The RCD tripped

[tom66]

I was unplugging a plug, and I heard a loud pop, then all the lights went out. Main 80A/30mA RCD tripped.

This is in university dorms, so I am fairly confident in the quality of wiring. But should I be concerned?

My housemates are a little confused, and I have no idea how I caused it (if I did at all?)

Could the pop I heard be related to an earthing fault?

[M. András]

small arcing in the socket? or someone on the same phase was switching something on which tripped the rcd

[Psi]

Check the cable, plug and socket in question for any signs of damage.

You could try to reproduce the fault and figure out which part is causing it,  the cable, plug or socket, but you need to be careful as there is obviously a problem somewhere.

[tom66]

I found the fuse to be blown in the extension cord I unplugged the device from, if that adds anything. It was a Nokia phone charger (genuine, early iron core transformer type), but that still works, and hasn't tripped it again. I have added the cord to my do not use list, a friend is giving me a spare.

[Zero999]

Perhaps you caught the plug with your finger when you unplugged it and the RCD tripped before you felt the shock?

[IanB]

Perhaps you caught the plug with your finger when you unplugged it and the RCD tripped before you felt the shock?

That wouldn't blow a fuse, though, as happened here.

[IanB]

I was unplugging a plug, and I heard a loud pop, then all the lights went out. Main 80A/30mA RCD tripped.

This is in university dorms, so I am fairly confident in the quality of wiring. But should I be concerned?

My housemates are a little confused, and I have no idea how I caused it (if I did at all?)

Could the pop I heard be related to an earthing fault?

I found the fuse to be blown in the extension cord I unplugged the device from, if that adds anything. It was a Nokia phone charger (genuine, early iron core transformer type), but that still works, and hasn't tripped it again. I have added the cord to my do not use list, a friend is giving me a spare.

If there was a loud pop followed by a blown fuse and an RCD trip, then all the signs point to a direct short between live and earth. The short circuit will cause the pop and blown fuse, and the earth current will trip the RCD at the same time.

It's quite possible the wiring is faulty inside the extension cord you were using. If it comes apart, unscrew the covers on the plug and socket ends and see if there are any loose wires. Also check that the strain relief at the cable entry is not loose as this can cause wires to pull out of the terminals.

[Monkeh]

This is in university dorms, so I am fairly confident in the quality of wiring.

You lost your lights because of a socket. You should not be so confident.

[tom66]

This is in university dorms, so I am fairly confident in the quality of wiring.

You lost your lights because of a socket. You should not be so confident.

There were two RCDs. One for just the sockets; one for the whole lot, including lights. They were the same device so presented with the same short one might trip before the other.

We still had the emergency exit light.

I am going to disassemble the strip and see if there's any damage.

[Monkeh]

This is in university dorms, so I am fairly confident in the quality of wiring.

You lost your lights because of a socket. You should not be so confident.

There were two RCDs. One for just the sockets; one for the whole lot, including lights. They were the same device so presented with the same short one might trip before the other.

As I said, don't be confident of the wiring. That's plain wrong.

[IanB]

You lost your lights because of a socket. You should not be so confident.

If the fuse blew in the extension cord the fault would be downstream of the fuse. Hence, unlikely to be in the building wiring.

[tom66]

I can see some dark colour in the earth pin for the particular socket that went bang. I think this must be a clue. I can also hear a rattling sound... hmm.

It is very difficult to disassemble it as it appears to be sealed, but I will try. I have succeeded before with wall warts.

I have never had a look at this extension before, but now I notice no CE marking at all. It must be at least 8 years old as I see a "2004" date partially visible and I found it in the spares pile while searching for stuff to bring when away from home. The only indication is:

13A 250V~BS1363/A
Total Load must not exceed 13A
EURO 9988

Hmm... want to bet that something has come loose and me moving the plug made the final contact? Next time I will be more vigilant.

[Monkeh]

It's very likely the cheap POS shorted internally.

[IanB]

Taking it apart may help with forensics to determine what failed, but it doesn't look like you ever want to use it again. Cut the cord off for potential re-use and throw the rest of it away after you have finished examining it.

[tom66]

I am destructively disassembling it. No chance of ever reusing this...

[Monkeh]

You lost your lights because of a socket. You should not be so confident.

If the fuse blew in the extension cord the fault would be downstream of the fuse. Hence, unlikely to be in the building wiring.

