Leakage current on installations with solar

Question

I have been looking at a domestic installation that has solar and batteries 
 
 The solar and battery ( Solar edge and Fox ESS ) are on their own sub board via henleys 
 
 The main house circuits on a split load board. 
 
 The leakage current is very high on the tails of the main board, round 350mA up to 700mA 
 
 With everything isolated, no leakage. With solar all off and isolated, and the main switch of the house CU on, but no circuits energised, 
 I get a bump in leakage of about 15mA 
 
 When I switch on individual MCBs, I get a leap of about 100mA on some circuits and 40mA on others 
 with all house circuits on, but no real current draw,(nothing switch on) 
 I'm up past 700/ 800mA and it fluctuates up and down rapidly. 
 
 I thought about disconnecting the isolated solar boards main earth, but a s I know little about solar 
 I do not want to start disconnecting anything really. 
 
 With isolated solar, can you still get leakage from it, which can find its way onto 
 the rest of the installation, and as circuits are closed, the leakage increases. 
 More parallel paths? 
 Any advice would be appreciated. Thanks

AJJewsbury · Accepted Answer

JohnE said: Close any MCB RCD side or non RCD side and I get a leap of 100mA or 40mA per circuit 
 if the RCD is working correctly (i.e. trips at no more than 30mA?) 
 If so and the RCD isn't continually tripping then the fault must be on non-RCD side. A N-PE fault could give those symptons - provided each of the circuits had some load connected (not a massive load, but something - a few tens of watts perhaps) - switching them on then increases the N current, hence raises the N voltage a little at the CU due to voltage drop in the supply which then pushes more current across the fault. Could be aggrivated if there was an extra impedance in the N somewhere (e.g. loose connection) - worst case if the N was entirely open circuit and all the N current was finding its way back via the fault. 
 - Andy.

mapj1 · Answer

CU and already 15mA with all MCBs open 
 Right, the only way that is possible is if you have a neutral earth leak between two objects either within the board before the RCD, or assuming the RCD is not defective, on one of the branches on the non RCD side.. (a fault on the RCD side is not possible, unless the RCD is busted...) 
 Note that unlike the latest compact RCBOs MCBs don't break the neutral, so this path is present even when the MCB is off.. 
 By having a second N-E path in parallel with the one that the DNO provides, the neutral current is split, most of it going down the low resistance meter tail of course, but a resistance dependent fraction going the other way and returning via a mixture of CPCs and bonding to plumbing etc, so when the balance is measured, the neutral tail is missing a few electrons relative to the live one. 
 Even when no load is applied, note that the supply neutral may be a few volts off true earth and also relative to any shared metal plumbing etc, due to the voltage slope in the street cables from the neighbour's loads, (the neighbours presumably share metal mains services, so current can dive down the plumbing and take a shortcut, rejoining the neutral at a house nearer the substation, )So even when you dont have any loads, that street voltage offset drives a small current through that NE fault 
 
 Buzz NE in the box with the mains incomer off, so the ohm meter should not see any NE bond, and let us know what you find, 
 regards Mike.

mapj1 · Answer

well you have detected your extra NE bond causing some of the neutral current to be diverted round the parallel path to the meter tails and show as if it was an earth leakage , more or less as predicted ;-) 
 I suggest this explains all the observed weirdness to date. 
 Now the hard part will be locating where that fault actually is! ~70 millohms is at the level that is very unlikely to measure accurately with general testgear, as the instrument tolerances and contact resistances and so forth all conspire against you. Also you will be lucky if the fault itself is zero resistance and all the measured resistance is down to the wiring distance (at one extreme it could be a 70 milliohm fault directly behind the CU for example or at the other a gold plated dead short as you deduce some 2-3 m away). So really all you can say is that there is a problem, and probably somewhere within the first few metres. 
 But, as well as the CU mounting screws, the nearest light switch, ceiling rose or curtain rail mounting screw above the light switch are all good contenders for the 'unintended connection of the year' award. ,,, 
 M..

mapj1 · Answer

It could have been a fault since first installation, but it would have made the installation cert and the IR values for that circuit interesting. By which I mean fraudulent ! Not completely unknown of course, 
 Id be looking to guesstimate the route and after looking behind boxes, to look for shelves or anything else that screws or in that may have been added over the years -perhaps even floor boards that have been nailed to de-squeak them 
 I have been known to run a new CPC or whole new T and E in trunking and abandon short lengths with faults that cannot be found - its often quicker - certainly all the detective skills will be needed. Don't assume that the fault is not a bit further than you think - not all cable has the maximum resistance, and not all meters round their readings down rather than up in the way you might expect. 
 you should also be able to measure the NE resistance from a few light fittings, and see it getting higher further away from the other side of the fault. As a quick temporary check that there is not a second fault lurking, and only if we are very sure that the supply polarity is correct, you could put the lights CPC under the same screw as its neutral, and confirm that all the weird leakage current effects in the meter tails are indeed greatly reduced, This would at least mean that lights fault was the only remaining problem. If you do that test, don't leave it like that though ! 
 Mike

