OK, time has come to sort out my house CU and associated wiring. The
purpose of the exercise is to provide termination for three new circuits
(2nd floor power and lights, and smoke detectors) and at the same time
bring the CU setup up to modern standards in the process. I am not
seeking to expand the scope of the job into loads of extra work right
Fuse -> Meter - > ELCB(VO) -> Henley -> Wylex 6 way rewireable CU
The henley also splits a submain off:
30mA RCD -> T&E JB -> 6mm sq SWA -> Metal clad 2W CU (rewireable) ->
2.5mm sq XLPE SWA -> Insulated CU
Have a look at the current plan, and see if you can spot any obvious
improvements I could make.
All CUs (appart from the insulated one in the workshop) will be
replaced. The outbuilding ones get a RCD each in their CUs rather than a
shared one in the house. The house ones cope with 100mA and 30mA
protected circuits (no space available for a wider split load CU, and no
need for a time delay RCD for discrimination this way)
I could not find much detail on protection of submains in the OSG or
other books I have. What do the regs have to say on the matter? With the
current setup the 6mm sq SWA could probably rely on the main supply fuse
and the RCD for some protection, but the 2.5mm section that it in turn
feeds is obviously at risk in the unlikely event of a short.
OK, I'll try. (NB much relies on the notes in the PDF file.)
1. Supply, main earthing and tails
Have you enquired of the supplier or supply network operator whether PME
is available on the overhead supply? Often it is nowadays and it would
enable an upgrade to TN earthing, although this probably would involve
paying for new service cabling.
If you continue to use your own (TT) earth you cannot continue to use
the gas pipe. As you say a new electrode probably won't provide a lower
impedance, but it will still be necessary.
The main earthing conductor needs to be 16mm^2 for either PME or an
unprotected run to a TT earth. (TT could use a smaller size only if
protected in conduit - see Table 10A in the OSG.)
I suggest that to allow for a (current or) future upgrade to a 100A PME
supply that you use 25mm^2 tails for meter to Henley block and Henley
block to the power CU. 16mm^2 should be OK for the feeds to the
lighting CU and the switch-fuse for the submain to outbuildings.
Similarly the main bonding should be in 10 or 16mm^2, not 6 (although 6
is compliant whilst the earthing remains TT).
2. House CUs
Not much to comment on here really, except that, being new, the MCBs in
the power CU will be 32A and not 30A.
You need both s/c and earth fault protection for the cables, in case
someone digs through an SWA cable. You also need overload protection if
an overload is possible, but that doesn't apply in your case since the
sum of the downstream MCB ratings (52A) is less than the rating of the
6mm^2 cable. (But in any case the 30/32A fuse provides it anyway.)
Now, whilst the 30A (32A?) fuse in the switch-fuse will provide
phase-neutral s/c protection, there's no earth fault protection in your
design (unless the main supply is upgraded to TN). For TT you will need
to either (a) use a 100mA time delay RCD ahead of, within, or after the
switch fuse, or (b) remove the 30mA RCDs from the workshops and use a
single one at the house end. The latter option is clearly less
convenient and arguably is non-compliant.
BTW I don't understand your comments about the main fuse protecting the
SWA cables. The smaller fuse in the switch-fuse does that, but only for
a P-N short. With an earth fault and your present gas-pipe TT earth (Zs
= 11 ohm) you'll only get ~20 A earth fault current and no fuse will
blow - hence the need for an RCD at the house end.
Andy Wade wrote:
Many good points, thanks!
(Much of this reply is my thinking aloud, but there is the occasional
question buried in there if you feel up to it ;-))
I have not enquired personally, but I know a close neighbour has gone
that route, so the upgrade certainly was (and probably still is)
available. Given that this will be a highish cost option (especially as
I have most of the RCDs I need already), do you think this would bring
Yup, I realised the gas pipe on its own was a no-no ;-) Hence why I was
going to add a spike as well. My comment was more an observation that
although the current arrangement is not to current standards, it does
provide for a reasonably decent earth since it is a long run of metallic
pipe in a heavy clay soil under concrete. Hence my assumption is that
the addition of the spike will not actually improve Zs much if at all.
