Adam Kuzack said:

I was asked to look at an installation of a family friend, who wanted lighting pendants replaced with led tubes.

What does, "look at" mean please? Perform an EICR?

If it was simply, can you install a new lamp fitting please, I think that you have been meticulous. Your average reasonably skilful householder would do the task without giving a moment's thought about the rest of the installation.

I think that knowing what you do now, you have to fix the RCD before doing any further work.

If you are going to issue an MEIWC, it seems difficult to avoid doing an IR test on the circuit, but I wouldn't risk a whole board check!

RCD functional test, trip time at rated current, polarity check, loop impedance, fault currents, bonding etc all  recorded on a minor electrical installation works certificate with a full description of what the certificate covers only (Replacement of RCD). 

I think that you have to do what is reasonable.

I would say that you test upstream of the replacement RCD, but are not obliged to test downstream.

If you were to test the lot, even at mate's rates for a family friend, you really could open a can of worms. The property could potentially become uninhabitable.

The MEIWC does not exactly help - see the paragraph at the top of page 514 of the BBB.

I would have thought the work fell under  replacing individual item of switchgear- replacing single protective device 

If you're going to replace a front end RCD , which sounds like it needs doing. I think it would be foolish to not carry out an insulation resistance test first. You could have all sorts of follow up problems after if you don't at least highlight any potential tripping problems down the line.

Gary

How about clamp test around the incoming tails and measure the mA leakage current

The RCD is a 30ma and tested at the board.

But was the main (downstream) CU main switch OFF at the time (or all the fuses pulled)? A consistent slowness in apparent RCD operating times seems a little unlikely if it's just due to it sticking - in my experience they tend to free up after a few operations. On the other hand, many loads (whether capacitive or inductive), if connected at the moment of the test, can result in the voltage on the load terminals not collapsing as promptly as expected - fooling the tester into thinking that the RCD hasn't opened yet when in fact it has. There might be that the RCD is good after all - which presumably would make life simpler all round.

   - Andy.

Seconded - before condemning the RCD, that should be checked with all loads removed- the meter can only look for voltage to fall away, it cannot know if the contacts have opened on time, but the voltage persists due to something elsewhere - things with motor-run capacitors can be especially odd at this.

Or you may already have done that.

An RCD that has not been operated often, sometimes needs a few shots to polish the moving parts, and gradually speeds up again, but total failures are more commonly never trip or occasionally trip without a fault.

Depending on the cost of the thing it may be simpler to replace it as a  lump.

Mike.

I agree. As I previously mentioned, provided that you are satisfied with the completion of all necessary tests in accordance with Guidance Note 3 to confirm the RCD is faulty