Interesting and sad case of electrical injury

[joeqsmith]

What actually happens to a CAT-III 600V Fluke if you connect continuous 4kV to it?

I assume you are asking about AC, 60/50Hz with unlimited current for all practical purposes?  And with the meter set to ACV mode only?   

It would certainly be possible to see if a meter could be damaged using low currents without too much risk.  My guess, most would be damaged but it's not something I have ever considered trying.

***
If you wanted to do it for real, maybe contact some of the test labs:
https://ep-us.mersen.com/sites/mersen_us/files/2018-11/BR-High-Power-Test-Lab-Brochure.pdf

[calzap]

What actually happens to a CAT-III 600V Fluke if you connect continuous 4kV to it?

I assume you are asking about AC, 60/50Hz with unlimited current for all practical purposes?  And with the meter set to ACV mode only?   

It would certainly be possible to see if a meter could be damaged using low currents without too much risk.  My guess, most would be damaged but it's not something I have ever considered trying.

***
If you wanted to do it for real, maybe contact some of the test labs:
https://ep-us.mersen.com/sites/mersen_us/files/2018-11/BR-High-Power-Test-Lab-Brochure.pdf

4kV with unlimited current is what happened in my original post.  Wouldn't have mattered if it had been a Fluke instead of an RS meter.  Up to 600 V with effectively unlimited current  ...  OK.  Beyond 600 V with unlimited current, you're holding a bomb.

Mike

[Marco]

4kV with unlimited current is what happened in my original post.  Wouldn't have mattered if it had been a Fluke instead of an RS meter.

With 1 Meg series resistance to the input voltage limiting, it's going to burn through pretty fast. With a couple Watt 10 Meg series resistance and assuming the selector switch and PCB don't arc over, it could probably hold off the voltage indefinitely.

I think it's 1 Meg though.

[calzap]

4kV with unlimited current is what happened in my original post.  Wouldn't have mattered if it had been a Fluke instead of an RS meter.

With 1 Meg series resistance to the input voltage limiting, it's going to burn through pretty fast. With a couple Watt 10 Meg series resistance and assuming the selector switch and PCB don't arc over, it could probably hold off the voltage indefinitely.

I think it's 1 Meg though.

Do you really think you could apply a continuous 4kV with essentially unlimited current to the input of any regular multimeter of any brand without internal arcing?  Doesn’t matter if a resistor is 1 gigohm if it arcs over.  With that kind of current, any tiny arc will almost instantly grow to catastrophic proportions.
Mike

[coppercone2]

what does this table mean?

it says 12kV for a 1000V rated meter, and 8kV for a 600V CAT IV meter. It says for impulse though

https://www.flir.com/discover/professional-tools/what-do-cat-ratings-mean/

would the 1000V rated CAT IV meter still explode like that?



And it made me wonder, if you can't fuse it there, I wonder if they could make a advanced AFCI that actually detects where the arc is. Like a arc TDR. I know they expect some arcing with HV circuits, maybe you can determine if arcing is happening in the correct location and shut it down if its not? If it had a resolution of a foot to determine if its in the switch or if its in the transformer case.

[shapirus]

Doesn’t matter if a resistor is 1 gigohm if it arcs over.

"If it arcs over" is key. The question is, will it?

[joeqsmith]

4kV with unlimited current is what happened in my original post.  Wouldn't have mattered if it had been a Fluke instead of an RS meter.

With 1 Meg series resistance to the input voltage limiting, it's going to burn through pretty fast. With a couple Watt 10 Meg series resistance and assuming the selector switch and PCB don't arc over, it could probably hold off the voltage indefinitely.

I think it's 1 Meg though.

Well, you did mention Fluke brand and not the bottom end.   I'll remind you of someone that had told me about how my testing was damaging the meter's MOVs.  I ran a test with a few kV for days measuring the current.  Person did not understand that the PTCs and series resistors limit what the MOV is exposed to. 

To be clear, there is typically a surge rated resistor in series with a large body PTC in series with some MOV combination.  The 1M you mention is basically looking across the MOVs.  Clamped typically under 2kV.   So, say 2.5kV across the surge rated resistor and PTC.   As the PTC heats (which it would in this condition) the voltage drop across it will increase.   I doubt very much that the 5mm PTCs I typically see in the low end meters would survive but I really have no feel for the large parts I see in the Fluke meters.

