Which 6.5 digit DMM has the better 0.1 to 10 Dc Volts or 100 to 100K Ohms specs?

[NANOJOLTS]

I narrowed the comparison to 0.1 to 10 Dc Volts or 100 to 100K Ohms just to keep this post from becoming too unwieldy.
 But if some other parameter(s) or other DMM'(s) can shed additional insight on these comparisons feel free to include.

 I'm sure there are other parameter(s) or 6.5 DMM's that warrant consideration?

1st  chart Obsolete NI PXI-4070
2nd 3rd  chart Keithley DMM 6500
 4th  chart keysight 34465A 

There seems to be a rather stark contrast in Specifications? Let's see what I've missed.

[Kleinstein]

Comparing just specifications for meters from different manufactureres can be tricky. The accuracy specs are nothing that can be measured directly, especially not when the meter type is new. It may need several years to get actual statistics on how good they hold the calibration.
The specs are more like best guesses on the expected performance with quite some safety factor to avoid a higher rate of failures / returns. How much safety factor can variy  (e.g. the Chinese tend to be rather optimistic).

The PXI4070 is only a card and needs an extra case / power supply and this can effect the performance (e.g. variations in temperature, fan or not, EMI). So the specs for the card are even more tricky.

[mendip_discovery]

There is more to measurements than specifications.

Noise/Flicker.
Repeatability
Reproducibility
EMF

These are just a few things to take into account.

Anyway, I just did a graph for you so you can see how the specifications compare.

[Mickle T.]

For the materials that I prepared for the second part of the article on the history of the creation of digital voltmeters, I developed a diagram of the “basic” error, which, among others, took into account 106 devices with a resolution of 6-1/2.
I don’t pretend to be accurate, but the first places in this category were taken by the Chinese Tianjin Zhonghuan Scientific HG1963 Autocal DMM, the Russian VK3-78A and the German Prema 5001 and 6001

[NANOJOLTS]

mendip, there several errors Fluke 1yr 10V spec OK , Keithley 6500 1yr 1V spec should be 2yr and 10V, keysight 34465a 1yr should be 2yr spec, 34401A 1yr spec OK,
4070 24HR 10V spec should be 0.0004% and 0.0002%,  and 2yr should be 10V 0.0025% + 0.0006%. I'm trying to compare apples to apples.

[NANOJOLTS]

Thanks Kleinstein,  One would expect (at a minimum) a DMM it to meet it's published  specifications, right out of the box, (assuming similar ambient conditions).
I have some experience qualifying electronic modules.
If the DMM does not meet the published specifications, the DMM is defective.

[NANOJOLTS]

Thanks Mickle T. nice chart, I also own some Prema DMM's 6001 6.5 and Prema 5017 7.5.

[NANOJOLTS]

 Mickle T. the Prema 1yr spec for the 5001 and 6001 should be 24ppm on the 10V and 1V ranges.

[Kleinstein]

The prema 5001/6001 have 2 V and 20 V ranges and the 20 V range is already with an input divider. To me the specs for the 20 V range look a bit suspect as the divider would normally add some uncertainy. Even without the 10 M input impedance makes it tricky with some signal source. The 20 V range definitely comes with downsides.

I have use such a meter for quite some time and I would consider it more a lower end 6 digit meter with less conservative specs, maybe a good initial adjustment. The error limits are a sum of contributions from noise, the initial adjustment / calibration, linearity error limit and allowance for drift.

The meters should meet the specs, but rather optimistic specs can be an issue if the meters over longer time tend to drift out of the 1 or 2 year specs and need readjustment when they get old. Especially for the old Premas this may be an issue, especially outside Germany.
With more conservative specs chances are high the calibration is just a verification and no adjustment needed over many years.

[NANOJOLTS]

Thanks Kleinstein,  yes error's on my Perma specs. Good DMM's > 30yrs old?, 0.2V and 2V > 1GOhm and 10M on 20V.

I notice the Premas refer to 24 hours accuracy as stability?

[mendip_discovery]

mendip, there several errors Fluke 1yr 10V spec OK , Keithley 6500 1yr 1V spec should be 2yr and 10V, keysight 34465a 1yr should be 2yr spec, 34401A 1yr spec OK,
4070 24HR 10V spec should be 0.0004% and 0.0002%,  and 2yr should be 10V 0.0025% + 0.0006%. I'm trying to compare apples to apples.

It is not much help that NI doesn't quote a 1yr spec like the other do.

[NANOJOLTS]

Hey mendip_discovery, I agree, I would like to see the pxi-4070 1yr spec. as well. But still some 2yr caparisons available.

[bdunham7]

Hey mendip_discovery, I agree, I would like to see the pxi-4070 1yr spec. as well. But still some 2yr caparisons available.

Unless you have stated confidence intervals, your comparisons are pretty much meaningless.  And even then, you are comparing specs between different manufacturers that all have their own methods and opinions as to how to state the specifications.  And even once you've accounted for all that, you still won't know much about the actual performance of the instrument--is it just barely within the specs or is it typically 10X or more better?

[mendip_discovery]

Hey mendip_discovery, I agree, I would like to see the pxi-4070 1yr spec. as well. But still some 2yr caparisons available.

Unless you have stated confidence intervals, your comparisons are pretty much meaningless.  And even then, you are comparing specs between different manufacturers that all have their own methods and opinions as to how to state the specifications.  And even once you've accounted for all that, you still won't know much about the actual performance of the instrument--is it just barely within the specs or is it typically 10X or more better?

Top Trumps for test gear. I can hear it now, metrology nut A "Number of digits 6 1/2", metrology nut B "same, 28 ppm spec of range", nut A "ha I win 18 ppm of range."

All depends on how brave the manufacturer wants to be. I think given the time that some of the older meters have been out it would be nice to re-develop the numbers and see what is the common reasonable specification for them.

Anyway back to coding snakes as I play with my 34401A and 10V ref.

[NANOJOLTS]

Started out, trying to see where my DMM's fits on the Plot?  There must be some validity to this Plot and similar DMM Plots.
 On my PXI DMM's, Vdc noise seems to be a function averaging?  Without averaging Just a bunch noise spikes.
More averaging less noise.  I see the 34401a on the plot.