Surge protection for Recom powerline psu.

[HackedFridgeMagnet]

I am just designing a board that needs 12v isolated DC from mains.

Because it needs to consume little power on a low load condition I want to use the Recom module RAC03-12SC/277. I only need about 1 watt power so I don't want to go crazy on input circuitry.

I want to add in any further protection/filtering required for Australian standards but the Recom app note give very little detail.

The images show a couple of PSU inputs that are used but have many components.
https://www.eevblog.com/forum/beginners/looking-for-a-good-transient-suppressor-circuit
Kremmen suggests some circuitry in his 101 pdf. 


I was thinking of something simpler see pic1.png. (But maybe a common mode choke). Considering that this Recom apparently meets  EN 55022 Class B Conducted Emissions.

There is also the possibility of a thermal fuse.

Anyone got a good suggestion as to how to play it safe but simple?
Anyone got anything to say about this series:RAC03-xxSC/277?

[codeboy2k]

that module looks pretty interesting.  It's small, about 2x3x4 cm

It has no connection to Earth ground. So without an earth pin, it cannot have the (2) Y-capacitors inside.  These normally go between Line-to-GND and Neutral-to-GND  (C2 and C3 in your pic2).  The Y caps are to shunt high-frequency conducted noise into the earth ground.  So if you are going to put any additional filtering, I would put (2) Y capacitors on the outside before your module, although it says it meets conducted EMI specs as it is, but it clearly has no Y caps inside.  It possibly has no X cap either, because its likely a design that depends on a choke and 2 DC filtering caps on the high-voltage DC side to make a pi filter that together with the transformer's interwinding capacitance is good enough to meet EN 55022 Class B Conducted Emissions

There is probably no MOV inside for transient suppression.  They don't say in the datasheet.  However, other datasheets of similar modules from the same manufacture specify an external MOV is required for 220 VAC sources only. So I would make sure that you protect the module from transients. It probably has enough EMI filtering built in, but not enough protection from wild transients. So a MOV outside as you've drawn it. Make sure  you put the fuse first, as I've drawn it, not after the MOV.  The MOV is good for short lived transients, but it needs to breakdown and blow the fuse for any longer over-voltages.

Probably in this order, a fuse, a MOV, a single 0.1uF X cap, a 22mH common mode choke, a 1 Meg bleeder resistor, and 2 4700pF Y caps.   This type of filtering is pretty standard, and would be a useful extra above and beyond what the module provides.  For added protection you can put 2 more MOVs after the fuse, between Line-to-Earth and between Neutral-to-Earth, but minimally 1 MOV is needed.

For connecting your earth ground to system DC ground I like to use a ferrite bead.  Some EE's like to use a small valued resistor < 100 ohms to connect the grounds, and still others just connect them directly together.  You may need the isolation and choose to float your ground.  Often times, with isolated grounds,  you will see a Y cap between the grounds (instead of the ferrite I drew).  I don't know if that's such a good idea, it would seem to serve only to couple high-frequency noise from the line to the DC side.  It seems dubious, I don't want that noise on my DC ground.  So when I don't need isolation, I connect the grounds directly for safety, using a ferrite to block HF noise from the AC side reaching my DC ground.

[digsys]

Totally agree with codeboy, I use exactly the same methodology.

[HackedFridgeMagnet]

Thats a great help codeboy. I think I might go with something like that.

I haven't decided whether to ground the secondary, but I probably will as one version will have cabled comms.

The circuit will be in an area prone to surges, so if I go with your additional MOVs I think I need to fuse the Neutral too, is that right?


Sorry Achmed I forgot to mention that pic2 and pic3 were just some random PSU schematics I found on the web, they seemed to me to be overly complex too for a 3W supply.
That is what prompted the question.
I put that app note into my collection.

[jahonen]

You might also want to consider thermally fused varistors (hopefully less destruction when MOV finally self-destructs):

http://www.epcos.com/web/generator/Web/Sections/ProductCatalog/ProtectionDevices/MonolithicVaristors/ThermoFuseVaristors/Page,locale=en.html

Regards,
Janne

[HackedFridgeMagnet]

Funny I saw them for the first time today, in one of the other surge threads.