It's certainly possible to check it (safely ish... sort of... assuming you have some idea what you are doing and take precautions.)

[Marco]

Do you really think you could apply a continuous 4kV with essentially unlimited current to the input of any regular multimeter of any brand without internal arcing?  Doesn’t matter if a resistor is 1 gigohm if it arcs over.  With that kind of current, any tiny arc will almost instantly grow to catastrophic proportions.
Mike

It already has to hold it off for a bit for CAT-III 600V. How long exactly? (I haven't read the standard.) If it's microseconds it likely won't arc, but if it's milliseconds and it can arc it will arc already.

It's not like the distance needed to hold off 4kV across a clean PCB is huge.

[Someone]

Do you really think you could apply a continuous 4kV with essentially unlimited current to the input of any regular multimeter of any brand without internal arcing?  Doesn’t matter if a resistor is 1 gigohm if it arcs over.  With that kind of current, any tiny arc will almost instantly grow to catastrophic proportions.
Mike

It already has to hold it off for a bit for CAT-III 600V. How long exactly? (I haven't read the standard.) If it's microseconds it likely won't arc, but if it's milliseconds and it can arc it will arc already.

It's not like the distance needed to hold off 4kV across a clean PCB is huge.

Impulse tests are the 1,2/50 μs test specified in IEC 60060 [...] If a.c. or d.c. tests are chosen, they shall be conducted for a minimum of three cycles in the case of a.c., or three times with a duration of 10 μs in each polarity in the case of d.c.

Different limits for the different testing regimes. But transient overvoltage is only over the 1,2/50 μs waveform.

[joeqsmith]

Standard surge is 50us FWHH.  For 4kV, I test using 100us FWHH with a 2ohm source, limited to about 15J.   

I had tested that Fluke 101 using 13KV peak, 100us FWHH, 2 ohm source and it was not damaged.  A member here repeated the test using an actual combo generator following the IEC standards and saw no damage as well.   

Still, there's a big difference between that and 50/60Hz 4kV.   With a 100us FWHH, I doubt the PTC would even respond.  So the voltage across the two (surge rated resistor and PTC) will more closely match.   Not the case with the low voltage AC mains.   

I would never suggest that the Fluke 101 would survive it without running at least some minimal test..

[robert.rozee]

I would never suggest that the Fluke 101 would survive it without running at least some minimal test

i'd certainly not expect any meter to survive it, but rather to be able to contain/absorb the 'event'.

attached below is the full schematic of the Fluke 77 series II. from this one can see that R2/E1 are largely there to protect the OHS pin of U1 (if S1 is closed) and pins 11 and 10 of  S1 (if S1 is open). at 1500v (E2 strikes) R2 would see approximately 1.5 watts, at 4kV it would see 16 watts; even at this power level i would expect it to survive for a few seconds, assuming it did not arc over.

the main show involves R1, RV1 and RV2.

to have any hope of containing/absorbing a 4kV continuous input i would suspect that both R1 (1k ohm) and R2 (1M ohm) would each need to be split across multiple parts that were good for 1kV each. it would then be necessary to add a fuse in series with the 'V/O HI' input connector such that our new fuse:
(a) was essentially rated at near to zero current. ie, a "10mA fuse",
(b) could break 4kV and contain/extinguish the resulting plasma,
(c) would go open prior to the R1 (1k ohm) combination overheating.
R1 would no longer need to be a fusible part, as our "10mA fuse" would replace this function.

now all these changes would add size to the meter, as well as cost. plus, they wouldn't help if our user tried to measure 10kV.

i'd still be keen to know how a bog-standard Fluke 77 reacted to a 4kV continuous input   


cheers,
rob   :-)

[NiHaoMike]

(b) could break 4kV and contain/extinguish the resulting plasma,

If the current can be sufficiently limited, that's surprisingly easy based on the videos of hobbyists playing with neon sign transformers. Probably would make most sense to have that fuse be in the probe, one design that would be sensitive enough would be a high voltage resistor soldered to a spring that would pull apart when it melts, with some sort of dielectric liquid to quench the arc. At the same time, it needs to give a visual indication that it has blown to not create another hazard of the meter silently indicating no voltage.

