or three external references only, i.e. 10V and 10k, and additionally 1Ohm for the 5720A.
Both instruments have internal circuitry for the complete range transfers (based on highly linear ADC or DAC) and also for the function transfers (DCV, Ohm => DCI and DCV => ACV) , so they are able to completely calibrate themselves to very good uncertainty.
For even lower uncertainties than those both instruments, transfer standards from Fluke as the Reference Divider 752A (10:1, 100:1 DCV transfers), the 720A (0.1ppm linear DCV transfers), and the AC/DC transfer standard 792A are used.
For ohm, there also exist decade transfer standards (e.g. SR1010).
Frank
-
why do you say the LTZ1000 is mistreated ???
I figured agilent knew what it was doing
-
why do you say the LTZ1000 is mistreated ???
I figured agilent knew what it was doing
They really do:
They designed the HP3458A for an environmental temperature up to 50°C, presumably for military usage instead of metrology purposes.
The internal heating of the instrument (+13..25K), plus the self-heating of the LTZ1000A (+10K), and the necessary overhead for temp. regulation (+10K) all sum up to a temperature set point for the LTZ of (50+25+10+10) = 95°C, the same as the LM399 reference.
This high heater temp. leads to a much higher drift over time, than the typical 1ppm/yr @ 65°C, as specified in the datasheet. Each 10K higher temperature will double the ageing rate, and so the mediocre 8ppm/yr. can be explained. That's what I call mistreating the LTZ1000.
Most other instruments have stabilization temperatures of 45..65°C and are therefore much more stable.
Please, read also the thread about the ultra stable LTZ 1000 reference, in the category Projects, Designs, ..
Frank
-
haha, can you cut a hole in the side and put a fan on it for better regulation?
-
No.
-
besides the Fluke 5720A a Wavetek 4808 is the right tool to use
and a Wavetek 4950 MTS to calibrate the calibrator
-
haha, can you cut a hole in the side and put a fan on it for better regulation?
Well, the 3458A already has a fan.
As Dave mentioned on EEVBlog #426, 3457A teardown, a fan would be no good for ultra precise instruments, and I agree, as the airflow might cause thermal gradients and thermo voltages on the internal circuitry.
A better thermal concept would have made the instrument more stable, especially the Ohm function.
Latter one is 7 1/2 digits only, and sadly for such an instrument, no transfer accuracy is defined.
But it's possible to make it more stable:
Use the 3458A in a stable, metrology grade environment, i.e. at 20..25°C only.
Environmental temperature should not vary more than 2K over the year, and not more than 0.2K during the measurement.
Don't put the 3458A in a rack, use it always as a desktop device, and always keep the fan filter clean, so that the interior is max. 13K above environmental temperature (i.e. 36°C max.).
Now it's possible to reduce the LTZ1000A heater temperature to 65°C, which will improve the 1 year stability by a theoretical factor of eight.
Calibration temperature should equal measurement temperature.
Under those conditions, it's also possible to get Ohm transfer uncertainties of about 0.2ppm.
I 'm able to demonstrate that on the comparison measurements of my DIY 10k standards, designed with Vishay Z201 resistors.
The absolute uncertainty of the Ohm function stays mediocre, but that's not very different to the other instruments.
Anyhow, the 3458A for me is the best instrument, due to it ultra linear ADC (typ. 0.02ppm of input), its ultra precise transfer accuracies, and its unique AutoCal feature.
Frank
-
After I spent a lot of money on haveing my 8,5 digit dmms calibrated, I decided I would rather spend this money on test gear that would enable me to do cal myself but traceable to national standards.
so what i bought was:
-732a reference
-a vishay 10k and other decade value hermetic foil resistors
-a 752a reference divider
-a 7 digit ratio transformer
-set of thermal converters
got all of these relatively cheap, total spend about 1k euro.
only the reference (and only its 10v output) and the 10k and one converter is calibrated. it has a stability of better 0,5ppm/a, so re-cal every couple years is sufficient, and with the 732 being cal'ed from time to time against national standards, you gain a history allowing you to cal less frequently.
in a nutshel, procedure is a follows (3458a):
-cal 3458a with 732a and 10k
-after cal, verification is required (you cannot assume the selfcal went ok, at least every second time you should do a verification, according to goverment guidance)
to do this:
-establish ohms values of all decade resistors from 10k and voltage divider analysis. this is possbile because of 3458a's linearity. establish uncertainties for these. this gives you a set of cal'ed resistors (these need to be stable, so hermetic resistors should be used as stated, the esi 1010 e.g. have a pretty high anual drift as per spec. this way you do not need to do the recal of all resistors very frequently)
-these can be used to do ohms and current (with a precise voltage source) verification
-use the cal'ed thermal standard to establish uncertainties for all other thermal converters. these can then be used to verify ac volts in high frequency ranges, in low frequency up to 10khz the ratio divider should be used.
additional advantage is, you can get a better cal than from the manufacturer. reason being, the standards used have uncertainties which are relatively high, e.g. 2ppm for voltage. you can have 10v calibrated to about 1ppm at relatively low cost.
-
@acbern
I'd love to get a PM about contact info on your instrument dealer. ;)
1K€ for that is a super price i think.
/Bingo
-
@bingo
you are right, thanks, typo, actually 2k, still pretty ok i think. if one can be patient things earlier or later show up on ebay at reasonable prices. many are overpriced, and never sell (at least not as long as I watch them), so you need to know what you want to spend.