That doesn't mean having every circuit in the building on a single RCD is correct. As I said, do not be confident of such an installation.

[tom66]

If I cannot be confident in the work of professional contractors, I do not know what to be confident in. You have to draw a line...

[Monkeh]

If I cannot be confident in the work of professional contractors, I do not know what to be confident in. You have to draw a line...

I draw the line where someone puts every circuit on one RCD. It's against regs, common sense, and good workmanship.

[tom66]

If I cannot be confident in the work of professional contractors, I do not know what to be confident in. You have to draw a line...

I draw the line where someone puts every circuit on one RCD. It's against regs, common sense, and good workmanship.

Every household installation I've seen has one main RCD for all circuits, then a separate RCD for the sockets only. What's wrong with this?

[Monkeh]

If I cannot be confident in the work of professional contractors, I do not know what to be confident in. You have to draw a line...

I draw the line where someone puts every circuit on one RCD. It's against regs, common sense, and good workmanship.

Every household installation I've seen has one main RCD for all circuits, then a separate RCD for the sockets only. What's wrong with this?

Well if you put them in series, which one trips? That's right, you don't know. They must be separate.

[tom66]

But it gives you a backup for the main sockets for essentially no extra cost (which is always a consideration) -- sockets which are far likely to have a ground fault than a light bulb or cooker.

[Monkeh]

But it gives you a backup for the main sockets for essentially no extra cost (which is always a consideration) -- sockets which are far likely to have a ground fault than a light bulb or cooker.

You don't need a backup. You need lights which work so you can go and turn the power back on without breaking your neck.

[tom66]

But it gives you a backup for the main sockets for essentially no extra cost (which is always a consideration) -- sockets which are far likely to have a ground fault than a light bulb or cooker.

You don't need a backup. You need lights which work so you can go and turn the power back on without breaking your neck.

There's an emergency light positioned very close to the breaker box which makes it possible to see the switches. I would recommend a torch in case it happens to be a lighting circuit fault though. Minor inconvenience At home, power trips were common on the pay-as-you-go electric meter -- often forgetting to top it up or activate the top up. I got quite adept at navigating in the dark and locating the charge key.

[Monkeh]

Doesn't change the regs. You don't put RCDs in series.

[tom66]

Doesn't change the regs. You don't put RCDs in series.

Eh. I'm no electrician. I stop once it gets beyond wiring a plug. There must be some reason they did it like this.
Remember, I am basing this all on the labels outside, as I have not lifted the cover (nor will I), to check the actual wiring.

[Monkeh]

Open it up, it's not going to reach out and kill you.

[AlfBaz]

Doesn't change the regs. You don't put RCDs in series.

Perhaps I misunderstand what you mean by "putting RCD's in series" but why would an electrical authority regulate against increased safety. Placing an RCD upstream from another gives you a back up trip in case the first one fails. Considering the first one would be lower amperage and therefore lower kA breaking capacity the chance of it arcing across on a high VA fault would be prominent. The larger breaker/rcd up stream will catch the fault and be better suited to disrupt power.

Normally you would have a cascading system in where upstream devices have higher trip values. Ideally the lower stream devices should catch the fault. You have to remember that all circuits need protection regardless of whats at the other end, particularly when it comes to short circuit and overload protection, all cabling must have protection

[Monkeh]

Doesn't change the regs. You don't put RCDs in series.

Perhaps I misunderstand what you mean by "putting RCD's in series" but why would an electrical authority regulate against increased safety.

They wouldn't. That's why you don't have a 30mA RCD which knocks out every circuit in a building!

Placing an RCD upstream from another gives you a back trip in case the first one fails.

They very rarely fail, that's what testing is for.

Considering the first one would be lower amperage and therefore lower kA breaking capacity

Mostly wrong and entirely wrong in domestic installations.

You have to remember that all circuits need protection regardless of whats at the other end, particularly when it comes to short circuit and overload protection, all cabling must have protection

And an RCD performs neither function except in certain installations (where a time-delayed unit is used to avoid the problem).

[AlfBaz]

They wouldn't. That's why you don't have a 30mA RCD which knocks out every circuit in a building!

RCD's are a relatively new thing and as such have been fitted retrospectively in a lot of buildings. Its quite commom (here in Oz at least) to have 1 RCD protecting protecting a bunch of sub circuits, usually leaving out things such as water heaters and stoves

They very rarely fail, that's what testing is for.

