RCD Tests

Hi, recently did quite a bit of work on a bungalow (DB change, kitchen wired, extra sockets, smokes etc.,).  Had some issues with installation but managed to locate faults and everything tested off fine.  A couple of weeks later customer contacted me to advise RCD tripping intermittently and when microwave / kettle being used. 

I moved the sockets over to the other RCD and everything worked fine - narrowed down to faulty RCD.

Now changed the RCD and it holds on when microwave / kettle being used but when doing auto RCD tests RCD trips but when reset for remainder of tests, doesn't trip.  Meter does not show any values.  Tried another tester and same happens so not the meter. Redid RCD test but not in auto and same again happens, RCD trips at x1 but meter shows no values.

All other tests come back good (continuity / earth loop etc).  Everything connected correctly in DB (ie. neutral links).  Just don't know why this is happening with the tester.  Same happens when testing RCD on another circuit - that's a brand new circuit and definitely nothing up with it.

Any advice appreciated.

Hi Leewood1973 have you tried testing the RCDs with nothing connected to them, from the outlet terminals? You might have to put in a couple of short lengths of wire since I believe some RCDs can't be tested from the terminals, too close to the gubbins.

What were the results of insulation tests?

Hi, thanks for getting back to me.  I didn't.  I had a problem on Friday with an RCD not tripping in time and switched off all circuits and did the RCD test.  Do you think there may be problems with things being plugged in, ie. fridge capacitor or something on other equipment?

Switching off the circuits protected by the RCD only disconnects the line conductor. There could be a N-E fault. Did you do insulation tests?

Yes did, an old installation. I don't have the results to hand as my colleague does.  Around 50mega ohms.  Not clear but well within the 2megaohms minimum.  Clearly a possible issue but I've had results similar on other installations but been able to complete the RCD test fully.

The general guidance for testing RCDs in the on site guide (11.1) does recommend the load is disconnected and the tests are done as near as practicable to the RCD. I would tend to assume for a load to be disconnected then both poles L&N are isolated between RCD under test and the load.

Doc H.

Doc Hudson said:

The general guidance for testing RCDs in the on site guide (11.1) does recommend the load is disconnected and the tests are done as near as practicable to the RCD. I would tend to assume for a load to be disconnected then both poles L&N are isolated between RCD under test and the load.

Doc H.

Click to expand...

yes, however, imo this is wrong. it should be tested as is. if something on the circuit is stopping the RCD working then it should be investigated as to why, not just disconnect it to get a number to put in a box, then reconnect knowing that if the RCD is required to operate because someone has got a shock, it probably wont

Andy™ said:

yes, however, imo this is wrong. it should be tested as is. if something on the circuit is stopping the RCD working then it should be investigated as to why, not just disconnect it to get a number to put in a box, then reconnect knowing that if the RCD is required to operate because someone has got a shock, it probably wont

Click to expand...

I agree. Would disconnect circuit only as part of investigation

The majority of the time RCD's will test out perfectly without disconnecting anything. However in this instance the OP has had some apparent inconsistent test results and has tried swapping RCD's. Yet from how I read it has not actually verified either RCD works within specification under no load conditions. We need to identify if the RCD, the circuit or the test meter is introducing these irregular results. The 11.1 guidance takes the circuit out of the equation. In the real world when genuine faults occur there is probably far more current flowing through the RCD than the amount generated under RCD test conditions. Sometimes a combination of meter sensitivity and various circuit factors can lead you up the garden path, unless you step back to check individual items first. I am guessing 11.1 must have been included for this reason, possibly the sort of issues the OP is referring to?.  

Doc H  

Andy™ said:

yes, however, imo this is wrong. it should be tested as is. if something on the circuit is stopping the RCD working then it should be investigated as to why, not just disconnect it to get a number to put in a box, then reconnect knowing that if the RCD is required to operate because someone has got a shock, it probably wont

Click to expand...

Then you could end up unnecessarily chasing your tail for no reason.  

I think the next step should be the RCD tests with all the mcb's switched off.

Me thinks there is a fault on a circuit.

IMHO the continuity and IR tests, test the circuit; and the RCD tests test the RCD

NozSpark said:

IMHO the continuity and IR tests, test the circuit; and the RCD tests test the RCD

Click to expand...

spot on...The purpose of an RCD test is not to establish the integrity of the circuit as this should of already been done if the correct sequence of tests have been followed. I wonder If the people who carry out RCD testing not at the RCD itself  do so at the extremity of the circuit and record individual results for every circuit.....IMO it would indicate they don't actually understand why they are peforming such a test.

Part of the problem is that in 99.9% of the 'real world' fault conditions there is far more fault current flowing than we use during our simulations with test meters and modern test meters automatically stop if they calculate that an earth touch voltage greater than 50v could be reached. So, although I haven't actually checked this, I would think that an RCD test on a fully connected circuit that appears to fail trip times, with a real world fault (electric lawn mower chops flex on the grass, or a drilled cable dead shorts live and earth etc.)  would actually operate correctly. There used to be a note in older issues of GN3 about RCD periodic inspection, stating that if a commercial RCD tester was not available then a 15watt Pygmy lamp wired across L and E would confirm the operation of an RCD. Clearly this will be a very random test fault current, greater than 30ma (but less than 150ma), depending upon where about on the circuit you plug it in.   

Also I assume that when manufacturers verify the tolerances of their devices it is not on every possible combination and permutation of circuit or circuits that it may be connected to. It is purely the device itself at the appropriate supply voltage.

Doc H. 

NozSpark said:

IMHO the continuity and IR tests, test the circuit; and the RCD tests test the RCD

Click to expand...

The 1/2 times rated trip current test with loads connected might also be considered a good test of the existing circuit. If there are already significant earth leakages from the loads which would not show up in the other dead tests, it could indicate why nuisance tripping is occurring.

Andy™ said:

yes, however, imo this is wrong. it should be tested as is. if something on the circuit is stopping the RCD working then it should be investigated as to why, not just disconnect it to get a number to put in a box, then reconnect knowing that if the RCD is required to operate because someone has got a shock, it probably wont

Click to expand...

Agree, should test as is first. If it fails, investigate further.

leesparkykent said:

spot on...The purpose of an RCD test is not to establish the integrity of the circuit as this should of already been done if the correct sequence of tests have been followed. I wonder If the people who carry out RCD testing not at the RCD itself  do so at the extremity of the circuit and record individual results for every circuit.....IMO it would indicate they don't actually understand why they are peforming such a test.

Click to expand...

I actually don't mind seeing that on a test result chart for a dual rcd board. They are testing, that is good.

I don't mind seeing the same result put down for each RCD either, it is tested.

IIRC.... If you have a look at GN3 then it says that the RCD tests should be carried out with the outgoing circuits isolated from it