Occasional mystery trip

[timhoward]

I'm writing this mainly hoping something occurs to me while writing it!

I got called to an occasionally tripping RCD. It's a Hager double height board, house is about 4 years old, 7 circuits on a type A RCD at top.
Global IR on top row of 0.07M. So fault not present, therefore I could only hope that the low IR was related to the actual fault.
Clamp meter around the RCD tails showed 4ma. Ramp test showed 23ma.

I IR tested all the circuits, L+N to E. Only circuit of interest was kitchen and utility sockets which was 0.07 (obviously still not RCD-trippingly-low).
I found the reason for that (old tumble drier). Then global IR rose to 85M.

Next thing I noticed was when the oven was on, the clamp meter around L and N RCD input shot up to 37ma. And the RCD still didn't trip....
So I thought I'd "help it" with a ramp test. But the ramp test STILL took 23ma to cause a trip. Brain turning to jelly now.
Faulty RCD?
It tripped this morning when they weren't using the oven, and the circuit has 200M IR when the oven is off.

Only other clue is that oil fired boiler was serviced last Thursday, first trip was Thursday night. The boiler circuit had >200M IR (it's wet underfloor heating).
It's tripped again tonight after I sorted out the sockets circuit, and drove away with a healthy global IR.

So I'm thinking
-this is a classic needle in haystack unless I manage to be there when fault is happening
-maybe faulty RCD, still confused by ramp test not tripping quickly when cooker is apparently leaking a lot.
-I should probably have a look inside the boiler next time
-how long do I fight this before converting top row to 7 x RCBO's at £21 a pop...

So far the only remotely suspicious thing has been the Beko oven range, and they have agreed to see if it still occurs with it's isolator off but that is rather clutching at straws.
Any other thoughts or diagnostic ideas most welcome!

 

[buzzlightyear]

The only way of finding the fault is to start switching thing off that is not used temporarily and keep the items on that is the main. And try and pin point where it is.

 

[timhoward]

Thanks Buzz.
It has the feel of some kind of 'background service' that can kick in on it's own, whether heating or lights that only come on at night or defrost heater on fridge feezer etc.

Heating is a strong contender. There's a room stat in each room and presumably a zone valve. I know next to nothing about UFH, whether the zone valves are accessible. One thing I didn't do was try activating each in turn.

The wall stats are all on the same boiler control circuit.
Maybe putting this on an RCBO (there's space for one where the SPD isn't fitted) would be a useful test.

Another thought, is that normally on a dual-RCD board I try moving things to the other RCD.
For some reason this didn't occur to me but I can obviously try moving circuits to the bottom row and see if the fault moves.

 

[Pretty Mouth]

Your thinking is about the same as my own. I'd start by simply swapping the RCDs over on the dual RCD board, this might indicate whether the RCD is faulty. If the tripping persists on the same group of circuits, then as you say, it's RCBOs, or swap some circuits from one side of the board to the other, until you narrow it down to a circuit.

 

[ferg]

It seems too much of a coincidence that the oil boiler was serviced on the same day.
I'd have the cover off it and inspect it thoroughly. something could have shifted or got wet etc.
Oil boilers seem to suffer from plumber wiring on a regular basis.

 

[baldelectrician]

I would be inclined to add a couple of RCBOS to the board-
One for the kitchen sockets, one for the boiler (or cooker)
Then take it from there

DId you split the kitchen ring whilst testing?

 

[timhoward]

It seems too much of a coincidence that the oil boiler was serviced on the same day.
I'd have the cover off it and inspect it thoroughly. something could have shifted or got wet etc.
Oil boilers seem to suffer from plumber wiring on a regular basis.

Thanks, will do. Circuit had good IR when off but as you say got to be worth a look.

DId you split the kitchen ring whilst testing?

Yes. Kitchen ring was 0.07M, I split ring, found tumble drier had N-E fault, then IR rose to something over 100M.

I'm still confused how the cooker was apparently leaking 40ma, the type A RCD didn't trip, and a further 22ma form a ramp test was needed to make it trip. It's got me wondering if I trust the clamp meter.

 

[mainline]

Thanks, will do. Circuit had good IR when off but as you say got to be worth a look.

Yes. Kitchen ring was 0.07M, I split ring, found tumble drier had N-E fault, then IR rose to something over 100M.

I'm still confused how the cooker was apparently leaking 40ma, the type A RCD didn't trip, and a further 22ma form a ramp test was needed to make it trip. It's got me wondering if I trust the clamp meter.

I had that the other day N+L supply from rcbo 20ma of leakage but still ramped at 21ma.
I didn't bother spending any time trying to measure the rcbo supply side or clamping the CPC, instead I just put it down to the clamp meter.

 

[timhoward]

I had that the other day N+L supply from rcbo 20ma of leakage but still ramped at 21ma.
I didn't bother spending any time trying to measure the rcbo supply side or clamping the CPC, instead I just put it down to the clamp meter.

I know this is a tangent and a rabbit hole, but you know me - I had to explore this idea further....!
A non-rcd protected socket, two different MFTs, a calibrated multimeter in series with CPC (set to Max), and a clamp meter (set to Max):

Conclusion: in this environment the the clamp meter is pretty close to the multimeter. The cause of the effect we both saw is likely not due to the clamp meters accuracy.

I set the clamp meter aside, and introduced an RCD that ramp tests at 25ma, then used one MFT in calibration mode to perpetually leak 15ma to earth, then the other MFT in ramp test mode. Interestingly the latter still ramp tests at 25ma even though it's counterpart is leaking 15ma. So I can reproduce the issue at home very easily.

I'm wondering if there is some sort of phase offset shift inside the MFT when ramp testing as what seemed like a perfectly logical idea simply doesn't work! So it seems we use one or the other but don't try and be clever with both, for reasons as of yet unknown.

 

[pc1966]

Conclusion: in this environment the the clamp meter is pretty close to the multimeter. The cause of the effect we both saw is likely not due to the clamp meters accuracy.

When you are trying to clamp-measure the imbalance (difference) with a lot of normal current flowing it can show a bit of an error as you move the two conductors around inside the clamp, not a huge amount usually, but I'm wondering if enough to explain the 40mA/22mA difference?