Why isnt the RCD tripping?

[HappyHippyDad]

I'll start with an embarrassed face
Davesparks was right.... again! Damn that man!

Took the outgoing conductors out of the RCD, RCD tests fine. In fact all I needed to do was switch the main switch off of the CU and it tested fine.
As soon as the main switch is switched back on then the RCD will not trip when tested.

I removed the neutral of the circuit with the fault (0.27ohms N-E) from the neutral bar and then the RCD tested as it should even with the main switch on.

Obviously the only reason my RCBO worked on another corcuit was because it wasn't being affected by the N-E fault on the faulty circuit, obvious now!
So..the RCD is fine but the short between N and E is stopping it from functioning when tested.

I shall code accordingly.

However, I do have one last question... Will this RCD still work as it should whilst the fault is there? Do the clients have working fault protection?

 

[davesparks]

I'll start with an embarrassed face
Davesparks was right.... again! Damn that man!

However, I do have one last question... Will this RCD still work as it should whilst the fault is there? Do the clients have working fault protection?

I have been reliably informed by my better half that it is in fact she who is always right, I am either wrong or not listening properly.

The RCD wont provide earth fault protection (well it will for a big enough fault, see my lats paragraph) however it will provide some shock protection.
If you received a shock to true earth the rcd should work, however if you recieved a shock to earthed metalwork I don't think it will work.

Apologies if you know this already but what is happening is that current flowing to earth is diverting via the cpc of the faulty circuit, it then travels via the fault back to the neutral bar and then via the RCD to the supply transformer.

When you established the Zs of 1.5ohms through the bonding how exactly did you test this? If you tested this with the bonding connected to the MET along with all circuits connected then you've more likely measured the return path via the fault than the bonding.

Of course some fault current will still flow to the earth rod in accordance with the rule of resistors in parallel, but as the Zs to Ra ratio is something like 1.5 to 300ish it will take a fair whack of fault current to cause the RCD to trip.

 

[HappyHippyDad]

I have been reliably informed by my better half that it is in fact she who is always right, I am either wrong or not listening properly.

The RCD wont provide earth fault protection (well it will for a big enough fault, see my lats paragraph) however it will provide some shock protection.
If you received a shock to true earth the rcd should work, however if you recieved a shock to earthed metalwork I don't think it will work.

Apologies if you know this already but what is happening is that current flowing to earth is diverting via the cpc of the faulty circuit, it then travels via the fault back to the neutral bar and then via the RCD to the supply transformer.

When you established the Zs of 1.5ohms through the bonding how exactly did you test this? If you tested this with the bonding connected to the MET along with all circuits connected then you've more likely measured the return path via the fault than the bonding.

Of course some fault current will still flow to the earth rod in accordance with the rule of resistors in parallel, but as the Zs to Ra ratio is something like 1.5 to 300ish it will take a fair whack of fault current to cause the RCD to trip.

Yes, Zs was tested with bonding connected to MET, so the blasted fault current was just coming back through the neutral due to the fault!

This is what was happening when the RCD was being tested as well. There was no imbalance because the current I produced when testing was coming back through the neutral which was a much more attractive path than the 300ohm rod, again due to the fault! I wonder what Ra would be needed to counteract this!

I do appreciate you explaining that Dave.

Now it's got me thinking more..

Has this happened in this case due to the fault being so bad, i.e 0.27ohm (and in effect joining the earth to the neutral)?
If the IR had been higher, lets say 500ohms or even 6000ohms (both of which should trip an RCD) would the RCD have tripped? I guess we'd have to do some maths, but I expect there would be an exact theoretical IR figure where the RCD would start tripping (depending on the relationship between the Ra of the rod and the degree of severity of the IR fault)?

 

[Risteard]

As above, the neutral/Earth fault has desensitised the RCD and therefore the nominally 30mA protection is not being effectively provided.

 

[HappyHippyDad]

Also, does this mean that on a TT with a poor Ra, if there is a dead short between N-E on one of the circuits, then an up front RCD will never trip when tested (with main switch still on)?

 

[davesparks]

Yes, Zs was tested with bonding connected to MET, so the blasted fault current was just coming back through the neutral due to the fault!

This is what was happening when the RCD was being tested as well. There was no imbalance because the current I produced when testing was coming back through the neutral which was a much more attractive path than the 300ohm rod, again due to the fault! I wonder what Ra would be needed to counteract this!

I do appreciate you explaining that Dave.

Now it's got me thinking more..

Has this happened in this case due to the fault being so bad, i.e 0.27ohm (and in effect joining the earth to the neutral)?
If the IR had been higher, lets say 500ohms or even 6000ohms (both of which should trip an RCD) would the RCD have tripped? I guess we'd have to do some maths, but I expect there would be an exact theoretical IR figure where the RCD would start tripping (depending on the relationship between the Ra of the rod and the degree of severity of the IR fault)?

What Ra would be needed to counteract this?
The short answer is that there isn't a value of Ra which could counteract this fault. The fault is a resistance in parallel with the fault, so assuming a theoretical textbook situation rather than the real world:
With the return path of 1.5 ohms via the fault an equal Ra of 1.5ohms the ratio of current flow is 1:1 so a 30mA RCD would need a 60mA fault to trip it.
With an Ra of 0.15 the ratio would be 1:10 so your 30mA RCD would need a 33mA fault to trip it (if my mental maths is working today)
Obviously in the real world RCDs trip at less than 30mA,usually 27ish, so you could in theory get Ra low enough to still have it trip at 30mA,but you'd be looking at an Ra of a fraction of an ohm which is quite tough to achieve without a lot of buried metal.