What testing regime do you precribe to here... Pressing the "Test often" button? I don't believe I've ever heard of testsing CB's at their rated kA fault levels at domestic or even commercial installations

Mostly wrong and entirely wrong in domestic installations.

??
Whilst I'm not ann expert in these matters, simply stating some thing without something to back it up doesn't make your statement any tuer than mine.

Are you saying that a 10A CB/FUSE would have the same kA breaking capacity than an 80A CB/FUSE?

And an RCD performs neither function except in certain installations (where a time-delayed unit is used to avoid the problem).

I would find it ectremely difficult to purchase an RCD that doesn't act as a CB protectinf against overcurrent.
Most time delay units will still trip once it sees the earth fault (even if the fault clears, the time delay allows downstream protection to trip first, allowing for locating of the faulty circuit

[Monkeh]

They wouldn't. That's why you don't have a 30mA RCD which knocks out every circuit in a building!

RCD's are a relatively new thing and as such have been fitted retrospectively in a lot of buildings. Its quite commom (here in Oz at least) to have 1 RCD protecting protecting a bunch of sub circuits, usually leaving out things such as water heaters and stoves

Unfortunately, that is common here as well. It's also against the current wiring regulations and common sense.

They very rarely fail, that's what testing is for.

What testing regime do you precribe to here... Pressing the "Test often" button? I don't believe I've ever heard of testsing CB's at their rated kA fault levels at domestic or even commercial installations

RCDs are not made for interrupting large fault currents except in certain conditions. Perhaps you've heard of the MCB?

There are appropriate testing tools for RCDs to ensure they break in the correct time at various levels of fault current.

Whilst I'm not ann expert in these matters, simply stating some thing without something to back it up doesn't make your statement any tuer than mine.

All modern protective devices for consumer units (fuse boxes) in the UK are rated for 6kA breaking capacity. The lot of them, whether it's 1A or 63A (usually the largest). Higher ratings are available but not necessary in domestic installations. And yes, this does apply to the RCDs and RCBOs, too.

Are you saying that a 10A CB/FUSE would have the same kA breaking capacity than an 80A CB/FUSE?

See above. Every circuit breaker in my installation is rated for 6kA. I do have two fuses on my side, I'm unsure of their rating. Certainly mid to high tens of kA, potentially as high as 150kA. The main fuse (which exists not to protect my installation, but to protect the cable feeding it) has a breaking capacity of at least 120kA. All of this is far overkill, as the PFC at my property is approximately 1.6kA.

I would find it ectremely difficult to purchase an RCD that doesn't act as a CB protectinf against overcurrent.

That is not an RCD (or more correctly RCCB), that is an RCBO. That is a different item entirely, and please see above regarding breaking capacity.

Most time delay units will still trip once it sees the earth fault (even if the fault clears, the time delay allows downstream protection to trip first, allowing for locating of the faulty circuit

They will only trip if the fault remains uncleared. They exist to allow line to earth faults with a poor earth connection (local earthing rather than a supplied separate or combined protective conductor) to clear before damage to cables occur, not to protect humans from shocks.

[G7PSK]

I think people here are confusing RCD's with MCB's. It is usual to have one RCD on the main input followed by an MCB very often found built into one unit then there will be further smaller MCB's for each circuit.

Example.

http://www.screwfix.com/p/mk-sentry-10-way-dual-rcd-board-with-2-rcds-10-mcbs/92791

[AlfBaz]

Yeah thats  my mistake, I call nearly every Circuit Breaker style device with earth protection an RCD.

Never the less I still beleive there's nothing wrong with having one feeding a bunch of Circuit Breakers.
If there's anything greater than a 30mA difference between active and neutlral it will trip power to them all, thereby afording the human protection these things were designed for. If anything there's likely to be a greater propensity for "nuisance" trips if the sum of all subcricuit leakage is greater than 30mA.

It is conceivable that if you had RCBO's on each sub curcuit and enough of them leaking to earth at something less than 30 mA without earth leakage monitoring at the main CB you could end up with currents substantially greater than 30mA flowing back, through paths unknown, to points other than your MEN link.

Of course you could go the route of placing RCBO's on each final circuit and an RCD/RCBO in conjunction with the Main Circuit Breaker or switch but this would be where you get into delayed earth leakage trip. I'm happy to stand corrected but I don't beleive that the purpose of delaying an earth trip is to allow time for "line to earth faults with a poor earth connection (local earthing rather than a supplied separate or combined protective conductor) to clear". The delay is usually there to allow downstream earth protection devices to do there job, its called discrimination.

