EICR Code for working RCD with non-functioning test button

[Lucien Nunes]

For whatever reason, test buttons seem to have more than their fair share of intermittent contacts even factoring in how infrequently they are used. The test button circuit is independent of the RCD proper, so an RCD with an inoperative test button that tests fine on the MFT is not necessarily less reliable than one with a working test button, unless some physical factor like corrosion or dirt ingress is the cause (in which case the rest of the RCD could very well fail prematurely). But as pointed out above, not only does the test button test the RCD, if used as prescribed it also provides some exercise, which can avoid sticky tripping mechanisms. Therefore, by preventing the exercise taking place, the faulty test button could (in theory, if not in practice) lead to impaired safety, not just lack of proof of that safety. That said, I see the logic in the OP and would also like to be able to code it C2.5.

We've had a discussion about wiring faults that disable the test button in the past. There are multiple ways of implementing a test button circuit on a normal 2-pole RCD. If the schematics on the front of bog-standard 30mA are to be believed, a resistor between the line on one side and the neutral on the other, creating an imbalance without actually leaking current to earth, is the most common. Obviously there are two ways round this could be done; a) Supply line to load neutral or b) Load line to supply neutral. Considering the case of an N-E fault downstream of the RCD with no circulating current, configuration 'a' would be de-sensitised by the fault, since some of the test current would go from supply L to E, never passing through the toroid. Arrangement b) does not suffer from this problem since hard L-E shorts don't persist on a live circuit in the same way as hard N-E shorts. Therefore it would seem to be the better of the two, but IIRC we found both in practice, which would negate any requirement for L & N to be a particular way round unless the contacts themselves were sequenced.

Another way of implementing the test button is with a separate winding on the toroid, energised from L & N of the same side. The same MMF is created in the toroid as with the intended imbalance in the main conductors, although as the test coil can have more turns, the actual current can be lower. This makes it a more suitable method for RCDs with high IΔn values and renders the test function completely indifferent to polarity.

But, and it's a big but, the point PC1966 makes about the relative phasing of any current circulating via an N-E fault current more or less overrides any other considerations. Since this can pass through the N conductor in the toroid in either direction, increasing or decreasing the net imbalance when the device is tested, the test button threshold will always be subject to an error up to IΔn when an N-E fault is present, regardless of how the test circuit is implemented.

But until someone makes an RCD with a built-in finger that pops out like a cuckoo from a clock to press the test button exactly on cue, I don't suppose any of this matters very much.

 

[Dartlec]

For whatever reason, test buttons seem to have more than their fair share of intermittent contacts even factoring in how infrequently they are used. The test button circuit is independent of the RCD proper, so an RCD with an inoperative test button that tests fine on the MFT is not necessarily less reliable than one with a working test button, unless some physical factor like corrosion or dirt ingress is the cause (in which case the rest of the RCD could very well fail prematurely). But as pointed out above, not only does the test button test the RCD, if used as prescribed it also provides some exercise, which can avoid sticky tripping mechanisms. Therefore, by preventing the exercise taking place, the faulty test button could (in theory, if not in practice) lead to impaired safety, not just lack of proof of that safety. That said, I see the logic in the OP and would also like to be able to code it C2.5.

We've had a discussion about wiring faults that disable the test button in the past. There are multiple ways of implementing a test button circuit on a normal 2-pole RCD. If the schematics on the front of bog-standard 30mA are to be believed, a resistor between the line on one side and the neutral on the other, creating an imbalance without actually leaking current to earth, is the most common. Obviously there are two ways round this could be done; a) Supply line to load neutral or b) Load line to supply neutral. Considering the case of an N-E fault downstream of the RCD with no circulating current, configuration 'a' would be de-sensitised by the fault, since some of the test current would go from supply L to E, never passing through the toroid. Arrangement b) does not suffer from this problem since hard L-E shorts don't persist on a live circuit in the same way as hard N-E shorts. Therefore it would seem to be the better of the two, but IIRC we found both in practice, which would negate any requirement for L & N to be a particular way round unless the contacts themselves were sequenced.

Another way of implementing the test button is with a separate winding on the toroid, energised from L & N of the same side. The same MMF is created in the toroid as with the intended imbalance in the main conductors, although as the test coil can have more turns, the actual current can be lower. This makes it a more suitable method for RCDs with high IΔn values and renders the test function completely indifferent to polarity.

But, and it's a big but, the point PC1966 makes about the relative phasing of any current circulating via an N-E fault current more or less overrides any other considerations. Since this can pass through the N conductor in the toroid in either direction, increasing or decreasing the net imbalance when the device is tested, the test button threshold will always be subject to an error up to IΔn when an N-E fault is present, regardless of how the test circuit is implemented.

