Gas Discharge tube based mains transient protector......upstream of mains fuse?

[treez]

Hello,

We have designed a 40W non-isolated , offline, 230VAC LED lamp.
For transient protection, we have a 470V gas discharge tube (Bourns, 2051-47-SM-RP-LF) in series with a 270V “surge absorber” (Panasonic, ERZVF2M271).

ERZVF2M271 datasheet:
https://industrial.panasonic.com/cdbs/www-data/pdf/AWA0000/AWA0000C4.pdf

2051-47-SM-RP-LF datasheet:
https://www.bourns.com/docs/Product-Datasheets/2051.pdf

This transient “block” sits across live and neutral, just downstream of the fuse. The fuse is the first thing after the mains input connector.

Our contractor tells us that because the Gas Discharge Tube will always fail open circuit, we should therefore move the fuse downstream of the Gas discharge tube based transient protector. –This way, when the gas discharge tube flashes over due to a transient, the fuse will not blow , and so the product will stay functional for longer.
Do you agree? Should we do this?

 

[betwixt]

Place a fuse first is sensible, it isolates everything 'live' in the event of a failure.
Why do you add both a GDT and surge absorber in series though? It gives worst case protection of around 770V with rather unpredictable clamping current. It wound make more sense to use only a suitable absorber with suitable working voltage, or both in parallel. I would suggest:

live ---> fuse ---> inductor ---> GDT/absorber to Neutral ---> live continues to EMC filter.

The initial inductor being to slow down very brief spikes to prevent unwanted fuse blowing.

Brian.

 

[FvM]

I don't recognize the Bourns discharge tube to be suited for power line applications. Why do you expect it to fail open circuit? Without a fuse, I rather expect a kA arc discharge, usually stopped by evaporated PCB traces.

This won't happen with a series varistor, but the unusual circuit has neither a useful protection effect, I fear.

 

[treez]

Why do you expect it to fail open circuit?

Thanks, we expect the Gas discharge tube to fail open precisely because it is gas filled.

- - - Updated - - -

incidentally, we applied some 600V peak transients to the product, using a spike generator, and noticed the Gas discharge tube flashing over (we could see it light up briefly)...We think the GDT was flshing over due to dv/dt because we were below the combined flashover voltage of the GDT and the varistor.

- - - Updated - - -

Without a fuse, I rather expect a kA arc discharge, usually stopped by evaporated PCB traces.

..Thanks, yes, thats what our contractor hopes will happen, ie, the pcb traces blowing will act eventually as the fuse...and hopefully this goes open before the mains supply to the entire installation goes down due to the overcurrent.

- - - Updated - - -

This won't happen with a series varistor, but the unusual circuit has neither a useful protection effect, I fear.

Thanks, so you mean the pcb traces wont blow due to the series varistor thats in series with the GDT?...ie the series varistor will blow open instead before the pcb traces blow.
The PCB traces are 1mm wide (20z copper) and short, because the transient block is (would be) right next to the connector.

 

[FvM]

I didn't yet hear a reasoning for the surge arrestor/varistor series connection.

Regarding usage of PCB traces as fuse, the idea surely doesn't comply with safety regulations. One problem is that an open electric arc might jump over to other conductors, effectively defeating all designed clearances.

A varistor might be designed fail safe under circumstances, refer to the specifications.

 

[mtwieg]

GDTs can definitely fail short, and you must always precede them with a fast fuse.

The series GDT/MOV makes no sense.

Do you have any common mode suppressors?

 

[treez]

The series GDT/MOV makes no sense.

The GDT gives good protection, as it flashes over due to dv/dt. So it has effect at voltages where normal varistors and tvs's just can't do it......we have to protect a 450V ic and this is so close to mains peak (373v) that you cant protect it with any tvs or varistor. Whereas the GDT flashes over as the transient quickly rises up above 400V.

 

[FvM]

The GDT gives good protection, as it flashes over due to dv/dt. So it has effect at voltages where normal varistors and tvs's just can't do it......we have to protect a 450V ic and this is so close to mains peak (373v) that you cant protect it with any tvs or varistor. Whereas the GDT flashes over as the transient quickly rises up above 400V.

Once the GDT ignites, there are essentially two problems
- As the GDT is effectively shorted during the transient (Varc ~ 10V), the MOV must be able to absorb the pulse energy
- The MOV must be able to clamp the pulse voltage

I have doubts that either of the objectives can be achieved.

Finally, I notice that GDT sparkover voltage tolerance of +/- 25% results in a range of 353 - 588 V. The former is below said mains peak voltage of 373V, the latter far above intended protection level.

 

[treez]

Thanks, presumably, when mains peak is 373V, the GDT and the MOV are both "high impednace", and both likely share the 373V between them....say ~186V each.....this is below the flashover voltage of each of them.
The problem would come if they had different off-state impedances.....then at mains peak (373V), the GDT could have say 353V across it and the MOV would then have......the remainder (20V) across it....in this case the GDT could breakover and then the MOV would be clamping mains peak, ...yes....you're right...bang!