[calzap]

In the video of the Eddie Adams accident, he was using a multimeter rated at either 600 or 1000 volts (both values are mentioned in the video).  The voltage he applied it to was mentioned as 2200 or 2300 volts; however, the equipment can be seen to have a label of 2300 volts.  The multimeter was almost certainly a Fluke.  Perhaps someone can identify the charred remains as a Fluke.  However, the electricians interviewed appear to be holding a Fluke at times.  The electrician doing the reenactment appears to be holding a Fluke.  They all worked for the same company; so likely all the company’s electricians were issued Flukes.  Someone familiar with Flukes can probably identify the model(s).

So, a Fluke rated at either 600 or 1000 V was applied to a continuous voltage of 2300 volts with huge current available.  And it exploded.

Mike

[johansen]

I have a capacitor bank which can be configured for 80uf at 8kv or 320uf at 4kv.

Its not enough to shrink quarters but it can bend them, self resonant frequency is 50khz

Please send me expensive test equipment to blow up.

[coppercone2]

I think the one in the picture is this
https://www.ebay.com/itm/175764036511

[johansen]

And remember, we are not talking about some wimpy 4KV supply from a CRT flyback or CCFL inverter!! We are talking here about (looking at the picture) a cabinet fed from mains 4160 three phase that maybe could sustain 30 amps or more and on a dead short could produce probably over 1000 amps for a short duration. You could have been staring down an arc flash of over 4 megawatts or over 5000 horsepower even if for a second or two. That will ruin anyone's day!!

You are off by a couple orders of magnitude.

An average 120/240 system on a larger building has a short circuit current of 10,000 amps depending on the size of the pole transformer and length do the utility drop. My parents house had a short circuit current around 5ka

4000 volt systems are typically fed from 100Kva and larger transformers of 4 to 10% impedance. Lets say a 10 megawatt transformer and 6% impedance, your short circuit current is 30 to 50,000 amps. Which my capacitor bank can do, for about 20 microseconds.

[jitter]

[...] those incredibly expensive Bussman KTK fuses [...]

those fuses are only active when the meter is measuring current. when measuring voltage it is just a couple of MOVs and a fusible resistor in circuit. for example, from the schematic for the Fluke 77 series II, see below...


cheers,
rob   :-)

That is very true, and normally DMMs don't use fused probes. And to be honest, I totally forgot about this. But I'm pretty sure that a high enough voltage (maybe not 4 kV, but probably 10 kV and higher) will have no problems destroying those wimpy MOVs and strike and sustain an arc across the pcb and the air where lower voltages would not have the 'reach' to do so.

Fused probes are available, however, and come as standard on power quality analyzers like the Fluke 1775. I still don't think those fuses will be rated high enough when it comes to high voltages. The fuses in our 10 kV grid are about the length of my under arm.

[jitter]

What actually happens to a CAT-III 600V Fluke if you connect continuous 4kV to it?

OK, not continuous 4 kV but 1 ms discharges, our own Dave has made a video in his early days that may be of interest:



This is done at 400 J per discharge and hence looks underwhelming. But you can probably imagine that continuous exposure (and hence sustained arc) is going to be a bit more impressive and maybe those pictures are actually correct for the article, after all.

[joeqsmith]

I would never suggest that the Fluke 101 would survive it without running at least some minimal test

i'd certainly not expect any meter to survive it, but rather to be able to contain/absorb the 'event'.

attached below is the full schematic of the Fluke 77 series II. ...

i'd still be keen to know how a bog-standard Fluke 77 reacted to a 4kV continuous input   

The Fluke 77 is a very old meter.  Fluke and other companies improved their input designs as I mentioned.

[joeqsmith]

...
An average 120/240 system on a larger building has a short circuit current of 10,000 amps depending on the size of the pole transformer and length do the utility drop. My parents house had a short circuit current around 5ka

4000 volt systems are typically fed from 100Kva and larger transformers of 4 to 10% impedance. Lets say a 10 megawatt transformer and 6% impedance, your short circuit current is 30 to 50,000 amps. Which my capacitor bank can do, for about 20 microseconds.

