Is MOV from L/N to PE safe?

[poorchava]

Hi there,

we're finishing a design on a product, that can be powered from mains. After initial EMC tests we've concluded, that we need MOVs across L-PE and N-PE against surge and burst events. Now the problem is, that MOVs are known for failing short-circuit and in such event, there will be mains voltage on PE, which is user accessible and this is obviously a dangerous situation. I could not find a single standard, document or any other similar source, that would elaborate on that issue. All the focus seems to go towards fire hazard. I've seen schematic diagrams, where MOVs from L and N are connected to a single point, and that point is connected through a GDT to PE, which would solve the issie, be we hardly have any more space to squeeze in a GDT.

Have you ever had any problems with MOVs between mains and protective earth?

[capt bullshot]

No, I've never had problems with MOV from L/N to PE. There's a simple explanation: Just don't do that and there will not be any problems ...

Back to serious: Check against the applicable regulations and ask a "certified body" (aka VDE / UL / whatever) that does such stuff on a regular basis. In general I'd consider a MOV in this place dangerous and not allowed by the regulations, but there might be exceptions.

[julianhigginson]

crikey..

can you use X and Y caps instead, like a normal person?

how about additional choking/filter?

[poorchava]

There are X and Y caps already.

X and Y caps are generally good for reducing EMI and stuff like burst. For a surge event they don't do anything.

[Huluvu]

No issue  with that
Just put both MOVs from L and N together and decouple it from PE  with a proper gas discharge tube and everything is fine   

[julianhigginson]

fair enough. was not much info on what exactly you had already. though you did mention in the original post that the MOVS were helping with surge AND burst testing.

so what's the standard and clause / test that you're having trouble with, anyway? and what's the test doing to your device?

I've never seen MOVs from live mains to PE. But if I was seriously considering such a thing, I'd be talking to my electrical safety test person about it.

[poorchava]

What is exactly happening is that some internal circuitry gets fried during surge tests when there is not MOVs. I don't know it it's ground bounce or what, but MOVs solved the problem. There are also X and Y caps, because the device is producing massive amounts of EMI and even with those X and Y caps we had a had time passing both conducted and radiated emissions tests.

In company where I work everybody is their own electrical safety guy, so no such person to ask. CE / FCC certification labs don't give a shit as long as the MOV is fused (which it is). We've also never had to add MOVs to PE before, but here the construction of the device is a bit peculiar because of the functionality and we cna;t really change laeyout of PCBs, wiring looms and so on to mitigate surge samage in any other way.

As for GDT: I assume, that in such case, MOVs should be rated at half the nominal voltage and the GDT similarly?

[Huluvu]

As for GDT: I assume, that in such case, MOVs should be rated at half the nominal voltage and the GDT similarly?

It depends on your requirement levels but imho you rate the GTD to a higher level than the MOVs to ensure they trigger only at high voltages (e.g. surges)
This ensures to have the MOVs out of normal operation voltages.

[Huluvu]

No, I've never had problems with MOV from L/N to PE. There's a simple explanation: Just don't do that and there will not be any problems ...

Back to serious: Check against the applicable regulations and ask a "certified body" (aka VDE / UL / whatever) that does such stuff on a regular basis. In general I'd consider a MOV in this place dangerous and not allowed by the regulations, but there might be exceptions.

We never had any issues with the certified bodies 

[NiHaoMike]

Every plug in surge protector I have opened up had MOVs going direct to ground.

[jitter]

Even though the use of MOVs between L and PE and N and PE is not that uncommon, one thing to keep in mind is that it will cause a problem with safety testing. You may want to look into the requirements for your particular device and see if the MOVs present a problem reaching those limits. A device that needs to be tested at 1500 V that is equipped with 500 V MOVs isn't going to pass the test.

[SparkyFX]

Wouldn´t that mean that MOV leakage current will bring the case of the device to the highest voltage in the system (e.g. from a charged capacitor) once disconnected? The current might not be high, but you might be able to feel it.

Which burst or surge relative to PE can be on L or N without being on N?

[rsjsouza]

Every plug in surge protector I have opened up had MOVs going direct to ground.

NiHaoMike, I have a vague recollection this is the case in most (if not all) surge suppressors I have seen as well, but what I can't recall is if the MOVs are placed after a fuse or a Pi filter.

[T3sl4co1l]

Depends on what safety certification you're aiming for. I'm not sure which UL and IEC standards apply to which kinds of products; I do have experience with 60950-1, which applies to ITE (information technology equipment, usually commercial office and server equipment), and requires MOV + GDT to PE.

The best example for which a single MOV (or single GDT) would be a bad idea, is anything with live neutral: accidentally swapped L/N, or L1-L2 of a three-phase system.  These configurations are among those listed in 60950-1.

There's also stuff about whether the ground is absent (insulated equipment), present but optional (cords may be plugged into ungrounded outlets, or ground-lift adapters used), or required (a bonded ground lead for fixed equipment), and what ground currents and faults are permissible in these cases.

Tim

[jbb]

L/N swaps are sadly common. So any MOV etc should be symmetrical around PE.

If you’re making a device with mains power it should probably comply with some standard. See if you have one listed in your product specification.

If you don’t have a product specification, then I sympathise...

[HackedFridgeMagnet]

No issue  with that
Just put both MOVs from L and N together and decouple it from PE  with a proper gas discharge tube and everything is fine   

^ This if I understand it correctly is mostly right.

