Rcbo tripping on 1/2 settings

[cliffed]

Rcbo tripping on 1/2 settings with load connectted but ok when load disconnected all other tests are fine… FI required but any thoughts on this

 

[timhoward]

I’d IR test the load and fixed wiring. Line and Neutral together to earth.
E.g If the load/wiring is leaking say 19ma and then your test leaks an extra 15ma then that’s 34ma and you’d expect a trip.

 

[UNG]

mA clampmeter around the outgoing circuit live and neutral to measure the circuit leakage and ramp test the RCBO would be my starting point
Given that an RCD / RCBO should trip at any leakage from 16 to 30mA as the spec requires it to do most of many I have tested over many years generally trip between 23 - 27 mA
In recent years I'm finding more RCD's and a few RCBO's that don't trip as the spec and some are brand new out of the box

 

[westward10]

To me the device is operating as it should.

 

[DefyG]

any thoughts on this

Always taught to test on no-load!

 

[loz2754]

Guidance Note 3 indicates RCD testing should be carried out with the load disconnected to avoid spurious results.

 

[pc1966]

Rcbo tripping on 1/2 settings with load connected but ok when load disconnected

The load is causing some leakage that, when added to your MFT's In/2 test current, takes it above the trip threshold.

Nothing to worry about!

You could do some of the above suggestions to clamp meter the leakage on-load and separately ramp-test the RCBO off-load if you really needed confirmation.

 

[UNG]

Always taught to test on no-load!

Guidance Note 3 indicates RCD testing should be carried out with the load disconnected to avoid spurious results.

I have found this to be a flawed concept as how do you differentiate between a spurious result and a fault
If you apply a measured fault at a point on a circuit and the RCD / RCBO does not operate as expected do you put it down as a spurious result and disconnect the outgoing circuit and test again to get the acceptable result

For many years I generally tested at a point on the circuit and got an acceptable result but around 12 years ago I did a board swap fitting a dual RCD CU for a customer and it all tested ok with RCD trip times recorded at points on the circuits, move on 6 months and I was doing some other work and during testing the RCD failed to trip at the point I had been working on. So I decided to investigate further based on the past testing 6 months before, testing the RCD with outgoing circuits disconnected it tested ok with acceptable trip times I then did a ramp test on the RCD and it tripped at 36mA so not within spec, spoke to the CU manfacturer who sent me out a replacement which when fitted and tested at a point on the circuit tripped as expected and a ramp test was within the 30mA spec

Since then I always ramp test RCD's / RCBO's as I install them as I have no idea how carefully they have been delivered to the wholesalers and have found quite a few that don't meet the spec

So the question is using the prescribed testing methods promoted by the IET how many potentially faulty RCD's are currently being used in blissful ignorance of the fact they may not do what they are ment to do, the old chesnut response I've had in the past when highlighting this is that the fault current will always be of a magnitude to trip the RCD but how do you guarantee it will be and what if it isn't,

 

[timhoward]

I have found this to be a flawed concept as how do you differentiate between a spurious result and a fault

It's actually quite simple - you drag a wall chaser through an illogically buried live cable that is out of zones and see whether the RCD beats the MCB, and whether you get a 30ma jolt or get thrown across the room. (I'm told)

But seriously, after managing to keep all my limbs I tested everything again. This showed the RCD behaved with no loads but was considerably slower with loads connected. I replaced the RCD and normality returned.

This was a few months ago and I've been reflecting ever since that the installation would have passed an EICR with acceptable results even though the RCD was not within spec. I actually think GN3 needs reconsidering. 'What actually happens' surely matters more than 'what might/should theoretically happen'.

I'd like to think that it's rare than no-load tests and real world tests differ in outcomes, but unless we start routinely trying this, how would we know the scale of the problem?

 

[UNG]

It's actually quite simple - you drag a wall chaser through an illogically buried live cable that is out of zones and see whether the RCD beats the MCB, and whether you get a 30ma jolt or get thrown across the room. (I'm told)

I must have missed that test in GN3

But seriously, after managing to keep all my limbs I tested everything again. This showed the RCD behaved with no loads but was considerably slower with loads connected. I replaced the RCD and normality returned.

As I mentioned in an earlier post I've found many odd results and replaced a few RCD's some only months old and now I take a more cautioned approach given the reliance many people place on an RCD / RCBO

This was a few months ago and I've been reflecting ever since that the installation would have passed an EICR with acceptable results even though the RCD was not within spec. I actually think GN3 needs reconsidering. 'What actually happens' surely matters more than 'what might/should theoretically happen'.

When I first had this issue I took the decision to delve a bit deeper having only fitted the CU 6 months prior and knowing that a test at a point on the circuit tripped the RCD at the time and now didn't I decided to test as GN3 suggests you should at the CU with the circuit disconnected when it tripped within the time limits I wanted to find out why so did the ramp test at the CU and found the RCD tripping about 6 - 7 mA over the 30mA limit
When musing on this I do wonder how long the RCD test results declared on an EICR are valid for if a serious injury or fatality occurs in the instance I've mentioned it was a second floor flat so not likely to have extension leads outdoors but there other risks present how long would or could it have been before this problem was found, ok the block management company require the flat owners to have EICR's done every 5 years but would the RCD remain viable under the GN3 test method while still having an underlying fault

I'd like to think that it's rare than no-load tests and real world tests differ in outcomes, but unless we start routinely trying this, how would we know the scale of the problem?

I routinely ramp test all RCD's / RCBO's since I found this problem and note the results on an EIC / EICR / MEWC.
With regard to the scale of the problem until there is some mechanism for fault / failure reporting and the testing changes it will be a hidden danger lurking in the cupboard

 

[pc1966]

I am glad that @UNG experience in lots of real-world testing back up my general paranoia about depending on a single device's electronic trip side for protection. So much more to go bad, and potentially in less obvious ways, than the thermal/magnetic trip and low Zs traditionally used for ADS.

Yes, fault currents will normally be far more than 30mA as even the TT Ze ought to be at least 230/50 = 4.6 times the RCD upper threshold (more at lower currents due to the 200 ohm for stability requirement), but poor quality parts are all too common and in practice folks never test using the built-in button to even know if the device ought to trip.

Rant over! Now I'll get my coat...yes its that one with the 100mA delay RCD in the pocket...