RCBO Tripping

[bernardgreen]

If one adds the multiple earth connections (see below) the PD between the incoming neutral (hence MET/CPCs) and true earth will be consdierably less than the voltage drop in the supply neutral cable (your "Vdrop").

The multiple earth connections do not signifcantly reduce the voltage drop along the neutral, they only pull the ground in the area around the earth rod closer to the neutral conductor's potential. If the ground was able to carry say half the neutral current from the rod to the star point at the substation then and only then would the voltage drop be reduced to half.

If the ground could carry the current then the DNO's accountants would be looking at using the ground as the neutral conductor to further reduce the costs of cabling. (*) It can't so the voltage gradient around a PME earth rod can be steep when there is there is a high current and therefore high voltage drop along the neutral.

(*) 25kv single phase railway traction used to rely entirely on ground conduction for the return which at 40mA per kilowatt was possible, In some areas it still does.

 

[JohnW2]

Yes this is a TN-S installation. I can't see the neutral fault being a candidate as I have measured the resistance and it is in the order or hundreds of MOhms for the earth neutral. I can't (even me) have cocked that measurement up.

In any event, with TN-S, Bernard's hypothesis centred on the (above earth) potential of the incoming N would obviously not explain any tripping due to a N-CPC fault (since the MET is connected, in the installation, to the supplier's earth, not the neutral) - that could only happen if (as in his diagram) there was a fault between Neutral and 'true earth' (not CPCs, and not bonded to CPCs) which seems very unlikely.

I think that there is a resistance between the live of of one circuit and the neutral of another. As this is the only thing that could have escaped my measurements to date.

Who knows - but that would be a very unusual state of affairs. With most (themselves all a bit obscure!) ways in which one might postulate that arising (e.g. the 'unfound' hypothetical TV booster), one would probably also have measurable resistance between some Ls & Ns.

It's the (seemingly consistent) 'asymettry' of what you described which still fascinates me. Unless that's due to differing operating characteristics of the two RCBOs, that does seem to defy any rational explanation.

Kind Regards,John

 

[JohnW2]

The multiple earth connections do not signifcantly reduce the voltage drop along the neutral, they only pull the ground in the area around the earth rod closer to the neutral conductor's potential.

Yes, that's true. However, it doesn't alter my main point - that your diagram assumed (and required) a fault between neutral and a 'true earth' which was not connected to the MET/CPCs - which, as I said, seems very unlikely in relation to what Martin has described.

Kind Regards, John

 

[martinxxxxxx]

Anyway am back there Thursday, considering making the fault-find into a live webcast - but worried that I have missed something so fundamental that I will become a Diynot outcast

Plan of attack:-

Isolate supply
Remove bonding conductors and earthing conductor.
Remove all neutral and CPCs from busbars
Unplug EVERYTHING in house
IR at 250V between all conductors with all switches in off position (tabulate results)
If all good IR at 500V as above. If not seek and find low resistance point.
Perform R1+R2 tests on all circuits starting with crossed rings (check r1,r2,rn 1st)
Perform Crossed L-N on rings

If all above show no abnormalities.

Connect one ring at a time and repeat tests starting with Down ring then Up ring (new out the box RCBOs to be used) Do Zs readings and plug in drummond at origin and at each socket. (Measure with multimeter and stray voltages on UP ring conductors, if found ground out all up ring conductors and repeat test, if it trips use circuit tracer to locate circuit path)

If that all passes add the kitchen ring then connect the other circuits one at a time using the above method.

If this fails, I will rewire the house!!!

 

[bernardgreen]

I think that there is a resistance between the live of of one circuit and the neutral of another. As this is the only thing that could have escaped my measurements to date.

I go back to capacitive coupling that cannot be measured with DC IR checks. A bunch of cables might just have enough capacity between the suspect live and other conductors to create a background earth leakage current that is not enough on its own to trip the RCD but when a few milliAmps of leakage from an appliance is added then the 30 milliAmp trip is reached.

I would use a milliAmp meter in series with the conductors at the output of the RCD to measure the background leakage from the cabling. Start on amps range to protect the meter or put a small wattage 240 volt lamp in series with the test leads as a protection. Even so use a cheap ( but safe to use meter ) as it is not impossible for something or some one to turn on an appliance and blow the meter.

With all other circuits connected disconnect the live and neutral of a single ring or radial and measure both the live and neutral currents. One lead to the conductor the other to the corresponding output of the RCD. Provided all appliances on that circuit are switched off or unplugged the only current would be due to capacitance between conductors.

