RCDs not operating as expected

[vwes59]

Hi
Ive looked around and i cant see this particular scenario being discussed.

I think i may have been a numpty, but bear with me.

I have just completed an electrical install into an outbuilding. Its distance meant i went with installing its own earth rod, thus not exporting the pme from the house. Installed a 2 core SWA, earthed back to the main house via its steel sheath. Thus the sub board has no earth connection back to the main house and is essentially TT.

All circuits in the main house including the new SWA are covered by a single RCD type AC. Circuits in new sub main are RCBO Type A.

All dead tests were fine, and earth rod came out at 56ohms. I confirmed there is no earth path from the main house into the sub main, it stops in a plastic junction box outside the consumer unit.

However when i tested the new RCBOs, sometimes they tripped, other times the main RCD in the house tripped. All the RCD tests passed, just it was random weather the main RCD or RCBO would trip.

Given i have separate earths i am surprised selectivity is an issue if that is what it is.

 

[plugsandsparks]

The main board is reacting to the imbalance in L and N from your tester. Why is the sub board fed from RCD on main board?

 

[vwes59]

Unfortunately the main board is covered by a single Type AC RCD, so i added an MCB to cover the SWA thinking the RCD would cover that cable. I didnt think for a second that even with earths separated that selectivity would be an issue. If that is the case then ill have to put in a 100ms time delayed RCD in the main board

 

[Julie.]

That's operating exactly as expected, you have two rcd in series, so both will see any imbalance between the line and neutral on that circuit.

If you have a sub board fed by swa, just use a mcb at the supply end

 

[Julie.]

Unfortunately the main board is covered by a single Type AC RCD, so i added an MCB to cover the SWA thinking the RCD would cover that cable. I didnt think for a second that even with earths separated that selectivity would be an issue. If that is the case then ill have to put in a 100ms time delayed RCD in the main board

You can’t do that, as if there is a single rcd at the main board this will be providing additional protection for all circuits within the house - which needs to be 30mA instantaneous, not time delayed.

 

[vwes59]

Yes you are absolutely right, must be this heat, my brain is swiss cheese right now. Surely there must be a way that i can ensure a fault in the outbuilding will not cause the RCD in the main house to trip but only the RCBO on that circuit, or is my design fundamentally flawed ?

 

[timhoward]

It may help to think that your MFT is 'creating' a 30ma load from L and N probes and leaking it to the E probe.
So if you follow the L and N right through the installation from the origin, through the Type AC RCD, down the SWA, through the type A, then you'll see that both RCD's will notice if some of it goes missing. (as explained above)

I think you have two options
1 - change the house board to an RCBO board, and use an MCB for the SWA. It doesn't need RCD protection as it has an earthed metalic covering and is mechanically protected.
2 - split the tails and install a switched fuse for the SWA

 

[plugsandsparks]

Just re-engineer it , there are options. Change main board config so you have the options necessary, if you really want to RCD protect armour, the RCD will need to be dedicated to that cct and then you can time delay it.

 

[mossep]

Assuming your sub-main doesn't require an RCD for fault protection, split the tails before the house board and install a small switch fuse to feed the out building. Having a sub main backed up my an MCB is unlikely to give you much in the way of selectivity if something goes bang in the out building, so feeding it by a switch fuse with a 1361 in it would probably be better anyway.

 

[loz2754]

Given i have separate earths i am surprised selectivity is an issue if that is what it is.

This is where some folks get in a bit of a muddle, and forget that RCDs do not require a connection to earth to function correctly.
As stated by @plugsandsparks, an imbalance between L and N is what the RCD reacts to, whether the imbalance is caused by leakage to earth or to anywhere else.

 

[vwes59]

Thanks for all the replies. Indeed I agree with all, complete brain freeze and forgot my original reasoning for splitting the earth was nothing to do with the RCD's, but to avoid using the PME over such a large distance to the outbuilding. The complication with trying to work out a solution, is the board has a combined main switch/RCD (For info its a Hager CDA 263U 63A/30mA in a Hager VML406AH board) which is giving me pause for thought on how to rewire the board to isolate the MCB for the outbuilding. I've noticed there is space for a 3rd terminal block at the top of the board, which would give me 2 neutral bars and an earth bar, but not sure that would actually help in this case.

Now I've reviewed Napits codebreaker book and having "Selectivity not achieved" is recommended as a C2 if safety is a concern, or a C3 if there are no safety concerns (regs 531.3.2 and 536.4.1.4). Given all my RCD tests passed I would give this situation a C3, but it's still not great for the client, so I do need to get a solution together of some kind.

 

[pc1966]

If enough free slots in the CU then a main switch and MCB could be fitted for it, alternatively splitting the tails and a fused-switch is another option. In both cases you would need to check the end of sub-main Zs (i.e. L to armour, not the TT case in the out building) is low enough for 5s disconnection time though.

 

[timhoward]

If enough free slots in the CU then a main switch and MCB could be fitted for it,

It's only a 6 usable way board, and as I was just typing....

The complication with trying to work out a solution, is the board has a combined main switch/RCD (For info its a Hager CDA 263U 63A/30mA in a Hager VML406AH board) which is giving me pause for thought on how to rewire the board to isolate the MCB for the outbuilding. I've noticed there is space for a 3rd terminal block at the top of the board, which would give me 2 neutral bars and an earth bar, but not sure that would actually help in this case.

....Ultimately I don't think this approach gets you far.
You can't do anything unless you change the main switch for a normal isolator.
Then you'd need to change the existing circuits to RCBOs', or at least sockets and lighting, and that assumes you know how the others are run. Realistically it will be all of them.
So you'd end up replacing every device in the board at a fair cost and end up with selectivity of RCD's but as pointed out using an MCB wouldn't be first choice as it would likely trip as well as the downstream OCPD devices. That is more liveable-with though.

Do you have any space to split the tails as this is by far the better option; it would certainly be cheaper and save messing around with the existing board.

Indeed I agree with all, complete brain freeze

I found a TT house this morning with an all RCBO board except an MCB sub main feed to a 2nd new board for the heat source pump. Two new boards. No fault protection at all on the heat source CU. Brain freezes happen - this was a decent large firm.

 

[Lucien Nunes]

When one is stuck with multiple RCDs for whatever reason, it is possible to test only the downstream RCD without risk of tripping the upstream one, by connecting the earth lead of the tester to the neutral on the line side of the RCD under test. This returns the test current to the circuit so that the upstream device does not see an imbalance.

 

[davesparks]

to avoid using the PME over such a large distance to the outbuilding.

Why? What's wrong with using a PME earth over a long distance?

Now I've reviewed Napits codebreaker book and having "Selectivity not achieved" is recommended as a C2 if safety is a concern, or a C3 if there are no safety concerns (regs 531.3.2 and 536.4.1.4). Given all my RCD tests passed I would give this situation a C3, but it's still not great for the client, so I do need to get a solution together of some kind.

This is irrelevant as that is for EICRs and not new work, those codes all indicate sometjing which does not fully comply with the regulations and new work should comply with the regulations.