What sort of earthing do I have?

[plugwash]

Well, for a start, this is a DIY forum where we are (should!) essentially be talking only about domestic installations. In any event ...

Sorry, when I skimmed through the thread earlier and saw the cutout, I somehow thought it was a large commercial supply, but looking closer it seems to "only" be a regular 100A TP supply.

... the effect of which is, when an RCD is used for fault protection in a TT system, to limit IΔn (albeit way above the devices IΔn) - e.g. 500 mA in the case of a 100 Ω TT earth.

Yes, so 1A with a 50Ω earth or 10A with a 5 ohm earth. No specific maximum, just a matter of what can be justified by your earth electode system.

In practice, of course, the RCD would have an IΔn of 30 mA, in order that it also provided 'additional' ('direct personal') protection, which majorly 'trumps' the much more modest requirement of 411.5.3.

My understanding is that the norm on a s TT installation is that you have 30mA RCDs on final circuits which require "additional protection" and then slower/higher trip current RCDs to provide protection to the distribution infrastructure. On a small install that likely means a 100mA or 300mA type S, but there certainly seems to be a market for larger.

Sometimes you will get a TT install where all the outgoing circuits from the main DB want/need 30mA RCD protection. In such cases some installers may decide to rely entirely on the 30mA RCDs/RCBOs and not have a main RCD.

While I think this just about complies with the letter of BS7671 I don't think it's a particularly good idea for two reasons.

* It provides no protection from faults inside the CU (including potentially faulty SPDs). A RCD upfront minimises the amount of unprotected wiring.
* It leaves you reliant on the correct operation of a single RCD as your only line of defence.

 

[JohnW2]

Yes, so 1A with a 50Ω earth or 10A with a 5 ohm earth. No specific maximum, just a matter of what can be justified by your earth electode system.

Indeed, but I really see no need for ones with particularly high IΔn in domestic situations, even though 'much higher' would be adequate for fault protection.

My understanding is that the norm on a s TT installation is that you have 30mA RCDs on final circuits which require "additional protection" and then slower/higher trip current RCDs to provide protection to the distribution infrastructure. On a small install that likely means a 100mA or 300mA type S, but there certainly seems to be a market for larger.

Yes, that's certainly traditional thinking/practice, but has a somewhat different significance now that virtually all circuits are required to have 30 mA RCD protection. In my TT installation, all of the final circuits in all of my CUs have always had 30 mA RCD or RCBO protection - so, in terms of the final circuits, I probably would have been happy with just that. However, there are some pretty long distribution circuits to some of my CUs, so I have up-front 100 mA Type S RCDs to provide fault protection to those distribution circuits.

I don't doubt that there is "a market for larger" but although I've seen plenty of 100 mA ones in domestic installations (and have some in my installation), II'm not even sure I've seen a 300 mA one in a domestic situation, let alone higher.

Sometimes you will get a TT install where all the outgoing circuits from the main DB want/need 30mA RCD protection. In such cases some installers may decide to rely entirely on the 30mA RCDs/RCBOs and not have a main RCD.

As above, that's what I would probably have done had it not been for the distribution circuits needing fault protection.

While I think this just about complies with the letter of BS7671 ....

I would think that it probably is compliant if there are not any 'distribution' cables of significant length - but, as with my installation, I don't think it would be compliant if there are significant cables upstream of the CU (hence RCDs/RCBOs) which need fault protection

I don't think it's a particularly good idea for two reasons.
* It provides no protection from faults inside the CU (including potentially faulty SPDs). A RCD upfront minimises the amount of unprotected wiring.

Traditionally speaking, I would have said that (L-E) 'faults within the CU' would be incredibly unlikely, but I suppose that the advent of SPDs (which I will avoid like the plague for as long as I can ) might change that a little. However, although I can see that an L-E fault in a SPD might trip an upstream RCD, I'm not so sure that there would be any significant hazard in the absence of such an RCD, would there?

* It leaves you reliant on the correct operation of a single RCD as your only line of defence.

It does (unless one 'doubles up' on one's finals circuits' RCDs/RCBOs ). However, before the days of 'additional protection' we were "left reliant on the correct operation of a single OPD as our only line of defence", despite the fact that (unlike the case with RCDs), ongoing correct operation of OPDs cannot even (practicably) be confirmed by testing!

However, this diversion onto TT does not help me to understand why (assuming such is what BS 7671 intends) primary fault protection ('ADS') provided by RCDs should not be acceptable in TN installations (i.e. in a final circuit in a TN installation which does not have a low enough Zs for OPD-mediated ADS.). Do you have an answer to that?

 

[ericmark]

My first introduction to RCD's was on the building of Sizewell 'B', Our supply would be from a WMDU (weatherproof main distribution unit) typically 500 amp with a moulded breaker and an add-on RCD unit set to 5 amp and 1-minute delay, our own WMDU had the same unit but set to 1 amp at 30 seconds.

