Tim S wrote:
John Rumm coughed up some electrons that declared:

Tim S wrote:

One thing I'm lost on, is why have an RCD (RCCB?) if we are satisfied
that the circuit will disconnect in less than 5 seconds (by my
calculations, it will disconnect in 3 seconds at worst) and if we are no
longer concerned about touch voltages.
I can't see the need for a RCD at the switchfuse position.

Me neither, but BigWallop's in that field so I won't discount anything just
yet, with no disrespect to anyone else.

To be fair, this comes up on the IET forum from time to time and the answers
are different every time, with the only common factor being the "3m rule".

2m and 3m come up quite often as the limit before a switch fuse is
required IME.

When (if) I get a reply from EDF Technical, I'll post a summary here.

Ta

Ignoring them for the minute, I can see no need for anything beyond a DP
isolator.

I can see the need for a fuse to offer fault protection, since a fault
on the submain may otherwise not be cleared in time, and would result in
taking out a fuse on the distributors side of the network.

The need for a RCD could only be driven by two possibilities, one being
risk of direct contact, and the second being a loop impedance that is
too high to reliably clear a fault. The chances of the former on a SWA
submain are to all intents and purposes nil, and the later we know not
to be true in this case.

Their 100A BS1361 fuse is quite capable of disconnecting my submain in 5s -
though it would not if the max permitted Ze of 0.8 Ohms was apparant (need
630A to meet 5s).

(Don't think I have actually met a TN-S system yet that was as bad as
the "notional" value).

A fuse is also probably better here in that it will discriminate with
just about any MCB you care to use.

I used a science principle in my calcs - my measured Ze was 0.19 +/-5%
+/-0.07 Ohm

That's the uncertainty in a Megger 1552 plus not knowing which pair of leads
from 2 sets the LI test was calibrated with (0.04 Ohms between them).

So I'm working to the max Ze with the worst case uncertainty applied, which
is 0.34. Even with this figure the disconnect times are fine. That gives a
bit of latitude for their supply varying a bit over time.

So I don't know why the DNOs have this rule.

Unless they merely want an isolator at that position... Or the right to prat
around with their cutout fuse whever the mood takes them, safe in the
knowledge it won't make the installation unsafe...

That is probably quite likely. They won't always be able to offer TN-S
with a very low impedance.

When you get articles written by the ECA saying "stick an MCB in" you tend
to assume there's a reason. Even the NICEIC Snags and Solutions books don't
cover a simple supply extension.

Quick bung from one of the MCB manufacturers perhaps ;-)

What I suspect happens in reality is noone gives a hoot and just does
something reasonable. Me, being me, wants to be able to justify stuff with
confidence. Silly rabbit, me...

Not really - I like to have all potential questions answered as well ;-)

Example: You want to install a separate CU for external lighting at a
later
date. A box will allow you to take other tails off without disrupting
everything. But that's my opinion.

It's a good idea, though I'm personally not keen on multiple adjacent
CUs. My leaning is to leave enough spare ways that it will never be
aproblem
Can be handy if you decide you want a totally separate one for
outside/garden circuits for example.



Just for context, here's my final circuit list so far (I'm pretty happy with
this):

2 x 32A ring circuits covering kitchen and utility area
1 x 32A ring for rest of ground floor
1 x 32A ring for first floor
1 x 10A lighting for left side of ground floor (inc kitchen/dining room)
1 x 10A lighting for right side plus upstairs[1]
1 x 40A cooker circuit (we use gas, but best have a circuit in place)
1 x 45A distribution circuit for boiler + backup thermal store heating (3 x
3kW)[2]
1 x 16A radial feeding two external 16A weatherproof commando sockets[3]
1 x 10A external lighting, including direct feed to shed lights[4]
1 x 32A (or 25A TBD) radial to shed sockets.

snip comments that seem fair enough

[4] Why not submain to the shed? Well, my calculations show it's impossible
to meet volt drop requirements on the lighting component without using
excessively oversized cable. I have a need for a few outside lights so it
seemed sensible to maintain separate circuits back the the main CU and lump
all the lights together. Shed will have a couple of local DP isolators for
safety.

One presumes the distance to the shed lights remains the same either
way, so house circuit fed shed lighting is unlikely to dim less than a
submain based installation (more in all probability!). This sounds like
an example of complying with the letter rather than the spirit of the
rule ;-)

I might use DP RCBOs for the outside circuits, or at least the sockets,
which are 2 module wide, so at the extreme the above occupies 14 ways from
a 20 way board.

I usually have a policy of making sure that anything outside does not
share a RCD with something inside. Since the outside circuits are a more
likely source of nuisance trips etc.

Here I have installed two split load CUs in the house[1] (preceded by
one main switch in a separate two module enclosure)). Both TT, the
second smaller CU is reserved for outside stuff. The 30mA RCD carries
outside sockets and lights, the 100mA type S RCD directly feeding
things like the garage and my workshop (via BS 88 cartridge fuse
carries). The garage/workshop has another split load CU with inside
lights on the main switch part, and sockets and outside lights on the
30mA RCD section.

[1] When we got here there was one CU with everything sharing a single
30mA trip RCD. Needless to say a horse farting would plunge the place
into darkness.

So you see, I'm not too worried about expansion

That's ok then!

--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/