Faulty RCD tripping on the button but not via the tester

[tony9i9i9i]

What should you do if a client has a CU that is a split board protected by 2 * 63Amp 30mA double pole RCCDs, you took Zs and tested both RCCDs by their buttons before starting work.

But on completion of work, one of the RCCDs, when tested via tester, would not trip, the other operated as expected!

 

[plugsandsparks]

Test without wiring connected to RCCB

 

[tony9i9i9i]

Test without wiring connected to RCCB

The button was working, but it would not trip in 300ms @ 30mA when I tested it. I've told the client it urgently needs replacing and that they effectively have no RCD on that side of the board and are not currently compliant. Is that me covered? It's not like I have the authority to turn their power off.

 

[westward10]

Did you test it with the outgoing circuits isolated from it?

 

[Chivers]

Did you do a Ramp test?

 

[UNG]

Test without wiring connected to RCCB

Did you test it with the outgoing circuits isolated from it?

If an RCD won't trip with a circuit connected what is the point of testing it without the outgoing circuit connected you might prove it trips but will it trip under a fault condition when the circuit is connected. I generally find that a ramp test finds the trip exceeds the 30mA threshold and the RCD is faulty

 

[plugsandsparks]

Simply to determine whether the RCD is faulty or there is a wiring fault. OP has not provided any more info so we have what we have. RCD trips on button but not with tester, no idea where it is being tested.

I do sometimes wonder why i even bother with replies, this is bread and butter.

I dont "generally" find anything, i have had RCDs blow up in my face when pressing the button, i have had RCDs non functioning because of wiring faults and RCDs that trip when i stamp on the floorboards, lol

 

[JoeMagin]

Might seem simple, but had it before where it was actually a 0.3A RCD and not a 0.03A. Easy to overlook if its been a long day

 

[HappyHippyDad]

As others have said, you need to test it with the outgoing terminals disconnected.
If it then trips, it is likely you have a bad fault somewhere. I have had this 3 times now. There is a long thread on it somewhere which @davesparks has given lots of useful information.

 

[tony9i9i9i]

Thanks for the replies all.
Sorry if this is bread and butter.

I tested from a ring main on that side of the board, the tester ran through the cycle 1/2 and 1/2 at 180, then 1 ran over 300ms and the tester aborted the rest of the tests. At that point, I assumed it was a faulty device. As I say when testing the other side of the board the RCCD operated properly.

So I understand you are telling me to test with the protected circuits disconnected (would turning the MCBs off be sufficient? Or would we need to disconnect from Earth and Neutral bars?) and then test.

If the RCD then operates as expected it's OK, if not it is faulty.

If the RCD is OK, then there is a fault somewhere. So I'd say add circuits back 1 at a time and retest, to identify the faulty circuit.

For my info, what type of fault could stop both the RCCD and the faulty circuits MCB from operating?

 

[westward10]

RCD may well be faulty. Try it with the mcbs off and the other RCD off and if no success ditch the circuit neutrals also, if it trips a circuit fault is causing the problem.

 

[tony9i9i9i]

RCD may well be faulty. Try it with the mcbs off and the other RCD off and if no success ditch the circuit neutrals also, if it trips a circuit fault is causing the problem.

Thanks Westward. Is there a typical fault I should be looking for?

 

[loz2754]

I have found that a high resistance neutral to earth fault can cause an RCD not to operate. I have also found that a doorbell transformer connected to one of the circuits can cause an RCD not to operate. And of course, a type AC RCD can be blinded by an appliance or other device that happens to be leaking DC current and thus the RCD will not operate.

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

 

[tony9i9i9i]

How is this for a procedure? I'd still appreciate some guidance on what to typically look for if it comes to a circuit fault that would give this issue.

"If RCD fails to operate when tested at 30mA, follow the sequence below,

Test with the protected circuits disconnected (MCB turned off)
If the RCD then operates as expected it's OK, then there is a fault somewhere. So add circuits back 1 at a time and retest, to identify the faulty circuit.
Rectify the fault and test again.

IF RCD does not operate, remove neutrals and test again.
If the RCD then operates as expected it's OK, if not it is faulty.

If the RCD does not operate, it is faulty replace.

If the RCD is OK, then there is a fault somewhere. So add circuits back 1 at a time and retest, to identify the faulty circuit.
Rectify the fault and test again."

 

[tony9i9i9i]

I have found that a high resistance neutral to earth fault can cause an RCD not to operate. I have also found that a doorbell transformer connected to one of the circuits can cause an RCD not to operate. And of course, a type AC RCD can be blinded by an appliance or other device that happens to be leaking DC current and thus the RCD will not operate.

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

Thanks Loz.

 

[timhoward]

I tested from a ring main on that side of the board, the tester ran through the cycle 1/2 and 1/2 at 180, then 1 ran over 300ms and the tester aborted the rest of the tests. At that point, I assumed it was a faulty device. As I say when testing the other side of the board the RCCD operated properly.

This is a very obscure possibility but it might be
-the tester is giving up at 300ms
-it actually would have tripped at say 600ms
-it is taking more that 300ms because there is a PV installation on an MCB that side of the board supplying the installation from the other side.

In any case, testing without loads as suggested above would lead you to discovering this anyway.

 

[tony9i9i9i]

This is a very obscure possibility but it might be
-the tester is giving up at 300ms
-it actually would have tripped at say 600ms
-it is taking more that 300ms because there is a PV installation on an MCB that side of the board supplying the installation from the other side.

In any case, testing without loads as suggested above would lead you to discovering this anyway.

but at 1 * delta it has to trip within 300ms.

 

[timhoward]

but at 1 * delta it has to trip within 300ms.

I agree.
My point is that if PV is incorrectly installed, it can back-feed an RCD until the inverter notices lack of grid supply and shuts down and this can take longer than 300ms. So in this (unlikely) scenario it could have possibly caused an RCD test to fail.

The first time I encountered this was when RCD's passed at the beginning of a job and then one of them didn't at the end. Though in my case it did trip, it just took far too long. But a different MFT might give up sooner.

 

[tony9i9i9i]

I agree.
My point is that if PV is incorrectly installed, it can back-feed an RCD until the inverter notices lack of grid supply and shuts down and this can take longer than 300ms. So in this (unlikely) scenario it could have possibly caused an RCD test to fail.

The first time I encountered this was when RCD's passed at the beginning of a job and then one of them didn't at the end. Though in my case it did trip, it just took far too long. But a different MFT might give up sooner.

I see, thanks again, but there is no PV.

 

[HappyHippyDad]

Thanks Westward. Is there a typical fault I should be looking for?

Have a thorough read through the thread I've linked below Tony. I had a similar problem to you (ignore the fact I wrote that the RCD was faulty in my first post in the linked thread, I was wrong as I suspect you may be) and this thread helped me to understand why a perfectly good RCD will not trip when tested from a socket or light rather than at the RCD itself (with outgoing terminals disconnected).

It is always unlikely that the RCD is faulty, not impossible, just unlikely. I have only come across a handful of faulty RCD's in the last 10 years.

Here is the link/thread Why isnt the RCD tripping? - https://www.electriciansforums.net/threads/why-isnt-the-rcd-tripping.176881/#gsc.tab=0