Driven mad by an intermittent electric fault

[Moonraker]

When I was working I maintained computers for my sins, intermittent faults were not terribly rare, though they could be a pig to repair. Anyway, my daughter has had her kitchen revamped, this included slightly moving the ceiling light. It was working fine, then not. This was a pain, it stills works intermittently, when it fails there is no power to the light, the big problem is I try to trace the fault then it starts to work again. I have tried wire wriggle, but cannot get it to change its state. It is not helped by the convoluted way the cabling goes under the floorboards of the upstairs, if only it went permanently bust I'd have a chance, as it is I am getting to my wits end. Anyone any bright ideas as to what I might try, apart from a complete rewire?

[TMC]

I presume that you have checked the connections in the switch?

when the fitting was moved was the wiring extended if so the join would be the next place to check

Is it a standard pendant fitting with a cfl or something fancy?

[Alan Deane]

Poor connection to a switch? Or junction box? Cable damaged by a floorboard nail or screw as it passes through a joist upstairs? (I've had this!)

[Bill]

In message , TMC writes

Was the switch removed and refitted, even if only to replace a cable? I have had switches fail intermittently due to plaster and general yuck getting in them during building work.

I fix radios for a living, I love intermittent faults, NOT have fun :-)

[Jules Richardson]

Check switch connections, and if they seem good then perhaps replace the switch (assuming they cost peanuts, like they do here) as I've known them to fail internally (particularly if old - I actually just replaced one in our kitchen yesterday which had probably been there since the house was built in '47)

I assume you've ruled out the bulb? I've always found that incandescents either worked or didn't, but I've had CFLs that are randomly flakey.

I also had a fault with our bathroom light which also went intermittent - that was the connection inside the light fitting not connecting with the terminals on the bulb very well (as I'm in the US though they're all Edison-type screw fittings here; I expect that bayonet-style sockets are less prone to that kind of fault)

If you've got a multimeter then it might be worth disconnecting wiring and measuring resistance at different points to narrow down where the fault is; it's possible that you'll find somewhere where it's obviously reading higher than it should be even with the fault still allowing the light to work. But I'd double-check all visible connections first (for tightness, and to visually make sure that the installer cut back the wire insulation properly so it's not caught in a joint)

cheers

Jules

[Dave Plowman (News)]

Most such problems are due to a poor connection. So start with the obvious things. Try a new switch - even the 'wrong' type, but known to be good. I assume you've re-made the ceiling rose connections? Give each conductor a good tug to make sure it hasn't just broken off inside the insulation at the end.

Once you've done the easy things, suspect the difficult ones like the JB or whatever they used to extend the cable, or cable damage.

The point being it's all to easy to assume the worst.

[Moonraker]

Thanks for the tips. The switch is actually a double one, so I will swap it with the other of the pair. I have checked voltage at the light fitting, it is not present when the light is not working. The cable was just (he says) gently pulled as there was plenty of slack. I am going to have to take up the floor boards and see if and where there is a junction box, however if the fault goes away during investigation, as it has to press, then I am just going to have to replace the cabling. Once the floor boards are up that wont be too difficult, sadly there is fitted carpets, then ply over the boards, so it will be a looooong job.

[TMC]

Is the wiring loop in and have you checked the connections in the celing rose rather than at the lampholder?

If the wiring is loop in then it is unlikely that there will be a junction box

How are you checking for absence of power?

If you are doing a live to neutral test for voltage then you may have a faulty neutral rather than a faulty live

Did you also try a live to earth test assuming that there is an earth?

You may need to check the connections at the previous ceiling rose on the circuit for the neutral

I guess a real expert will be along shortly

Regards

[Bill Wright]

Assuming it isn't the switch or the fitment, and assuming there is just the one switch controlling the light, I suppose if you're desperate you could (a) Make a temporary connection to the supply side of the switch, at the switch. Take the L from there but find an independent neutral somewhere else (beware of triggering RCDs by connecting to the neutral on an RCD protected circuit). Fit a lamp on that circuit and leave it one all the time. If it stays on when the fault occurs you've eliminated the supply to the switch. (b) Do the same thing but take the L from the switched connection on the switch. If the lamp stays on when the fault occurs you've eliminated the switch. (c) Do the same thing but make the temporary L connection at the fitment. If the lamp stays on when the fault occurs you've eliminated the supply, the switch, and the wire from switch to fitment. (d) Find an alternative known good L. Make a temporary connection from it to a lamp. Connect the other side of the lamp to the neutral connection at the fitment. Leave the lamp on. If it stays on when the fault occurs you've eliminated the neutral connection to the fitment.

Bill

[Brian Gaff]

Well the only way is to isolate the various parts and see if hyou can induce the fault. if it is the cable run, they its not a great thing, as pulling new cable through is going to be a pain. You don't give much detail on what kind of lighting it is. Brian

[Brian Gaff]

A while back I had something like this and it was merely a sticky spring in the bayonet socket. It did not jam if you pushed it, but if you applied some side force as you would to twist a bayonet type bulb, it could be induced to jam down. Brian

[Moonraker]

It is a fluorescent strip light, only Live, neutral and earth come through the ceiling, so there must be a junction box somewhere. I am up at my daughters on Monday so will explore further. What I will do if it working ok I am not sure, I keep hoping that the fault will go solid.