Weird fault that no-one can work out the issue.

[Raptor0014]

Evening all.

Been at a property the last couple of days that had suffered damage due to a surge. I got the power back on all circuits apart from the upstairs lights as it immediately trips.

Long story short I broke the circuit down and re-added room by room to narrow down the problem area, as no matter what test I did there were no apparent faults. Yet turn the lights on and the circuit trips. I ramp tested the RCD and that tripped at 24mA. The earth leakage with all the circuits on and under load with the particular RCD was 6mA.

The light in question is a weird GX38 CFL that sits in an alcove. But I removed it and wired a brand new pendant in and tried that. Immediately the circuit trips. The light works when wired in place of another light on the circuit. So the light itself isn’t faulty.

I removed all working loads in the circuit again
and had just this one light wired, circuit still trips.

I removed all other circuits from the RCD and had just this one light wired to the switch to rule out cumulative earth leakage and again circuit trips.

I tried the light and the pendant wired directly to the switch and both work fine.

So the problem seems to be a peice of cable approx 5m long that is buried in 2 walls and runs under the floorboards across the landing.

I wondered if the problem was maybe rodent damage on just one conductor that was giving a fault to earth via the rad pipes. But testing between the rad and the conductors gives open circuit. So it’s not that either.

I spent 20mins on the phone to technical support and they were at a loss as to the fault.

I did a R1+R2 and RN+R2 to check continuity. All fine.

I IR tested at 500v L-N, N-E and L-E and all gave >999MΩ

So there is a problem with the cable it seems. But testing it is not showing what it is.

At the moment I’ve left the suspect piece of cable between the load side of the switch and this one light disconnected. I’m going back in a couple of weeks to change the CU as the owner now wants a full RCBO surge protected unit. When there the only thing I can think left to do is run a new 5m long piece of cable directly from the switch to the light and see if the circuit still trips.

If it doesn’t trip then problem solved. But it still doesn’t make sense that any test I do shows no apparent problem. If it still trips then I haven’t a clue what the fault is.

So over to you knowledgeable folk. I’d appreciate any advice as it’s a real head scratcher and I can’t think of anything else.

 

[timhoward]

Fun!
Have you tried continuity between the Neutral for this circuit and every other neutral in the CU?

 

[cireland]

Evening all.

Been at a property the last couple of days that had suffered damage due to a surge. I got the power back on all circuits apart from the upstairs lights as it immediately trips.

Long story short I broke the circuit down and re-added room by room to narrow down the problem area, as no matter what test I did there were no apparent faults. Yet turn the lights on and the circuit trips. I ramp tested the RCD and that tripped at 24mA. The earth leakage with all the circuits on and under load with the particular RCD was 6mA.

The light in question is a weird GX38 CFL that sits in an alcove. But I removed it and wired a brand new pendant in and tried that. Immediately the circuit trips. The light works when wired in place of another light on the circuit. So the light itself isn’t faulty.

I removed all working loads in the circuit again
and had just this one light wired, circuit still trips.

I removed all other circuits from the RCD and had just this one light wired to the switch to rule out cumulative earth leakage and again circuit trips.

I tried the light and the pendant wired directly to the switch and both work fine.

So the problem seems to be a peice of cable approx 5m long that is buried in 2 walls and runs under the floorboards across the landing.

I wondered if the problem was maybe rodent damage on just one conductor that was giving a fault to earth via the rad pipes. But testing between the rad and the conductors gives open circuit. So it’s not that either.

I spent 20mins on the phone to technical support and they were at a loss as to the fault.

I did a R1+R2 and RN+R2 to check continuity. All fine.

I IR tested at 500v L-N, N-E and L-E and all gave >999MΩ

So there is a problem with the cable it seems. But testing it is not showing what it is.

At the moment I’ve left the suspect piece of cable between the load side of the switch and this one light disconnected. I’m going back in a couple of weeks to change the CU as the owner now wants a full RCBO surge protected unit. When there the only thing I can think left to do is run a new 5m long piece of cable directly from the switch to the light and see if the circuit still trips.

If it doesn’t trip then problem solved. But it still doesn’t make sense that any test I do shows no apparent problem. If it still trips then I haven’t a clue what the fault is.

So over to you knowledgeable folk. I’d appreciate any advice as it’s a real head scratcher and I can’t think of anything else.

That's a cracker.
Just for the sake of clarity, when testing to earth, did you leave the earth in the bar?
Does it go with a bang or merely trip?