You can apply a similar answer to your next question.
If the fault was 300ohms and your Ra is 300ohms you would again get the 1:1 ratio. So as the resistance of the fault goes up you will reach a point where its effect becomes negligible on the tripping of the RCD.

With this fault on the system most of the circuits may now appear to have a Zs low enough to allow the mcbs to provide fault protection, apart from the minor issue that all the fault current will be going via the 1mm cpc and neutral of the lighting circuit,provably causing quite a bit of thermal stress.

 

[davesparks]

Also, does this mean that on a TT with a poor Ra, if there is a dead short between N-E on one of the circuits, then an up front RCD will never trip when tested (with main switch still on)?

Even on a TT with what is generally accepted as a good Ra this will happen. And it's not just that it won't pass a test, it won't trip on fault.

I think one of the reasons why I object to the 'TTing' of outbuildings without a very good reason is becoming clearer.
And also why I get very annoyed with the 'regs say 200ohms' 'one rod is enough' attitudes towards TT systems.

Dare I raise the subject of SPSN RCBO's at this point?

 

[HappyHippyDad]

Even on a TT with what is generally accepted as a good Ra this will happen. And it's not just that it won't pass a test, it won't trip on fault.

I think one of the reasons why I object to the 'TTing' of outbuildings without a very good reason is becoming clearer.
And also why I get very annoyed with the 'regs say 200ohms' 'one rod is enough' attitudes towards TT systems.

Dare I raise the subject of SPSN RCBO's at this point?

I was thinking that a little earlier. I've always thought whats all the fuss with 200ohms when 1667 is all that is required. 200 is almost definitely going to be stable enough not to exceed 1667 even with the driest of conditions. However, even though a much smaller Ra (say <20ohms) would still not have been sufficient in my example to enable to RCD to trip, it would in far more cases than 200ohms.

 

[davesparks]

I was thinking that a little earlier. I've always thought whats all the fuss with 200ohms when 1667 is all that is required. 200 is almost definitely going to be stable enough not to exceed 1667 even with the driest of conditions. However, even though a much smaller Ra (say <20ohms) would still not have been sufficient in my example to enable to RCD to trip, it would in far more cases than 200ohms.

Yes you can't guard against it 100% as it becomes an endless cycle of 'what if' and 'how many faults can you protect against at once'.

On balance I think I'd rather see a focus on using SPSN RCBO's on all circuits. Your fault would still not trip the RCBO for that circuit, but as soon as another fault occurs it should trip.

 

[Lucien Nunes]

However, even though a much smaller Ra (say <20ohms) would still not have been sufficient in my example to enable to RCD to trip,

It's actually Zdb that matters for how much the N-E fault affects RCD operation, not the Ra. If there is any low impedance path from the MET, then the current will flow through that just as well as through the actual earth electrode, so a high Ra doesn't inherently mean the RCD is completely disabled. Also important is the resistance of the circuit between the DB and the N-E fault. The higher the resistance, the less the leakage (on other circuits) will divert to the neutral and the less impact the fault will have.

Here is another peculiarity to think about. Shock faults to true earth not involving the CPC, e.g. touching a damaged lawnmower cable standing barefoot in the garden, create a normal unbalanced current in the RCD because in the absence of the CPC there is no parallel path to neutral via the N-E fault. However, the RCD is measuring the algebraic sum of line and neutral current, including leakage in both conductors. This will be made up of the shock current (via the RCD line terminal) and some current from the neutral bar going to earth at the fault (via the RCD neutral terminal). These will tend to be in phase, both flowing through the RCD in the same direction, in which case they will cancel out. The RCD might therefore trip without leakage occurring from the line, or not trip with leakage occurring from the line, or anywhere in between, depending on the magnitudes of the currents.

 

[HappyHippyDad]

Always good to give an update!

Identified the problem circuit, broke it down in a couple of hours to between 2 points.

It was between a wall switch for a wall light (containing PL and N) and a ceiling pendant. Approx 50cm (DIRECTLY) above the switch was the wall light, with a screw that looked suspiciously directly in line with the PL+N cable. Removed the screw and the fault disappeared! Always feels good doesn't it when you find it, especially when it's a very visible problem that you can show the customer.

Oh yes, and the RCD trips as it should do now when tested. Actually, I better rephrase that otherwise I'll get a few damns from DaveS . The RCD always tested fine when tested correctly with all outgoing cables disconnected. It now even tests with them connected .



A big thankyou to all those who replied, especially @Lucien Nunes and @davesparks who as usual gave very useful advice.

 

[Grant1987]

Always good to give an update!

Identified the problem circuit, broke it down in a couple of hours to between 2 points.

It was between a wall switch for a wall light (containing PL and N) and a ceiling pendant. Approx 50cm (DIRECTLY) above the switch was the wall light, with a screw that looked suspiciously directly in line with the PL+N cable. Removed the screw and the fault disappeared! Always feels good doesn't it when you find it, especially when it's a very visible problem that you can show the customer.

Oh yes, and the RCD trips as it should do now when tested. Actually, I better rephrase that otherwise I'll get a few damns from DaveS . The RCD always tested fine when tested correctly with all outgoing cables disconnected. It now even tests with them connected .

View attachment 51925

A big thankyou to all those who replied, especially @Lucien Nunes and @davesparks who as usual gave very useful advice.

Just read all of this topic and this is one of the main reasons why I joined this forum to learn from the more experienced and intelligent sparkys had to read the posts a few times to understand but now I do another tick off the box ?