I was planning on a protected run - The Meter and CU are located in an
under stairs cupboard adjacent to an outside wall. The supply comes down
the wall in a conduit and goes through it about a meter above ground. I
was going to drill through the wall low down and run a earth wire
straight down in conduit to ground level and then position the spike
close to the wall (with luck dodging the gas and water pipes!)
I did consider that, but then thought I could do that upgrade at the
time if I wanted. I was not too bothered since I don't really have need
for a 100A supply. The meter currently has 16mm tails leaving it and was
proposing leaving those as is.
Yup, my design was assuming that it was staying TT.
Yup, true. Was obviously still thinking of my old Wylex ;-)
Yes good point. The current setup has the RCD (the only one in the whole
setup) at the house end, and this can be a PITA. Since the workshop is
where I typically power any outside equipment from (partly convenience,
partly because it is the only RCD protected supply), If I ever do get a
trip (typically when I borrow either a dodgy cement mixer from a
neighbour, or a dodgy thicknesser from a friend) it can entail lots of
trips to the kitchen to reset the RCD. Hence my desire to move the RCD
to the destination end...
The easy fix I suppose would be to replace the incomer switch on the
switch fuse unit with an RCD and be done with it (but position the thing
so you don't have to sprawl over the kitchen worktop to find the thing),
or put a time delay RCD in there and have it in addition to the
destination end ones.
In fact if I were going for the latter solution I could just as well
scrap the idea of a switch fuse for the outbuilding supply, and put the
time delay RCD in the 8W CU and run the outbuilding supply from a MCB on
That was me thinking my way through the rationale for the current setup
(although wrongly I now realise). The current setup has RCD protection
for the SWA, but *no* over current protection at the source end. I was
thinking that the 6mm cable size was such that over current protection
would be provided by the main supply fuse, however as you point out,
that would only be true for P-N faults and not P-E.
The outbuilding supply was added much later to the install (I would
guess the CU is original equipment since the house was built in '56),
the outbuilding feed was probably added 15 years ago. The quality of
wiring looks like a pro job (and having met the previous owner, and seen
some of the very limited DIY attempts he made, I am pretty sure it was
not done by him). However there are two aspects of that setup that seem
odd, the aforementioned lack of over current protection on the SWA, and
also the separate TT earth for the outbuilding but used in conjunction
with a metal clad CU at the shed. (I am not certain that the cable
armour is connected to earth at the house end, but it seems likely.
Obviously I will need to test this to make sure!). Were the regs circa
1990 notably different in these respects?
With the current setup (RCD at source) I suppose the only fault scenario
that is not accounted for is an P-N overload on the SWA. It is protected
by rewireable fuses at the destination end, so you would need to stick a
spade though it in such a way as to cause a P-N short but not trip the
RCD in the process which I guess is pretty unlikely
The RCD positioning is not a "solution" to the tripping problem. An RCD
should trip if faulty equipment is connected. BS 7671 requires the
electrical installation to be designed to provide safety and
convenience. Inappropriately placed protective devices are not
convenient, so don't comply with the regs.
Yes of course, but there's an SEP-field round it :-)
True, but there occasions when you don't look a gift cement mixer in the
mouth (but you do make sure you are wearing gloves!) ;-)
(To be fair to it, it is fine in nice dry weather, but as it gets damper
in the evenings you can get the odd trip).
The thicknesser may well be protected by a SEP field, although next time
I am feeling charitable I will replace the NVR switch which I think is
the culprit... it would be nice to borrow a machine that actually turns
on every time you push the button ;-)
Cheers. I'm suffering more from the cold I've got at the moment than
any excess of Christmas...
I only mentioned it for completeness really. The only immediate
advantage would be that it would solve the submain earth fault
protection problem by making it go away. And you'd probably get a lower
impedance supply, which could be an improvement if you have any problem
with lights dimming when large loads are switched on. On earthing alone
though there's probably nothing to justify the cost. Your TT earthing
system will clearly be adequately designed and maintained, so no
It will probably worsen it on paper. Remember that you do the tests
with the main bonding disconnected, so that any fortuitous parallel
earths are excluded. It did occur to me BTW that if the house remains
TT then there's not really any reason to separate the house and
outbuilding earths. They could be paralleled up (via the armour of the
6mm^2 SWA) to lower the overall Ze. The down-side is the need to make
sure that they get separated again if the house is ever converted to PME.
