Potentially nightmare RCD tripping issue!

[JohnW2]

In my large cellar, I have three 2 x 4ft (Class 1) fluorescent fittings, all of which were retrofitted with LED tubes a few years ago (without bothering to remove ‘ballasts’), all ofwhich have Indi dual switches (but usually all left ‘on’ with control by a single upstream switch), plus a few individually-switched LEDs in batten holders. The only other loads on the (6A) circuit (in a mini-CU, with 30 mA RCD in cellar) are three PIR detectors.

For years there have been no problems and the lights ate usually switched on and off a good few times every day. However, for the last few weeks, the RCD has occasionally (probably only once per week, or less) tripped when I have switched (all) of the lights on. When this happens, there are often two or three further trips when I switch the lights on after resetting the RCD, but the trips usually stop after that.

A few days ago, it was worse - every time I reset the RCD, there was a trip when I tried to switch the lights back on. I therefore went around switching the fluorescent fittings off one at a time, and found that the trips stopped when I switched one of them off. However, when I switched the fitting back on, there were still no trips, so it could have been just a ‘co-incidence’ that trips had stopped when I switched it off.

The fluorescent fittings are all Class I, with CPC connected only to the casing and nothing else. None of the CPCs at the batten holders and PIRs are connected to anything - and the cellar is generally ‘dry’. On testing, the RCD consistently trips at about 25 mA.

The only thing I could think of was that there may have been a leakage to earth (metal case of fitting) from one of the chokes (‘ballasts’) or Power Factor Correction capacitors (both of which are metal-bodied) in one of the fittings. Starting with the fitting I had ‘identified’, I removed the chokes and PFC capacitor. That was a few days ago, and there have been no trips since - but, given the intermittent history of the problem, that doesn’t prove much.

On doing IR testing on the removed chokes and PFC, I found that one of the chokes had an IR to its case of about 300 kΩ (at 250V, 500V and 1,000V) and that one side of the capacitor had an IR of about 150 MΩ to its case, but only at 1,000V. However, none of that is really not enough to explain trips, hardly enough to be ‘the last straw’ if similar (but worse)) leaks existed in other fittings.

So, I’m currently in ‘wait and see’ mode (and probably will remove all the other chokes and PFCs, anyway). In the meantime, has anyone got any other observations or suggestions, since this has the potential to become a nightmare if the problem persists?

Kind Regards, John

 

[Harry Bloomfield]

So, I’m currently in ‘wait and see’ mode (and probably will remove all the other chokes and PFCs, anyway). In the meantime, has anyone got any other observations or suggestions, since this has the potential to become a nightmare if the problem persists?

That is as much as you can do - divide and conquer.

 

[333rocky333]

How did you test the rcd, may be worth testing
first with the load connected and all switched on, then with the load disconnected from the Rcd.
See if theres a substantial difference.
Before the days of rcds it was guite common for magnetic chokes to breakdown, sometimes even burning up and eventually shorting to the case and tripping mcb

 

[333rocky333]

Would you have a spare rcd rattling about, 2 pole latching like in the old consumer units, maybe fit temporary half way round the circuit, if lucky may trip and help break down the faulty, if any, part of the circuit, as its your property sure it can be fitted in a safe manner.

 

[Adam_151]

On doing IR testing on the removed chokes and PFC, I found that one of the chokes had an IR to its case of about 300 kΩ (at 250V, 500V and 1,000V)

While, as you say that value is not low enough to cause a leakage that would trip an RCD, 300k / 0.3meg is more than low enough to indicate that things are not as they should be, and if that value were measured on the circuit initially, we'd be wanting to break it down and locate where it ws comming from. Just out of interest what is the IR of the circuit as it stands (Live and Neutral tied together and tested to earth)?

(Bear with me here.... memory might be a little fuzzy.... this was a few years ago)
I've seen an issue where a choke was testing fine on IR (even on 1000v) but would take the RCD every time the lights were switched on, it was found by a process of elimination by splitting the circuit in half, and then half again, etc. Luckily it was only in a site tea hut so it wasn't a very big circuit (But was a big problem....the kettle was loosing supply.... and that was important equipment XD). Replacing the choke solved the issue, but it still refused to test at anything other than clear. So if that one could cause issues and seemingly test fine, your one that this giving results that defintaly don't seem right (but don't look anywhere near low enough to cause a trip) I could easily see as being able to be able to do similar. I put it down to Magnetostriction, when its in circuit and its energised there will be some physical forces acting on the windings, which is probably enough to turn your marginal fault into much more of one when the damaged winding moves by some thousanth of a mm. WIth an electronic load I suppose there is scope for the issue to be reduced if switched back on immediatly as the inrush current could be less if capactitors on the DC side of rectifiers still had some charge. Also scope of the issue to be affected by temperature, or have a seeminly randomness about it.

