Protective Devices Sticky

2011

Over the years there has been a lot of discussion about various protective devices on this forum including:
1) MCB’s
2) Fuses
3) RCD’s
4) RCBO’s

Other protective devices have been discussed, but to a lesser extent. Many of the same questions keep cropping up especially in terms of how they work and how to troubleshoot. Because of that I thought it would be good to start this sticky that could contain links to some of the more informative posts we have had as well as new contributions.

I would like this thread to focus more on understanding the issues and the science behind these devices than the rules.

Happy posting

Bruthal

Single phase RCD schematic with MS paint

Bruthal

We often hear questions about intermittent tripping of RCDs controlling socket circuits.

Often it is a N to earth fault.

One procedure to find which circuit such a fault is on is to switch off all the socket circuit MCBs.

Disconnect all the socket circuit neutrals from the neutral bar.

Test all the neutrals to earth for faults. If any faults N to earth are present, this will find them.

Once one is found, continuity test from the affected neutral to the load side of the switched off MCBs to show what circuit it was on.

Now the circuit with a fault has been identified.

brightspark

That schematic looks wrong?

It should really show the conductors wrapped around a common armature as opposed to to just passing through a sensor coil

Bruthal

brightspark wrote: »

That schematic looks wrong?

It should really show the conductors wrapped around a common armature as opposed to to just passing through a sensor coil

Might be worth looking into alright, they certainly were originally constructed with main conductors around a common core a few times. Later ones I suspect are like a CT setup with conductors straight through.

But the main aim was a basic diagram of current in and current out being compared.

Edit:
Looks like straight through in this photo


Not sure what type it is etc, but anyway, the oul MS paint wouldnt be great at drawing coils:)

2011

brightspark wrote: »

It should really show the conductors wrapped around a common armature as opposed to to just passing through a sensor coil

That would be one way of doing it, obviously the more turns the higher the induced current.

So here is a question for you:
Should the neutral and phase conductors be wrapped around the ring in the same direction or in opposite directions, and why?

brightspark

opposite so they cancel each other

Bruthal

I wired a house when I was an apprentice. Everything working fine for a while.

Later, RCD started tripping even when everything plugged out, but the cooker was used.

I was too busy to look immediately, so they got electrician out. When he seen the cooker tripping the socket RCD, he just said you better get the fella that wired it to look.

Any suggestions as to how this could happen? Or maybe it was all a dream:D

H.20v3

Bruthal wrote: »

I wired a house when I was an apprentice. Everything working fine for a while.

Later, RCD started tripping even when everything plugged out, but the cooker was used.

I was too busy to look immediately, so they got electrician out. When he seen the cooker tripping the socket RCD, he just said you better get the fella that wired it to look.

Any suggestions as to how this could happen? Or maybe it was all a dream:D

Fluorescents on non-rcd lighting circuits can also trip the RCD, seen it happen

The fix was to remove the capacitor, didn't know how it was happening until my inspector told me

H.20v3

^ not saying you should remove the cap

Bruthal

H.20v3 wrote: »

Fluorescents on non-rcd lighting circuits can also trip the RCD, seen it happen

The fix was to remove the capacitor, didn't know how it was happening until my inspector told me

When I returned to the house to look, I noticed new coving in. I isolated the faulty socket circuit, then the faulty leg. Checked coving above. Nailed through the t&e shorting the N and E, on both sides of the wall at 2 separate sockets.

When I removed the nails and then the length of t&e, it had 2 holes through between the N and E, one for each socket drop, on the same piece of cable.

Bruthal

H.20v3 wrote: »

Fluorescents on non-rcd lighting circuits can also trip the RCD, seen it happen

With N-E faults on RCD circuits, a load not on the RCD can trip them, particularly if it is a decent size one.

2011

brightspark wrote: »

opposite so they cancel each other

I would think they would be wound the same way. As the currents are flowing in different directions they would cancel each other out.

H.20v3

The 4-pole RCD can be used for 3 or 4 wire 3 phase

You must connect the neutral if 4-wire

It will also work for single phase supplies but one of the line poles may have to be linked.to N for test button.to operate

Bruthal

2011 wrote: »

I would think they would be wound the same way. As the currents are flowing in different directions they would cancel each other out.