OK - then OSG Table 10C applies, and your original 10mm^2 is more than
I'd bite the bullet and go for the delayed RCD option.
Except that an HRC fuse is a better option, discrimination-wise, than an
MCB for protecting the submain. Suggest you look at either a delayed
RCD in a separate 2-module enclosure or using one of the CU systems with
a DIN-rail fuse carrier option (MEM and Wylex, AFAIK).
Hmm, that's living on borrowed time. A short at the far end of the
2.5mm^2 section would probably destroy the whole cable and move the the
fault to the end of the 6mm^2 leg. I don't know whether SWA cable would
catch fire or not (T&E probably would in those circumstances, but the
armour might act as a sort of fire shield).
Not really. The (amended) 15th edition was still in force then and the
advice on use of RCDs now in the OSG hadn't appeared in the form we now
know and love, but overcurrent protection requirements weren't really
any different. Installer ignorance is a more likely explanation of the
design you've inherited.
That's unlikely, but a disaster affecting one of the outbuilding CUs
could conceivably cause a P-N short. The lack of protection could then
propagate the disaster housewards...
Allot of it about it seems, sprogs recovering, SWMBO ailing, me trying
to keep out the way ;-)
It would, although I expect a time delay RCD will be cheaper ;-)
You get a slight blip on starting of a load, but not enough to be
irritating. All the computer gear is on line interactive UPSs so that
don't care either.
True... it may well rise to the 22 ish ohms of the outbuilding spike
then (unless I get it through the gas pipe!)
That could well have been the design intention of the original
Ah! realisation dawns!
Now that is a point, when I tested the loop impedance at the
outbuildings, I did so with the SWA armour still connected. That would
suggest that the reading I got should have been from both earth
connections in parallel, however I got a different (i.e. double the
figure) reading which would suggest that the SWA armour is isolated at
the house end.
Hmm looking that way.... shame thought I had managed to avoid that...
Discrimination with what, the MCBs in the destination CUs? Since they
are current limited at 20A on their power circuits, and the source end
was protected at 32A, I figured this would not lead to a discrimination
problem, or have I missed your point?
You suggest using a HRC Fuse, what about a Type C MCB at the house end
of the outbuilding supply? (and I suppose if that is OK, then there is
no reason to not move the time delay RCD to the 100mA RCD CU and use a
spare way for the outbuilding feed, thus dispensing with the extra
switch fuse unit).
Looked at in those terms perhaps it would have been better not waiting
10+ years to upgrade this lot!
(at least the shed is brick/concrete built and less likely to go up in
smoke than some).
No, that was exactly my point: discrimination with the upstream MCB.
IME MCBs with ratings one or two steps apart don't discriminate if the
fault current is high enough to be on the instantaneous (magnetic trip)
part of the characteristic curve of both devices - which I suspect it
will be here. ICBW, but I don't think having 'current limiting'
characteristics guarantees discrimination - but do check with the
manufacturer's technical support people. OTOH fuses don't do instant
magnetic tripping and are far more likely to survive a fault cleared
'instantly' by a downstream MCB.
Or even a Type D! That might work, but you need to study the
characteristics, calculate fault currents and reach a conclusion.
There's also the need to make sure that the submain cables remain
protected. I'd worry particularly about the 2.5mm^2 leg since the fault
current will be getting quite low at the far end of that. There may be
a case for sub-fusing it at the junction in the shed.
If you haven't encountered the technique you can plot the 'adiabatic
line' for the cable on the fuse or MCB characteristic curve and rapidly
determine the minimum fault current at which the cable remains
protected. From Table 43A of BS 7671 the 'k' value for 90 deg XLPE
insulation is 143, so the adiabatic 'fault withstand' line for 2.5mm^2
conductors is defined by I * sqrt(t) = 2.5 * 143 = 357 (i.e. it will
stand 357 A for 1 s, 1130 A for 0.1 s and so on). The adiabatic lines
are straight lines with a gradient of -2 when plotted on the usual
log-log format of fuse/MCB characteristics.
Yes, you could do that. (I'm assuming you mean making it a split load
unit with one 100mA RCD for the lighting and another 100mA delayed RCD
for the submains. You wouldn't want them on the same RCD, for obvious
Indeeed, if it wern't for the upstream RCD on its incomer it would be
simplest to stick a wider CU in the first shed, and take the workshop
supply off a way on that.