I understand that there may be differences in rules and regulations from country to country, but most of these regulations are derived from Internaltion standards so they should bare some resemblance. End of line earth leakage protection devices (No delay device) need to trip within a certain time to be able to fall into the no delay class. Delay types come in either fixed or adjustable delay times who's uses are generally described as allowing discrimination of trip circuits. MOst of the ones I have come accross will delay the trip but still trip even if the earth fault has cleared

[G7PSK]

If you have two RCD's in series the one further down the line needs a lower trip current than the main one, I once brought a whole factory to a standstill when I plugged in my pressure washer to clean up a standby generator, I pressed the test button on the RCD plug (30mA) that caused the main 30mA RCD to cut out. The test button is seen as an earth fault on all the RD's in the circuit.

[AlfBaz]

You are probably right. Most of the delayed earth leakage protection devices i have come across not only have adjustable time but also adjustable current trip values. Their intent though is not for shock prevention but to prevent arcing in hazardous areas and to protect the cable. They're the ones that trip even after the fault clears.

I'd have to look up data sheets on domestic/commercial delayed RCD's to see if they didn't trip if the fault cleared before the delay time, but I suspect Monkeh is right in his description of how they operate, in that they don't trip if the fault clears before the time delay

If this is the case, then having delayed rcd's further upstream might have prevented what happened to you
The domestic ones I have dealt with usually show a schematic representation of the internals and the test button is normally shown as placing a resistance from the load side of the active to the incoming side of the neutral, going "outside" of the "CT". Therefore only causing a localised imblance

[Monkeh]

Never the less I still beleive there's nothing wrong with having one feeding a bunch of Circuit Breakers.

You're working up a ladder with an angle grinder. A faulty kettle trips the one RCD feeding the entire building. It is now pitch black, you are two meters up a ladder, and you have a 9" angle grinder still spinning in your hands.

You're walking down a steep flight of stairs, trying to avoid a very hungry cat. The faulty kettle trips the one RCD. It is now pitch black and you can't see the cat.

Nope, nothing wrong there.

It is conceivable that if you had RCBO's on each sub curcuit and enough of them leaking to earth at something less than 30 mA without earth leakage monitoring at the main CB you could end up with currents substantially greater than 30mA flowing back, through paths unknown, to points other than your MEN link.

Which doesn't matter in the least.

I'm happy to stand corrected but I don't beleive that the purpose of delaying an earth trip is to allow time for "line to earth faults with a poor earth connection (local earthing rather than a supplied separate or combined protective conductor) to clear". The delay is usually there to allow downstream earth protection devices to do there job, its called discrimination.

That is not what I said. The time delay is to allow for discrimination, the presence of the time-delayed RCD is to allow line to earth faults to clear before damage to cables or potential fire occurs.

MOst of the ones I have come accross will delay the trip but still trip even if the earth fault has cleared

They are unsuitable for protecting a dwelling without emergency lighting, simple as.

[AlfBaz]

Never the less I still beleive there's nothing wrong with having one feeding a bunch of Circuit Breakers.

You're working up a ladder with an angle grinder. A faulty kettle trips the one RCD feeding the entire building. It is now pitch black, you are two meters up a ladder, and you have a 9" angle grinder still spinning in your hands.

You're walking down a steep flight of stairs, trying to avoid a very hungry cat. The faulty kettle trips the one RCD. It is now pitch black and you can't see the cat.

Nope, nothing wrong there.

We are talking about protection against electrical shock. You are talking about emergency lighting issues, the two subjects have very little to do with each other. What will doing it your way do to prevent the exact same situation if there's a black out?

It is conceivable that if you had RCBO's on each sub curcuit and enough of them leaking to earth at something less than 30 mA without earth leakage monitoring at the main CB you could end up with currents substantially greater than 30mA flowing back, through paths unknown, to points other than your MEN link.

Which doesn't matter in the least.

Really? you can't think of a situation where these currents are summing through a common frame or structure and becomes high resistance the circuit can then potentially be completed by somebody stepping from one structure to the other. Return currents should always return via the return conductor.

I'm happy to stand corrected but I don't beleive that the purpose of delaying an earth trip is to allow time for "line to earth faults with a poor earth connection (local earthing rather than a supplied separate or combined protective conductor) to clear". The delay is usually there to allow downstream earth protection devices to do there job, its called discrimination.