This level of testing is NOT recommended for DIY

 

[JohnW2]

Anyway am back there Thursday, considering making the fault-find into a live webcast - but worried that I have missed something so fundamental that I will become a Diynot outcast

If it were me, I wouldn't worry about that! Whenever one is faced with a seemingly 'impossible' situation such as you have described, the most common explanation is, indeed, that one has done something very silly or overlooked something very obvious - but we are all human, and very few of us "have not been there" - so I (and, I would suggest also you!) would not feel bad or in any way ashamed! It's all part of the 'learning experience'.

I certainly look forward to hearing what you discover.

Kind Regards, John

 

[JohnW2]

I go back to capacitive coupling that cannot be measured with DC IR checks. A bunch of cables might just have enough capacity between the suspect live and other conductors to create a background earth leakage current that is not enough on its own to trip the RCD but when a few milliAmps of leakage from an appliance is added then the 30 milliAmp trip is reached.

As I've said, I personally think that capacitive coupling is a most unlikley explanation (apart from theoretical considerations, has anyone ever heard of this happening in a domestic installation?) - but I accept that one has to consider all possibilities. As you say, if this were the case it would have to be that the capacitive coupling was resulting in a sub-30mA background imbalance in the RCBO, which was then being 'pushed over the 30mA threshold' when a lest load with a L-E leak was connected to the circuit.

I suspect that the test load (a hammer drill) Martin has been using is very probably 2-wire (in which case could not add any N-E leakage) but the possibility you mention is one reason why I suggested that he tries with a simple, resistive, 2-wire test load. If that still results in tripping, then the above mechanism obviously could not be the explanation.

Kind Regards, John

 

[bernardgreen]

http://www.scribd.com/doc/62863090/35/Capacitance-to-Earth#page=36

The RCD Handbook, A BEAMA guide to the Selection and Application of Residual current devices.

9.2 Capacitance to Earth
The capacitance of 1.0,1.5,and 2.5mm2flat thermoplastic insulated twin and earth cable isapproximately 150pF per metre.It would not be unusual for a domestic installation to have100m of 2.5mm2cable and 250m of 1.0 or 1.5mm2cable,which would result in acapacitance to earth of up to 52.5nF.This would allow a standing protective conductorcurrent of 11µA/m or a cable leakage current of nearly 4mA for the whole installation (at230V,50Hz).

Bunched cables would have multiple stray capacitors with a live in each cable forming a capacitor with the earth in ajacent cable(s) as well as forming a capacitor with its own earth. Thus a higher overall capacity than for the same length of cable un-bunched.

 

[JohnW2]

The RCD Handbook, A BEAMA guide to the Selection and Application of Residual current devices.

9.2 Capacitance to Earth
The capacitance of 1.0,1.5,and 2.5mm2flat thermoplastic insulated twin and earth cable isapproximately 150pF per metre.It would not be unusual for a domestic installation to have100m of 2.5mm2cable and 250m of 1.0 or 1.5mm2cable,which would result in acapacitance to earth of up to 52.5nF.This would allow a standing protective conductorcurrent of 11µA/m or a cable leakage current of nearly 4mA for the whole installation (at230V,50Hz).

Bunched cables would have multiple stray capacitors with a live in each cable forming a capacitor with the earth in ajacent cable(s) as well as forming a capacitor with its own earth. Thus a higher overall capacity than for the same length of cable un-bunched.

All this theory is all very well, but I remain incredibly doubtful about any of it as a probable explanation of what Martin has described. Apart from anything else, as the figures above illustrate, one is unlikely to get as much as 1 mA such leakage in the cables of any one circuit, and your 'bunching theory' is unlikely to make that a lot worse. The anatomy of T&E cable is such that the distance from the L conductor to the CPC of any other cable would be appreciably greater (hence lower capacitance) than between L and CPC within the same cable, and it would be next-to-impossible for the L of one cable to be significantly close to more than, at most, one or two CPCs in other cables - and incredibly unlikley that such a close relationship would exist over any significant length of a socket circuit cable run (why on earth would such a cable have multiple other cables bunched with it for most of its length?!).

In any event, none of this could, in itself, explain the main 'mystery' of Martin's experience - namely 'cross-circuit' tripping.

As Martin himself has essentially acknowledged, I will be incredibly surprised if the problem, once identified, is not something fairly 'obvious' and far less esoteric than these theories which are being postulated.

Kind Regards, John

 

[martinxxxxxx]

I put both circuits on the same RCBO and it tripped, not sure how this could happen if stray capacitance between circuits

 

[JohnW2]

I put both circuits on the same RCBO and it tripped, not sure how this could happen if stray capacitance between circuits

As I understand it, Bernard isn't really addressing the 'cross-circuits' issue. He was merely postulating that capacitance with L's and CPCs (in general) were resulting in a background imbalance (<30mA) through the RCBO and then you were plugging in a test appliance which had a L-CPC leak which added enough to bring the imbalance to the 30mA trip threshold.

