Handheld meter robustness testing

[bdunham7]

When cold, I am expecting with a kV applied, the initial draw is somewhere around 1kV/2000ohms or 500mA.   Higher for some of the cheap meters.   I doubt many electronics hobbyist have that sort of power supply on hand.   I would think the use case was you touch the leads to the live source with the meter set to each function, except for the current.  Allowing a minute or so for the PTC to cool between tests.   Creeping up on the voltage would seem like a rare case.

The 116 appears to be ~3k initially on Lo-Z ranges.  My DC supply does sag a bit on the initial load, so the transient effect is blunted.  I have two of them, so perhaps I can set them up (carefully) in parallel.  It's the AC that it takes me about 5 seconds to bring all the way up.  I'll have to work on that. I'm also trying not to make stupid mistakes like accidentally giving the 8842A a full 1000VAC (it is only rated for 700).  It didn't kill it, it just blinked at me--while reading  the voltage correctly-- until I realized what was going on.

I also tried a Fluke 27, a CAT III/1000V version.  It has an initial input resistance of ~5k so my DC didn't sag as much and I was able to start the AC higher and jump up in one step.  It only has mVAC, mVDC, Ohms, Continuity and Diode ranges and it withstood the full 1000VAC and 1000VDC on all of them. 

As for range switching, I think they do put some thought into that.  For example, the Auto Lo-Z function requires you to cross the OFF position to get in or out of it.  But things like hooking up to 600VAC, realizing you are in mV and just switching to VAC, that seems like a potentially destructive test to me.  Maybe you could do a combined lifecyle test with using your switch lifetime tester with the meter connected to 1000VAC! 

[joeqsmith]

The concern would be engaging a low voltage clamp and opening the contact with a kV DC applied.  I don't see too many meters surviving that but I've never tried it. 

For AC, things could get interesting if you are suggesting that any frequency is open game.  We just need to stay below the printed voltage on the meter.   I cooked that UT61E+ doing just that and suspect that many of these cheap meters using only a PTC/s to limit the current would be damaged well below their claimed levels.    I'm not sure how that fits into your criteria of being robust. 

***
Consider the 121GW uses two PTCs in series for the low Z mode.   Where the UT61E, has a low voltage clamp which will shave off some insignificant amount, the 121GW is right across the inputs. 
***
Your Fluke 116 unlike my 115 (which has never been damaged and I still have) has a LowZ mode.  Maybe like the 121GW, it uses a couple of PTCs as well?  The plot thickens.   
***

As for range switching, I think they do put some thought into that.  For example, the Auto Lo-Z function requires you to cross the OFF position to get in or out of it.

That off position could be the end of the meter.  You had the meter in the lowZ, attached to a high current DC kV source, then turn it to the OFF position.  I doubt that will end well. 


A couple of demonstrations from John Ward's channel:

[bdunham7]

The concern would be engaging a low voltage clamp and opening the contact with a kV DC applied.  I don't see too many meters surviving that but I've never tried it. 

I think the meter's main defense against that would be that the PTC quickly lowers the current to a level that a switch can maybe handle.  If you spin the dial with the power on, you may defeat that feature.  I suspect that even if the meter doesn't burn up on the spot, there would be cumulative damage.

For AC, thing could get interesting if you are suggesting that any frequency is open game.  We just need to stay below the printed voltage on the meter.   I cooked that UT61E+ doing just that and suspect that may of these cheap meters that only use a PTC to limit the current would be damaged well below their claimed level.    I'm not sure how that fits into your criteria.

I would say 1kHz and below is fair game, but unless the meter is specifically rated for more, I wouldn't expect them to really deal with HF AC.  My calibrator only goes to 1100VAC @ 1kHz (20 and 50kHz at lower voltages) so I also don't have the means to give them the jqs MeltCal^TM treatment.  However, I think it would be reasonable for us to expect manufacturers to explicitly list these limitations in the specs.  Maybe if they were forced to think things all the way through and completely specify their products, they'd figure out why some manufacturers spend all that money on those expensive resistors.  I wonder how some of those meters would stand up to even something like 1kV@10kHz.