That is not what I said. The time delay is to allow for discrimination, the presence of the time-delayed RCD is to allow line to earth faults to clear before damage to cables or potential fire occurs.

You've lost me completely here...

MOst of the ones I have come accross will delay the trip but still trip even if the earth fault has cleared

They are unsuitable for protecting a dwelling without emergency lighting, simple as.

again with the emergency lighting

[Monkeh]

We are talking about protection against electrical shock. You are talking about emergency lighting issues, the two subjects have very little to do with each other. What will doing it your way do to prevent the exact same situation if there's a black out?

It's adding a risk in the installation which need not be present. See common sense, good workmanship. A fault on one circuit which can be cleared by a circuit-specific protective device has no business interrupting other circuits.

Really? you can't think of a situation where these currents are summing through a common frame or structure and becomes high resistance the circuit can then potentially be completed by somebody stepping from one structure to the other. Return currents should always return via the return conductor.

Three keywords for you: Testing, equipotential bonding, high integrity earthing.

You've lost me completely here...

Apparently so.

A time-delayed RCD exists to allow the disconnection of supply in the event of a line to earth fault with a high earth impedance which would prevent the MCB from opening the circuit. It is not used for shock protection*, and it must not trip should the downstream fault be cleared by another device.

Note *: With the notable exception of clearing the fault quickly enough to reduce the risk of someone touching an 'earthed' (used loosely if it's high impedance) item with a high voltage present. This is different from direct shock protection as an RCD is normally used for.

again with the emergency lighting

Which you won't have in a dwelling.

[AlfBaz]

It's adding a risk in the installation which need not be present. See common sense, good workmanship. A fault on one circuit which can be cleared by a circuit-specific protective device has no business interrupting other circuits.

So we should hard wire the lighting circuit to the incoming feed... after all if the cable melts it can start a fire, at least we can still see thanks to the flames. Its all about protecting against shock, if it achieves that for the price of one RCD as opposed to many RCD's then more households will be able to afford the protection therefore more installations will have protection, win, win.

Three keywords for you: Testing, equipotential bonding, high integrity earthing.

Obviously you've never worked at a building site. I can tell you now, none of the other trades are going to go and make sure any of those things happen because you are a dodgy electrician and if they did none of those purported "keywords" will do anything to mitigate current flow through earth. Its there for protection not to carry load current

Apparently so

A time-delayed RCD exists to allow the disconnection of supply in the event of a line to earth fault with a high earth impedance which would prevent the MCB from opening the circuit. It is not used for shock protection*, and it must not trip should the downstream fault be cleared by another device.
or
Note *: With the notable exception of clearing the fault quickly enough to reduce the risk of someone touching an 'earthed' (used loosely if it's high impedance) item with a high voltage present. This is different from direct shock protection as an RCD is normally used for.

Please don't get smug... This is the first time you have stated anything about the fault being cleared by downstream protection, that and the previous post using the word "discrimination", you are slowly pinching from my original statement and trying to make it sound like this is your original thought
 

again with the emergency lighting

Which you won't have in a dwelling.

You dont need it. see my first comment about the fire lighting things up for you:))

[Monkeh]

Three keywords for you: Testing, equipotential bonding, high integrity earthing.

Obviously you've never worked at a building site. I can tell you now, none of the other trades are going to go and make sure any of those things happen because you are a dodgy electrician and if they did none of those purported "keywords" will do anything to mitigate current flow through earth. Its there for protection not to carry load current

What does it have to do with other trades? It's the electricians job to do all of that, and it's not done until it's all been tested after all other work is complete.

Please consider what correctly installed equipotential bonding will mean to a few tens of milliamps of leakage or fault current and its risk to human life.

And before you go around throwing petty insults, try reading our regs.

Please don't get smug... This is the first time you have stated anything about the fault being cleared by downstream protection

They will only trip if the fault remains uncleared. They exist to allow line to earth faults with a poor earth connection (local earthing rather than a supplied separate or combined protective conductor) to clear before damage to cables occur, not to protect humans from shocks.

If you can't figure that one out, step away from the mains electricity. Anyone familiar with fuses or breakers will understand they will not clear a line to earth fault in a timely manner if the earth impedance is sufficiently high.

that and the previous post using the word "discrimination", you are slowly pinching from my original statement and trying to make it sound like this is your original thought

Pinching? I'm fully aware of discrimination and it's been my point throughout this thread. I apologise for not having used the word before while posting at 3AM.

if it achieves that for the price of one RCD as opposed to many RCD's then more households will be able to afford the protection therefore more installations will have protection, win, win.