For the reasons I explained earlier, I really don't believe this theory and, in any event, it would not, in itself, explain the cross-circuits effects you're seeing.

To suggest that capacitive coupling between the two circuits could be the explanation would be close to ridiculous - since, by definition, very little of the cable runs of upstairs and downstairs circuits are going to be in very close proximity to each other - so any capactivive coupling would be so small as to be irrelevant.

As I've said, I would put a lot of money on the answer (when you discover it!) being something 'normal', simple and straightforward ('obvious') and not in any way 'esoteric'!! One of the very first things I was taught in my higher education is that "common things are common"

Kind Regards, John

 

[RF Lighting]

Personally I'd disconnect both rings entirely, and power down / disconnect the rest of the installation.

Then connect one line leg of the first ring to the RCBO and power it up.

Check all the wires and see that only the line leg two of the first ring comes live. Use a high impedence volt meter.

This will prove the ring is not crossed. It will also show any interconnections to the other ring, and if the neutrals come live on any of the installation, it will show a borrowed live.

If no neutrals come live, then plug in a resistive load such as a filament lamp to a socket on ring one and check that only the neutrals for ring one come live.

Repeat the process for ring two.

This should show any interconnects, crosses, hidden loads etc.

Again, this is not DIY testing.

 

[Peet]

So the DOWN RCBO in case 1 or 2 (below) saw an imbalance and tripped. The UP RCBO is capable of doing its job once the DOWN live has gone, suggesting no L to CPC or GND leak or no N to CPC or GND leak on the up circuit.

If the neutral for the DOWN circuit was somehow going to GROUND (not the CPC but GROUND) then I can see a way that all 3 conditions (see below) you list could coexist.........


1)

If any socket on the Up circuit is used - the RCBO for down trips, but the up circuit stays on and continues to work

The send & return currents for the UP circuit are fine and there's no connection to another circuit, however when the neutral current returns through the CU busbar, part of it returns to the incoming neutral and part goes through the down RCBO neutral to GND, tripping the down RCBO. JohnW2's suggestion of using a small resistive load to see what maybe happening here I think is a good one. As neutral current increases due to the load on the UP sockets the amount of current going to GND would also increase to where it would trip the DOWN RCBO.
The UP RCBO never saw an imbalance because there wasn't one.

2)

If any socket on the down circuit is used the down RCBO trips.

Essentially the same although the current path is a little different and some of it goes to GND directly now, again tripping the down RCBO.

3)

put everything onto one rcbo so that both circuits shared the same RCBO - and guess what - the RCBO tripped straight away.

Is possibly the most interesting combination as now you don't even need to plug in a load anywhere to get it to trip. Previously the system was managing to 'deal' with a 'fault' that appeared with the plugging in of a load by tripping out. Now it can't even do that and never stays in. It might be this configuration that's easiest to troubleshoot. In this state I would check its resistances L to E and N to E and then also L to ground if possible, and N to Ground if possible, with the two circuits connected together like you've said but not into the RCBO.

I was going to say that (and this links to the other RCBO tripping discussion) by having the neutral and GND (two low impedance sources) connected together, small voltage differences cause large currents. But that wouldn't explain why the down RCBO ever stays in. Maybe the connection of the 2 circuits joined together is enough wiring to pick up enough noise to do the job of tripping it? Noise pickup will be directly proportional to loop area.

Once the down RCBO has tripped ( the only RCBO that monitors a path to the fault) the fault is still present ( as the RCBO only disconnected the line) and none of the other RCBOs can respond to it as everything else balances as far as they're concerned. On an old board with no RCD/RCBOs you'd never of seen the fault.

http://www.scribd.com/doc/62863090/35/Capacitance-to-Earth#page=36

The RCD Handbook, A BEAMA guide to the Selection and Application of Residual current devices.

9.2 Capacitance to Earth
The capacitance of 1.0,1.5,and 2.5mm2flat thermoplastic insulated twin and earth cable isapproximately 150pF per metre.It would not be unusual for a domestic installation to have100m of 2.5mm2cable and 250m of 1.0 or 1.5mm2cable,which would result in acapacitance to earth of up to 52.5nF.This would allow a standing protective conductorcurrent of 11µA/m or a cable leakage current of nearly 4mA for the whole installation (at230V,50Hz).

Bunched cables would have multiple stray capacitors with a live in each cable forming a capacitor with the earth in ajacent cable(s) as well as forming a capacitor with its own earth. Thus a higher overall capacity than for the same length of cable un-bunched.