[joeqsmith]

Your 116 specs the frequency to 50kHz.  Dave's 121GW, 1MHz.  Seems like the 121GW should handle 1MHz within the voltage limits.   Maybe...   Of course, I would think a meter with a spec'ed freq counter at 220MHz would be able to read 220MHz before the solder starts to reflow and the case melts.   

The concern would be engaging a low voltage clamp and opening the contact with a kV DC applied.  I don't see too many meters surviving that but I've never tried it. 

I think the meter's main defense against that would be that the PTC quickly lowers the current to a level that a switch can maybe handle.  If you spin the dial with the power on, you may defeat that feature.  I suspect that even if the meter doesn't burn up on the spot, there would be cumulative damage.

Maybe.  It would be easy to setup a simple test without a meter and just see if an arc could be drawn.    I would use a 1k resistor in series with the PTC, apply the 1kVDC across it, let the PTC heat up, then open the circuit.  It's nothing I have ever tried.     I'll make a clip of it. 

[bdunham7]

Your 116 specs the frequency to 50kHz.  Dave's 121GW, 1MHz.  Seems like the 121GW should handle 1MHz within the voltage limits.   Maybe...   

I don't see that in my manual and I know it won't measure HF AC--it works well to 2kHz and then starts to drop off.  The manual specifies accuracy only to 1kHz for TRMS. It lists the frequency counter function to 100kHz but doesn't specify a maximum input voltage as far as I can see, nor is there any explicit mention of maximum V-Hz, like there is with the F27.  Apparently mentioning V-Hz isn't in vogue any more, perhaps dV/dt is too complex of an issue...    Now as for giving the 121GW 6*10⁸ V-Hz, I suspect that won't go well, but I wouldn't want to be around if you try it.  If you have a low-impedance 600V_RMS 1 MHz source, that is.   

Of course, I would think a meter with a spec'ed freq counter at 220MHz would be able to read 220MHz before the solder starts to reflow and the case melts.

Silly you for believing anything printed on a UNI-T box.....    Did you ever figure out what the output voltage and power of that RF amp was? 

[Neutrion]

The 20J may be overkill but all you safety experts posting here already know where that number came from. 

Hmmm.  Is it because that is about the energy used in a defibrillator or because it matches up with the output of an electric fence charger (a powerful one)? 

You may not agree but hopefully you at least now have some understanding why I will continue to call my transients low energy.

Thanks for the detailed explanation.  Low is a relative term, of course, and relative to MOV ratings, IEC transients, etc, 20J is obviously much lower.  The comment I responded to was talking about damage to semiconductors, etc and there 20J  will pop the lid on quite a few devices.   I think the low energy events he was referring to are the sorts of noise, ESD or spikes that might get through normal filtering for reasons other than that they overwhelm the protection systems by simply exceeding their voltage or energy limitations.  Piezo igniters and 220MHz RF come to mind....

Also thanks for the explanation Joe, but I was also talking similar small energy surges, where basically only the fast rise time is what counts, and the only interesting parameter of the protection circuit is how fast they act, and what is the max voltage with the small energy spikes which will be let through.
The Up parameter is mostly the only thing specified at all, but that is at the max amp. rating so usualy no info about the reaction to the much weaker stuff.
Because even if a surge protector meets its Ka and thousend joule specs, there is no word about how it filters out the small surges which eventually leads to failures in electronics.
The surge protector manufacturer is not interested in building a circuit which clamps everything all the time, because than the MOVs etc eventually wear out, so I suppose most of the small spikes will hit the protection circuits of the electronics. But this is only a theory because not much testing going on on ACTUAL surge arrestors and input protection circuits on the net. But I understand if you are not interested in using the... Joe-gen for it.


And here is an important announcment for The Uni-t Ut60e Protection Circuit Variations Collector Society:
(short: TUUPCVCS)
There is a possibly really unique version aviable in Sweden, the Clas Ohlsson edition. Which is a kind of store chain. Being a swedish version I suppose maybe it comes with REAL fuses from the Nobel factories.
(Note for the Fluke test guys: Reinforce the blast shield if testing those!)
Being a rarity and a collectors item soon, I consider sending it to anyone for a symbolic 2000 Euro price. (Shipping included) It would be a solid investmen in these hectic days! (No I don't have it yet, so can't post pictures.)