How about the price of two RCDs, with lighting and sockets kept separate (if multiple lighting circuits, split across the RCDs so light may be available if one trips). That's all you need to do. Or just put the lights on RCBOs by themselves. Neither option is expensive, either will comply with regulations. One 30mA RCD for all circuits will not comply with regulations and is piss-poor design.

[AlfBaz]

Hey what time is over there? 2:10am here (just south of sydney)

The other trade thing I was talking about was in reference to structures. I can't possibly think of how many things are not an electricians concern. Surely you are not suggesting that every conductive item in a building be earthed or have its continuity checked? Sure, metal structures housing electrical equipment will be earthed and tested but thats about it.
As for reading your regs, there are enough here to keep me excited

With regards to line to earth with a high impedence earth, both upstream and dwonstream devices are in the same boat aren't they?You never use a delayed one unless you have an instant one downstream

As for the rest of it, I'll have to take your word for it, that it doesn't comply with your regulations

[Monkeh]

Hey what time is over there? 2:10am here (just south of sydney)

Right now, 1530. Last night when I was posting, 0300.

The other trade thing I was talking about was in reference to structures. I can't possibly think of how many things are not an electricians concern. Surely you are not suggesting that every conductive item in a building be earthed or have its continuity checked? Sure, metal structures housing electrical equipment will be earthed and tested but thats about it.

Exposed conductive objects capable of introducing an equipotential zone (plainer english: a different earth potential) need to be bonded to the main earth terminal. I think that's the wording, anyway. If it's not exposed, it doesn't matter. Steel framework buildings are very often designed over here to use the entire frame as an earth. It's to everyone's benefit, especially as we use PME a lot (who doesn't want a warehouse sized earth rod?).

With regards to line to earth with a high impedence earth, both upstream and dwonstream devices are in the same boat aren't they?You never use a delayed one unless you have an instant one downstream

Typically you would have a 30mA RCBO on each circuit in such installations, yes. The main RCD protects against faults on non-RCD protected circuits (they still exist and between that RCD and the rest of the installion. The time delay provides discrimination and prevents nuisance tripping of the entire installation (A Bad Thing).

Such an RCD is also suitable to handle large earth currents produced across a number of separately protected circuits (at which point you undertake testing, high integrity earthing for IT equipment, and modify the installation if such leakage currents cannot be avoided).

As for the rest of it, I'll have to take your word for it, that it doesn't comply with your regulations

Indeed.

[SeanB]

I will bite in with the regulations here in South Africa.

Incoming mains is fed to a master disconnect switch, then to a bus bar in the board. There it is fed to MCB's for lighting, water heating and for the stove. The bus also feeds a RCD that then feeds a separate RCD protected bus bar and a separate neutral bar. This bus then feeds the socket outlets via 20A MCB devices, as the requirement is that all socket outlets that can take a standard 16A SA standard plugtop must be protected by a RCD device and a MCB.

The alternate uses the RCD as the incoming isolation switch, with all loads connected downstream via MCB units of the RCD being protected by the RCD unit. This is allowed as the RCD devices certified for use here are usable as disconnectors ( large enough contact spacing when off, denoted by a green trip handle, there are some with both RCD and overcurrent combined, but they are both uncommon and a special order item). While used often in low cost Rediboards ( a complete solution with incoming power, prepayment meter indoor unit, a set of 4 breakers using 7 of the 10 available positions in the board, a single bulkhead light unit with a switch, and 2 16A switched socket outlets all fed by a 10mm split concentric cable with 2 signal wires in the bundle) the first option is preferred in larger houses, as this allows you to have a sub board for the swimming pool which will not trip the whole house if it gets wet, or if the water heater or stove elements get a little leaky to earth ( safe as they are earth bonded to the supply earthing point, and have a MCB for overcurrent protection) but are still functional.

I prefer to have separate RCD units for the stove and water heater, and separate ones for socket outlets. A little more expensive ( OK a lot more) but a lot safer, as a single leakage will not switch all power off, but will isolate an island of load so that it is easier to do basic fault finding to isolate faulty equipment. I have distribution boards with up to 4 RCD devices, 1 per phase of standard sockets and the last for the 3 phase socket outlets.