But until someone makes an RCD with a built-in finger that pops out like a cuckoo from a clock to press the test button exactly on cue, I don't suppose any of this matters very much.

Thank you! I think this qualifies as a comprehensive and knowledgeable reply!

I'm guessing that whoever made these RCDs for Screwfix didn't do anything more than the absolute minimum.


Out of interest, taking the 'exercise' benefit, does flicking the main switch off manually give the same benefit in unsticking the tripping mechanism, or is that separate from the 'switch' disconnection part?

One day I hope to meet a home owner or tenant who says they press the button - it's only been 15 years so early days

It's good to know that my initial view was not completely out of place. I think I'll discuss with the customer, given that his son is to be the new tenant and may want some additional sockets added before he moves in - in which case it may all become moot and I can persuade them to upgrade the main board.

Failing that, swapping the RCD and single MCB for a bigger 'name' will mean I don't need to worry about it for another 5 years.

 

[Lucien Nunes]

I don't think operating the lever is equivalent in terms of exercise, because it does not move the armature of the tripping solenoid. Of all parts of the linkage, correct and free movement here is most impotant as (on a traditional non-electronic RCD) the sensitivity depends directly on it responding to a relatively small magnetic force. Once it has released the trip-free linkage, there's a strong spring to do the rest of the work.

 

[pc1966]

As Lucien said, the only way to really exercise / test the RCD is by simulating a trip event so the magneitc part is moved. Having said that, cycling switches and MCBs periodically as part of a general check is not a bad idea as it will show up some sticky mechanisms that need replacing, and keep some from going bad in the first place.

For socket circuits you can trip without a functioning test button with some of the fancier plug-in socket testers that also have a button to divert current L-E to simulate a fault, sort of like the MFT test but without timing it, but again hardly anyone is going to do that!

 

[davesparks]

some of the fancier plug-in socket testers that also have a button to divert current L-E to simulate a fault,

And if the RCD doesn't work, or if there is no RCD installed then the current continues to flow and the tester gets quite hot relatively quickly, until the tester stops working, or it did on the early versions anyway.

It also requires someone to be daft enough to stand there holding their finger on the button, sadly I have a mate with the required level of daftness so have watched this happen first hand.

 

[pc1966]

That is a good point, as 30mA corresponds to around 7W which is going to toast fairly soon. I guess they could put in a PTC thermistor next to the main current setting resistor to save damage but as you found out they are normally built very much down to a price.

 

[Vortigern]

I would say F.I. and move on. I know a couple of others have suggested this and I agree it seems logical?

 

[pc1966]

But what further investigation is there to do?

The unit has a fault. It may or may not imply longer term safety issues, but no one is going to dissect the RCD to find out and put it together and back in service if it is deemed to just be the button's circuit. So realistically you have two choices:

• Code it as C3 and most likely it will be ignored for 5 years until the next EICR.
• Code it as C2 and the landlord has 28 days or whatever to replace it.

My own feeling is the only acceptable choice becomes C2.
[automerge]1595675371[/automerge]
I know that may seem a little incompatible with the coding of no RCD on older installations as C3 when not for outdoor sockets, etc.

But to me it is a bit like the arguments on MOT testing of older cards. If you have a 60s car it probably had seatbelts as an extra-cost option as they were not yet mandatory. So if you presented such a car for MOT without seatbelts it would pass, but if it had seatbelts fitted that were faulty it would fail.

 

[Vortigern]

I suspect all cases of "test button not working" should be investigated.

But what further investigation is there to do?

Perhaps you could answer that question yourself?
In any event my view is you are not there to repair just report. In order to deal with a non working button you would have to work out why it is not working. For instance I have come across this a few times. Where the RCD did not operate under test or the test button. Puzzled I approached the RCD in question and switched it off manually. After this I found it operated under test and the test button worked. My investigation showed (to me) that due to lack of operation of test button and age/dust/contaminants the mechanism was stuck and manual operation cleared the "stickiness" If it had not worked I would have concluded it was shagged and got a new one. Investigation concluded.

 

[pc1966]

Perhaps you could answer that question yourself?
In any event my view is you are not there to repair just report. In order to deal with a non working button you would have to work out why it is not working. For instance I have come across this a few times. Where the RCD did not operate under test or the test button. Puzzled I approached the RCD in question and switched it off manually. After this I found it operated under test and the test button worked. My investigation showed (to me) that due to lack of operation of test button and age/dust/contaminants the mechanism was stuck and manual operation cleared the "stickiness" If it had not worked I would have concluded it was shagged and got a new one. Investigation concluded.

That is a fair point that it should not be a case of "pressed once, did not work, replace".

However, in the OP's case they have tried it out several times and found it would trip with the MFT test but not the button. So here I would say they have done as much investigation as it reasonable to do.