[joeqsmith]

Silly you for believing anything printed on a UNI-T box.....    Did you ever figure out what the output voltage and power of that RF amp was? 

The amplifier's output voltage will be load dependent.  Part 2 shows how the PTCs behave at higher frequencies and how there was nothing else to limit the current.  It also shows how the Brymen BM78x would easily handle this because Brymen engineers were smart enough to use a device that was more stable (a resistor).     

To melt the solder what, 20W - 40W?  A bit of damage to the PCB, 80W?   Say the meter presents 5 ohms @ 120MHz,  20V would get you 80W.   Just a ballparkish guess.  Let's say double that.   Does 40V seem like a lot to you?  I mean, I thought you were all about kVs?!      

***
Looks like I was able to load it to about 70W in that range.   Highest seems around 100V with a 50ohm load at 100MHz.  It makes a decent lab amplifier.   

https://www.eevblog.com/forum/projects/home-made-pa-and-load/msg635912/#msg635912

[bdunham7]

To melt the solder what, 20W - 40W?  A bit of damage to the PCB, 80W?   Say the meter presents 5 ohms @ 120MHz,  20V would get you 80W.   Just a ballparkish guess.  Let's say double that.   Does 40V seem like a lot to you?  I mean, I thought you were all about kVs?!   

4.8*10⁹ V-Hz seems a bit much for direct connection to almost any test instrument, let alone a DMM.  Even my 8505A would like to stay below 2*10⁷.  And your 8506A, despite being a specialized HF AC instrument, is only good for half that.  Having a wirewound resistor or a common-mode toroid right at the front end might help in these extreme cases, but I suspect that level of RF will burn up lots of things.  If you're doing it some more, I'll watch from here.

[Fungus]

There must separate path injecting current into DUT for resistance and capacitance, and this composite 1 MOhm resistor is measurement path for these ranges. There will be transistor clamp after it somewhere before going into DMM chip. ...

I would have said the complete opposite: The "voltage path: is 10MOhms impedance so even a 10,000 volt input is only going to pass 1mA - very easy to clamp and not going to hurt much. It wasn't complete;y crazy to think that chain of resistors was the voltage input, and indeed it is the voltage range protection on my "cheap meter" and many others.

The Ohms/Diode/Capacitance range requires the meter to apply a current to the DUT so it's much lower low impedance. It makes sense to have all the PTCs and MOVs and stuff on those ranges, and some meters appear to do it that way..

My LIDL meter is interesting because it mixes both, it has a 10MOhm chain for the voltage input and a 1MOhm chain for whatever the slightly lower chain turns out to be.

Nice job sir.   Do you have the tools/skills needed to remove one of the resistors that make up the chain R8-R2?

I actually bought a cheapo hot air gun a couple of weeks ago. Maybe this is the opportunity to take it out of the box. 

[floobydust]

Consider the newer small Fluke multimeter input protection circuit, instead of desoldering things.
You can see the three main channels, the 1MEG, 10MEG and how they are used.

[joeqsmith]

...
And your 8506A, despite being a specialized HF AC instrument, is only good for half that.  Having a wirewound resistor or a common-mode toroid right at the front end might help in these extreme cases, but I suspect that level of RF will burn up lots of things.  If you're doing it some more, I'll watch from here.
...

I have no plans to do much testing with that old Fluke after saving it from scrap.  It's only job is to hang out next to my 8000A and act as a reminder of days of old. 
 
When I showed Part 2, I talked about the resistor becoming more inductive while the PTC became more capacitive.  No doubt that added resistor is what saves it.   

It seems like you are backing down about surviving the max marked voltage on the face of the meter.   Rather we need to actually read the manual and understand the specs to save the meter.   I guess that is true for the ESD and other faults as well.   After all, who in their right mind is going to try and measure the output of their amplifier directly with a meter ....  oh wait... there was that radio hobbyist who connected their low cost VNA directly to their radio to advance the art.       They were special.

