Hi Graham

Welcome.

You need to isolate/unplug all loads and try again. You should then find the leakage is substantially zero. Then it is a case of finding the high leakage loads one at a time, and working out why they leak. However, if this is worse in damp weather, I think you may have a damaged cable problem or something similar allowing water to get the the electricity! If all the wiring seems to leak somewhat then insulation testing is the order of the day, you should find the problem fairly quickly.


Regards

David

Your approach in isolating the leakage in each cct is correct, the next step is to divide to the device level. A test extension lead with the 3 cores split out for a clamp meter may be useful.


Be wary that in a PME house the current down the CPC may not be the same as the leakage an RCD sees, try to measure the way it does threading L and N together through the core of an AC clamp meter.


17mA is quite a lot for a house at idle, note that it does not stack with the 37k ohms (230/37 is about 6mA - high but not the full story ). For IR and leakage currents not to be consistent  is common, and is telling you that you have a mix of capacitance between L and E, and resistance.


Subdividing over more  RCDs or retrofitting RCBOs in place of the MCBs  if available from that maker,  or at the top of the scale  a whole new CU, are the nuclear option(s). Lets keep them in mind at however many pounds an hour,  but try to avoid it. The halfway is to fit Henlys in the tails and have another CU for some of the stuff.


Things that leak to earth resistively,


1)anything with a metal sheathed heating element, cookers, (spilt liquid getting inside the works ) immersion heaters , kettle, dishwasher etc. (sometimes the jacket of the element develops a pin-hole corrosion and the inner mineral insulation, often magnesium oxide powder tight packed then slowly gets wet.)

2) Any appliance where water could get in, things with outside lights, outside cable joints ... poor sealing where cables go through walls. Almost anything under a leaking radiator or pipe.


Things that leak to earth capacitively by design,

motors with suppression capacitors, IT equipment like laptops and monitors (a few mA each is possible) Some LED drivers - the ones with an 'electronic transformer' - really a switch mode power supply.


How much of this sort of stuff is there ?

Let us know how you get  on.

Mike.


PS EDIT

I'd not be happy  with more than about 10mA total leakage on any one  30mA RCD, ideally less, and I''d like a much higher DC result for the insulation test - ideally a few megohms, or at least being able to point to one appliance. As an aside, some designs of induction hob are quite bad and end up being on their own RCBO.

"Insulation resistance test of the whole installation (L & N to E at 250V) gives 37kohms."


Well, that's not right. Assuming that only one side has a low resistance, that is 6 mA. As a first step, find out which circuit is responsible.

37K global L+N to E whilst not a trip in itself is not so good. Any " leakage" in  addition to that on any circuit or appliance whilst running current could easily throw any RCD into trip therefore with a few circuits intermittently adding a bit can cause a "random" problem, fridges, freezers and CH to name a few. I usually talk the end user into RCBOs if possible. Having said that I`ve had a front end RCD set up for decades now with very few problems causing RCD trips, usually a kettle or a steam iron but watch ou for toasters full of toast crumbs and even the odd ali milk bottle top 

+1 for 37kΩ insulation resistance being a red flag. I'd expect something at least three orders of magnitude higher than that in healthy domestic (at least with appliances disconnected).


That low, I'd expect something semi-conducting - a lot of damp having got into somewhere it shouldn't (outside lights/PIRs?) or rodent damage or a charred cable - just the sort of thing not to be stable but vary with the oddest conitions (e.g. after rain, but only with the wind in a certain direction, or at the whim of a mouse or when an odd corner of the house gets to a certain temperature).


   - Andy.

There are another couple of things ref loose connections on supply side adding their tenpennorth too

Thanks all for your thoughts and suggestions - really helpful.

I'm going to re-test circuit-by-circuit to find out where the leakage current, both capacitive and resistive, is going - then decide on the correct fix.  It should be an interesting piece of technical detective work!

I'll post the results as soon as I have them.


Graham

Yes, that's what I meant really. I'm just interested in seeing how the earth leakage current I measure with the clamp meter reduces once all the appliances are disconnected, then comparing that measurement with the IR test results.  I will be IR testing circuit by circuit.


Graham

Also note the make and model of the consumer unit - you may (or not  ?)be able to get an RCBO to retrofit  for it, if say you find that it is just the outbuilding power that is touchy,

Mike.