Lighting circuit short

[Jeff Layman]

I went to turn on a bedroom ceiling light last night and it didn't come on. Nor did some other ceiling lights. The kitchen lights are on a different circuit, and they could be switched on. The CU is in the kitchen, and it was obvious from it that the lighting circuit MCB had tripped. I reset it, expecting it to trip immediately, but it didn't. About 20 minutes later, it tripped again, this time taking the RCD on that side of the split CU with it. I left the MCB off, and reset the RCD.

This morning, I switched off the two permanently-on PIR security lights in case it was one of those causing the trip in the lighting circuit. That meant all the lights controlled by that MCB were off as their switches were off. I reset the MCB, and once again it didn't immediately trip. An hour later, the MCB and RCD tripped again! I left the MCB off and reset the RCD.

Any ideas why there should be such a delayed trip? And why did the first trip (which could have happened hours before I tried to switch on the light) only take out the MCB, and not the RCD?

I think this will have to be a job for an "Adam" to sort out. The lighting wiring here is strange to say the least. As the place was built in 65/66, half is earthed and half isn't (the regs changed). As I've mentioned, the kitchen wiring is on a different circuit, and I found out last night that our en suite lighting was not affected by the MCB trip. It is connected to the kitchen circuit!

[Alan Lee]

Turining off the switch, unless it is double pole, does not open the neutral contacts, so if you have any water between Neutral and earth, it will still trip.

[nothanks]

As a first step I would completely disconnect the supply (L and N) to anything that is outside.

[David Wade]

When I had similar issues it was water ingress. The water was soaking through capillary action into a taped joint (made by the previous house owner). When there was enough to create a short the current vaporised the small amount of water tripping the MCB and not the RCD as it wasn't reaching the earth.

If the wiring is as dodgy as you say it could be your switch for the external PIR only switch neutral not live....

... I would certainly be checking for water ingress in all the external circuits....

Dave

[Jeff Layman]

Wouldn't the MCB trip be an overload (I assume live to neutral), and the RCD be a live-earth or neutral-earth fault?

I'd forgotten about the bathroom fan being on the lighting circuit and disconnected that before resetting the MCB. This time, the MCB tripped immediately, and so did the RCD.

I'll get an electrician in with decent testing equipment.

[Jeff Layman]

That's only the security lights. Even so, I can't get at the cables easily, which are in the loft near the eaves (awkward enough at the best of times, but with a dodgy knee...).

[Rod Speed]

Likely something fried in a junction box or lamp socket or even just mouse having pissed in it etc.

Depends on where the remains are wire wise.

If you can look in the junction boxes involved, it will normally be obvious.

[Animal]

yup, suggesting a large L to E current. If you don't know a great deal, looking in the accessory boxes might show you the problem. With power off of course.

[John Rumm]

A MCB trip implies a high current load rather than just a current leakage (there may be that as well - but that is obviously not the main issue).

If a RCD is already sensitised, then a large transient (like a MCB cutting the load of a whole circuit) could be enough to cause it to trip as well.

That the MCB trip is not instant - e.g. tripping as you try to reset it, suggests you are not looking for a hard short. However, sinking enough current to trip a MCB (A B6 MCB would take a constant ~9A for probably

20 mins to trip on the thermal part of its response curve) - does not sound like a damp socket either but a more substantial L to N load. Perhaps something has crawled into an accessory box and died!

So you have a potential fire risk by the sounds of it.

I would use a clamp meter round the live wire from the top of the MCB in the CU, to look at the current draw on that circuit. Chances are it will be abnormally high. You could then turn the power off, and break the circuit at the mid point (in a ceiling rose for example) and see if the fault goes away when you re-test. If it does, you know it is in the second disconnected half of the circuit, or if not, then in the first half.

Keep sub dividing until you find the culprit.

[Jeff Layman]

Yes, that was my main consideration, so now I'm leaving the MCB off. Perhaps originally an insect shorted across terminals in a box and that led to insulation damage which has now shorted. Whatever, the tripping is now instant for both MCB and RCD.

I'm afraid I'm past crawling around the loft - particularly above where the CU is in the kitchen. It's not even 60cm from the corner below this bungalow's hipped roof eaves. The CU was replaced 10 years ago, and I remember the electrician connecting up all the wires taken from the old unit. It's possible that the faulty lighting circuit has more than one cable going into the CU, so that would help in starting to isolate the short. But then it would need someone quite agile to follow that cable from where it enters into the loft and spreads out. Annoyingly, there are lights in the loft, but I think you can guess which MCB they are supplied from!

So it makes sense to get an electrician in. I'll firstly try the guy who installed the CU; his company is still around, but I think he must be retired by now.

[nothanks]

a thought ... you could try replacing the MCB, or swapping it with another of the same rating from within the CU, as a quick and cheap way to eliminate that as a problem. Also, if you find there are several wires going to the MCB, you could experiment to see whether leaving one out stops the trip.

[Jeff Layman]

That's an option, but only with a new MCB. There is another MCB in the CU that I could try, but in the admittedly very unlikely situation of a live-neutral short /and/ a faulty MCB which doesn't trip, the result could be a potential fire. And, of course, unlike RCDs you can't test an MCB to see if it is working.

[Brian Gaff]

Hmm, I'd suggest testing the actual lights one at a time then. From what you say its not clear if you have other permanently on lights, but I guess having everything off at the switches if it trips then, you can at least isolate what is not involved. These circuit breakers these days are quite complex compared to what we used to use, and I suppose its possible to have these themselves become faulty.

I bet if we put in a new CU here with wonderful new forms of protection, it would be always tripping. Brian

[Brian Gaff]

That would seem counter intuitive. I wonder what the conductivity of a spider is? Brian

[Brian Gaff]

I do think that when houses are wired or rewired, not enough thought is given to service access. I have a lot of wires going from the CU/meter cupboard up a wall to the upper floors, and as far as I know nobody really knows the actual route or what wire goes where, and how. So lets hope nothing ever goes wrong. It was done in the, when the old 1939 wiring was removed, that was that rubber covered with a woven outside sheath. Brian

[Jeff Layman]

I'm sure that information is available somewhere on the web...

[Andrew]

Well you could, by disconnecting the circuit, and adding a flying external length of 2.5mm cable going into a plug with a fuse of the same rating as the MCB then connect the L and N on the plug and turn on the MCB. I would expect the MCB to trip immediately and presumably just before the plug fuse would blow.

The only cable being 'exercised' is the length of 2.5T&E connecting to the shorted plug.

[Max Demian]

Once or twice a month my amplifier decides to change the input channel randomly, usually repeatedly. Almost as if a spider is crawling around inside...

[John Rumm]

:-)

[Jeff Layman]

The electrician had a quick look yesterday afternoon, but nothing obvious was found. He swapped the MCB for an RCBO so that when it trips it doesn't take out half the CU when the RCD also trips on that side.

He's booked next Tuesday to work his way around the loft opening connection boxes and disconnecting circuits in turn. Could take 5 minutes, could take 5 hours. Hopefully he'll be able to detect and isolate the faulty wiring, and even if it's not obvious what's causing the short, he'll be able to substitute new cable for the faulty one.