Put each MOV in series with a fuse hence known as a "Fused MOV".
At least one Fused MOV between each phase and neutral.
Fused MOVs may be paralleled to give greater energy absorption.
The fuses need to indicate they have blown somehow, so people can at regular intervals inspect and replace the shorted MOVs.

Then a gas filled spark gap between PE and N. Needs to be gas filled as otherwise it may not suppress properly when the fault condition is removed.

[Cerebus]

As for GDT: I assume, that in such case, MOVs should be rated at half the nominal voltage and the GDT similarly?

Eek! Don't do that. With the GDT in the non-triggered state it will present a very high impedance (1G minimum is a common specification), leakage current through the MOVs will place the full line voltage across the GDT (more of this latter). Rate it low and it'll fire at the slightest provocation. Once it fires, it goes to a very low holding voltage, a few volts, and the MOV will see effectively the full surge voltage across it.

Generally, to get the GDT to fire reliably it's a good idea to put a highish resistance (say 1M) in parallel with the MOV (i.e. don't rely on MOV leakage current), and in parallel to a speed-up capacitor (suitably voltage rated).

This was from an input protection circuit, but it shows the principal.

[DBecker]

Every plug in surge protector I have opened up had MOVs going direct to ground.

NiHaoMike, I have a vague recollection this is the case in most (if not all) surge suppressors I have seen as well, but what I can't recall is if the MOVs are placed after a fuse or a Pi filter.

The cheapest surge suppressors have a single MOV L-to-N.
The 'good' ones have three in a delta configuration.  Power strips have a circuit breaker / power switch, but the stand-alone ones have no additional protection.

[poorchava]

The device is being certified for IEC 61010-1 - generally T&M and lab equipment. We've went through the standard and there is no mention whatsoever about MOVs other than fire hazard, so they have fuses in series.

I think we are simply gonna built some prototypes, go medieval on them with various transients and we will see.

[rsjsouza]

Every plug in surge protector I have opened up had MOVs going direct to ground.

NiHaoMike, I have a vague recollection this is the case in most (if not all) surge suppressors I have seen as well, but what I can't recall is if the MOVs are placed after a fuse or a Pi filter.

The cheapest surge suppressors have a single MOV L-to-N.
The 'good' ones have three in a delta configuration.  Power strips have a circuit breaker / power switch, but the stand-alone ones have no additional protection.

Thanks for the input. I looked at two clips online that showed a delta configuration interspersed with fuses.

[Zero999]

As for GDT: I assume, that in such case, MOVs should be rated at half the nominal voltage and the GDT similarly?

Eek! Don't do that. With the GDT in the non-triggered state it will present a very high impedance (1G minimum is a common specification), leakage current through the MOVs will place the full line voltage across the GDT (more of this latter). Rate it low and it'll fire at the slightest provocation. Once it fires, it goes to a very low holding voltage, a few volts, and the MOV will see effectively the full surge voltage across it.

Generally, to get the GDT to fire reliably it's a good idea to put a highish resistance (say 1M) in parallel with the MOV (i.e. don't rely on MOV leakage current), and in parallel to a speed-up capacitor (suitably voltage rated).

This was from an input protection circuit, but it shows the principal.

I agree. I'd even go far as to say: never connect a MOV between the earth conductor and phase or neutral, certainly not on an appliance with a mains plug. There's just too much of a risk of the MOV passing a deadly leakage current. I'm aware of the fact that the leakage current isn't an issue if the earth conductor is good, but it shouldn't be relied upon as a routine current path.

[CopperCone]

I think that if you add a gdt in series with the mov you have a longer responce time.

[Cerebus]

I think that if you add a gdt in series with the mov you have a longer responce time.

Hence the speed-up capacitor.

[T3sl4co1l]

I wouldn't bother with the resistor or capacitor, because the MOV has enough of both already (whereas GDT is gigs and a pF or two).  The GDT in and of itself has slow response, because of ionization time; typically for a surge, I think this is 10s of ns, but it can be as much as us for slower rising signals.

So, it will have less effect on ESD and EFT, and more effect on surge.

There will always be leftovers, which your circuit must tolerate.  In common mode, this is handled by filtering and isolation.  In diff mode, this is handled in part by filter resistance, and bulk capacitor size.  The bulk cap should also have low enough ESR, or an X1 film cap be used before the rectifier, to help filtering.

Tim

[Kjelt]

I used to put two MOVs in series between line and neutral and on the midconnection (both movs meet) a sparkgap to PE. Then a fuse in the line and another mov between line and neutral.
That is a common lightning protection scheme.

[TimNJ]

You say internal circuitry gets fried...like what exactly? Part of the power supply? The "actual product"?

If you have a PFC front end in particular, maybe a "bypass surge diode" wold help? basically bypass the PFC inductor and boost diode with an ultra fast rectifier. This probably does not address your problem, though.

https://www.powersystemsdesign.com/images/articles/1519761954_Figure%2003.jpg

[CopperCone]

I think that if you add a gdt in series with the mov you have a longer responce time.

Hence the speed-up capacitor.

Yes but regardless of how fast you can get the gdt to fire it is still much slower then the mov, so the circuit could get whacked

My recomendation is parallel movs with kapton tape and a thermalfuse. This ishow its typically handled in better power strips and such.

Then you can protect your mov by putting a gdt and a impedance between the mov and the gdt to ensure proper firing sequence.