Weird?! Neutral-earth fault

[ADelectrical]

Hi was recently at a property and was installing a Solar PV system. was doing my testing and i came across a weird result when doing the insulation resistance stage.

testing between Neutral and earth I got a dead short. so split the circuit and narrowed it down to a 1ft long length of cable bewteen the DB and a generation meter. everything else was clearing no issue but this seciton was still dead short.

now heres the thing when testing just the cable i.e taken out of the bus bars. the cable cleared but when they where connected in it was a dead short.

at the time this was on the RCD side of the a split load board. the solar was using and MCB and the RCD had been changed for an A type the neutral was installed on the RCD neutral bar and the earth was installed in the rest of the earth bus bar.

any ideas what might have made this happen?

in the the end I installed the solar circuit onto it own RCBO and the fault was cleared but would like to know any suggestion as to why i might have been

 

[timhoward]

Welcome to the forum.
Especially as you are only up the road from me - I'm in Oswestry.

at the time this was on the RCD side of the a split load board. the solar was using and MCB

Best not to do that. If the RCD trips, the solar can still feed every other circuit on the RCD side until the inverter notices the lack of incoming mains, which I've known take 500ms.
So I'm glad you went for the RCBO approach in the end.

testing between Neutral and earth I got a dead short. so split the circuit and narrowed it down to a 1ft long length of cable bewteen the DB and a generation meter. everything else was clearing no issue but this seciton was still dead short.

Forgive me asking the obvious question - what was the earthing arrangement? TNCS?
And was your RCBO single pole or switched neutral?

 

[ADelectrical]

It was on a TNS earthing arrangment and the RCBO was double pole.

Thats good to know about the inverter potentially back feeding into other circuits wasn't aware that could happen. I normally put everything on RCBO or install its own RCBO consumer unit. but in this case the guy was replacing his board and his only spare ways were on the RCD side.

 

[timhoward]

It was on a TNS earthing arrangment and the RCBO was double pole.

Sorry - it was too obvious not to ask!
So you basically had 0 Mohms between CPC bar and the N bar for that side.
But the N and CPC for the PV circuit was clear.

Some random possibilities...
1 - N-E fault on another circuit, normally 0.00 Mohm would mean instant RCD trip unless it wasn't working?! Did you happen to try a continuity test too?
2 - bonding to services effectively created a link to another property which is TNCS; RCD and main switch would both have to be on/closed for this to work.
3 - for whatever reason a mistake was made identifying the earthing type. Do you remember if PSCC and PEFC were the same?

 

[Lucien Nunes]

N-E fault on another circuit, normally 0.00 Mohm would mean instant RCD trip unless it wasn't working?!

Not by any means a certainty. A reading of 0.00 MΩ displayed on an insulation test could be as high as a few kilohms, or over over 10kΩ if you allow the +/- one count. E.g. 5kΩ= 0.005MΩ. With the few volts likely to exist between neutral and earth, the resulting leakage current is likely to be less than a milliamp.

 

[rewireIT]

Best not to do that. If the RCD trips, the solar can still feed every other circuit on the RCD side until the inverter notices the lack of incoming mains, which I've known take 500ms.

Do you mean you're not supposed to RCD protect solar if the RCD protects other circuits?

 

[timhoward]

Do you mean you're not supposed to RCD protect solar if the RCD protects other circuits?

Yes, it’s not a good design. It’s common to find it installed like this unfortunately. It can increase disconnection times after the RCD trips.

 

[Risteard]

Do you mean you're not supposed to RCD protect solar if the RCD protects other circuits?

The RCD must not be shared with other circuits. If RCD protection is required then it still must be fitted.

 

[davesparks]

Do you mean you're not supposed to RCD protect solar if the RCD protects other circuits?

Definately not, this is potentially lethal!

A solar system is a source of supply, it effectively generates power.

If a person comes into contact with a live part and receives an electric shock then an RCD will disconnect the supply before that shock becomes fatal.
But if the outgoing side of the RCD has a second source of supply connected to it, such as a solar system, then the RCD will only disconnect the DNO supply whilst the solar continues to generate power and kill the person who would otherwise have survived.

 

[davesparks]

The RCD must not be shared with other circuits. If RCD protection is required then it still must be fitted.

Surely it would need to be fitted at both ends of the circuit if RCD protection is required otherwise the solar will still maintain that circuit in a live state if the RCD trips.

 

[timhoward]

Surely it would need to be fitted at both ends of the circuit if RCD protection is required otherwise the solar will still maintain that circuit in a live state if the RCD trips.

Grid tied inverters are required to shut down if the grid supply fails. But they don’t always do it quickly.