The WMDU would feed a load of smaller DBs, normally with either 300 mA or 100 mA RCDs, S type and often one would feed a load of wet weather cabins, which would have 30 mA RCDs fitted, and I know a nail to hang a coat hitting the cable would take out the lot. So much for delay.

At this time, we had never heard of type AC, A, F, or B. And the type which looked like a current transformer which sent a 230 volt pulse to the moulded breaker to trip, were clearly not fail-safe. It was in fact three units, the breaker, the CT and the pair of dials to set trip current and time. Don't think ever allowed for control by an ordinary person.

By the time working on T5 Heathrow, we were seeing the RCM, you don't want to stop the drum on a batching plant when full, it would mean getting in with a shovel to empty it, so just a light to show there was an earth fault, so it could be investigated before the next batch. And the operator points to the big red lamp and says, "What's that for?" so that idea didn't work.

As to 10 mA RCD sockets for MK, where when you pressed the test button, it would also trip the 30mA and 100mA RCDs feeding it, clearly some lack of thought on the design there. A big sign "Don't Press" was like a red flag to a bull.

But if an RCD is a secondary protection, not that important if type AC, A, F, B, or S, it should never be required, it is just belt and braces. But on a TT supply, very different, I watched the John Ward demo, but even a type B is not pure DC, think in fact quite low, something like 10 mA, to trip within the allowed time. So should they include a DC detection unit?

Even the idea of a metal consumer unit, with no RCD before it, raises questions. Bring back the old days, when we were just careful.

 

[JohnW2]

..... But if an RCD is a secondary protection, not that important if type AC, A, F, B, or S, it should never be required, it is just belt and braces.

It "should never be required" IF the OPD (which we cannot test) works as intended and IF the LE fault is of negligible 'impedance'. However, it would definitely be "needed" if the OPD failed or, more importantly, if the L-E fault current were 'just a few ohms' or more, including the situation when the fault path is through a human body.

But on a TT supply, very different, I watched the John Ward demo, but even a type B is not pure DC, think in fact quite low, something like 10 mA, to trip within the allowed time. So should they include a DC detection unit?

I need to re-visit his video but if I recall correctly (maybe not!) the main effect of a large DC current was to increase (if I recall, only fairly 'modestly') the (residual current) 'trip threshold' of the device. If that is the case, then even an 'appreciable' increase in the 30 mA trip threshold would still result in sensitivity to L-E currents which was dramatically 'better' (lower) more than would be the case (e..g. up to 160 A for a B32) when protection was provided by an (allegedly 'primary') OPD.

Another point I've never seen mentioned is that, since we (including John Ward) cannot really test MCBs, we (at least I) don't know what effect, if any, a DC component will have on the magnetic trip threshold of an MCB

 

[SimonH2]

Another point I've never seen mentioned is that, since we (including John Ward) cannot really test MCBs, we (at least I) don't know what effect, if any, a DC component will have on the magnetic trip threshold of an MCB

That's probably easy to answer. A DC component will add to the magnetic field on half the mains cycle, and subtract from it on the other half.
So in extreme it could delay tripping by half a cycle ... approximately.
Worst case the fault occurs late in a half cycle - so doesn't trip. The next half cycle doesn't trip due to reduced current - zero if a pure DC fault. Then the next half cycle there'll be a sum of the AC and DC which should trip.

For the thermal trip, I don't think it'll matter. The heater element will get the same I²t heating as the cable so if the trip threshold does change, it'll still match the thermal effect on the cable.

I have sometimes mused about knocking up a test rig. Bypass DUT with a contactor and solid state switch. Use welder to provide test current at low voltage - adjusting the current with contactor closed. Have something like an Arduino open the contactor, then the SSS at a known point in the mains cycle, and time how long it takes for the MCB to trip - detecting that could be the trickiest bit.

 

[JohnW2]

That's probably easy to answer. A DC component will add to the magnetic field on half the mains cycle, and subtract from it on the other half.
So in extreme it could delay tripping by half a cycle ... approximately. .... Worst case the fault occurs late in a half cycle - so doesn't trip. The next half cycle doesn't trip due to reduced current - zero if a pure DC fault. Then the next half cycle there'll be a sum of the AC and DC which should trip.

Is it necessarily as simple as that? Could one not say the same thing in relation to an RCD? Is not the perceived problem (with RCDs) that a DC current may (partially or completely) magnetically saturate the sensor coil, thereby impairing its ability to respond to superimposed AC currents? If so, could the same not happen with an MCB?

For the thermal trip, I don't think it'll matter. The heater element will get the same I²t heating as the cable so if the trip threshold does change, it'll still match the thermal effect on the cable.

Sure, I'm not suggesting that a DC component would affect the thermal trip - but that may not be fast enough to satisfy disconnection time requirements - and, of course, in contrast with the situation with RCDs, since we effectively cannot test MCBs, we cannot even be sure that the thermal trip will work when it should (the mechanism could be 'jammed'!).

