Earthing question.

[The Natural Philosopher]

Due to the fact that apparently the 11KV that feeds my house has got a short in it, I finally got a look inside the cabinet of the 'all me own' substation that feeds the house.

Basically I get power down a coaxial cable, the outer of which is strapped to neutral and earthed via a bloody great earth stake in the cabinet.

Now for historic reasons, I also have a bloody great earth stake outside, with the house earth tied to it.

I cannot remember whether or not this earth is connected to the neutral at the consumer unit..

I am getting - now I am on a generator, generously supplied by EDF, also with te neutral earthed vua another bloody great stake in the ground,

- a lot of nuisance RCD trips on the main 100mA house RCD.

I have never understood how one should earth a house properly, or if mine is to standards, or why one way is better than another.

Anyone care to educate me with the REASONS why one way is correct, or not?

[meow2222]

[The Natural Philosopher]

think..in which case my earth should NOT be neutral connected at the consumer unit..

Not sure whether it is or not.

If it is, I'll have to disconnect it.

[Andy Wade]

Why would you want to do that? You appear to have a PME supply (TN-C-S) which has the benefit of providing an all-metallic earth fault current path. Why change to TT?

There should not be any neutral-earth connection in the consumer unit

*ever*. On the consumer's side of the meter earth and neutral must be separate - that's the S in TN-C-S. On the supply side earth and neutral are the same conductor - that's the C (for combined). The point of separation is the sealed PME earth terminal inside the supplier's cut-out (the main fuse assembly, sometimes called supply head).

The fact that you're getting RCD trips indicates that something is wrong in the installation or an appliance, possibly a neutral-earth short or low-resistance path. Either that or 'legitimate leakage' from L to E due to capacitance in the wiring and EMC filter capacitors etc. is bringing the RCD close to its trip threshold - in this case subdivision of leaky loads between more RCDs is the best solution.

[meow2222]

IIUC your mains supply is/was PME, which is safer than TT. When running off a gen things are different, gens are normally IT or TT. Either way I dont see any need to unbond the two earths.

RCDs normally only monitor downstream L&N wires, so questionable upstream earthing cant upset them AFAICS. Hence there must be a fault in or after your rcd, not before. And guess what:

[The Natural Philosopher]

I do not think its PME as such. I laid the cable from the transformer to my house myself. It is well insulated and armoured, and IIRC the inner sheath is not connected to the armouring, which is purely mechanical.

This transformer replaced a transformer on a pole, which had no earth at all, so the house had been provided with a stake and earth stuff bonded to that.

When the house was rebuilt the existing installation was kept..it is the only part of the old house that was..until construction was almost complete, and it was moved into a new meter cabinet, and reconnected. I cant remember what the useless electrician said at that point, but it was something about 'not sure about the earth here' anyway.. they got their marching orders for being useless and I finished off the wiring myself.

Yes. That I know. I am just not sure if its bonded at or close to the meter or not.

On the consumer's side of the meter earth and neutral must be

Or a mixture of all of the above.

I am wondering whether a clamp ammeter on the actual earth wires would be sensitive enough to identify which are carrying the currents.

[John Rumm]

A photo of the main cutout or box on the end of the incoming service cable would be handy...

Failing that:

TT no earth wire anywhere near the cable head end.

TN-C-S earth wire connects to the head end directly - often next to the neutral tail, but sometimes to an earth tag on the side.

TN-S cable often connects to a clamp on the incoming cable at the base of the head end.

Yup there are such beasties designed for the purpose, note that the fault currents may not be flowing in the earth wires though - they could be finding an alternative path. Measuring the current imbalance between line and neutral will find either. See the wiki RCD article for more info on locating the cause of nuisance trips:

[Fred]

Noisy mains or spikes from the generator may cause some issues with RCDs. If all the appliances are disconnected does the RCD still trip?

Nuisance trips often are a result of suppression caps in equipment. The strangest I've know is a power factor correction cap in a fluorescent tube, even when the body was isolated from the earthed fitting. Removing the cap, though not ideal, solved the problem.

[Andy Wade]

Is this is the cable you referred to before as "coaxial"? "Concentric service cable" is the usual description. The inner line conductor is solid aluminium - 35 mm^2 for a 100 A service - and the outer combined neutral & earth is stranded copper totalling 25 mm^2. It doesn't have armouring, AFAIK. ("Split concentric" where the outer strands are divided to give separate neutral and earth paths is used for TN-S services.)

That sounds like protective neutral bonding (PNB) which is a variant of PME used where a transformer supplies only one customer. Electrically it's the same as PME in the sense that the service cable is 2-core, but the neutral is earthed only at the load end and not at the transformer end. Arguably it's safer than PME since a broken neutral on the service doesn't leave your earthing floating up at phase voltage.

[RCD trips]

See John's reply. With a little improvisation you can increase the sensitivity of a clamp meter by a factor of N by passing the circuit under test through the jaws N times. N=10 is very convenient if there's room. . .

[John Rumm]

Oddly I was going to say that in my initial reply, but decided to delete it to reduce complexity ;-)

For doing tests on things which can be plugged in, I have a short extension lead that has the outer stripped away an the wires coiled and cable clipped into bunches of ten turns. Hence it is easy to clamp round a set. I use it usually for reading running loads of kit when sizing UPS' etc, but it might be worth experimenting to see how good it is clamped round a pair of coils sensing imbalance... if I can find the lead I will get my RCD tester out and have a play.