Nuisance tripping - faulty RCD?

[I Trippet]

Had a real headache of a fault finding job which has left me scratching my head - RCCB on a split load board, supplying 3 circuits (shower, immersion and downstairs lights) constantly tripping. The frustrating thing about this was while it would sometimes trip within seconds, mostly it would take anything from ten minutes to several hours before tripping, which made it very difficult to pin down to anything in particular. Insulation resistance readings for those circuits were fine, RCD not tripping at 15mA. Checked every light switch and fitting, shower and isolation switch, immmersion heater switch, no sign of anything that would cause it to trip. Tried leaving shower and immersion heater MCBs off, still tripped. Switched shower off at the isolator, still tripped (immersion heater was hardly ever used so that was already switched off). Disconnected the shower and immersion heater circuits at the board leaving only downstairs lights running off it, still tripped. To take a closer look at the downstairs lighting obviously involved lifting parts of the floor upstairs which was easier said than done. There were 2 or 3 junction boxes on the circuit which I opened and were fine, and all cables were in good condition. Only thing I could think left to try was swapping the RCCB with a spare one I had. Reconnected everything, tested the new RCD and it hasn't tripped since! All the fault finding I've ever done it's been because of the wiring or a faulty appliance, never come across a faulty one RCD before and wondered how common they were. I took the old one back with me and did an experiment to see if it tripped with no load connected and it didn't, so maybe it wasn't faulty. Either way, replacing it cured the nuisance tripping.

 

[I Trippet]

If an rcd is tripping with no load connected then you have an N to E fault somewhere in the installation.

Even with all conductors on the affected circuits disconnected? The other two circuits on the non RCD side are RCBO protected, a N to E fault on either of those would have surely tripped the respective breaker?

 

[LastManOnline]

Even with all conductors on the affected circuits disconnected?

I understood from your post that you disconnected the shower, the immersion but not the lights?
If you did of course disconnect all the affected conductors (lives and neutrals) then by default you would have isolated any N to E fault and the rcd should not trip.
Whenever I have experienced an rcd tripping with no load connected it was always because of an N to E fault in the installation combined with a DNO neutral that had a significant volt drop, (10 to 12 volts) causing a circuit to flow between N and E in the installation.

 

[pc1966]

If tripping with no load then it has to be the RCD/RCBO's fault one way or another. Either it is actually faulted inside, or it lacks EMC protection against smart meters.

The whole EU (and thus UK) EMC directive came in to force around 1996 and was unusual compared to older rules on causing RF interference in that it also required certain minimum standards of RF immunity as well. But as always there was some "wriggle room" in how you interpret the rules and many companies paid only lip service to it. In particular you could define what constituted "failure" in a way that would allow you to get out of certain things - so you could say that a Walkman style unit need not work with a mobile phone adjacent to it as you would not be using both at the same time, and so close mobile phone operation causing transient degradation was allowed as a pass.

I have no idea if the makes of the RCBOs assumed nobody would have a phone 0.3m away all the time and so did not take sufficient steps to filter it, after all those extra capacitors costing £0.01 would seriously impact on the profits...

As well as newer RCD/RCBO hopefully being more immune by design due to the knowledge that RF transmitters adjacent to the CU are the "new normal", the move to metal CU also should help there. Though the cables going in/out mean the reduction in RF is not nearly as large as you might hope for, but it will still be better than plastic CU!

 

[LastManOnline]

This is a totally unknown area for me.Smart meters have only started to be really rolled out here in the last 2 years.Have never yet been called out to a tripping rcd fault that transpired to be caused by a smart meter.A question that crops up in my mind is this.Would the older more "traditional " rcd,s /rcbo,s ( so less electronics) be less suseptable to this issue?

 

[pc1966]

This is a totally unknown area for me.Smart meters have only started to be really rolled out here in the last 2 years.Have never yet been called out to a tripping rcd fault that transpired to be caused by a smart meter.A question that crops up in my mind is this.

I don't do most of this "professionally" so can't comment on the percentage of faults due to this reason or other reasons.

Would the older more "traditional " rcd,s /rcbo,s ( so less electronics) be less suseptable to this issue?

The very original sort with just a sensitive trip coil and differential transformer would not be bothered by RF. But if you start having electronics in there the issue of RF causing upsets has to be addressed as basically and diode junction (part of transistor, etc) will rectify the very high frequency RF and generate a DC or low frequency term from it.

Many products from older times suffered badly from EMC issues as mobile phones came in as they were never tested for it, and in the past you would not have had a 0.1W or so transmitter only 30cm away (at least outside of a radio amateur's home, which is equivalent to a 11kW transmitter at 100m, or 10MW at 3km away). By to 2000s that was common and so companies had to start fixing stuff, but how quickly products were checked/modified is anybody's guess!

 

[LastManOnline]

The very original sort with just a sensitive trip coil and differential transformer would not be bothered by RF. But if you start having electronics in there the issue of RF causing upsets has to be addressed

That's exactly what I wanted to hear.Thanks pc1966