I now wonder what is the off state impednace of each of these devices? I suspect the GDT has no leakage current, and so its off state impednace would indeed be higher than that of the MOV, but i dont know how much higher.

 

[Kajunbee]

When you applied the 600 volts to flash gdt were able to get any information about how much energy passed through before it flashed. From what I have read tvs diode is faster than gas discharge. I also noticed the 25% tolerance, and if there's a chance the GDT could go as low as 353 volts why not put a 400v tvs. The GDT just seems a little to unpredictable.

- - - Updated - - -

The datasheet also says the 2051 is ideal high speed/ high bandwidth application. What would be the difference between mains and high speed/high bandwidth applications?

 

[FvM]

What would be the difference between mains and high speed/high bandwidth applications?

Much higher pulse energy rating of a dedicated mains surge arrestors.

 

[Kajunbee]

Been doing a lot of research trying to figure out what makes the series MOV/GDT superior to other methods. To me it seems they are redundancys for one another. From what I've read the GDT works well in signal lines because the circuitry is generally high impedance and low current!
1. Low impedance higher current applications will cause a " follow through current " that will keep the arc maintained. The MOV will help to extinguish the arc quickly.
2. MOV will degrade after repeated cycles and will leak. The GDT is basically and open switch and will not allow MOV to leak and accelerate the degradation under normal conditions. I guess the same can be said for the MOV extending the life of the GDT. Not sure !!!

I'm still unsure whether it protects the circuitry any better though. Even though the MOV is faster it still has to rely on the slower and more unpredictable arc time of the GDT before it conducts. There's probably something obvious for others that I'm overlooking!

Something else that I've found out is the minimum arc voltage of GDT should not be below the normal peak mains voltage.

 

[treez]

When you applied the 600 volts to flash gdt were able to get any information about how much energy passed through before it flashed.

We hit our circuit , which has the gdt/mov combo, ....we hit it with a step input to an LC circuit by suddenly switching the mains on at mains peak, and the ringing in the LC circuit rang up to 600V......the L was 1mH and the C was 1uF.

....this made the GDT flash visibly...for just a fraction of a second.

....The combo should not have flashed over until 353+270V went across it....and the voltage never got that high......so i reckon the gdt must have flashed over on dv/dt....this is advantageous, as it means it de-energises transients as they are rising up....stopping them getting to damaging levels.....at least in cases of weak transients.

 

[Kajunbee]

There is one possible advantage to this that I can see. Since the GDT is normally a open switch you can go with a MOV with a lower clamping voltage. Instead of using the 271 surge sorber you could use the 241 that has a lower clamping voltage.

 

[treez]

Thanks, i can see that the problem is that if the transient occurs at mains peak, then the GDT will flashover and go short...then the transient vanishes after 50us (they are all very short duration)....but the GDT is still short circuit....so then the MOV stays there clamping the normal mains peak......thats going to wear the mov out very quickly. We will have to increase the voltage rating of the mov, ..but then our transient clipping capability gets worse, and we get exposure to higher voltage transients.

What we need is a different circuit but we have no room in the product.

We overlooked that the GDT goes short circuit when it flashes over....unlike a TVS or a MOV.

...woops not, actually the GDT needs 1A to sustain its arc...and it wouldnt get that much current through the series mov when at mains peak...so therefore we will be ok.

 

[Kajunbee]

Not sure if I am understanding the difference between DC sparkover and impulse sparkover correctly. The way I uderstand it is the faster the rise time, the higher the voltage will get before the GDT arcs across. I believe this is based on a 50/60 hertz system. If the rise time in your test setup is less than 50 hertz you would probably expect it to flash at a lower voltage. Maybe someone could verify this for me.

 

[FvM]

Not sure if I am understanding the difference between DC sparkover and impulse sparkover correctly. The way I understand it is the faster the rise time, the higher the voltage will get before the GDT arcs across.

Yes. In detailed application literature, the ignition over voltage is explained with the finite gas ionization speed.

...woops not, actually the GDT needs 1A to sustain its arc...and it wouldnt get that much current through the series mov when at mains peak...so therefore we will be ok.

May be the design is safe under normal line voltage conditions. But does it provide useful protection and perform better than a single (higher voltage) MOV?

I also believe that you don't foresee the dynamic behavior of the series combination, e.g. the effect of MOV capacitance.

 

[treez]

Thanks, for the GDT do you know what the minimum current needed to sustain arc is? (like the 'hold' current)
Also, the minimum current needed to start an arc?
The datasheet doesnt say

 

[Kajunbee]

The datasheet shows 1 amp at 10 vac. I take this mean the lowest v/I to maintain and arc. It also gives glow voltage and current.

- - - Updated - - -

The second question is confusing. There is no current until it arcs so I would think as long as the voltage is high enough it will arc. It would depend on the source as to whether it could deliver enough power to maintain the voltage and current in its shorted state and sustain and arc. But this is just my guess and not based on facts.

 

[treez]

Thanks, so the glow voltage is whats needed to start an arc?
I wonder what the holding current is?