The highest ASCC I measured in our home for the 110 outlets was 0.39kA. 
https://www.eevblog.com/forum/testgear/trashy-meters-redux/msg4835948/#msg4835948

[joeqsmith]

...
OK, not continuous 4 kV but 1 ms discharges, our own Dave has made a video in his early days that may be of interest:
...

When looking at the Fluke 28, he points to the large PTC and identifies it as a MOV, that is basically the front end I would expect to see.   

https://youtu.be/M-FZP1U2dkM?t=663

But is does seem to suggest that a modern Fluke would not survive at 4kV 50/60Hz

...
So, a Fluke rated at either 600 or 1000 V was applied to a continuous voltage of 2300 volts with huge current available.  And it exploded.
...

As I mentioned, Fluke hasn't remained stagnant.  Some of their old products could easily be damaged. 

Looking at my own tests, I ran 1.6kVDC into a modern Fluke for several minutes while monitoring the current.  I think the MOVs were rated to 800 ea.  So now we are only talking about 700V across the PTC and surge rated resistor.  Seems like it might survive.

700V/2500 Ohms or about 280mA.  My DC supplies would all fault out before the PTC would start to heat up. 

[jitter]

DC power supplies are one thing, but the grid another.

A (e.g.) 10 kV distribution ring is protected by relais that receive data on current (and sometimes also voltage) from transformers, so a secondary measurement. When an overcurrent fault is detected, there's a preset time that needs to expire before the protection kicks in and operates a switch to isolate the circuit.
The further away from the main station, the shorter these delay times. This is done to create selectivity in order to have the least amount of customers lose service when a fault occurs. Basically, these delay times are staggered and could end up being as long as 2 seconds but no shorter than 300 ms.

Cables and switchgear must be rated to carry short circuit currents for this amount of time. A 1 ms discharge of 400 J is enough to blow up a DMM, how about 2 seconds of continuous discharge.
These protection relays are there to prevent damage to equipment directly and personal injuries or death indirectly. They cannot protect against someone probing with the wrong equipment.

When you look at the specs of multimeters supplied by the manufacturers, they simply tell you the max working voltage you can use it on. The CAT ratings take transient spikes into account, depending on the use (and that depends on the source impedance, the lower, the higher the cat rating).
I am required to have a cat IV meter for my work. Measuring high voltages (or rather: proving live or dead) are done with capacitive testers like the Pfisterer KP-5 Test on a long stick. Basically a Voltstick on steroids.

[johansen]

The highest ASCC I measured in our home for the 110 outlets was 0.39kA. 
https://www.eevblog.com/forum/testgear/trashy-meters-redux/msg4835948/#msg4835948

At 4000 volts the resistance of your 100 feet 14ga wire is negligible, and the short circuit current is 16,000 amps instead of 480 on 120vac. Ive shorted 120vac outlets out while changing them live, and didnt even see the spark when the breaker trips.

Not sure why you thought to add to the conversation.

As for dmms, ive put 2000 volts across two 10m ohm input dmms and they survived a long time.

[joeqsmith]

The highest ASCC I measured in our home for the 110 outlets was 0.39kA. 
https://www.eevblog.com/forum/testgear/trashy-meters-redux/msg4835948/#msg4835948

At 4000 volts the resistance of your 100 feet 14ga wire is negligible, and the short circuit current is 16,000 amps instead of 480 on 120vac. Ive shorted 120vac outlets out while changing them live, and didnt even see the spark when the breaker trips.

I don't have 4kV running to the house.  240 single phase is common. 

Not sure why you thought to add to the conversation.

Because you claimed to have measured more than 10X higher at your parents home. 

As for dmms, ive put 2000 volts across two 10m ohm input dmms and they survived a long time.

I will assume what you are describing is two meters in series.   That should come as no surprise, assuming the two meters are well matched so the voltage balances evenly.   

[joeqsmith]

I have a capacitor bank which can be configured for 80uf at 8kv or 320uf at 4kv.

Its not enough to shrink quarters but it can bend them, self resonant frequency is 50khz

Please send me expensive test equipment to blow up.

I've actually had people claim this is how I test hand held meters (connecting them directly to a bank of capacitors).   In my case, I shape that waveform without a meter attached.  If I were only charging up a bunch of caps with only the meters at the load, it's more like pushing DC through them (if they don't break down).