 

[JohnW2]

How did you test the rcd, may be worth testing first with the load connected and all switched on, then with the load disconnected from the Rcd.
See if theres a substantial difference.

Yep, did both - and there was no substantial difference. However, I necessarily did this at a time when there was no tripping problem (which is the case for 99.9% of the time), so it doesn't necessarily prove much.

Before the days of rcds it was guite common for magnetic chokes to breakdown, sometimes even burning up and eventually shorting to the case and tripping mcb

Yes, I remember that - but, as my measurements demonstrate, this one was minimally damaged. Indeed, for all I know that 300 kΩ 'leakage' to its case may have been their since its birth (very many moons ago)

 

[ericmark]

I had a problem with parents house before rewire, and the insulation tester showed no fault, and same with my old house, insulation tester showed no fault, it may have been an intermittent fault, or it could have been spikes on the supply, neither house had a SPD.

I now have a clamp on which does measure 0.001 amps, whole house around 8 mA, however since all RCBO protected no one circuit gets near that limit, but the boiler was on a RCBO and when solar panels fitted changed to FCU without any RCD protection, so tried swapping for a RCD FCU but kept tripping, so swapped back, I had intended to investigate, but never got around to it, however odd it tripped the RCD FCU but not the RCBO.

I have had it in past, RCD tripping, did test shows OK, but swapped RCD anyway and problem cured, and yes tested replacement and passed all 6 tests. I did consider fitting the

auto resetting type, or type with built in warning the X-Pole rotten name to google, get anything but RCD's.

However moved house and RCBO's seem to work great, only problem they said type B on the box, but after they were fitted realised all type AC curve B, not seen a single module width RCBO better that type A, but since the RCD is secondary protection as house TN-C-S can't see the point changing to type A now. After all type A only good for 6 mA so unless some thing trips at over 6 mA no real point fitting.

The was a John on this forum who would tell me he had two RCD's to cover whole house and he had no problems, however lost the contents of 2 freezers just before we moved due to a RCD trip, so did not want a repeat of that, so for peace of mind fitted all RCBO's except for freezers which use RCD sockets, and are on an UPS supply.

I suppose my old clamp on with 10 mA increments should still show if there is a high back ground leakage to left, most RCD's need over 20 mA to trip, problem with insulation tester is it uses DC, and we have an AC supply, so there will always be some capacitive and inductive leaking, which means instead of needing around 26 mA to trip it, only 18 mA would trip my house with a back ground leakage of 8 mA.

 

[JohnW2]

Would you have a spare rcd rattling about, 2 pole latching like in the old consumer units, maybe fit temporary half way round the circuit, if lucky may trip and help break down the faulty, if any, part of the circuit, as its your property sure it can be fitted in a safe manner.

I've got plenty of spare RCDs knocking around, but since the problem has only been arising once every week or two, it could take a long time before I could be at all confident about anything - and I would probably have to bypass the existing RCD - because, as we know, if I left the two in series, there's no telling which would trip, even if the spare one had been inserted in a place where iot should catch the 'culprit.

 

[JohnW2]

While, as you say that value is not low enough to cause a leakage that would trip an RCD, 300k / 0.3meg is more than low enough to indicate that things are not as they should be, and if that value were measured on the circuit initially, we'd be wanting to break it down and locate where it ws comming from.

Indeed, if one was measuring IR of the circuit. However, the 300 kΩ I measured was, of course, the choke alone, sfter removal, 'on the bench'

Just out of interest what is the IR of the circuit as it stands (Live and Neutral tied together and tested to earth)?

Sorry, I forgot to tell you about that. Prior to removal of the chokes and capacitor from one fitting, the IR of the entire circuit [ L+N joined - CPC ] was again around 300 kΩ - suggesting that (maybe by luck ) the one choke I had removed was responsible for nearly all of the circuit's 'leak'. This was further confirmed by the fact that after I'd removed that choke, the IR of the circuit rose to around 12 M - not 'perfect', but easily explained Ωby slight leaks in some of the other chokes and capacitors which I had not removed.

