Potentially nightmare RCD tripping issue!

[JohnW2]

It's occurred to me that I forgot to tell you all one additional thing, which may be relevant ...

... although the RCD trips (when they have occurred, only very occasionally) have occasionally occurred more-or less 'immediately' on switching on the lights, far more commonly the trip happens after a small delay (probably usually in the range 5-15 seconds) - which might suggest that something is 'warming up' (albeit not much in such a short time).

 

[ebee]

I think you have followed pretty muck the correct logical procedures to investigate the problem. I ntermittent faults can be a right beggar to find with any degree of certainty .
Coils an capacitors can sometimes conspire alongside insulation resistances and load/switching variations to help cloud the issue further,
You could even remove all of the most likely culprits and think you have cured it then 5 years later it starts happening again. Divide and conquer by using as many separate RCDs for end units might even be worth the hassle eventually.

 

[ericmark]

Of of interest what is back ground leakage?

 

[JohnW2]

I think you have followed pretty muck the correct logical procedures to investigate the problem. I ntermittent faults can be a right beggar to find with any degree of certainty .

Glad you agree. I've tried to approach the problem as logically as I can!

Coils an capacitors can sometimes conspire alongside insulation resistances and load/switching variations to help cloud the issue further,
You could even remove all of the most likely culprits and think you have cured it then 5 years later it starts happening again. Divide and conquer by using as many separate RCDs for end units might even be worth the hassle eventually.

All agreed.

Something has clearly changed (maybe 'deterioration' of the choke I have removed). These are very old fittings. I've been living with them for best part of 40 years, during which period they have always been protected by 30 mA RCD(s)., and it is best part of a decade since I changed the tubes from fluorescents to LED ones, but there have never been any RCD trips until the last few weeks.

I may well never know 'the answer'. It could be that what I have done already has been enough to stop the problem recurring, and that will be even more likely once I have removed the chokes and capacitors from the other fittings - as I've said, once I've done that there will be very little (essentially nothing!) left that could trip an RCD. Although the cases of the fittings remain metal, and 'earthed' there is really no obvious way in which there could be a leak to earth in any of them (which could really only happen if there were a breakdown of the 'sockets' into which the LED tubes are 'plugged'.

Kind Regards, John

 

[JohnW2]

Of of interest what is back ground leakage?

Unfortunately, I don't have a means of measuring it. When the lights are switched off, the only loads protected by the RCD in question are one small fridge and three (Class II) PIR detectors.

 

[ericmark]

Have you got an RCD tester with ramp? Would not normally ask this, as rather expensive kit, but electrically wise your not normal. So can you ramp with nothing connected, and then with all normal loads connected and see the difference, which would be back ground leakage.

My RCD tester died, and my wife promises me a new one for Christmas I will be interested to see how good it is, as rather cheap. However before my old one failed, tested dad's house, no fault found, RCD gave A1 results, I had spares so changed it anyway, and fault vanished. My first thought was replacement faulty, but no, that also tested as OK.

The X-pole RCD when it came out had a data sheet singing its praises, saying how it would not trip with spikes, and had a built in RCM, I was temped to get one, however since my fuse boxes (with fuses swapped for MCB's) were in the garage, I was not likely to see if the RCM reported a leakage.

I have seem a monitoring device built into a batching plant, no RCD as stopping a batch mid mix would mean getting in drum with shovels to empty it, it would not restart during a batch, but can't see what help they would be with domestic?

I have a 10 mA MK socket, but never fitted it, problem was the test button would trip the 100 mA RCD as well, and too many people would press the test button to turn off socket. If you lived near me you could have it, but postage would be silly, I am not likely to take a trip to Buckinghamshire.

But I am sure you will find the problem, so watching with interest to see what it turns out to be.

 

[JohnW2]

Have you got an RCD tester with ramp?

Of course I have - how else do you think that I measured the trip threshold?

Would not normally ask this, as rather expensive kit, but electrically wise your not normal. So can you ramp with nothing connected, and then with all normal loads connected and see the difference, which would be back ground leakage.

Fair enough. As I've said, there is very little scope for 'background leakage'. Apart from the lights, the only other 'normal loads' protected by the RCD (unless I plug something into a socket!) are three PIR detectors (Class II, with no earth connection, hence no scope for L-E leaks) and a fridge.

I've just tried several times. With all lights and the fridge on, the IΔn is 26 or 27 mA. With lights off and fridge disconnected it is also 26 or 27 mA. This therefore seems to confirm that, as expected,the 'background leakage' is roughly zero!

 

[ericmark]

Just in case, I assume no auto defrost on the fridge? I know how I would test, and it is a very naughty method, I would ensure no one had access but me during the test, and I think you can work out what I would do, without spelling it out, which I would not do on a forum open to anyone. I am sure you can read between the lines and do it in a safe way.

 

[JohnW2]

Just in case, I assume no auto defrost on the fridge?

You assume correctly.

I know how I would test, and it is a very naughty method, I would ensure no one had access but me during the test, and I think you can work out what I would do, without spelling it out, which I would not do on a forum open to anyone. I am sure you can read between the lines and do it in a safe way.

I'm struggling a bit to think of what sort of fault/situation such a test would reveal that could result in an RCD tripping as a resuly of switching lghts on. Are you perhaps postulating a very high 'background leakage' under some conditions?

 

[333rocky333]

Could you not wire that suspect choke fitting on its own to one of your spare rcds and see if it trips that on its own

 

[JohnW2]

Could you not wire that suspect choke fitting on its own to one of your spare rcds and see if it trips that on its own

Well, yes, I could, but the problem with all this is the intermittency - I might have to wait 2 or 3 weeks to discover whether there are any trips of that spare RCD.

I'm therefore not convinced that such an exercise would achieve much - if, having removed the choke (which I have), I just wait for that same 2 or 3 weeks and experience no trips during that period then, in practical terms I will have discovered much the same, wouldn't I?

 

[Harry Bloomfield]

and it is best part of a decade since I changed the tubes from fluorescents to LED ones, but there have never been any RCD trips until the last few weeks.

If they have all been changed to LED, surely the chokes can all simply be removed/isolated?

 

[JohnW2]

If they have all been changed to LED, surely the chokes can all simply be removed/isolated?

Quite so. As I've said, that's exactly what I'm going to do ...

..... Fortunately, now all the tubes are LEDs I no longer need any of the chokes of PFC capacitors, so if I remove them all, there really will be virtually nothing left which could possibly trip RCDs

.... I may well never know 'the answer'. It could be that what I have done already has been enough to stop the problem recurring, and that will be even more likely once I have removed the chokes and capacitors from the other fittings - as I've said, once I've done that there will be very little (essentially nothing!) left that could trip an RCD.