Do you just use them as you would a normal MOV?

Digikey has them in stock but they are non-stock.

[jahonen]

Funny I saw them for the first time today, in one of the other surge threads.

Do you just use them as you would a normal MOV?

Yes, except for that indicator connection, which can be ignored if not needed. They are just a MOV and a thermal fuse built into the same package.

Regards,
Janne

[codeboy2k]

The circuit will be in an area prone to surges, so if I go with your additional MOVs I think I need to fuse the Neutral too, is that right?

Short lived (transient) overvoltages (i.e. surges) across L-G or N-G would be handled just fine with additional MOVs and without the need of a fused neutral.  If your device is not hardwired (i.e. pluggable) and the plug is non-polarized, then you need to fuse both the Line and Neutral.

Longer lived overvoltages cause the MOV to heat up, short partially, heat up more, out-gas and maybe catch fire. The thermally protected MOVs described by Janne will break the connection (to the MOV) when it overheats, thereby preventing a catastrophic failure.  But the overvoltage condition may still exist, especially if the MOV can thermally pull itself out of circuit before it fully shorts and blows the main line fuse.  So what I am trying to say is, with thermally protected MOVs the plan of attack is different, and you can no longer rely on the Line fuse to blow in the event of a sustained overvoltage event (causing MOV failure, which causes a short circuit).   If you are concerned about sustained overvoltages, you should add something to protect against that.

In North America where I live, we have a 120/240 volt system, which means that a sustained overvoltage can exist L-N or N-G, and can theoretically reach 240V (typically in the event of a Neutral break somewhere between the service entrance and the device being powered).  This overvoltage in and of itself would not harm the device because it's a universal input device and can handle that voltage level.  However, if the device was protected with 150V MOVs, these MOVs would either instantaneously burst if there was nothing to limit the current, or they would be current limited by devices on the other leg of the split 120/240 V system. In the current limited case, they would slowly heat up, out gas and possibly catch fire if they were not of the thermally protected variety. A 150V thermally protected MOV across a 120V line that has a sustained overvoltage would eventually heat up to the point where it takes itself out of circuit, without catching fire, and that is good. However, the overvoltage still exists. If the sustained overvoltage is a concern than it needs to be considered. I would likely use a triac across Line to Neutral, with an RC time constant of 10 ms or so at the gate, to crowbar the ac source and blow the Line fuse.  The choice of 10ms should be longer than any transient overvoltage, so as not to blow fuses willy-nilly, but short enough to be effective.

In summary, one fuse on the line is good enough, unless your line is unpolarized, then you need 2 fuses.  Use thermally protected MOVs (they are great!) and protect additionally against a sustained overvoltage if it's needed.

[David_AVD]

For the 1M resistor, do you use something like a Vishay HVR37 type?

http://www.vishay.com/docs/30260/hvr25.pdf

[HackedFridgeMagnet]

Thanks codeboy for the info, well explained too.
We dont have universal plugs in Aus so I will go with just the one fuse on the phase. I have done the schematic but not the layout yet.

One thing that I dont understand is what happens if there is a surge that shorts the MOV between Neutral and Earth.
I understand that MOVs normally fail short, if they don't burn.
So we are left with a some sort of short circuit between the Neutral and the Earth. Is this what would happen? Is it considered a problem if undetected?

[David_AVD]

One thing that I dont understand is what happens if there is a surge that shorts the MOV between Neutral and Earth.
I understand that MOVs normally fail short, if they don't burn.
So we are left with a some sort of short circuit between the Neutral and the Earth. Is this what would happen? Is it considered a problem if undetected?

I'm pretty sure that will usually cause an RCD protected outlet to trip.

