I know before I ask this question that “it depends” will be part of any
answer. However, I will ask it anyway.

How stable does a lab temperature need to be maintained, in order to
warrant using an 8.5 digit multimeter such as a 3458A? The same question
for 6.5 and 7.5 digit meters.

Dave.

Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@kirkbymicrowave.co.uk
https://www.kirkbymicrowave.co.uk/
Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

In message CANX10hCp5Y3pF11qQJWRtERydPgAO8OpkNNg3uDpdunq+HZxLw@mail.gmail.com
, "Dr. David Kirkby" writes:

I wont attempt to answer your question, but I will make a note about
measuring the temperature.

Once you get into sub-Celcius temperature regulation, you need to pay
attention to all of convection, thermal radiation AND humidity.

Not all ways of measuring temperature give a proper account of the
radiation component (ie: any encapsulated sensor has delayed reaction
to radiation) and even minor changes in humidity changes the thermal
conductance and radiation propertis of the air.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

David I think the the benefits of a meter such as a 3458A can be utilized even in an environment where temps and humidity are not controlled. barometric pressure can also have a substantial effect but would be extremely difficult to control. The 3458A has ACAL which can (to some degree) compensate for temperature and other factors.
As for "it depends" the function and range can be very big variable. Also  if traceability is an is important the the meter must be kept within specified parameters. That said I have been amazed at how little a my 3458A monitoring a 732B is affected even over a temperature change of as much as +/-10F. It easily stayed within manufacturers specs. But again "that depends" for although I have not really experimented with measurements more dependent on internal resistors over a similar time and temp range I imagine that would show a much greater uncertainty. Most of this is detailed in product specs. So from my practical experience I would think you would see benefits from a 8.5 digit meter in almost any environment . But obviously everything makes a difference so reducing variables where practical is important. What I find interesting -  In this modern era of 8.5 digit meters my guess is more uncertainty is lost during shipping of the instrument then the rest of the year sitting on a bench. I see a future where precision instruments calibration is not only based on TIME but the environmental conditions the instrument was exposed to DURING that TIME utilizing an environmental package built in to the instrument continuously monitoring  1.hours power up, 2.temp, 3.humidity., 4.barometric pressure, and 5.vibration shock using an algorithm to providing the user with a real time estimates of uncertainty and notifying the user if the instrument has been exposed to conditions that exceed the instruments rated specs. I have actually experimented with this idea a bit although it has been on the back burner for the last few year because of other priorities. I  believe there needs to be a rethinking of this aspect of calibration. But whatever the reason this has not yet been researched by groups such a NIST - simply is a lack of awareness or because it does not involve any sexy physics- this approach has the potential to improve uncertainties in actual work environment which is the en tire point of calibration. To start studying this is relatively simple, it would basically require a environmental chamber with capabilities to generate all the needed conditions on the device under test DUT and collecting data on both temporary uncertainty that recover, conditions that change uncertainty requiring re-calibration to correct, , and conditions that would permanently  damage the instrument. Then studying real world conditions such as during shipping to build a model. I believe Fluke are the only group to have played with this studying the 732B's and building models on uncertainties during there rotation between metrology lab and changes in uncertainties of these standards when they arrive back at Fluke.  Fluke has achieved some impressive results, but that is a long way from the real time monitoring I am speaking of.
Enjoy;

Tom Knox

303-554-0307

actast@hotmail.com

"Peace is not the absence of violence, but the presence of Justice" Both MLK and Albert Einstein

From: volt-nuts volt-nuts-bounces@lists.febo.com on behalf of Dr. David Kirkby drkirkby@kirkbymicrowave.co.uk
Sent: Thursday, October 31, 2019 6:48 PM
To: Discussion of precise voltage measurement volt-nuts@lists.febo.com
Subject: [volt-nuts] Lab temperature stability required for X digit multimeters.

I know before I ask this question that “it depends” will be part of any
answer. However, I will ask it anyway.

How stable does a lab temperature need to be maintained, in order to
warrant using an 8.5 digit multimeter such as a 3458A? The same question
for 6.5 and 7.5 digit meters.

Dave.

Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@kirkbymicrowave.co.uk
https://www.kirkbymicrowave.co.uk/
Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

volt-nuts mailing list -- volt-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/volt-nuts_lists.febo.com
and follow the instructions there.

I think Tom Knox is right on the money.

I am quite interested in logging environmental history, particularly during transit. I participated in the EEVblog USA Cal club Round 2 and have proposed adding an MSP430 data logger to the kit for Round 3.

I'm  adding  support for a real time clock, temperature and humidity to the AR488 GPIB-USB package using an Arduino Mega 256. AR488 is an extension of the Prologix commands, so it is easy to implement logging in any scripting program. I also plan to add relay control of HP 44421A boards as well as SMA relays for RF work.  At the moment I'm focused on developing an Arduino Uno shield which can be profitably sold on ebay for $15 + $5 for each GPIB device in excess of 1 and handle 10-12 devices.  All in a nice plastic enclosure using ribbon cable and IDC Centronics connectors.  I've built one hand wired unit, but it did not use a 2x12 0.1" header.  So making it up was rather time consuming and error prone.

I spent several years working in the area of mechanics of materials and I think that it is possible to characterize the time and environment dependent behavior of voltage references accurately enough to significantly reduce the uncertainty.

I asked previously for time series of voltage references logged from initial start up, but never  got any data. If you have such data you're willing to share, please send it to me.  The graphs I've seen look to be of the form A +/- B*exp(f(t)). This is a classic relaxation process.  I'm fairly experienced at modeling that.

My numerical experiments have found that predictions based on historical measurements can be made to a few per cent error for a period equal to the length of the history. But I don't yet have real data to test the techniques with.

So, please, if you have data, send me it to me as ASCII text such as CSV format.

Have Fun!
Reg