I'd say it's opposite way to each other myself. If you were to look down the axis of both coils from above, the direction of the current through windings would at any instant have current going around the core in same direction with opposite wound windings. But the current through the actual windings would be in opposite directions.

So my opinion would be they are wound in opposing directions if they are not simply straight through.

Although it's mainly because they would be wound on opposite sides of the core, they appear as opposite wound. In reality, the load carrying conductors must go through in same direction, and if being wound around the core at all, they just go around it and back through again.

H.20v3

An issue that came up at one time with an inspector was the orientation of protective devices

I saw a crabtree link stating that their RCDs can be mounted sideways.

H.20v3

Wasn't me but he contractor mounted the distribution board 90° off

The inspector pulled him up on it , I can't remember how it worked out in the end

2011

Bruthal wrote: »

I'd say it's opposite way to each other myself.

Suppose I were to wind a two core cable three times around an iron ring. If I were to pass a 20 A sinusoidal current at 50 Hz down core A of this cable it would induce a current in that ring. Now if I also were to pass a 20 A sinusoidal current at 50 Hz down core B of the same cable that had a 180 degree phase shift with respect to the current in core A then I would expect this to induce an equal but opposite current in the came iron ring. Both currents would cancel each other out. The resultant current in the ring would be zero.

Although it's mainly because they would be wound on opposite sides of the core, they appear as opposite wound.

Appearances are not what matters. Whether they are on opposite sides of the core or not makes no odds (once the cables are equidistant from the core).

In reality, the load carrying conductors must go through in same direction, and if being wound around the core at all, they just go around it and back through again.

Yes.

2011

H.20v3 wrote: »

I saw a crabtree link stating that their RCDs can be mounted sideways.

From a technical perspective if the manufacturer states that it will work I would accept that. However I would not accept this.

From a a technical perspective if the manufacturer states that it will work properly I would accept their word on this.

However I would sign off on RCD's, MCB's or RCBO's being mounted sideways unless the distribution board was specifically designed for mounting these devices in this orientation (this was the case with some Square D boards in the past).

I wouldn't accept light switches being mounted sideways either.

H.20v3

The issue I would have is everything except the physical operation of the protective device

It's just wrong to orient a device or board sideways, legends and manual operation of the device are off

H.20v3

BS7671 will have regulations for this scenario

As usual ET101 will have no mention of it anywhere

2011

H.20v3 wrote: »

The issue I would have is everything except the physical operation of the protective device

It's just wrong to orient a device or board sideways, legends and manual operation of the device are off

+1
My last post was ambiguous.
I edited it to clarify, see above.

Bruthal

2011 wrote: »

Suppose I were to wind a two core cable three times around an iron ring. If I were to pass a 20 A sinusoidal current at 50 Hz down core A of this cable it would induce a current in that ring. Now if I also were to pass a 20 A sinusoidal current at 50 Hz down core B of the same cable that had a 180 degree phase shift with respect to the current in core A then I would expect this to induce an equal but opposite current in the came iron ring. Both currents would cancel each other out. The resultant current in the ring would be zero.

If you have the phase conductor wound around the ring on one side in a clockwise direction, the neutral conductor on the other side will be wound in opposite direction.

So it depends how you view it, as I said.

2011

Bruthal wrote: »

If you have the phase conductor wound around the ring on one side in a clockwise direction, the neutral conductor on the other side will be wound in opposite direction.

So it depends how you view it, as I said.

Ok, I see your point. Do you accept the part of my post that you quoted in your last post? (WRT the 2 core cable)

Bruthal

2011 wrote: »

Ok, I see your point. Do you accept the part of my post that you quoted in your last post? (WRT the 2 core cable)

I think that is like asking is 220v more than 210v...

But in fairness you quoted a specific part of my post. I have said the critical part is both conductors must go through in the same direction to start with. And I said they will be wound in opposite directions on if on opposite sides of the core, which has to happen if they both go through from the same side and are placed on. Opposing sides of a core ring. After all, the question asked was, would they be wound in same direction or opposite direction. Answer. Depends on how you look at it.

