www.beama.org.uk/.../BEAMA-Technical-Bulletin-Connection-of-Unidirectional-and-Bidirectional-Protective-Devices.pdf

Thank-you that's interesting, very interesting. I wonder what the exact mechanism for causing damage to the device is. I'm having difficulty in imagining how arc chutes would be at all bothered if the voltage waveform happened to be out of sync.

   - Andy.

I may be able to offer some clarification - all "electronic RCD" designs have a fairly bosky operating solenoid, and a series triac or similar high power semiconductor that acts as a switch to apply the L-N voltage to it to fire it.

However, to make a coil that pulls in strongly with anything from the worst case drooping supply during fault, to the full 230V+10% needs quite a bit of space - but  the saving grace is that it does not need to be continuously rated, as after 40msec or whatever, the power is removed when the contacts open. (similar economies of rating  apply to the test resistors in some designs, and if for any reason the contacts do not open the resistor does instead...) so a coil  is wound that is OK for 60V steady state or something, and on full mains it is grossly over driven but for an acceptably short time.

If the electronics that operates the solenoid is on the load side, then of course this all works really well, and the trigger circuit can simply come on and stay on, until the power goes off. Except, of course,  that if the supply is on the wrong side it never does go off, and the driver electronics and the coil slowly cook even after the contacts have done their thing.....

It is not too hard to design electronics that removes the trigger signal after a certain interval and that can be relied upon regardless of supply direction, but it is a deliberate design decision to do that, and not one that all designs incorporate.

(The apparently easy way would be to wire the trigger coil in the same way as the test resistor actually so that it gets its live from one side of the contacts and its neutral from the other - but there are also problems with  that sort of design in that it is possible to wire it up in a way that the neutral of the 'dead' side is then still connected to the live via the coil and the trigger after  it is supposed to have tripped. )

Mike.

Thanks Mike - that makes a lot of sense.

I've had a look at the type of RCBO that's on my PV system - looking  at the front  there was no indication of in/out or line/load - terminals just marked 2/N and 1/N so according to the BEAMA document I should be OK. I then noticed a label on the side that seems to suggest that 1/N should be line and 2/N load though (so should I start worrying again?)

Looking closely the test circuit does look to go through a 3rd contact so I guess they're recognised the problem and sorted it.

Anyone got any ideas on the "arc extinguishing / short-circuit characteristics impaired" problem that the BEAMA document mentions?

    - Andy.

Refer to the BEAMA guidance, it shows terminals marked Line and Load in your photo.

www.beama.org.uk/.../BEAMA-Technical-Bulletin-Connection-of-Unidirectional-and-Bidirectional-Protective-Devices.pdf

As opposed to:

I think you should start worrying again. Contact the makers - it may just be a labelism or it may actually matter.

is it the Garo one ?

If so you are out of luck I am afraid.

should have used  the two module width one

Mike

I am getting a bit confused here. The BEAMA guidance refers to both MCBs and RCDs. I can see how an RCD in a split load CU could be run backwards (for want of a better description) but if the fault is before the RCD, it will not trip; and if it is after it, a fault in the consumer/meter/DNO tails seems pretty unlikely.

Why would you connect a PV array to a final circuit which is protected by an RCBO? Is it not a distribution circuit, which accordingly, does not need RCD protection?

What you might do is to put an RCBO downstream of the PV array, but the array would be connected to the supply terminals and the distribution circuit (to the CU) to the load side, so no problem there.

Now think of night time. The supply side of the RCD is not energised. The load side is energised by the grid. So what are we worried about now? Is a fault on the wrong side of the RCD a realistic proposition?

My apologies if I am missing something.

Well if the inverter makers instruction require it, it needs to be protected by an RCD (or double pole RCBO equivalent)...   But part of the problem  of inverters is that they do not generally provide enough over current relative to normal load current to operate conventional ADS in any sensible time, if at all. So for faults to earth, when running on an inverter derived supply, the RCD is needed to meet disconnection times. However, for systems that cannot run as an island, i.e, without grid power, it is not so important, as the grid always provides enough fuse-blowing current.