 

[SimonH2]

Is it necessarily as simple as that? Could one not say the same thing in relation to an RCD? Is not the perceived problem (with RCDs) that a DC current may (partially or completely) magnetically saturate the sensor coil, thereby impairing its ability to respond to superimposed AC currents? If so, could the same not happen with an MCB?

The detection mechanisms are different.

MCB, the magnetic field pulls on the trip mechanism - more field, more force. It also operates at high currents, so would need a massive DC component to come anywhere clise to the AC component.

RCD, there is transformer action where rate of change of flux imposes a current in the sense coil - at relatively low currents (e.g. under 30mA differential in a single turn primary.) Higher AC currents may saturate the core - but reversing every 10ms so still generating a current in the sense coil.
It wouldn't need much DC to saturate the core, and then the transformer action for a low level AC fault would fail

Unless I've missed something !

 

[JohnW2]

The detection mechanisms are different.....

Fair enough. What you go on to say makes sense. I'd forgotten/overlooked that the magnetic field trips an MCB 'directly' (without any electronics etc.)

However, none of this really helps me to understand why BS 7671 seemingly doesn't want us to rely on an (appropriate type of) RCD to provide ADS in the face of L-E faults in a TN installation, apparently preferring 'reliance' on a device whose sensitivity to L-E fault currents is a couple of orders of magnitude inferior to that of an RCD, and despite the fact that (unlike RCDs) the ongoing correct functioning of an MCB cannot, in practice, be confirmed by testing! Any thoughts?

 

[davelx]

Doesn't 411.whatever have something to say about RCDs providing ADS?

 

[JohnW2]

Doesn't 411.whatever have something to say about RCDs providing ADS?

Indeed it does. As I quoted in post #29, in relation to fault protection, it says ...

411 PROTECTIVE MEASURE: AUTOMATIC DISCONNECTION OF SUPPLY
.....
411.4 TN system
....
411.4.5 The following types of protective device may be used for fault protection:
(i) An overcurrent protective device
(ii) An RCD

... which seems to fairly clearly indicate that an RCD is an acceptable 'primary' (or 'sole') device for providing fault protection by ADS. However, as has been discussed that contradicts what 415.1.2 appears to say ...

415.1.2 The use of RCDs is not recognized as a sole means of protection and does not obviate the need to
apply one of the protective measures specified in Sections 411 to 414.

Furthermore, as also discussed, 415.1.2 also fails to acknowledge that, in a TT installation, one usually has no choice but to rely on an RCD as the sole means of providing fault protection by ADS.

 

[davelx]

Exactly. 411 says you can, 415 says you can't and refers you back to 411, which says you can. USW. So the regs are self contradicory are need to be corrected in a future addendum.

Ps I missed your post #29, there are so many long posts in this thread

 

[Ragnar_AT]

I‘d read 415.1.2 as „if there are no CPCs an RCD is not enough to make the installation comply“. Apparently in some countries that used to be an option.
It’s definitely not a UK-specific reg, I‘ve seen it elsewhere too.

 

[JohnW2]

Exactly. 411 says you can, 415 says you can't and refers you back to 411, which says you can. USW. So the regs are self contradicory are need to be corrected in a future addendum.

Quite - that's what I've been saying.

As for needing to be corrected in a future addendum", I think the regs in question have been essentially unchanged through many additions/amendments of the regs, so I'm not holding my breath.

 

[JohnW2]

I‘d read 415.1.2 as „if there are no CPCs an RCD is not enough to make the installation comply“.

If there were "no CPCs" then OPD-based ADS would presumably also not work in most situations?

 

[SimonH2]

Fair enough. What you go on to say makes sense. I'd forgotten/overlooked that the magnetic field trips an MCB 'directly' (without any electronics etc.)

However, none of this really helps me to understand why BS 7671 seemingly doesn't want us to rely on an (appropriate type of) RCD to provide ADS in the face of L-E faults in a TN installation, apparently preferring 'reliance' on a device whose sensitivity to L-E fault currents is a couple of orders of magnitude inferior to that of an RCD, and despite the fact that (unlike RCDs) the ongoing correct functioning of an MCB cannot, in practice, be confirmed by testing! Any thoughts?

Note it only says an RCD can't be the sole protection. I can think of a number of reasons, or possible reasons :
* An RCD doesn't give any overcurrent/fault protection for a L-N fault. I haven't looked at the regs, so in this context is that a valid reason ?
* As I've mentioned, while there is a significant beneficial difference in testability, could it be that when these regs were written, there wasn't confidence in the reliability of RCDs for what is a safety critical function ?
* While an RCD does provide a much more sensitive detection of L-E faults, does this really matter in a properly designed circuit (and assuming low enough fault impedance) ? I.e., OC protection will disconnect the supply anyway.
* I guess there's an acceptance that MCBs are fairly reliable (very simple) so the lack of testability is acceptable.