I agree that working with RF is a bit out there and I would also say even a kVDC with 100s of mA wouldn't be common but I don't test these.   I've seen people post about damaging them with a MOT and their lawn mower mags but I won't go there.   I was really more interest in how they held up with low energy short transients as it seems that has been how I have damaged them in the past.  ESD is a problem where I live but again, I wasn't expecting to see any meters be effected by it.   

As you can see, testing robustness isn't so cut and dry.   I still find it an interesting topic. 

I actually bought a cheapo hot air gun a couple of weeks ago. Maybe this is the opportunity to take it out of the box. 

I've soldered with a heat gun for many years.  Your boards may be lead free and require more heat than you can get out of your gun.  I would practice with the cheap meter first.  Maybe blend in some tin lead on the Fluke to help pull them until you get an idea how to use the new gun.  Always new things to learn. 

[bdunham7]

Consider the newer small Fluke multimeter input protection circuit, instead of desoldering things.
You can see the three main channels, the 1MEG, 10MEG and how they are used.

Series PTCs and a trimpot?  What model(s)?

[floobydust]

That's the Fluke 15, 17 family. The 101, 107 only uses two MOV's. Point is it's been optimized beyond the classic 87 input protection, and for cost as well.

[bdunham7]

It seems like you are backing down about surviving the max marked voltage on the face of the meter.   Rather we need to actually read the manual and understand the specs to save the meter. 

Not 'backing down'.  Clarifying! 

Yes, as I was going on and on about V-Hz, I would reasonably expect the CAT rating on the face to apply to DC and mains, so say at least up to 400Hz?  Unless there is a type-N connector on the front.  The problem with V-Hz limitations is that it can't be easily condensed into something that you can print on the front of the meter.  Even just a V-Hz number, as if the average user can do math in their head, won't help.  I'm pretty sure that almost any meter can take 10mV @ 1GHz, but not many will do well at 1GV @ 10mHz!  So yes, that is always going to be an RTFM specification, its just a shame that most manuals omit this entirely, even when the meter has specifications that might invite disaster.

A while back I was looking at some HVDC power supplies used for plasma deposition systems.  I ended up not working on them at all because they don't have much market value, but I did wonder how I was going to safely measure even the primary side.  These are 480V 3PH input SMPS using big IGBT modules, the supplies are the size that you can carry them in a pickup truck but you would use a forklift to put them in.  So you have a 600-700VDC bus and a SMPS probably operating in the tens of kHz.  If you carefully RTFM for a Fluke 289, you would see that over 50VAC and 10kHz, accuracy is 'not specified', but the 2*10⁷ V-Hz limitation is hidden away elsewhere and not easy to find.  Most other meters do not have specifications either on the front or in the manual that would tell you not to use them for this application. 

[joeqsmith]

Your 116 specs the frequency to 50kHz.  Dave's 121GW, 1MHz.  Seems like the 121GW should handle 1MHz within the voltage limits.   Maybe...   Of course, I would think a meter with a spec'ed freq counter at 220MHz would be able to read 220MHz before the solder starts to reflow and the case melts.   

The concern would be engaging a low voltage clamp and opening the contact with a kV DC applied.  I don't see too many meters surviving that but I've never tried it. 

I think the meter's main defense against that would be that the PTC quickly lowers the current to a level that a switch can maybe handle.  If you spin the dial with the power on, you may defeat that feature.  I suspect that even if the meter doesn't burn up on the spot, there would be cumulative damage.

Maybe.  It would be easy to setup a simple test without a meter and just see if an arc could be drawn.    I would use a 1k resistor in series with the PTC, apply the 1kVDC across it, let the PTC heat up, then open the circuit.  It's nothing I have ever tried.     I'll make a clip of it.

[bdunham7]

That's the Fluke 15, 17 family. The 101, 107 only uses two MOV's. Point is it's been optimized beyond the classic 87 input protection, and for cost as well.

That's market segmentation for you!  Still, depending on what the actual components are, it looks passable, especially if they're certified in China.    

I'd prefer the resistor to the PTC, it almost seems bizarre specifying either or.  I'd guess you can get by with 2 MOVs, a PTC and a resistor as long as you have a simple meter (no Lo-Z) and use good parts.