H.20v3

Is this "Blinding" of type AC RCDs by DC fault currents a big issue

H.20v3

^^In domestic installations

Steve

Bruthal wrote: »

I wired a house when I was an apprentice. Everything working fine for a while.

Later, RCD started tripping even when everything plugged out, but the cooker was used.

I was too busy to look immediately, so they got electrician out. When he seen the cooker tripping the socket RCD, he just said you better get the fella that wired it to look.

Any suggestions as to how this could happen? Or maybe it was all a dream:D

Was the cooker wired through the RCD of not?

Damp elements are notorious for leaking to earth.

Steve

H.20v3 wrote: »

The issue I would have is everything except the physical operation of the protective device

It's just wrong to orient a device or board sideways, legends and manual operation of the device are off

It's quite common.

The Americans in particular do it a lot.

Bruthal

Steve wrote: »

Was the cooker wired through the RCD of not?

Damp elements are notorious for leaking to earth.

No it wasnt.

In a similar one a couple of years later, I seen an rcd tipping on a DB I had just fitted in place of an old fuse board. Not unusual when an rcd is fitted to a house for the first time. Anyway a socket had N and E connected together.

It was similar to my earlier one in that the RCD kept tripping even with power off to the house. Similar in cause rather than symptom.

2011

Steve wrote: »

It's quite common.

Yes,, I did say "However I would sign off on RCD's, MCB's or RCBO's being mounted sideways unless the distribution board was specifically designed for mounting these devices in this orientation (this was the case with some Square D boards in the past)."

Bruthal

Steve wrote: »

Was the cooker wired through the RCD of not?

Damp elements are notorious for leaking to earth.

As a further update to my earlier post, if the cooker was through the RCD, it wouldnt have been any great unusual mystery, would it.....

When the piece of twin and earth that was nailed between N and E was removed, no more tripping of RCD at all.

With a N-E fault, there is an alternative path through the RCD to the N-E fault from the main Neutral bar for neutral current. Usually too little current will go that way to trip an RCD via such a fault.

On non neutralised installations, this may be more likely to divert more than the 30ma required to trip an RCD, particularly with larger loads on, in certain circumstances.

In the case here, there were overhead lines supplying houses along the road, suitable for loads in the 70s. So the N can end up at a higher potential than the earth bar when no neutralising is done in the house. So some current will take the path from the main neutral bar, to the N-E short on the RCD controlled circuit, and back to the MET in the DB, tripping the RCD.

This was also the case in the house where the RCD was tripping even with power to the house off (overhead lines supplying houses).

I would say it is unusual to see it.

Steve

Bruthal wrote: »

When the piece of twin and earth that was nailed between N and E was removed

Was this on the T+E supplying the cooker or was it downstream of the RCD?

Can't get my head around the scenario

Bruthal

Steve wrote: »

Was this on the T+E supplying the cooker or was it downstream of the RCD?

Can't get my head around the scenario

Downstream of the sockets RCD,

Cooker was not on any RCD,

Any N-E short on an RCD controlled circuit is providing another path for main neutral bar current to flow through. It will usually be very small.

Basically, if you have 40 amps flowing through the main neutral bar, for 30ma to be diverted to a N-E short, the alternative (fault path) has to be greater than 1/1333th of the main neutral path ohms.

This is unlikely in the case of modern underground cable feeding minipillars and neutralized installations, but is possible with old undersized overhead supplies, and probably more likely with non neutralized installations too, as neutralizing locks the N and E bars to almost identical potentials, but the alternative path is still there.

So it would be unusual in newer installations.

Steve

Is this what you are describing?

Bruthal

Yes except no neutralising link

Steve

Bruthal wrote: »

Yes except no neutralising link

So this then?

Bruthal

Yes thats about it.

The first one wouldnt really be right, as there is no earth conductor from the house to the esb transformer, but that is academic.

