RCD tripping - intermittent fault finding.

[Rednadnerb]

My sister has an intermittent fault with her electrics that trips the RCD. We're think that the fault is with the ground floor ring as it sometimes trips when the kettle is plugged in but more usually does it all on its own. The RCD is on the upstairs and downstairs power but not the lighting. I know this has been covered in previous posts and the suggested solutions have usually involved calling an electrician with a megger, installing secondary consumer units, MCBO's etc but I was wondering if there is any mileage in the following to isolate the problem: Split the downstairs ring main into two radial circuits by disconnecting the wires at the back of a socket around the midpoint, I suppose everything should carry on working but please let me know if this is dangerous. Install something like a lower rated RCD on each leg of the radial circuit, would need to think a bit more about how to do that. What do you think?

[Harry Bloomfield]

Rednadnerb explained :

I think a better first step, is too keep a log of what is not only turned on, but what is plugged in at the times of actual tripping. Least likely will be a fault on the fixed wiring, but it can happen.

Be aware that several small leakages will add up to enough leakage to cause an RCD to trip. Several items all with input filters (computers, displays, printers, TV's etc.) can cause such problems, unplugging one might seem to cure the trip - but that might not be the entire problem solved.

[Rednadnerb]

Thanks Harry but this has been a problem for years. We have been unable to establish any pattern.

[Harry Bloomfield]

Rednadnerb formulated the question :

So, if all else has been taken out of the equation - a faulty RCD or a wiring fault. Could a cable have been nailed or a screw put though one? Any sockets in a damp area or where the damp course has failed?

Leakages between N + E can cause a trip, just as easily as between L + E. You can buy old Megger type testers fairly cheaply, which should make finding the fault very much easier.

[John Rumm]

As long as the load is reasonably well distributed around the ring, then its unlikely to be dangerous for a short term test.

Have a look on the section on nuisance trips here:

covers a fair bit of detail on the ways to track down these problems.

[NT]

It should halve the number of appliances affected by the tripping, but probably won't solve it. It might though.

A lot more likely to find the culprit is to test phase to earth resistance on all appliances using a standard multimeter. Simple, quick, easy, cheap and often finds the problem.

NT

[Harry Bloomfield]

NT was thinking very hard :

A neutral to earth problem is just as likely to be tripping it, so also needs to be checked.

[Peter Parry]

"Poltergeist" faults, where the condition is intermittent and no amount of altering load seems to affect it are indeed often earth to neutral faults. They can also be difficult to find as the location of the fault and what causes it to manifest itself are not always in the same place.

Some years ago I came across one where plugging a long extension lead into any socket downstairs sometimes tripped the (whole house) RCD. Plugging anything else in didn't, and the extension lead had nothing plugged into it. The RCD was also prone to random tripping for no particular reason.

Plugging the same extension lead in upstairs didn't do anything and both the downstairs ring circuit and extension lead tested out perfectly.

Eventually we disconnected and isolated the line, neutral and earth to every circuit and tested each quite independently. The fault was then found to be a neutral/earth short circuit in a lighting circuit - at some stage the wiring had been renewed and when nailing a cut piece of floorboard back the nail had pierced the cable and shorted neutral to earth. This obviously created a slight current flow insufficient to trip the RCD but enough to sensitise it so that any further imbalance, no matter how small, would trip it. The extension was presumably triggering the RCD because when downstairs the coiled lead was lying on the ground and this must have created sufficient capacitance to create a pulse of a milliamp or so to trip the already very sensitive RCD. Doing the same upstairs didn't have the same capacitative effect.

Usually no particularly special test equipment is required to find these faults, a normal multi meter is all that is needed.

[NT]

No, it doesnt. Neutral connects to live through the appliance via a resistance far lower than the leakage resistance youre looking for.

Make a note of any imperfect resistance readings, and the lowest value on the list is the most likely to be the culprit. Fault resistances can change a lot from 9v to 240v ac though, so a standard meter isnt a perfect detector.

NT

[Harry Bloomfield]

NT laid this down on his screen :

Any imbalance between the current which goes in through the L and comes out via the N can cause an RCD to trip. A short between N to E can cause some of the current to flow via the E and cause a trip. Even where there is no actual load placed on the sub-circuit, leakage between N and E - due to differences in their voltages can cause enough current to flow and the RCD to trip.

That I can agree with, which is why a Megger can be so very useful in many circumstances.

[Peter Parry]

That isn't the problem. An RCD senses imbalance in the current flowing in both the line and neutral. If they are not the same it assumes the missing bit is going somewhere it shouldn't and disconnects the circuit.

