Non-precision multimeter function tester - which properties are required?

[Bzzz]

Hi guys,
I was wondering which physical measurements are required to fully check a multimeter. Not for testing if they're in spec, but to see if every function is generally working or not.
We had an incident at work this week where we tried figuring out the insulation test procedure of an older Gossen electrical safety tester unit, and it turns out the multimeter used is dead on AC volts. We do not have any reference gear at work and I only own a AD584 quad DC "reference" (not that bad, actually) personally, so currently the only ACV testing option is sticking the probes into a wall socket (that also yields a 50Hz signal...yay)

From my understanding, having a working DC volts readout is essential for basically all measurements. However, there's little "X is working -> Y will be working, too" derivation beyond that.

DCV: Well, needs DCV, d'oh. Are, let's say, µV and kV ranges fundamentally different?
R: DCV + current source
Continuity: DCV + current source + buzzer (or other alerting circuitry)
Diode test: DCV + current source
T (thermocouples): DCV + LUT
T (resistive): DCV + current source + LUT
f: some sort of timing circuit
C: DCV + current source + timing
L: DCV + current source + timing

Not exactly sure if ACV is handled the same between TRMS and averaging meters, same for ACA. Both ACA and DCA will require knowledge about the meter since there might be two or more shunts (and fuses) that all need to be tested. All advanced features like e.g. 4-20mA readouts or Low-Z voltage measurements will require the base function plus some woo-woo on top.

Judging from that list, I'd already say measuring each of these things from one or more physical representations would be easier than somehow deriving it by plugging into cunning circuitry. When researching this on the forum and the rest of the interwebs, I already came across units like the DMMcheck (plus), but these are often positioned to be references instead of simple testers, so they're limited to doing a subset of these measurements in a precise way. Is there any commercial or non-commercial unit out there that just provides a couple of resistors, capacitors, inductors, a diode and a DC+AC voltage and current source? Shouldn't be that difficult if sub-1% precision isn't a requirement, one just needs to find a suitably large switch to change between all these options?

[Fungus]

This is why the mantra here is always: "One meter isn't enough"

It doesn't matter how awesome or expensive your multimeter is, you need another one.

If you have more than one meter you can contrast/compare them using any old components you have lying around.

[joeqsmith]

I would say it very much depends on the meter. 

I've run potentially destructive tests on multimeters and testing them for failures is certainly part of it.   To do this, I made up a custom test box.  It's evolved over time but covers most of the basics.   

You can see it here:

[HKJ]

You can get something here: https://dmmcheckplus.com/shop

It can be used to check many of the ranges.

[Bzzz]

This is why the mantra here is always: "One meter isn't enough"
[...]
If you have more than one meter you can contrast/compare them using any old components you have lying around.

Guess how we found out

Exactly that, but in a box as shown above. Just a little nicer (sorry pal), so probably planned from the beginning and with two banana jacks and internal switching to access all components/functions. Now that we've established that more than one weirdo has use for such a thing, chances are someone else already made a PCB or at least schematic for it.

[Fungus]

Exactly that, but in a box as shown above. Just a little nicer (sorry pal)

You'll still need more than one meter  - to check if your pretty box is working or not!

[joeqsmith]

True.  Switches fail, which is why I chose to not use one.  Eventually though, I ran out of space...

[Bzzz]

You'll still need more than one meter  - to check if your pretty box is working or not!

While I do not understand where your idea of "this guy only has one meter" comes from, I can assure you that there are dozens of low to lowest class handheld meters. The more I can identify as malfunctioning or right away safely blow up during testing, the better.

Switches do fail, but I'd rather exchange a faulty switch every couple of years than move around crocodile clips 20 times per device tested. Relays or MOSFETs would need a µC with lots of I/O and aren't bulletproof either.

[Fungus]

While I do not understand where your idea of "this guy only has one meter" comes from, I can assure you that there are dozens of low to lowest class handheld meters.

Good!

one just needs to find a suitably large switch to change between all these options?