Bruthal

So the brown is the cooker or any neutral return path, red is alternative path created by sockets N-E short. Current will flow that route, it is just a question of how much. Influences will be magnitude of the actual load current, and the ratio of N path ohms:Fault path ohms, which is governed by several factors including network supply line sizes, condition of earth rods in non neutralized houses etc.

I used the first diagram, but the second is more accurate in terms of earth rod at house.

Bruthal

The esb transformer will have its neutral earthed. But there is no link to house earth apart from the neutral itself if the house is neutralised.

The Neutral from the ESB sub is a combined NE.

But again, not overly relative to the overall point.

John.G

I posted a version of this ~ 6 months ago but am doing so again in case there are different views on its cause:

I had a very unusual RCD tripping problem a few years ago, I tried all the usual remidies but even with the RCD tripped I was still getting a full (o ohms resistance) between neutral and earth in any of the sockets in the house, the fault would disappear but over a 2 to 3 day period would re occur at random times, I even swapped the RCD for the shower RCBO and still no good, eventually the problem disappeared (neutral to earth resistance; infinity) and has never returned, original RCD still in use.
I have a 1972 built house with fuses, TCNS system? with a earthing rod in the garage.
I know it would seem to be impossible with a double pole RCD tripped to still measure a full neutral to earth fault except that it is on my (the consumer) side but is there any conceivable way get a apparent full short, neutral to earth, from a external source in the conditions described above?

Bruthal

John.G wrote: »

I know it would seem to be impossible with a double pole RCD tripped to still measure a full neutral to earth fault except that it is on my (the consumer) side but is there any conceivable way get a apparent full short, neutral to earth, from a external source in the conditions described above?

If a N-E short on sockets as example, trips an RCD, the N-E short will still be on the sockets with the RCD tripped.

If a socket circuit with no faults is tested N-E at a socket while the RCD is on, a N-E short will be seen via the neutralising link, but with the RCD off, the N-E short would clear, unless the socket circuit has a N-E short (fault) on it , (downstream of the RCD)

John.G

Thanks Bruthal, that's the way I would see it myself but it was uncanny, at one time while watching the M.meter it was swinging from 0 to infinity every few seconds as if someone/something was switching the short in/out, you replied to my original post that it could be a nicked cable against the corner of a steel socket box which of course it could have been but one would think that it would have reappeared again at some stage.
I sort of hoped that it might have as I would then have removed the neutrals from the neutral bar one at the time to track it down.

Bruthal

John.G wrote: »

Thanks Bruthal, that's the way I would see it myself but it was uncanny, at one time while watching the M.meter it was swinging from 0 to infinity every few seconds as if someone/something was switching the short in/out, you replied to my original post that it could be a nicked cable against the corner of a steel socket box which of course it could have been but one would think that it would have reappeared again at some stage.
I sort of hoped that it might have as I would then have removed the neutrals from the neutral bar one at the time to track it down.

Yes thats why a meggar is better to find such faults than the multimeter, set to high voltage and test the individual neutrals at the neutral bar after disconnecting them.

For an annoying fault like that, if it is not showing in testing, splitting the sockets over 2 RCDs can eliminate some circuits, and so on, to narrow it down.

Also as mentioned earlier in the thread, some N-E faults will trip only as the load gets higher, especially if its on a circuit other than the faulty one, while it may be on the same RCD too, so it can seem even more intermittent then.

These problems are made more tricky with multi circuits from 1 RCD.

2011

Posts such as the ones above make a strong argument for one RCBO per socket circuit.
It is something I would strongly consider if ever wiring a domestic installation again.

John.G

How does one know which one of these RCD's is being purchased as RCD 1 with the test coil resistance taken from the load side would appear to be the better choice if one is doing a routine button test as all the test current flows through the sensing coil but only the neutral full test current passes through with RCD 2 which may not happen if the load side neutral has a earth fault. I just read somewhere that the (at least initial) button test always be done with the L&N load side disconnected, I now understand the reason I think.

Bruthal

Type 1 can have some test current flow through the N coil as there is a path to the N-E fault, thus also reducing the effective test current.

N-E shorts reduce effectiveness of rcds if someone receives a shock also.

Rcd will have a diagram on the side of it showing its internal setup.