[codeboy2k]

For the 1M resistor, do you use something like a Vishay HVR37 type?

http://www.vishay.com/docs/30260/hvr25.pdf

That's an appropriate component for a very safe system, yes. It has a flame retardant coating and is pulse withstanding up to 7KV.  Whether or not you need it depends on what standard you are trying to meet.   I'm not up on all the standards requirements as much as I used to be, and they're changing and/or having decisions being published on them all the time. Specifically, I've used regular carbon film/metal film resistors here without any problem.  In this application, the MOVs should shunt any pulses before the resistor sees it, but really it depends on what standard you are trying to meet.

Also, if I was putting this component near another component that might catch fire, I'd probably at least use a flame retardant resistor.  They are about the same price for extra protection, and that's good in my book.

[codeboy2k]

One thing that I dont understand is what happens if there is a surge that shorts the MOV between Neutral and Earth.
I understand that MOVs normally fail short, if they don't burn.
So we are left with a some sort of short circuit between the Neutral and the Earth. Is this what would happen? Is it considered a problem if undetected?

I'm pretty sure that will usually cause an RCD protected outlet to trip.

Yes it should, but then you are depending on the RCD / GFI  for protection. Actually, in practice, short lived transients are handled just fine by the Neutral to Earth MOV.  If this is a thermally protected MOV it will disconnect itself if the current is sustained long enough that it overheats. Once that happens, there is no more short circuit Neutral-to-Earth, so it's not a problem. But if the Neutral-to-Earth voltage differential persists you've got bigger problems elsewhere in your power distribution network.  In this case, your Line will likely rise above it's nominal voltage, and trigger an over-voltage protection if it's there. If there is no overvoltage protection, the other thermally protected MOVs will slowly cook with a slight overvoltage, and disconnect.  But the overvoltage persists unless it's taken care of (if it even needs to be).

[HackedFridgeMagnet]

I did some talking to another engineer about surge protection in Australia. He is very experienced and knowledgeable on this subject.

He suggested only something like this diagram.

the reasoning is:
the noise suppression is done in the Recom device.
the common mode choke will saturate and be useless in a surge.
the neutral and earth in the MEN (TN-C-S) system are shorted together very close to where the device will be. So any caps or movs here are not going to do anything surge wise.

The MOV should be rated at 50kA which should be ok for most surges.
380vac is a worst case scenario for a broken neutral on the Australian grid (230-240vac) so a nominal operating voltage for the MOV should be just that.
The roll off point should be around 1000Hz for the filter.

I am not sure why a common mode choke would saturate before a single line choke, it probably doesn't but I am not sure on this.

I think the point is that the choke is not necessary on the neutral for surge protection if the MEN link is nearby.

I am also not sure how to size the inductor, so it doesn't saturate. I will look into that.


 

[codeboy2k]

Yep. It's all down to how much safety you need, and can budget for. 

I wanted to add one more point about the 1M bleeder resistor.. 1Meg is a common value that works for up to 1uF of capacitance across the mains.
If you are looking for a lowest quiescent current design in standby, then IEC-60950 — Chapter 2.1.1.7 “discharge of capacitors in equipment”, says that the bleeder resistor should be sized so that the X cap across line to neutral has a time constant of < 1s ... so tau = RC; 1s = RC, R = 1/C , so for C=100nf, as in my example,  R <= 10Meg ohm. I would choose 10Meg bleeder if I wanted the most efficient design.

[_spirit_]

I am also not sure how to size the inductor, so it doesn't saturate. I will look into that.

@HackedFridgeMagnet, did you find the right way to size this inductor? 

[HackedFridgeMagnet]

The short answer is no, I guess to get a 3db at 1000hz you need to go big on the inductor, and cap.

The reasons being, the client baulked at the cost of surge protection as he worked out he could pass that responsibility onto someone else.
The Recom got ditched, at the suggestion of the manufacturer who said he could do it cheaper, he was wrong.
He delayed the project for 6 months with his design, I wasn't in a position to argue as he is a legit psu designer.
Anyway the revised project is in the field and last I heard fine.
The design ended up with a purpose wound transformer isolated to 8kv rms, due to the compliance guys.
8kv rms why do they even test with that? why rms not a surge.