Looking for a good transient suppressor circuit?

[FenderBender]

Hi guys,

I've been searching and I've been seeing a whole ton of different ways to go about surge suppression. Is there an "ultimate" surge suppressor? combining all techniques of circuit protection? MOVs, PTTCs, fuses, filters, TVSs, GDTs. All those good things. Any design guides?

Thanks.

[ftransform]

You can take a look at daves multimeter videos but I don't know how well that applies to power supplies and whatnot. I am very curious about this too.

[ftransform]

I would love to see a tutorial for input protection to a equipment transformer...

[FenderBender]

Same. I've been scouring the Internet, but I have yet to find a source that is concrete. Some explain the different types but they don't explain how to actually use them in a circuit. Some have circuits, but no insight onto how to pick the values etc.

[ftransform]

Same. I've been scouring the Internet, but I have yet to find a source that is concrete. Some explain the different types but they don't explain how to actually use them in a circuit. Some have circuits, but no insight onto how to pick the values etc.

Yea. I should do a tear-down on my rather nice power strip in order to demonstrate a somewhat decent mains protection circuit but I'm sure there are tons of tricks and traps involved.

[FenderBender]

Yeah I was going to do the same thing. Take apart a decent surge protector and reverse-engineer it. But...I feel like I'd get something wrong and then KABOOM.

[Kremmen]

I have done multilevel protection schemes for off line converters, industrial machinery and the like. I might have something to say about the matter, but not if i am going to be called an asshole again. Up to you.

[ftransform]

I have done multilevel protection schemes for off line converters, industrial machinery and the like. I might have something to say about the matter, but not if i am going to be called an asshole again. Up to you.

Lol knowledge is power on a internet forum eh asshole?
Either help the COMMUNITY or don't.

[Kremmen]

Just referring to an earlier evaluation by the OP. Otherwise judge yourself based on my posts so far.

But OK, unless someone else manages it first, i can put something together in the evening after i get my day job done.

[madires]

First determine what spikes or overvoltage you expect. Then choose the matching parts and place the coarse ones (which take the large hits) near the source followed by the finer ones (which take smaller hits, but more precisely) towards your circuit. E.g. GDT first, then a MOV and a TVS at the end.

[ftransform]

Thank you. I got annoyed because this is the type of circuit that could perhaps save someones home from burning down if made right. This kind of stuff should be made way more clear.
Then again, maybe you should just buy a good surge protector.

[FenderBender]

Kremmen I apologize for overreacting a bit the other day. I did not like how you responded to my post. You had a different opinion on my post, and perhaps you think I'm the asshole, and that's your opinion, but personally I felt a bit taken aback by your retort so I let you know. But no hard feelings here. I'd hate to dwell on something as silly as that.

Otherwise, in regard to this thread, I'd (we'd) be very excited to hear what you have to say. Madires seems to have a good idea of what to do, but I'm not exactly sure if you just throw these all in parallel with the AC or if there are special configurations. I've seen several designs with all sorts of series-parallel combinations and other things. If you know the "proper" way to do it, we'd love to hear it.

Thank you.

[Kremmen]

Apology accepted and no hard feelings. For my defence i wish to say that my comment was meant to be more humorous than offensive. But the written text does not really convey the emotions like physical presence does and i can see how my comment can easily be interpreted in a different spirit than it was written. So i am sorry as well for the unintended offense and let us leave it behind at that.

I wrote a quick text around some core issues. For presentation reasons i made it into a PDF so that images are more easily included where they are wanted. This must be considered a quick draft - all kinds of errors are possible as it was really rather quickly written. Still i include it in the hope it will be at least a bit useful, and improvements are always possible if desired. So here it is attached.

[FenderBender]

Looks great. Thanks a lot.

I'll definitely give it a thorough read when I'm done with all of my work.

[FenderBender]

Also found this webpage. A bit dated but the information is good.

http://www.data-acquisition.us/sss_7-0.html

http://www.data-acquisition.us/sss_8-0.html

[FenderBender]

I've never seen a dual fused system before. Is that typically not used because it's more expensive and generally ruled out as unnecessary, or what?

I'm slowing learning about all of this.

[beatle]

Thank you, Kremmen! This is very helpful.

[Kremmen]

I've never seen a dual fused system before. Is that typically not used because it's more expensive and generally ruled out as unnecessary, or what?

I'm slowing learning about all of this.

Like i wrote in the text, not all of the features in the example circuit are always needed. Dual fuses are not common but there are cases when they make sense. Some reasons for having them are:
- protection against common mode surges. In some types of installation it is possible for the neutral wire to raise to a high voltage against the earth or PE wire. In such cases the N to PE MOVs need protection just as do the L to PE MOVs.
- Sometimes it is not guaranteed which of the supply wires actually is live and which neutral. For example the Schuko we commonly use in Europe is unpolarized. That means that 50% of the times the nominal L and N wires are reversed. The connected devices are not supposed to care, but if you open the gadget for any reason, there is no fuse protection if you only have one and it happens to be in the currently N wire. Accidentally grounding the currently L circuit (i.e. the nominally N one) will cause fireworks.

In today's consumer world of safety isolated class 2 gadgets and RCDs behind all wall sockets the dual fuse is maybe not that important in a domestic environment. I wanted to include it so that the reader is aware of this possibility and for surge protection in harsher environments it has its place.

[robrenz]

Thanks Kremmen, that paper makes things much clearer.

[MartinX]

Hello Kremmen I  read your paper and it is very informative but the energy calculations at the end cannot be right.

Average MOV power during the 8 us surge risetime is whopping 12 MW resulting in a ridiculous 101 MJ of dissipated energy. And that is just the rising surge without accounting for the decay phase at all. The MOV will be long gone before the surge is over.

12MW for 8µs, I get 96J and that is realistic to handle. Further more I think you have forgot to include the impact of the 2ohm internal resistance of the surge generator, at the high currents we have here the voltage drop over the internal resistance of the generator is substantial and the voltage over the MOV will never reach the value the generator is set to. The voltage will be distributed between the MOV and the internal resistance of the generator. If we take the I/U curve of the varistor in your paper as an example even 10kA trough the varistor will not allow the voltage to rise above 1000V (that curve is not for a small varistor). And the dissipation in a 385V varistor at 385V will be tiny as it is just starting to conduct, it seems to me you did not get out of gas discharge thinking mode when you wrote that last segment .

[Kremmen]

Yep i never checked my calculations after making them like 1:30 am and us vs MW should produce joules now that i think about it. So you must be right.
As to the voltage the MOV sees in real life - it is practically hopeless to try and calculate. Every installation is unique and the indicated resistances and impedances are just those used as "official" values.
The MOV curve was not really meant to represent any single physical device so don't read it with calipers in hand. I drew that curve off the cuff at the same time i wrote the text.
At exactly the limit voltage a MOV will pass a small current but at that point it will also trigger into conduction where the resistance drops sharply. I did the calculations in the defensive way that produces results that are maybe biased towards the overkill. But at least the devices won't evaporate at the first surge.

[Christe4nM]

That's a well written paper Kremmen 
Thanks a lot!