One thought about their figure of 150pF/M. I've just stuck a capacitance meter across an old, regular 6.8M long piece of 2.5 T&E. L-E measured 880pF and N-E 875pF, so it's coming up at about 130pF/M. Fine, not a bad match for their 150pF/M. But that's for 2 wires in intimate uniform contact along their entire length. ie for the same piece of cable or a radial leg.
The capacitance will tail off with cable separation, I was going to guess that the rate of tail off would be inverse square law but it turns out that parallel wires are an inverse natural log function, so perhaps inverse square like, if my dodgy maths holds up. But either way the capacitance will fall away fairly quickly with even a small distance from cable to cable. So for that reason even if these cables were sandwiched together in conduit, I'd guess their capacitive coupling wouldn't approach 130pF/M. Add to that that they'll probably not be close coupled for that far in a house before they go their seperate ways, so I doubt the effect you see would be inter cable capacitance. Unless the system was so close to the edge that the RCDs were on a hair trigger for other reasons.

 

[JohnW2]

If the neutral for the DOWN circuit was somehow going to GROUND (not the CPC but GROUND) then I can see a way that all 3 conditions (see below) you list could coexist.........

As you go on to say, sort-of. A N-GND fault (or an N-CPC fault with TN-S or TT) could explain Martin's first two situations. However, as you say, such a fault would not, without other factors, explain the third situation (RCBO trips immediately, even without load, if both upstairs and downstairs circuits are connected to it simultaneously). As you say, for that to happen with both circuits connected to the RCBO, but not with just the downstairs circuit (with fault) connected requires one to postulate a 'marginal' fault plus some additional (unknown) factor (your 'noise' suggestion sound pretty unlikely).

However, I do agree with you that it seems unlikely that there is anything wrong with the upstairs circuit and therefore that investigations might most profitably concentrate on (or, at least, start with) the downstairs circuit.

Kind Regards, John

 

[JohnW2]

[nerd mode] - for nerds only - others please ignore

One thought about their figure of 150pF/M. I've just stuck a capacitance meter across an old, regular 6.8M long piece of 2.5 T&E. L-E measured 880pF and N-E 875pF, so it's coming up at about 130pF/M. Fine, not a bad match for their 150pF/M. But that's for 2 wires in intimate uniform contact along their entire length. ie for the same piece of cable or a radial leg.
The capacitance will tail off with cable separation, I was going to guess that the rate of tail off would be inverse square law but it turns out that parallel wires are an inverse natural log function, so perhaps inverse square like, if my dodgy maths holds up. But either way the capacitance will fall away fairly quickly with even a small distance from cable to cable. So for that reason even if these cables were sandwiched together in conduit, I'd guess their capacitive coupling wouldn't approach 130pF/M. Add to that that they'll probably not be close coupled for that far in a house before they go their seperate ways, so I doubt the effect you see would be inter cable capacitance. Unless the system was so close to the edge that the RCDs were on a hair trigger for other reasons.

For reasons I explained yesterday, I agree with you that, except in the most marginal of situations, it is incredibly unlikely that there would be enough capacitive coupling between conductors in domestic wiring cables to significantly affect the operation of an RCD or RCBO on a single circuit – as I said, one would be very hard-pressed to get as much as 1 mA such ‘leakage’ on any single circuit. Furthermore, I certainly agree that capacitive coupling between cables in two different domestic circuits is going to be essentially negligible (with any such ‘leakage’ likely to be very much less than 1 mA.

However, in terms of the maths of what you say, I don't think that it is quite right. The relationship between the capacitance between two wires and the distance between them is not quite as simple as ‘inverse natural log’. As can be seen in the equation below, it is the ‘inverse natural log’ of a function of the ratio of wire separation to wire radius. Interestingly, it’s only this ratio that matters, not the absolute dimensions – hence the capacitance between 1 metre lengths of two 1 mm radius conductors with their centres 4 mm apart will be the same as the capacitance between 1 metre lengths of 10 mm radius conductors with their centres 40 mm apart, or 100 mm radius ones 400 mm apart.

Anyway, my point is that, far from being “inverse square like” (as you suggest), the relationship between capacitance and wire separation is actually a surprisingly ‘weak’ one, in fact significantly ‘weaker’ than even a ‘inverse linear’ relationship – as illustrated in the graphs below. For example, if one doubles the separation/radius ratio (e.g. double separation without changing wire diameter) from 4 to 8, the capacitance only falls to about 64% of its original value, whereas it would fall to 25% with an inverse square relationship or to 50% with an inverse linear relationship. Intuitively, I would not have expected an inverse square relationship (since the ‘lines of electrical force’ are all essentially in one dimension), but I would have expected it to be fairly close to inverse linear (which is the case for two, relatively close ‘plates’) – which I suppose it roughly is.

However, as I said, despite all this (which makes capacitances between different cables a bit higher than you were imagining), I certainly don’t think that ‘between-cables’ capacitance is even remotely significant in the context we’re discussing.

Untitled

• JohnW2
• 27 Sep 2012
• 1

[/nerd mode]

Kind Regards, John