However, again, the trips are extremely intermittent (hence the 'nightmare') so all these measurements were necessarily undertaken when no trips were occurring - so may not tell us anything useful about the situation when trips are occurring.

(Bear with me here.... memory might be a little fuzzy.... this was a few years ago)
I've seen an issue where a choke was testing fine on IR (even on 1000v) but would take the RCD every time the lights were switched on, it was found by a process of elimination by splitting the circuit in half, and then half again, etc. Luckily it was only in a site tea hut so it wasn't a very big circuit (But was a big problem....the kettle was loosing supply.... and that was important equipment XD). Replacing the choke solved the issue, but it still refused to test at anything other than clear. So if that one could cause issues and seemingly test fine, your one that this giving results that defintaly don't seem right (but don't look anywhere near low enough to cause a trip) I could easily see as being able to be able to do similar.

One does hear these strange stories about things causing RCD trips when it's hard to think of a credible mechanism - particularly when they are stories about trips occurring when things are switched off. However, these things clearly do sometimes happen so, as you say, it's not impossible that I am experiencing something like that.

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

Kind Regards, John

 

[JohnW2]

... however lost the contents of 2 freezers just before we moved due to a RCD trip, so did not want a repeat of that, so for peace of mind fitted all RCBO's except for freezers which use RCD sockets, and are on an UPS supply.

As I've said so often, provide that one powers one's freezer from a circuit whose loss would be rapidly noticed (like one which supplies a TV or internet router ) and NOT by a 'dedicated' supply, one will never be unaware of loss of power to a freezer for an appreciable period of time whilst a house is occupied, and one would be incredibly unlucky if, by pure co-incidence, a freezer lost power whilst the house was unoccupied for an appreciable period of time.

In the many decades I've lived with freezers, I have 'lost the contents' on two occasions, both a very long time ago (before I had 'over-temp alarms) and both due to failure of the freezer itself (not its power supply) - something which no 'dedicated circuit' or UPS can prevent ... which is why I always advocate (and practise!) having 'over temp alarms'(again, obviously only of use when a house is occupied)!

 

[ericmark]

At the time we were looking after my mother, so house was unoccupied when the RCD tripped. In the main a RCD trips due to some thing in the house having a leakage, and when unoccupied items are unplugged so less likely the RCD will trip, however I think the trip was due to spikes on the supply, like some one using a welding set, as we could go for years with no trip, then trip 5 times in two weeks, then go years again before next trip.

 

[Harry Bloomfield]

I've seen an issue where a choke was testing fine on IR (even on 1000v) but would take the RCD every time the lights were switched on, it was found by a process of elimination by splitting the circuit in half, and then half again, etc. Luckily it was only in a site tea hut so it wasn't a very big circuit (But was a big problem....the kettle was loosing supply.... and that was important equipment XD). Replacing the choke solved the issue, but it still refused to test at anything other than clear.

The problem is that a choke, on switching, can generate much higher voltage pulses, than the mains, or even the 1000v you were IR testing with.

 

[JohnW2]

...however lost the contents of 2 freezers just before we moved due to a RCD trip, so did not want a repeat of that, so for peace of mind fitted all RCBO's except for freezers which use RCD sockets, and are on an UPS supply.

It's just occurred to me. I presume that you made very sure that the RCD sockets were not 'active' ones (which many/most are) - since that would clearly be potentially disastrous for a freezer!

To anyone else reading this ... never use an 'active' RCD socket to power something (such as a freezer) which needs to remain on continuously - since power will not be restored to the item after a 'power cut' until it is reset manually! The packaging or instructions should indicate whether it is 'active' or 'passive' - and it's the latter one needs for things like freezers.

 

[JohnW2]

At the time we were looking after my mother, so house was unoccupied when the RCD tripped. In the main a RCD trips due to some thing in the house having a leakage, and when unoccupied items are unplugged so less likely the RCD will trip,

Exactly - it's extremely unlikely that an RCD will trip ('correctly') in an unoccupied house in which almost all loads have been disconnected or switched off - essentially leaving just spontaneous/'nuisance' trips (which are very rare for most people) as possibilities.

however I think the trip was due to spikes on the supply,

No-one will ever know, but I personally think that would be incredibly unlikely.

 

[JohnW2]

The problem is that a choke, on switching, can generate much higher voltage pulses, than the mains, or even the 1000v you were IR testing with.

Yes, valid point - but wouldn't that only really apply at 'switch off', rather than 'switch on' ?