 

[littlespark]

Any electronics on the circuit that could be confusing the tester? You say you disconnected loads, what about dimmers or smart switches?

High resistance short like a fried rat?
Testing with you mft not giving enough juice to read as a fault.

Try 1000v if tester does it.

 

[plugsandsparks]

Have had to do that once in 12 years, lol - basically run a temp cable to bypass the suspect cable to prove the installation is stable. Sooner you than me nowadays

 

[Raptor0014]

Fun!
Have you tried continuity between the Neutral for this circuit and every other neutral in the CU?

No I haven’t as I removed every circuit bar the light circuit in question when testing for cumulative leakage.

All the other circuits were tested individually as part of checking everything after the surge. But I’ll give that a test when I go back to change the CU in a few weeks.

It’s been a real head scratcher. Good job I’m bald as I’d have scratched it all off by now anyway.

 

[Raptor0014]

Any electronics on the circuit that could be confusing the tester? You say you disconnected loads, what about dimmers or smart switches?

High resistance short like a fried rat?
Testing with you mft not giving enough juice to read as a fault.

Try 1000v if tester does it.

There are some Hive lights etc. but no dimmers. Plus I disconnected every switch and wago’d the links between for the common as it’s all looped through the switches. They’re all metal so took them out the equation as they’re earthed but just to make sure nothing spurious.

My tester will do 1000v so will give that a try next time.

 

[Raptor0014]

That's a cracker.
Just for the sake of clarity, when testing to earth, did you leave the earth in the bar?
Does it go with a bang or merely trip?

It just trips the RCD not the MCB. And no the earth was removed from the bar to rule out parallel paths.

The piece of cable was also removed at both ends when I test the suspect piece.

 

[cireland]

It just trips the RCD not the MCB. And no the earth was removed from the bar to rule out parallel paths.

The piece of cable was also removed at both ends when I test the suspect piece.

Could be the parallel path causing the trip.

 

[timhoward]

When there the only thing I can think left to do is run a new 5m long piece of cable directly from the switch to the light and see if the circuit still trips.

You've proved a pendant directly on the switch works, and the original fitting works directly on the switch, so logically adding some fresh copper in the middle will work.
(You might want to consider terminating the feed to this switch, sending a new feed to the fitting in the loft from somewhere else and adding a quinetic switch.)

High resistance short like a fried rat?
Testing with you mft not giving enough juice to read as a fault.

Try 1000v if tester does it

If he's testing clear at 500v I don't see any value in sending 1000v through it, it's not going to suddenly jump from millions of ohms to under 7666.6 ohms surely?

 

[timhoward]

Could be the parallel path causing the trip.

From what's been said, I think this is the most likely reason.
I'm suspecting a N-E fault to something that isn't the CPC of that circuit.

I'd repeat IR tests with all parallel paths connected. I know you tested low resistance to the pipe near it but it might be something else.
You could also try running a temporary N conductor from switch to light, using original switched live.

Ultimately though you've proved it's that particular cable upsetting things somehow, and unless it's easy to run another one I'd start focussing on a solution that doesn't involve it, as it's probably unlikely you'll be able to repair whatever the issue is very easily.

 

[Pretty Mouth]

How old is this property? Does the RCD trip with a loud 'pop', or does it just click off? Is there any delay at all, or does it trip immediately on energising?

3 possible faults that I can think of:

1 - Regular earth fault, but perhaps you're doing something not quite right with your testing, so it's not showing up.

2 - Borrowed neutral/interconnected circuits. This won't necessarily show as low IR to earth.

3 - Fault between L and E that is not quite touching, but arcs intermittently. I have come across this only once. It gave no indication whatsoever on an IR test at 500V, but would intermittently trip the RCD (and sometimes MCB) with a loud 'pop'. Traced to a nicked live conductor that was arcing to the backbox.

Here's what I'd do:

Main switch off. Put everything back together, as it was. Lamps must be in place, nothing disconnected/bypassed. All light switches etc in the 'on' position.
Disconnect the suspect circuit L+N conductors from the MCB and N bar, and join together. Leave CPC in earth bar.

At 250V, IR test between the faulty circuit and E bar, then N bars, then each and every MCB outgoing L terminal. If all results are reasonable, try again at 500V. If you're still getting reasonable results, then 1 and 2 are eliminated, and it has to be 3.

Good luck!

 

[pc1966]

When IR testing are you sure the CPC is linked back to the MET?

If you have narrowed it down to a length of cable are you sure it has no junction box to some other thing?