I wouldn't do it that way.

You're supposed to disconnect the meter from the device when you turn the range switch. That way you don't turn the dial to resistance when the AC test voltage is coming out of the test box.

You could do it with a push button for each range to be tested that only connects the signal when it's pushed but then you might push two buttons at once or a button might get stuck.

Much better to have separate test points for each signal.

[joeqsmith]

It takes me a couple of minutes to run through everything manually.   

[J-R]

For work use I do think a box like the DMMCheckPlus makes good sense.  You can mail it back every year and receive an updated printout of the values.  This spring they released the L/C add-on board, so it covers a lot of measurements in a compact package.  Having a business-to-business relationship seems better than an employee building something.  It's safe to say most regulars here can run a DMM through its paces quite easily but at work you might have different skill levels.  So you need more...

Maybe someone who does this for a living could chime in, but it seems to me you'd have a go/no-go procedure that would be followed EVERY time a measurement device was pulled off the shelf for use, and then regularly you would ship the devices off for certification.  Anything less and it could open the company up to liabilities if an employee was injured?

[joeqsmith]

For work, I would have thought that the meters were sent in for annual cal, or cal'ed in-house.   For electrical work, I assume the electricians carry a checker.   Maybe not depending where you work. 

[Bzzz]

Imagine sending off a 10€ multimeter for annual calibration. Gonna piss my pants 
Guys, quite a lot of them are in fact 10€ class meters that, if found faulty or dropped correctly, can be replaced with 50€ class meters like UT61s. The one with the ACV defect was an UT139C, same price bracket. Those do not earn money, so nobody will spend money on them. But they're wasting my time and everybody else's if that old banger causes any misdiagnosis on a 12V/24V "R&D" test rig, sensors, heaters, you name it.

The electricians use their Duspol which technically is a multimeter, but only ever used for testing voltages and continuity.. Since I assume they know what they're doing and it's their assigned device that is not shared with others, they would pretty quickly find out if it's faulty. But I also doubt those will be sent in for calibration or any other checks.

You're supposed to disconnect the meter from the device when you turn the range switch. That way you don't turn the dial to resistance when the AC test voltage is coming out of the test box.

You could do it with a push button for each range to be tested that only connects the signal when it's pushed but then you might push two buttons at once or a button might get stuck.

Much better to have separate test points for each signal.

I expect a multimeter to survive 60V or whatever is allowed (SELV) and practical from a DC-DC boost converter from a 9V battery. If it dies while range switching, see above.
Having a separate button for everything surely would work, but someone will need to make a case for it. A box with 20 buttons and wires hanging from it will always have the crazy scientist look.
The tester will never be used by the random dude that uses his favourite wire color for everything, so I very much accept the risks of having a button stuck or being able to press two at once. I understand that you are 'muricans with all the liability BS forced upon you, but we're not building a reactor control center here. It's resistors, capacitors, inductors, diodes, thermocouples and low voltages/currents.

[joeqsmith]

Imagine sending off a 10€ multimeter for annual calibration. Gonna piss my pants 

...

I understand that you are 'muricans with all the liability BS forced upon you, but we're not building a reactor control center here. It's resistors, capacitors, inductors, diodes, thermocouples and low voltages/currents.

I suspect at most companies, anything that has to do with controlling a process is under calibration.   For the companies not making reactor control centers, it may be more a matter of quality control than liability. 

[bdunham7]

I don't know of any commercial calapproximator as I'm going to dub your idea.  You just need a about a 30 volt AC transformer, some resistors for voltage dividers, rectifiers and filter caps to make the DCV, DCA, ACV and ACA sections.  Then the rest is just a bunch of individual components.  I think a row of banana jacks is just as easy to use as as switch and then you have a place to mount the components as well. 

[Fungus]

I think a row of banana jacks is just as easy to use as as switch

Me too.

Simplest solution: A piece of perf board with a 12V supply, some resistors/diodes/capacitors on it, a 555 timer-oscillator for AC tests and a row of test points.