However, one of the increasingly popular features of inverter PV, especially systems that include a storage battery, is the ability to operate as a generation island, and power essential functions in a power cut. In such a case, the RCD performs an ADS function, perhaps the only ADS function, depending on the strength of the inverter.

Mike

Indeed. I still think there's room for confusion with my unit though.

From the front you could believe it's bi-directional (no arrows or in/out or line/load markings) - although the 1 and 2 notation (which I think comes from a German convention for terminal markings might suggest line and load - on your bi-directional example both L terminals are marked 1/2). But then the label on the side clearly says line and load:

So the marking on the device itself seems contradictory. Personally, noting the 3rd contact interrupting the test circuit, which would seem to make the device immune to the problem Mike described, I'm inclined to believe it is bi-directional, but I'd be hard pressed to argue the case if someone pointed to the label and said otherwise.

   - Andy.

more importantly - the data sheet for it says in bold that the single module one isn't bi-directional but the double width one is - see my screen grabs from two posts above ;-)

taken from this PDF.... www.garo.co.uk/.../uk pricelist 2022 april .pdf

I suspect this is a subtelty of the electronic mechanisms that is only just becoming common knowledge, as text book, passive RCDs and  of course MCBs have no sense of supply and load-side.

M.

more importantly - the data sheet for it says in bold that the single module one isn't bi-directional but the double width one is - see my screen grabs from two posts above ;-)

taken from this PDF.... www.garo.co.uk/.../uk pricelist 2022 april .pdf

I suspect this is a subtelty of the electronic mechanisms that is only just becoming common knowledge, as text book, passive RCDs and  of course MCBs have no sense of supply and load-side.

M.

see my screen grabs from two posts above

apologies - for some reason I didn't see that before I posted (not sure if due to my poor eye sight or the forum software).

I'm pretty sure none of the manufacturer's instructions said that at the time - so might be a change of design or the later recognition of an older problem. I think I've got a 2-module one of a similar vintage so I'll see if I can dig a bit deeper (but probably not the right rating, so probably can't do a simple swap).

   - Andy.

The forum threading is a right pain actually, but that is a side track to the main discussion. I was not trying to tell you off, just to indicate that for GARO there is (in the current datasheets now at least, maybe not earlier ones) very clear advice .

The earlier datasheet has the same terminal drawing you have on the side of your unit but is nothing like as clear - there is not the explicit statement about the line load allocation actually mattering that appears in the later document.

It may be fun to power it from the wrong end and see how much current it draws when open and when shut, and  it may even be fun to trip it but I'd have a hand hovering ready to kill the power if it fails to trip in a second or so.
M.

My enquiry to the manufacturer of my particular device has yielded an interesting reply - while they confirm it is not bi-directional, they suggest that as it's still working after several years of service (I've just run a full set of RCD tests to confirm that), they don't see any immediate safety issue - so basically no need to worry about it - but if I wanted to upgrade to meet current regs they'd recommend their two-module version.

Which sort of leaves me wondering what the problem really is?

(I'm reminded of my enquiries to a combi boiler manufacturer about converting the system to open-vented - they initially said no - when I asked why, the answer wasn't any particular technical limitation, just that they hadn't specifically tested it in that situation so were unwilling to guarantee that it would ... in practice it was of course fine. I might be being unfair though.).

   - Andy.

How often has the RCD ever tripped, and in what direction was power flowing when it did ? I'd not expect any damage under any conditions until it tries to trip. Have you tested it trips OK on the tester both ways round,.
You may be right and it is just an example of stickerism and a lack of understanding by the makers of their own product,

Mike.

As I think that Mike implies, that can perfectly simply be explained by no fault having occurred!

How often has the RCD ever tripped, and in what direction was power flowing when it did ? I'd not expect any damage under any conditions until it tries to trip. Have you tested it trips OK on the tester both ways round

Many times, but only by T button or RCD tester - and almost certainly only when the PV wasn't generating (I wouldn't want to miss out on some of my home grown!). I may well have an experiment, but after I've swapped it for a two module one, so it won't be a disaster if I do fry it. Possibly a saving grace is that the PV system will shut down automatically on loss of main (after a few seconds) so that might limit the potential for burn out too. The 3rd contact in the test circuit also eliminates some failure modes.

   - Andy.