Multimeter Resistance Accuracy Calculator

[Nogtail]

I've put together a tool to check the accuracy of multimeter resistance measurements: https://checkohm.code7.au/ I'd love to get some feedback/suggestions, especially if someone could double-check my maths!

The source is here if anyone is interested: https://github.com/ZoidTechnology/CheckOhm

[LordXaos]

Didn't check the code, but the outcome looks plausible. Makes me wonder why benchtop DMMs with graphical displays don't show some kind of confidence interval on screen (at least as an option)...

[Sensorcat]

Accidentally entered 'Reference Resistance' and 'Measured Resistance' of 100 Ω for a 'Multimeter Range' of 20 Ω. Input accepted! Either don't understand how you mean it (=> more text would be useful) or input checking should be added. Otherwise, good and useful idea.

[srb1954]

Didn't check the code, but the outcome looks plausible. Makes me wonder why benchtop DMMs with graphical displays don't show some kind of confidence interval on screen (at least as an option)...

I agree. With some high-end multimeters including temperature sensors and real time clocks they could easily include the aging effects and temperature coefficient into that confidence interval.

The manufacturers of these multimeters could possibly get more business for their calibration labs when users see how much worse the confidence level gets over time.

[Nogtail]

Accidentally entered 'Reference Resistance' and 'Measured Resistance' of 100 Ω for a 'Multimeter Range' of 20 Ω. Input accepted! Either don't understand how you mean it (=> more text would be useful) or input checking should be added. Otherwise, good and useful idea.

Thanks for the feedback! That's actually intended. On a lot of multimeters, the full-scale reading is slightly higher than the range. I think on my Keithley 2000 it's around 120%. I could show a warning if the resistance is significantly higher than the range, but I'm not sure where I'd set the threshold.

[Kleinstein]

AFAIK the expensive Fluke 8858 give some info on the confidence interval, but that is an exception.
It is mainly a software only thing, so we may see it more frequent in the future, though many use cases don't really care.
It would make sense at the higher end, when the quality of the calibration standards could make a difference - so no fixed specs but confidence range  dependends of the CAL grade.

Still the specs and confidence intervals are only crude estimates made in the initial phase. So don't take the specs to litteral and as absolute accurate. Some instruments later turn out to barely hold the specs, while other usually do much better or for a longer time. Especially the very high resistance range that is implemented with a prallel 10 M resistor may not really fit the simple format of percent of reading plus percent of range.

The over-range is kind of limited and it can come with an increased error. So a warning at more than the nominal range makes sense, though this is also easy to see for the user.
Ideally one would have a database with specs - so one would enter the meter and not that specs. That step would be the real benefit of a computer program.

[guenthert]

Thanks for the feedback! That's actually intended. On a lot of multimeters, the full-scale reading is slightly higher than the range. I think on my Keithley 2000 it's around 120%. I could show a warning if the resistance is significantly higher than the range, but I'm not sure where I'd set the threshold.

Or a lot higher.  On long-scale Fluke meters (and those of Datron/Wavetek they descend from) overrange is 200%.

[andrewtaylor]

, but I'm not sure where I'd set the threshold.

My proposal:
add an extra box in the GUI,
with a default of 5% -- user can overwrite it for special as  gunthert  mentioned.

Shoul be easy to add in your programming.

[mendip_discovery]

Thanks for the feedback! That's actually intended. On a lot of multimeters, the full-scale reading is slightly higher than the range. I think on my Keithley 2000 it's around 120%. I could show a warning if the resistance is significantly higher than the range, but I'm not sure where I'd set the threshold.

Or a lot higher.  On long-scale Fluke meters (and those of Datron/Wavetek they descend from) overrange is 200%.

I have kept away for a bit but can't stay quiet for long...

If it is a 100 Ω range that can go to 200 Ω then you put the range down as 200 Ω. No need to cope with your edge case meters.

May I suggest that "Measurement Accuracy" has a digits option.  0.005% + 5 digits

[Nogtail]

Ideally one would have a database with specs - so one would enter the meter and not that specs. That step would be the real benefit of a computer program.

I love the idea of a specs database. I'll try to put something together when I get a chance.

If it is a 100 Ω range that can go to 200 Ω then you put the range down as 200 Ω. No need to cope with your edge case meters.

I think I'd have to include a separate input for the case where accuracy is specified as a percentage of the range. It'd be nice if datasheets could provide overrange or full-scale reading in a consistent way. For some of Fluke's meters, the full-scale reading is 10000 counts lower than what the range is specified as. The worst I've found is an East Tester meter where it's listed as "overrang".

May I suggest that "Measurement Accuracy" has a digits option.  0.005% + 5 digits

Measurement accuracy has a counts option. Is this the same thing? I could rename it digits, but I didn't want to cause confusion with the "digits" unit in multimeter resolution.

[mendip_discovery]

May I suggest that "Measurement Accuracy" has a digits option.  0.005% + 5 digits

Measurement accuracy has a counts option. Is this the same thing? I could rename it digits, but I didn't want to cause confusion with the "digits" unit in multimeter resolution.

Least Significant Digit aka LSD. I rarely refer to meters by counts other than to say it goes up in 3s or 6s but that is just my experience.