 

[davesparks]

Grid tied inverters are required to shut down if the grid supply fails. But they don’t always do it quickly.

Exactly my point, if a person receives an electric shock the RCD trips within 200mS but the.circuit remains live for another 100mS then that could be the difference between it becoming an anecdote or a funeral.

 

[timhoward]

Exactly my point, if a person receives an electric shock the RCD trips within 200mS but the.circuit remains live for another 100mS then that could be the difference between it becoming an anecdote or a funeral.

We are in agreement - that is why I raised this point back in post #2 and OP took it on board.

 

[Risteard]

Surely it would need to be fitted at both ends of the circuit if RCD protection is required otherwise the solar will still maintain that circuit in a live state if the RCD trips.

I certainly don't disagree with your logic, but it doesn't seem to be done this way.

 

[Risteard]

Grid tied inverters are required to shut down if the grid supply fails. But they don’t always do it quickly.

I seem to recall that the anti-islanding protection operates within about 5 seconds (could be wrong about exact length of time allowed), so certainly it would be well beyond the permissible disconnection times of an RCD.

 

[timhoward]

I seem to recall that the anti-islanding protection operates within about 5 seconds (could be wrong about exact length of time allowed), so certainly it would be well beyond the permissible disconnection times of an RCD.

It seems to vary a lot between inverters. The first time I encountered this I’d been working on a circuit and did RCD tests which failed at about 450ms. ( There was no signs of PV at all at the CU and no two sources of supply sticker which didn’t help)
I’ve known other ones (Solar edge) be very snappy but it still is against the regs and an instant C2.

As you say it’s more normal to split the tails and feed from own CU. I think RCDs both ends would still contravene the reg that prohibits a shared RCD and multiple sources of supply but I’m not looking it up tonight so could be wrong.

 

[rewireIT]

Wow you guys, i've seen hundreds of installs just installed directly on the RCD. They try to do them on RCBO's but sometimes they would bang it in on the RCD because it was a lot of council work where the council wouldn't pay for an RCBO board, so they'd just buy a breaker and match what was already there. Will be giving the company in question an email later to explain this to them.

So for future reference - solar has to be on an RCBO or have its own dedicated RCD?

 

[timhoward]

Wow you guys, i've seen hundreds of installs just installed directly on the RCD

I see it all the time too, and frighteningly companies are still fitting them like this. It tends to be when there isn’t room for another board or a board change would be difficult, and it’s never a straightforward fix but I always find some way.
The biggest problem is usually a 5 way rcd main switch board on TT earthing in a tiny cupboard.
Yes, either installed so RCD protection isn’t required, or an RCBO, or split tails and own CU.
So NOT ever sharing an RCD. I’ll dig out the reg this afternoon.

What gets me particularly is that the cost of PV is high and sorting out the CU isn’t significant in the context of materials for the whole job.

 

[rewireIT]

I see it all the time too, and frighteningly companies are still fitting them like this. It tends to be when there isn’t room for another board or a board change would be difficult, and it’s never a straightforward fix but I always find some way.
The biggest problem is usually a 5 way rcd main switch board on TT earthing in a tiny cupboard.
Yes, either installed so RCD protection isn’t required, or an RCBO, or split tails and own CU.
So NOT ever sharing an RCD. I’ll dig out the reg this afternoon.

What gets me particularly is that the cost of PV is high and sorting out the CU isn’t significant in the context of materials for the whole job.

Awesome thanks for the explanation - like the other poster i've done tonnes of these installs and just done them as instructed by the main contractor. Will refuse to do that from now on. Should have been a red flag when said company's 'QS' found a fault on a circuit and fixed it by changing the MCB type from B to C without doing a single calculation.

 

[timhoward]

Regs-wise I think the most relevant area is section 551 (even though it's called "LV Generating Sets" it includes PV cells according to 555.1.1). Reg number and my attempt at a plain English version below:
551.4.2 - any RCDs anywhere in the installation needs to work properly whether the supply is from grid, PV, or both. (This can't be assumed to be the the case with a shared RCD)
551.4.1 - fault protection needs to work properly for either supply (relevant in the case of an inverter messing up disconnections times with TT earthing)
551.7.1 (ii) - if the final circuit to the inverter needs RCD protection it needs to disconnect live and neutral so if an RCBO is used it needs to be SP+N not single pole.
551.7.2 - It's very simple if the inverter is connected on the supply side of the final circuits. But If connected on the load side the OCPD for a final circuit the requirements are more complex regarding conductor size, and ensuring the disconnection time for the final circuit still complies, taking into consideration the time taken for the inverter output to fall to under 50v.