I'd agree with your assessment that the installation is likely safe for use. The problem is likely a combination of all these things mentioned: damp in outside stuff ( do you have DP isolation for all outside circuits, could you just try leaving these all off for a couple of weeks), not the best insulation values overall),  but a single RCD is always going to prove problematic, and although RCBOs seem expensive: how many times can you go back? If you could switch some or all of it over- RCD to control internal fixed appliances: Boiler, shower, RCD; yes these can still cause tripping but can usually be double pole isolated if they fail, and RCBO for all socket and lights circuits; possibly only 5 RCBOs for a modest house. Are your meters calibrated? can you try a ramp test? If you're sure you're right, you need to discuss properly with the customer, or ask them to get a second opinion. Their system is safe, the RCD is doing it's job; Its going to cost them money if they want it to stop it  tripping.

Thanks for the update, sounds quite frustrating.

There is something odd about the lights and power both showing 67k ohms-  perhaps the neutrals are conjoined somewhere - are there wall lights ?


Neons draw a current that is very voltage dependant. so if you see the same resistance reading at 500V and at 250, and if you have an ordinary meter at half a dozen volts as well, then it is not a neon indicator . (typically neons are open circuit up to about 70 to 90V, then they strike and start to conduct - the rest of the voltage is then dropped across a series resistor of  100k ohms or so for the small ones, and a bit less for the big ones.) So if you want to know if it is a neon to ground instead of to neutral, vary the test voltage.


That conduit lark sounds like a total hack - the sort of thing that was always supposed to be temporary that becomes permanent.


Personally I prefer to test L+N to E - as even if there is some hidden load wired L-N in a cupboard, it will not be damaged by the test voltages, as both sides of it are at the test voltage at once , and there is no question that nothing much should be wired either L-e or N-E so testing with L and N in parallel together is useful.



Certainly before changing the CU, you will need to make all reasonable efforts to find the problem, you should not in good conscience leave it with known faults, unless there are exceptional circumstances that justify a departure.

Under floor faults  can be very variable - puddles when it rains, mice if the airbricks are damaged,  all out of sight and hard to inspect - and maybe the odd suspended joint box or chock-block flapping in the breeze. And if it is  Victorian there really should be a noticeable breeze or it will have damp and  rot sooner or later. 

It may be possible to gain at least an inspection access under a kitchen  cupboard or in the under the stairs even with a small hole a mirror and a light can sometimes reveal a great deal.



Also if there are conjoined neutrals between circuits then while the affected circuits are on the same RCD you will not realise, but if any of the current that leaves via the live terminal on one RCD or RCBO comes back via the neutral on another, then both will see an imbalance and  pop off.- better check circuit to circuit isolation  for that before fitting that all RCBO board...

For breaking and rejoining circuits quickly for divide and conquer testing, where there is a mass of looses ends, and for things where the probes do not quite fit,  I find these
3 pole lever connectors,  very useful.


M.

There will always be nuisance R.C.D. tripping causing much inconvenience and disruption with consumer units with just one R.C.D. as the main switch, and many final circuits. A change to an all R.C.B.O. board will be much more reliable and cause less trouble. If you tinker about and future nuisance tripping occurs you will most definitely fall out with your customer. Do a good reliable  job and charge the going rate and they will be happy.


Z.

Installing a replacement consumer unit with a RCBO for each circuit could conceal over 150 mA of leakage, which is not always desirable.