 

[Raptor0014]

How old is this property? Does the RCD trip with a loud 'pop', or does it just click off? Is there any delay at all, or does it trip immediately on energising?

3 possible faults that I can think of:

1 - Regular earth fault, but perhaps you're doing something not quite right with your testing, so it's not showing up.

2 - Borrowed neutral/interconnected circuits. This won't necessarily show as low IR to earth.

3 - Fault between L and E that is not quite touching, but arcs intermittently. I have come across this only once. It gave no indication whatsoever on an IR test at 500V, but would intermittently trip the RCD (and sometimes MCB) with a loud 'pop'. Traced to a nicked live conductor that was arcing to the backbox.

Here's what I'd do:

Main switch off. Put everything back together, as it was. Lamps must be in place, nothing disconnected/bypassed. All light switches etc in the 'on' position.
Disconnect the suspect circuit L+N conductors from the MCB and N bar, and join together. Leave CPC in earth bar.

At 250V, IR test between the faulty circuit and E bar, then N bars, then each and every MCB outgoing L terminal. If all results are reasonable, try again at 500V. If you're still getting reasonable results, then 1 and 2 are eliminated, and it has to be 3.

Good luck!

Thanks for the tips. Will give them a try.

To answer your questions.

Property is 12 years old and RCD clicks off immediately on energising. Ie I can’t even energise it as it trips straight away if the light is in the on position. If it’s off then it will energise but immediate trip when the light is turned on.

I don’t think I’m doing anything wrong with my testing and when I rang tech support and talked through exactly what I’d done they didn’t say I’d done anything wrong.

I don’t think there are any borrowed neutrals. Certainly not at any of the switches and I’ve had most of the light fittings down.

A lot of the back boxes have a lot of cables in due to the loop being at the switches. But I had all the switches off and the common was linked through with Wago’s to take out all the other lights in the circuit. So I can only think I that scenario there is rodent damage that is arcing once the circuit tries to energise.

I’ll do the 3 steps you mentioned and see how it goes. Thank you. I do appreciate the advice as I’m far from a font of all knowledge and always happy to learn.

 

[Raptor0014]

When IR testing are you sure the CPC is linked back to the MET?

If you have narrowed it down to a length of cable are you sure it has no junction box to some other thing?

Yes it’s linked as R1+R2 tests fine.

I’d be surprised if the cable has a junction box as it goes down the wall from the switch, underfloor across landing and up the wall to the light. But as it’s buried in the walls I will only be able to check the underfloor section unless the owner decides to have to pulled out and new cable chased in.

 

[Bob Geldoff1234]

I have had similar type of faults and they all have turned out to be faulty cable shorting out under mains power.Your test meter cannot supply enough power to leap the faulty area but mains power certainly can.Sometimes 1000v on the tester might just about show it up but still no guarantees.
Mains power has much more potential to jump gaps than your tester can.

 

[LastManOnline]

I have had similar type of faults and they all have turned out to be faulty cable shorting out under mains power.Your test meter cannot supply enough power to leap the faulty area but mains power certainly can.Sometimes 1000v on the tester might just about show it up but still no guarantees.
Mains power has much more potential to jump gaps than your tester can.

Fully agree.The OP is basically describing what is a "high resistance" fault for a test meter but that becomes a "low resistance" fault once mains power is applied.Typical of water faults and other foreign matter

 

[Lucien Nunes]

Mains power has much more potential to jump gaps than your tester can.

It's not power that decides how far a spark will jump or how insulation will break down, it's voltage, With the peak voltage of 230V mains being 325V, a 500V insulation test will create more stress and a 1000V test much more stress and is more likely to cause it to break down to an extent that shows as a bad reading. The power behind the mains is likely to do vastly more damage and create a much more noticeable effect if it breaks down catastrophically, but a 230V RMS sinusoid won't jump as big a gap in the first place.

When we get into the realm of fast-risetime pulses, things get a bit more complex as dv/dt is important as well as voltage itself, but not applicable to normal mains or DC insulation testers. A 4kV surge on the mains could show a lot of things up but again that's out of the realm of the comparison being made.

As far as the circuit fault is concerned, I agree that a short between this circuit and another is a likely culprit. When testing for an elusive fault like this, far from separating the CPCs etc my first move would be to link everything together that can be linked, all L's, N's, CPCs, other than the conductor under test. Anything to prove that the insulation of that one conductor is breached, before splitting it down again to localise.