[Per Hansson]

And here is an important announcment for The Uni-t Ut60e Protection Circuit Variations Collector Society:
(short: TUUPCVCS)
There is a possibly really unique version aviable in Sweden, the Clas Ohlsson edition. Which is a kind of store chain. Being a swedish version I suppose maybe it comes with REAL fuses from the Nobel factories.
(Note for the Fluke test guys: Reinforce the blast shield if testing those!)
Being a rarity and a collectors item soon, I consider sending it to anyone for a symbolic 2000 Euro price. (Shipping included) It would be a solid investmen in these hectic days! (No I don't have it yet, so can't post pictures.)

I only found a UT61D, not UT61E: https://www.clasohlson.com/se/Multimeter-med-USB-UNI-T-UT61D/p/36-4717
That said your post is funny.
However it seems to be the bog standard "Chinese" version and not the "GS" branded version because the listed spare fuses for it are 1A 240V for the mA range.
But they are 0.5A 600v in the "GS" version.
And the linked manual on their page also claims the instrument is CAT-IV rated at 600v and CAT-III at 1000v which we all know is blatantly false.
(The GS rated meter is CAT-III 300v and CAT-II 600v but even that is a stretch if you follow the link in my quote below from Joe).

[Per Hansson]

Left on for 10 seconds, survives just fine at the nominal 245V here. Readings a smidge low as the PTC recovers.

So I'm curious about why these tests would be conducted at those levels on meters with a CAT-anything label at.  Should the meter not withstand, without damage, the application of full rated voltage (1000V in this case) to any input on any setting?  AFAIK, even supposedly fragile CAT I bench meters will pass that test.  Some older pre-CAT bench meters have separate, lower specifications for voltage protection on ohms and such, but even the old Fluke 8842A will pass that test.

Link to the persons posts who had asked about this may be found here:
https://www.eevblog.com/forum/repair/uni-t-ut61e-diode-mode-repair/

Some of the smaller PTCs are only rated for 500V.  Some meters have only a single PTC which if the low voltage clamp is active, will have well over 900V across them....  Maybe....  Turn the dial, I suspect you will get a light show. 

***

I'm referring to meters that some claim to be properly rated, not ones where there isn't any dispute that the CAT info printed on them is a bad joke.

Sorry, I missed that part.   If a KVDC were applied with unlimited current and the function switch were rotated, I suspect you would burn the contacts on most meters.  I wouldn't recommend anyone attempt it.

It was me who requested the test
Thanks to both Joe and Dave for performing them!
It is nice to see this model survives at least, as can be seen in the link from Joe's quote above the old "GS" model did not fare so well, at least the revision I have!

[floobydust]

I'm not sure what loophole is being exploited.
UT61A,B,C,D,E 61010 certificate report #17014567 003 by TÜV Rheinland (China) Ltd. in Beijing is for the "GS-Mark" to Cat. II 600V, Cat. III 300V with 600V fuses, three MOV's.
Many Uni-T tests are TÜV Rheinland (Shenzhen Office) results then carried over to TÜV Rheinland Nurenburg.

Yet the products are in the Intertek database as well. I think you can carry over certificates between agencies for further laundering of details.
It might be a case of the Golden Sample to TÜV, with smoke and mirrors about the National Differences product builds.
So the fact the GS special is a different build for i.e. North America gets lost.


On another note
From another TÜV 61010 test report, they added tests:
16.2DV.1
"Multifunction meters and similar equipment shall be tested by changing the function /range selector to all possible settings while connected to the maximum rated source."
16.2DV.2
"Compliance is checked by testing to verify no HAZARD occurs when switching selector settings."

[Neutrion]

And here is an important announcment for The Uni-t Ut60e Protection Circuit Variations Collector Society:
(short: TUUPCVCS)
There is a possibly really unique version aviable in Sweden, the Clas Ohlsson edition. Which is a kind of store chain. Being a swedish version I suppose maybe it comes with REAL fuses from the Nobel factories.
(Note for the Fluke test guys: Reinforce the blast shield if testing those!)
Being a rarity and a collectors item soon, I consider sending it to anyone for a symbolic 2000 Euro price. (Shipping included) It would be a solid investmen in these hectic days! (No I don't have it yet, so can't post pictures.)

