Cost effective DC current drift checking ideas?

[binary01]

Hi all - long time browser of EEVblog and have learnt plenty - thank you all.

I work in metrology, but in the area of low uncertainty pressure measurement/calibration, unlike the many electrical experts found here. I have a basic capability in DC current and voltage source/measure in ranges typically used for industrial instrumentation (4-20 mA, sometimes up to 50mA and 0-10 V, sometimes up to 100 V).  For some instruments, such as high accuracy pressure transmitters, my capability in electrical measurement can become the dominant uncertainty component so I am always looking for ways to minimise my uncertainty in electrical measurement as much as possible (such is the aim for everyone in metrology).
With a reasonably long calibration history on my references I am able to demonstrate good stability and suitable uncertainty in DC voltage measurement via a Keithley 2000 (drift < 10 ppm).  However most pressure transmitters output current rather than voltage (to enable long distance signal transmission) but I am finding it much more difficult to demonstrate and check for good stability in my DC current capability.  Primarily I use a Fluke 7526A for current source/measure.

Can anyone offer suggestions for a cost effective DC current drift checking method to enable me to keep a handle on my current measurement ranges?  I have been tossing the idea of purchasing a reasonable quality current shunt to anchor my current measurement on my voltage?  Or could I perhaps get away with a historic standard resistor (or two) found on ebay, perhaps 10 Ohm? If I could perform regular checks on my current range to approx 20 ppm it would be ideal.  Any ideas appreciated.

BTW, are there any members from the Metrology Society of Australasia in here?

[ap]

The easiest and most cost effective for low uncertainty current measurement, if you have an accurate volt meter (seems your K2000 suits you) is to use a shunt, i.e. a reistance standard with low uncertainty and low drift (temperature and aging). The 'bottle' type shunts available have a pretty high TC, and are inteneded for use in oil baths, so would not use those, unless you can operate them that way. Also, you should strive for achieveing a voltage drup accross the resistor of 1V, to limit the impact oif EMF voltages and environmental noise and so. A 10Ohms at 10mA will only give you 100mV, not desirable (although possible). 100Ohms would be the recommendable velue here. So you may need a few resistors depending on your load current. Also keep in mind to minimize the self heating effect of the shunts.

[binary01]

Thanks ap.  I suspected that would be the case.  I was thinking 10 Ohm initially so that I could do full scale checks on the Fluke mA ranges (100mA output and 50 mA input), but thinking again it is probably best to check in the important measurement portion so I might need 100 Ohm also.

[czgut]

You could consider Vishay Power Hermetic resistor with heatsink. Hermetic resistors drift usually less than other.
I would try to fit to 1V on Keithley 2000. If more important is 4..20mA range, I would use 50 Ohm.
If 50mA is more important, then 50 Ohm version. Or two resistors..
Stability of Vishay hermetic not heavily loaded should be below 10 ppm/year. Your Keithley @ 1V has specification of <30ppm/year, but probably is better (drift slowers with time). 

[borghese]

Perhaps you would like to test the zero of the scale, so 250-ohm foil resistor does the job (1volt @ 4 mA)

[ArthurDent]

A shunt is the way to go. Although an application of mine is different than yours, it shows the principle works. I have a precision D.C. power supply that has calibration points between 18 Ma and 3 A and I use an 80-year old L&N 4360 0.1 ohm/15 A .04% shunt like the stock photo below for the entire range. I use a 6.5 digit HP 3457A meter to measure the drop across the shunt. The low resistance of the shunt eliminates any voltage drop concerns I'd have if I used a much larger value resistor and the HP 3457A has the accuracy and resolution to measure the 18 Ma to the accuracy I need in this application.

Depending on the accuracy you're looking for you could still consider an old 4 connection standard resistor as the shunt. They are very  age stable and if your room temp is fairly stable you won't have to worry that much about TC of the resistor. A Vishay shunt resistor would work as well with a lower TC and you could build it into a box with terminals to suit your needs. The Vishay VCS232Z has .02% accuracy and 2 watt power handling.     

[binary01]

Thank you all for your suggestions. Plenty of food for thought. The Vishay resistors certainly sound like impressive units for their cost and size. The lab is controlled to 20 degC +/- 1, but humidity is not controlled and varies 40-60% which I have read to be a concern for some shunts.
I have a very old/stable JL Williams standard resistor but it is 1000 Ohm so not very useful for this current range.  I had a brief thought of simply locating some equivalent units with lower R values and eventually performing my own calibration of my mA ranges, but I can now see that it's not so straightforward due to the impact of loading.  That is unless I have the standard resistor / shunt calibrated at every expected load - an expensive excercise each year.
Nonetheless, I think I'll go ahead with a standard resistor or two to use as check devices, even if I can't use them directly in the calibration chain.
My Keithley 2000 has been performing very well for voltage drift (less than 5ppm on almost all DC V ranges) but as expected it is quite poor for mA drift so it's not so useful as a  current check device. I'm hoping to replace it with a 3458A or 8508A eventually, which could lend a hand with mA checking too, and offer some redundancy.
Out of interest, do you have any comments on the true drift performance of the 3458A vs 8508A for DC mA measurements? I understand the specification on the 8508A suggests it is superior, but in pressure instruments I have learnt to not always trust manufacturer specifications...