Thanks for that... since I don't have a full copy BS7671, I am missing
those tables. I shall go dig out a data sheet for some MCBs and see what
I thought about that, but was thinking of keeping it as a non split with
just the time delay RCD. My logic being, that the overload protection of
the SWA issue is an important one, but having said that, the chances of
a fault condition occuring that is not handled by the downstream RCDs /
MCBs is pretty low. There is the spurious trip issue, but experiance
would suggest that it not a problem in this case, since I have never
experianced a trip of this type in ten years with the whole outbuilding
installation fed from a single 30mA RCD (both buildings are solid,
waterproof etc, and wired to a decent standard). Swapping that for a
100mA rime delay RCD at source would lower the chances of a spurious
trip still further. There is the other issue of a split load unit with
enough ways will not fit in the cupboard!
Been reading through a few data sheets. I think for MCBs one step away
you are not going to get discrimination on the "instant" trip part of
the curve as you said, but for two steps away you probably will. That
may vary between manufacturers as well though.
Yup, probably the path of least complexity as well.
I presume that for a TT system with RCD, the only fault current we are
interested in is the prospective short circuit current rather than the
prospective phase earth one? How does one make a realistic stab at
estimating that, when you don't know what the supply impedance is? Or is
is simply taken as supply voltage over combined resistance of phase and
neutral conductors? (given the rather feeble looking PBJ insulated
overhead wires from the pole, that is going to give and estimate well
over the reality!)
For mainstream urban underground supplies you can assume that the supply
impedance (P-N) won't exceed 0.35 ohm. (This is the max. Ze value given
for PME supplies and used as the basis of the standard circuits in Table
7.1 of the OSG, along with Ze = 0.8 ohm for TN-S.) FWIW, my experience
is that the great majority of supplies are in the 0.1 - 0.25 ohm range.
Old rural & overhead supplies can have higher impedances, but from
your previous comments yours is fairly 'stiff', so the 0.35 ohm
assumption will probably be reasonable.
To confirm any assumptions it's best to measure the impedance at the
main switch terminals or CU busbars with a loop-tester. If you don't
have one of those then you can get a good-enough estimate by measuring
the voltage drop when a large known resistive load is connected - e.g.
several rings on a cooker or two 3kW heaters together - and applying
Ohm's law. Measure the voltage at the CU, or on a different final
circuit to one feeding the load, so as to exclude the voltage drop in
the post-CU wiring from the measurement.
Once you have a supply impedance figure the rest is fairly easy. Add on
the resistance of the circuit P and N conductors and divide the sum into
230 to determine the fault current at any particular point. All
impedances can be assumed to be resistive and added arithmetically -
reactance will be negligible in this context. There's a table of
conductor resistances on page 158 of the OSG. These are at 20 deg. C
and should be multiplied by about 1.4 to allow for the increase in
resistance due to heating during a fault.
I noticed that you tend to see those figures repeated fairly often in
various books etc, however estimates for TT systems seem to be far less
It seems reasonably stiff, but that may just be a reflection of not
usualy loading it that hard. Something like a kettle will get a slight
step change in brightness, but not enough to cause the reduction itself
to be noticable.
I shall do some experiments and see. I don't think I am actually that
far from the sub station. If it is the closest one, then it could be as
little as 500m LOS away. The weakest link will be the overhead wire
which looks quite thin, feeds both our house *and* the non attached semi
next door. I would not swear to it, but it may also be an ali cable
rather than Cu.
I have a Megger LT5, although I only have the standard 3 pin plug lead
for P-E measurements for it. I also can't remember what test current it
uses though off the top of my head.
Yup, easy enough to do. Just bought a new 3kW kettle which should do the
job nicely (and the CU is in a cupboard in the kitchen to all intents
Yup that bit was simple enough, it was more a case of if there was a
supply impeadance starting point one could assume.
(Ta for the page number, I was looking through the thing for ages last
night for that table, being sure that I had seen it in there - but could
I find it? Too much turkey, too little sleep is my excuse!)
I have also seen mention of a 3/4 "rule of thumb" used here...
HomeOwnersHub.com is a website for homeowners and building and maintenance pros. It is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.