The problem with AC RCDs is that they have no defined response to d.c. currents at all - the standard simply doesn't mention anything on the subject - so there's no generic safe level. Individual devices well well be safe - if a 30mA normally trips at say 24mA you might imagine that even with 6mA of d.c. bias it should still trip at 30mA - but of course individual devices can vary, and a ramp test only tells you how it responds in one set of circumstances - test on another day of the week, at a different ambient temperature, or with a slightly different variation in supply voltage, the result may well be different - as long as it trips by 30mA it would be compliant, and you'd have no margin at all. There was even a rumour a few years ago that RCDs supplied into the UK market labelled as AC types were really A types under the hood (to simplify manufacturing as other countries were demanding A types, while meeting UK expectations for AC types). I guess the least worst option is to ask the manufacturer of the particular device concerned - as they may be able to define d.c. performance that's above and beyond the BS EN.

  - Andy.

Or if you have a tester that can create a test with the lumpy DC (unsmoothed rectified) superimposed that an A type should respond to, there is no harm in seeing what that specific RCD does to either set your nerves jangling or give a warm feeling.

And you could consider how serious is it if that RCD is blinded - if the whole site loses RCD all cover, including all the sockets, then it is more serious matter  than if a only lighting circuit that until a few years ago would not have needed an RCD anyway is affected.

There are not that many credible failure modes of LED lights or induction hobs that actually would trip a type A and not a type AC it needs a fault to earth after the rectifiers that does not actually blow the diodes to smithereens or open a series fusible resistor or similar weak link. Possible, but rare.

I can see the dilemma.

I do not think that your average reasonably confident DIYer would give a second thought to updating a fluorescent batten to LED, or indeed any other shape of incandescent or fluorescent fitting to an LED one.

Where does this stop? Should I not have replaced my incandescent lamps with fluorescent ones and now LED? (Let's forget that I don't have RCDs save on a few sockets.)

On the other hand, knowingly increasing a risk is a different matter entirely. Do the manufacturers of the new battens specify the type of RCD which should (must?) be used?

There are not that many credible failure modes of LED lights or induction hobs that actually would trip a type A and not a type AC it needs a fault to earth after the rectifiers that does not actually blow the diodes to smithereens or open a series fusible resistor or similar weak link. Possible, but rare.

The other consideration is standing d.c. currents desensitizing a 30mA RCD intended for additional protection - so for example the RCD tripped at ≤40mA rather than ≤30mA - so someone touching a bare conductor or a picture nail driven into a concealed cable elsewhere on the circuit could now receive a >30mA shock without the RCD tripping (the actual shock current being limited by body resistance plus any anything else in the path - e.g. shoes).

   - Andy.

An associated issue when replacing conventional fluorescents with LED versions would seem to be if there are any occupancy sensors controlling the fittings being replaced.  Sensors which have coped for years with, say, a couple of twin 1500mm battens tend to give up the ghost when asked to control two single tube LED units - or are we just unlucky?

There won't be standing DC current to earth in either of the types of device described, unless faulty,. Leakage to earth through undamaged filter capacitors has no DC competent - it may not be sinusoidal but the capacitor will charge up to any mean DC offset and leave only the AC component flowing.

EVs actually use the CPC to carry a 'pilot' DC that detects the charger is connected to a real car, so there is slightly more possibility for confusion there. but even so to blind an RCD still requires a fault between true earth and the pilot wiring in the charger lead.

Mike

Maybe not just the filer capacitors to consider though. A lot of LED fittings are Class I with the LEDs in close contact with earthed metal - (I'm thinking of one bulkhead in particular where the light consists of LED tape stuck to a combined heatsink/reflector) with what I suspect is not an entirely galvanically isolated  driver. Any poor insulation between the two (damp, dirt...) carries a risk I would have guessed.

   - Andy.

perhaps - but unless it fails an IRtest /PAT then it should be good for at least 500VDC between L//N and the CPC,   I agree there is some dreadful non-compliant rubbish on sale out there, but anything class I and legitimately CE/ UKCA  marked as suitable for use on the mains, will be fine on a 500V  insulation tester, and a high potential test for at least a minute will have been made at factory on samples of the product as part of the 'proof' it meets the low voltage directive - a requirement for CE and similar.

The tests to meet the low voltage directive are described informally here, as the actual standards themselves are surprisingly opaque and expensive for what they are.

And a surprisingly thin layer of insulation will hold off mains if not mechanically abused. Figures like 10-20kV per mm for polythene and PVC mean that even a single turn of undamaged cling film can be an adequate mains insulator, not that I'd actually recommend it. The hard part with such thin layers is the 'undamaged', as almost any handling introduces weakened regions from which failures can propagate.

Mike.