[ap]

While I do not have personal experience with the 8508A, there are a few general things to consider: The 3458A is self calibratiing its DCI, (ACAL as well as during external CAL), and derives cal constants from its known values of intereal voltage and resistance references during ACAL, while the 8508 needs external CAL for all values (but also does some internal recalibrations). In the essence, if the standards in a 3458A are stable (and they are on old aged units, and you can also get, at a high price, a 2ppm/a internal voltage reference), ACAL does the rest on a daily basis. You can get the 3458A voltage drift further down by reducing its reference set temperature (voltage reference drift contributes to DCI spec). No problem in a cointrolled environment. Some more details are written in the orginal HP brochure published when the meter was launched. You could even do your own artifact calibration (i.e. adjustment) if you have a 10V and 10k standard available (verification is a different story; according to a US government document, they permit/recommend verification only every second time; may not really be an option to everybody, depending on its use).
And, more improrantly probably, the 3458A (used) comes at a fraction of the cost of a 8508A (compeared to a new one, essentially only few used ones used available, it any). New, price factor is about 2.

[e61_phil]

I think the need for external CAL for every single range is more an advantage of the 8508A than a drawback. One of our 3458As was out on the 100mA some time ago. It was spot on on 10V and 10k. That shows, you cannot rely on ACAL only. ACAL cannot replace a proper calibration. To bring this 3458A back to specifications you have to run an artficat calibration. That will destroy your calibration history of all ranges, even if only a single range is out of spec (100mA in our case).

The 8508A will outperform the 3458A in almost every function. If you want fast DCV measurements with low noise than choose a 3458A. In most other cases (especially in measuring high voltage or resistance) the 8508A will be better, if not much better. But as ap already said it costs as twice as the 3458A.

I attached a small example. The DUT was a temperature stabilized Caddock USF370 20Meg resistor. It is a bit unfair, but it is something I have to measure very often.

[Echo88]

But why is the 8508A such a rare beast on the used market? Arent there as many units compared to the 3458A because of metrology-lab-acceptance/price? Maybe Plesa or CalMachine can answer. Also couldnt find teardown-pictures of it anywhere.

[binary01]

Some interesting info there - thanks.

For a while I have intended on picking up a good used 3458A eventually, as the need arose and when seeing a trustworthy unit popping at a good price.  It appears to be somewhat of a "gold standard" and when I did an electrical measurement course at my NMI they spoke very highly of the 3458A, and had them everywhere.  However, as I am more often pushed for accuracy in mA rather than voltage, the 3458A's primary capability as an excellent and fast voltmeter is maybe not exactly what I need.  If ACAL was very reliable it would be great to continually trim up the mA range.

I have revisited it recently and the mA ranges on the 3458A of 10mA and 100mA are not ideal for my interest up to 20 mA, as the measurements above 10 mA immediately leap up to the 100 mA range - for example measurements at 16 mA would be spec'd at around 70 ppm, using the 100 mA range.  On the other hand, the 8508 has ranges of 20 mA and 200 mA, so would offer a better specification up to 20 mA, but worse above.  I am not one to rely on manufacturer's specification and instead develop a drift history and combine with my calibration uncertainty to get an expected performance over 12 months.  So I wondered if someone that sees these regularly in a calibration environment might have a sense of the true drift of a 3458A and 8508A in mA measurement.

As we are an accredited lab, we will be obliged to have external annual calibration in any case (and 6 monthly checks - as per the reason for my original post), unless an extensive drift history can support an extended interval.

[e61_phil]

I don't want to say the 3458A is bad. It is a great meter.

You have to be careful with the "2" ranges of the 8508A. The 3458A allows 120% overrange. The 8508A doesn't. You can't even reach 2... on some ranges of our 8508A before overrange. That means, If you're lucky you can measure your 20mA on the 20mA range. If not you have to measure it in the 200mA range. If you can reduce it, than everything is fine, but sometimes that isn't possible (resistors for example).

[quarks]

Fluke 8508A full scale for 20 mA range is 19.999000 mA

[chickenHeadKnob]

But why is the 8508A such a rare beast on the used market? Arent there as many units compared to the 3458A because of metrology-lab-acceptance/price? Maybe Plesa or CalMachine can answer. Also couldnt find teardown-pictures of it anywhere.

I am not a cal machine or even a love machine, but I think your question has a simple answer. Hp3458  is purchased by the truckload  (slight exaggeration ) by non-metrology type users like semi-conductor manufacturers who put them in racks in their production/test plant because they want a DC voltmeter with high dynamic range and fairly fast sample rates. It is these units that end up on the used market.

[binary01]

Fluke 8508A full scale for 20 mA range is 19.999000 mA

Yes, this annoying when I will often be measuring just under and just over 20 mA.

[ap]

A cost effective option to the 8508A covering 20mA as well would be the Datron 1281. Older, worse specs than the 8508A, lower cost.

[e61_phil]

Fluke 8508A full scale for 20 mA range is 19.999000 mA

Thanks for this info. I thought the range limit depends on the calibration constant of the individual meter. I verified this a couple of minutes ago. The limit of our 8508A is exactly 19.999mA


Another advantage of the 8508A in current mode is the lower burden voltage compared to the 3458A.

[binary01]

A cost effective option to the 8508A covering 20mA as well would be the Datron 1281. Older, worse specs than the 8508A, lower cost.

Thanks ap.  I can see the influence of the 1281 on the 8508A visually, however the 1281 looks to have 10 mA / 100 mA range split like the 3458A, rather than the 20/200 mA on the 8508A?  The 1281 also looks quite well spec'd but a little worse than the 3458A's 100 mA range, so maybe I should just continue planning for a 3458A and except that I will need to jump on to the 100 mA range above ~12 mA.

[quarks]

I can see the influence of the 1281 on the 8508A visually, however the 1281 looks to have 10 mA / 100 mA range split like the 3458A, rather than the 20/200 mA on the 8508A? 

1281 has the same range and full scale as the 8508A

only the datasheets show different terms in uncertainty with the same meaning
1281   ± (ppmR + ppmFS)
8508A ± (ppm Reading + ppm Range)