Reference resistance

[franklin]

Hello everybody.

I finished a reference resistance box a while ago in the same way as Dr. Frank have described his own reference in this forum. I used the Vishay metal foil series resistor VHP101 (Y407810K0000V9L, Digi-Key Part Number 804-1059-ND) which is 10 kΩ, 0.005% and a TC of 2ppm.

My 3458A is from 2013 and last calibration was in 2015. The 3458A has been on 24/7 for the last 2,5 years, works well. I’ve been monitoring it as per Service Note 18 and A/D drift seems to be low. I cannot say anything about the 3458A’s internal 10 V reference as I have no external reference of quality.

I’ve measured the reference resistance at ambient temp. 23 ⁰C ± 2 ⁰C (same temp in the resistor box) and Rh 47%. I used the MATH STAT function and got the mean value (and deviation) of 16 readings each time and there is a total of 100 samples of that kind. The samples are in the range 10.000 295 to 10.000 310 kΩ; the mean value is 10.000 304 kΩ. To me this seems nice and repeatable.

However, my question is, can the readings be trusted when using a 3458A “out of cal” for resistance measurement and is the result good enough to make it a reference resistance? Or are the measurements just rubbish and can be discarded? Any views on this are appreciated.

Franklin

[dietert1]

A reference you derive from an uncalibrated instrument is uncalibrated, too.
Still it can serve the purpose of checking your instrument for unexpected change. Doing this check regularly (e.g. once a year) can protect you from using your uncalibrated instrument giving wrong results.
And you might get your reference resistor checked by somebody else. That may be easier than sending out the meter for calibration.

Regards, Dieter

[mendip_discovery]

Record your readings. Try it again in a few months see if you get drift. You might get some from the resistor due to aging and some from the meter. Use it for yourself or find some other voltnuts and ask them to measure it as well. Eventually you find someone with a meter with some traceability and all will be good.

[martinr33]

One thing you can do is use the resistor to check your 3458A for drift. Just take daily readings - without autocal - and see how much the resistance changes. If you have a drifter, you will see a trend - and ACAL DCV should fix it. 

The voltage reference, internal standard resistor, and VHP are all sub ppm stable over long periods (weeks) if they are good.

[Kleinstein]

The resistance ranges (especially the current sources) have a few parts that can have relatively high drift. How much the resistance ranges drift without ACAL is not really relevant for the 3458 quality. The important part it the drift after ACAL.

Even without (expired by quite some time) calibration the reading from the 3458 still has some value. The 3458 internal resistance reference is a similar grade resistor and for the 10 K range there are not many ACAL steps involved. The reading would not give you tracability, but it can still be used internally. For just internal use in most cases the drift relative to the meter would be what matters, as this is your best method of measurement. If the get the 3458 calibrated in the future, one may apply the calibration also backwards - it kind of validates the use for the last year or so.
Ideally one would like calibration before and after use. Normally the calibration before is considered good enough for practical purpose, though the calibration after the use would be the slightly more valuable, as defects are rarely reversible.

[ap]

You said you have your 3458A on 24/7. Unless this is for reaons other than obtaining stability, ok, but the drift will increase if the meter is on always compared to a meter that is on from time to time only.
So you may consider switching it on only when needed or maybe monthly for a day or so to get lower drift.

[mendip_discovery]

AP that is the opposite of what I have been told. Keep them on to keep the ppms inside.

Heat cycles tend to cause failures and drift.

[guenthert]

You said you have your 3458A on 24/7. Unless this is for reaons other than obtaining stability, ok, but the drift will increase if the meter is on always compared to a meter that is on from time to time only.
So you may consider switching it on only when needed or maybe monthly for a day or so to get lower drift.

    That in itself is remarkable.  Naïvely I would have expected that the largest share of drift is caused by the greatest stress, which I would expect to occur in the warming/cooling during powering on/off.

    Otoh, is the drift really of greatest concern?  Isn't it rather the variation over time of drift?

[mzzj]

Hello everybody.

I finished a reference resistance box a while ago in the same way as Dr. Frank have described his own reference in this forum. I used the Vishay metal foil series resistor VHP101 (Y407810K0000V9L, Digi-Key Part Number 804-1059-ND) which is 10 kΩ, 0.005% and a TC of 2ppm.

My 3458A is from 2013 and last calibration was in 2015. The 3458A has been on 24/7 for the last 2,5 years, works well. I’ve been monitoring it as per Service Note 18 and A/D drift seems to be low. I cannot say anything about the 3458A’s internal 10 V reference as I have no external reference of quality.

I’ve measured the reference resistance at ambient temp. 23 ⁰C ± 2 ⁰C (same temp in the resistor box) and Rh 47%. I used the MATH STAT function and got the mean value (and deviation) of 16 readings each time and there is a total of 100 samples of that kind. The samples are in the range 10.000 295 to 10.000 310 kΩ; the mean value is 10.000 304 kΩ. To me this seems nice and repeatable.

However, my question is, can the readings be trusted when using a 3458A “out of cal” for resistance measurement and is the result good enough to make it a reference resistance? Or are the measurements just rubbish and can be discarded? Any views on this are appreciated.

Franklin

Assuming linear drift over years your 3458A could have drifted 7x 1-year specification or about 70ppm.
More reasonable assumption could be RSS (root sum square) of 7x 1-year drift or same as 10ppm*sqrt(7) =26ppm
3458 datasheet points also to this direction as 1 year specification is 10 ppm and 2 year specification is 15 ppm (pretty close to 10*sqrt(2)

TL:DR so for non-critical home lab use I'd consider your measurement result good to +-0.00026 kΩ