If there is a neutral/earth fault some of the current in the affected circuit flows via the earth wire rather than the neutral and creates an imbalance in the line and neutral currents. What often happens is that that imbalance is too small to directly trip the RCD (say 25mA with a 30mA RCD) but effectively creates a "5mA sensitivity" RCD - almost any transient condition on any other circuit protected by the RCD (not just the one with the fault) then causes the RCD to trip.

[js.b1]

Clamp meter on the earth wire - establish what the leakage current is. Disconnect final circuits & unplug appliances one by (must unplug, not just turn off).

Capacitive (and inductive) effects can be a PITA with leaky equipment. Have a dodgy old storage heater which trips the RCBO every 5-7 years - despite new wiring and RCBO, it is most likely a leaky element (25yr old heater). In summer an IR test gave N-E & L-E of 2.1megohm, this morning straight after an RCBO trip both were infinity. When doing a continuity test just running multiple cables side-by-side in a wall creates irritating capacitive effects - Robin continuity test gave

2VAC reading on N-E with final circuit disconnected at each end (ie, floating).

There is always enough of a voltage differential between N & E to cause enough current to flow to knock out an RCBO/RCD. The kettle similarly might have a leaky element, just enough with "background" leakage from all other circuits (mains filters in washing machines, immersion elements) to occasionally take the RCD/RCBO over the trip threshold (23mA).

[John Rumm]

A neutral / earth short on a circuit *is* quite capable of causing nuisance trips even without any contribution from faulty appliances. So no amount of appliance testing will find these.

They can be particularly difficult ones to find since the actual leakage will vary with applied load. In the case of TN-S installs, the load in question is not only that of your installation, but potentially that of neighbours as well (which will affect the potential difference between your installation earth and neutral).

[Mike Barnes]

Peter Parry :

I took him to mean that a test of phase-earth leakage would also show up any neutral-earth leakage. If that were so you wouldn't need to test neutral if you'd already tested phase.

[tony sayer]

In article , NT scribeth thus

Nope wrong. The RCD monitors the current through it on the one leg and what returns on the other. Any imbalance above its rated leakage current will see it trip. Think about it!..

Now a Neutral to Earth wiring fault won't trip the RCD until theres some current flowing in the circuit..

Seen this happen much more than the once;)..

Eh?..

[Mike Clarke]

Yes it will. In our previous house the (whole house) RCD would merrily trip if earth and neutral conductors touched each other on a circuit with the fuse removed from the CU. It could be a serious pest working on a lighting circuit in the loft where the only light came from an inspection lamp fed from a 13A socket. Admittedly that was a TN-S installation where neutral could float a few volts above earth, I don't have that problem in the current house on a TN-C-S system.

[Roger Chapman]

It has happened to me on a modern installation (exact nature unestablished) when working on a disconnected circuit. ISTM that if there is more than one circuit on the RCD side of the box the neutrals are commoned together and linking the earth and the neutral of the 'disconnected' circuit together will provide an alternative return for the other circuits and an automatic trip if there is any current flowing in any of the RCD protected circuits. That of course could be interpreted as "theres some current flowing in the circuit..".

And if you are paranoid about safety it is best to remember the advice previously given on this ng to never ever work on household electrics unless everything has been turned off by the double pole isolator in the CU.

[NT]

I'm puzzled that so many people appear to have substantially misundersood what I was saying. But the only common factor is me, so let me try again...

When insulation resistance testing an appliance (L to E) using a multimeter, the appliance load (from L to N) has a resistance that's very low in comparison to any problematic level of insulation resistance (due to insulation failure) that may be present. Thus when insulation testing, as long as any appliance power switch is on, there is no need to put the probes on N/E as well as L/E.

Trust me, this is correct. :) In fairness it is possible to come up with exceptions, such as an appliance with a blown fuse in its live line, or a discharge lamp that remains o/c at 9v, but these produce only a low level of missed faults, and testing appliance insulation at only 9v doesn't pick them all up anyway. Its just quick, easy, very cheap and does detect a lot of faults.

NT

[Harry Bloomfield]

tony sayer submitted this idea :

...and the current flowing does not necessary need to come from the actual circuit or load which is protected by the RCD - it can come from other circuits or even from your neighbours circuits.

[Dave Liquorice]

I'm not convinced that a SMPSU offers a low impedance L-N, mains through filter, bridge rectifier with a C across the ouput then to the chopper transitors. A transformer or heating element will be low few tens of ohms.

It's a simple matter of swapping a probe to L then N when testing and avoids the problems of a blown fuse or an ambiguous power switch.