I only found a UT61D, not UT61E: https://www.clasohlson.com/se/Multimeter-med-USB-UNI-T-UT61D/p/36-4717
That said your post is funny.
However it seems to be the bog standard "Chinese" version and not the "GS" branded version because the listed spare fuses for it are 1A 240V for the mA range.
But they are 0.5A 600v in the "GS" version.
And the linked manual on their page also claims the instrument is CAT-IV rated at 600v and CAT-III at 1000v which we all know is blatantly false.
(The GS rated meter is CAT-III 300v and CAT-II 600v but even that is a stretch if you follow the link in my quote below from Joe).

True, its the D, I just saw it in the shop remembering it was the E. Was even thinkering buying one, because the special edition could be someting nice.
But see, you are also not perfectly accurate in this case, as when you open the first user manual, Ver. 20140523 than you get the 240V fuses listed. BUT!
If you open all the other manual versions, than you get 1000V fuses listed. 
Now I have to raise the price to 5000Euro, because I either brake into the shop during the night, or without being noticed have to brake all the plastic boxes during the day, tear down the meters, and check what's inside. And possibly do that in different shops in the country.
Which meter has which protection within the Clas Ohlsson Edition? (of the D version...)

[AndrewBCN]

From the video: "... I think it would do the meter in." Proceeds to film in slow motion plasma with a 5mm arc from 1000V DC power supply. 

I am no expert but I too think "it would do the meter in".

[bdunham7]

From another TÜV 61010 test report, they added tests:
16.2DV.1
"Multifunction meters and similar equipment shall be tested by changing the function /range selector to all possible settings while connected to the maximum rated source."
16.2DV.2
"Compliance is checked by testing to verify no HAZARD occurs when switching selector settings."

I don't see any reference to AC or DC.  I wonder if there might be a presumption that CAT voltages only refer to AC at mains frequencies.  Either might damage a meter in this case, but AC would surely be less likely to pose a hazard.

[bdunham7]

So I was contemplating testing my Fluke 289 (CAT III/1000V CAT IV/600V) with 1000V on all ranges and I decided to RTFM first.  The F289 is a very nice instrument and I don't mind the menus, but the manual and specifications are complex and not straightforward or intuitive--the type of thing you need to have on the bench to refer to, not the type you can read once and remember it.   I found this gem:

[Fungus]

Consider the newer small Fluke multimeter input protection circuit, instead of desoldering things.

I was thinking I could maybe take the spring contact out of the range selector - it breaks the circuit, it's much easier!

[joeqsmith]

...
Even just a V-Hz number, as if the average user can do math in their head, won't help.  I'm pretty sure that almost any meter can take 10mV @ 1GHz, but not many will do well at 1GV @ 10mHz!  So yes, that is always going to be an RTFM specification, its just a shame that most manuals omit this entirely, even when the meter has specifications that might invite disaster.
...

...
Even my 8505A would like to stay below 2*10⁷.  And your 8506A, despite being a specialized HF AC instrument, is only good for half that.  Having a wirewound resistor or a common-mode toroid right at the front end might help in these extreme cases, but I suspect that level of RF will burn up lots of things.  If you're doing it some more, I'll watch from here.
...

I would expect that my first Fluke DMM was highly susceptible to being damaged by RF (or looking at it sideways).   I suspect as the front ends of these handhelds evolved, the need to derate them was less important.  Of course, you still end up with meters like the UNI-T UT61E+ having some sort of frequency counter speced at 220MHz with a couple of PTCs in series, but I suspect most modern robust meters would not have a problem.   Like I had shown with the BM78x, with the same setup that damaged the UT61e+, the PTCs had very little heat.   As I demonstrated, that resistor was fairly stable to 50MHz and a 1kohm just doesn't present much of a load to the small amplifier. 

I don't mind running other tests if you had something in mind.  I could take that same resistor/PTC we just used and run them any way you like.  The same with that Brymen BM78x prototype.   Let me know.