AJJewsbury · Answer

JohnE said: With a reading of 0.07Ω for 1.5mm2 T&E (30.20mΩm) i think that would that give me a loop of about 2.4m. Which theoretically would suggest that my short is quite close. Unfortunately no obvious N/E connections at all accessible points. 
 My next guess would be a nail through the cable then (with flat T&E it's common enough for a nail or screw to hit the c.p.c. and one of the conductors while leaving the other with perfect insulation). Any new pictures or other DIY jobs? 
 - Andy.

mapj1 · Answer

yes - that's the sort of lateral thinking sort of thing I meant - then the defective section can be cut off short so no-one tries to use it in future, and in ten years time when the floor is lifted for some totally unrelated reason, the original problem can be exposed. I cant see it of course, but it sounds like you have identified a reasonable way out . 
 There was lots of this sort of thing in the 90s as RCDs first became common, as NE faults that had lain dormant for decades on hot wire fuses suddenly prevented the new consumer unit from working - with much wailing and gnashing of teeth, as of course in the eyes of the householder, 'it worked just fine before you put in that new fangled fuse-board' 
 I've also left someone with most of their lighting circuit on a flex plugged into the socket circuit for a week while waiting for other building work to be completed - there are a great many 'tricks' to get out of trouble - I don't do domestic wiring except friends and family now, but the principles have not changed. 
 Mike.

Chris Pearson · Answer

JohnE, well done for persisting. 
 Here was my N-E fault ( posted here 8 months ago ): 
 
 mapj1 said: There was lots of this sort of thing in the 90s as RCDs first became common, as NE faults that had lain dormant for decades on hot wire fuses suddenly prevented the new consumer unit from working 
 My fault would have become apparent when I switched the circuit over to a new DB, but most people don't go around measuring earth leakage, so it has been there a long time: it is 40-odd years since the place was re-wired. 
 It really does not say much for the quality of initial verification.

alanblaby · Answer

mapj1 said: It could have been a fault since first installation, but it would have made the installation cert and the IR values for that circuit interesting. By which I mean fraudulent ! Not completely unknown of course, 
 It goes on more than you'd think possible. New house builds are the worst from my experience. If it doesnt trip the RCDs when turned on, it passes. 
 I brought one up here a few years, a new council house, the Tenant had connected the new cooker, but it didnt work. I go there, test the terminals, indeed, it was dead. I go to the CU, to find the 10mm T+E cable to the cooker lying loose inside the cu, not connected on either LN or E. But dont worry, it had been tested a few months previously, as the EIC was there, and all readings were perfect.

alanblaby · Answer

T-N-C-S presumably. Fault downstream on a neighbours property causing neutral current diversion through your property? That's a total guess. 
 Totally isolate the PV, even the CPCs, to see if it is that causing the fault. Presumably there is an isolator on the DC side to turn that off? 
 Narrow down where it is coming from, as you say, anything with a CPC or a class one accessory could be the cause with a N-E fault.

AJJewsbury · Answer

The leakage current is very high on the tails of the main board, 
 
 The main board then, not the solar/battery one? I take it you're clamping the L&N tails but leaving the earth out of it, Sounds more like a conventional fault. If it increased with load I'd suspect a N-PE fault somewhere (although it could still be that, if loads in neighbouring properties happened to be pulling the supply N up or down). The symptoms you mention sound more like rodent damage to cables to me though or damp ingress. Was you IR test done with the c.p.c.s still connected to Earth? Other possibilities might be a shared N between the sub-db and the main one - hopefully unlikely for a recent install, but maybe if there are controls/sensors involved. 
 - Andy.

lyledunn · Answer

So we keep a record of leakage on every inspection we do. This one is fairly typical of what we are seeing in a small single phase installation. All circuits on, but not necessarily with load. Top photo complete installation, bottom photo measuring separate tails to the PV board which then supplies the inverter.

AJJewsbury · Answer

I think what I need to do is isolate the main board including earthing conductor, and split the neutrals and do some proper IR tests 
 
 Generally you leave the c.p.c.s connected to the means of earthing - otherwise you're more likely to miss damaged cables that are in contact with the general environment (e.g. damp masonry or structural steel) but not the c.p.c.. 
 - Andy.

Chris Pearson · Answer

JohnE said: When I switch on individual MCBs, I get a leap of about 100mA on some circuits and 40mA on others 
 with all house circuits on, but no real current draw,(nothing switch on) 
 I'm up past 700/ 800mA and it fluctuates up and down rapidly. 
 
 I think that I am about as confused as you are. 
 If nothing is switched on, where is the current flowing? 
 One possibility is FCUs (with the appliances' functional switches open), but surely not on every circuit. 
 If the leakage occurs when the solar is isolated (with a 2-pole switch), I cannot see how it can be relevant, so perhaps forget that it is there for the moment.

Chris Pearson · Answer

Like the OP, JohnE, I claim no expertise in solar tackle. 
 My understanding is that a 2-pole switch is usually provided. If not, the possibility of a N-E fault in the solar tackle is not excluded.

Leakage current on installations with solar

Community Rules & Guidelines