I think we've been the victims of history a bit on this one.

When I first saw a decent sized computer installed (back in the late 1980s or early 1990s), it had a 'clean earth system' - i.e. it's c.p.c and and additional 10mm² G/Y went directly back to the MET and connections with other c.p.c.s in the area (e.g. air conditioning) were avoided. The idea was to try and segregate it from all other equipment that might somehow introduce "interference" onto the earthing system. (If it wasn't for the fundamental requirements of EEBADS I dare say some would have wanted an independent electrode to connect the computer to).

The problem was that it didn't really work as intended - the main computer might have had a nice clean earth, but it was connected by (earth referenced) RS-232 cable to terminals (as in CRT teletypes, rather than bit of brass to join wires) scattered all around the building, which in turn were just plugged into the local ring circuits. So not only did the clean earth get polluted by all the other connections, the 0V lines in the data cables themselves could end up carrying unwanted current, so possibly even making the situation worse.

So these days a completely different approach is taken - rather than try to keep different "earths" apart, have just one and make it a low impedance as possible by having as many parallel paths as possible. Hence all the talk of meshes and rings and stars of meshes in BS 7671 and BS EN 50310. Some of the old language about separate "clean" earths does seem to cling on in some quarters though.

Likewise with METs. If you'd asked a few years ago, I think the general consensus of opinion was that you had one MET per installation, and if an installation spanned  two more buildings, the subsidiary buildings had earth marshalling terminals (BEMTs). More recently the BS 7671 requirements for main bonding have changed, reworded, and are now very much more clearly 'per building' (and only internal to a building too), rather than per installation (and applied to each building) - which in a way makes things rather simpler, and the naming each building's earth terminal a MET makes more sense. But again much of the older language persists in many texts (and likely will continue to do for a while yet).

   - Andy.

Changes in terminology and wording and interpretation again!

I do remember as a young apprentice in the very early 80's and one of the best electrical lecturers I ever had litterly "Bounced" the blackboard duster ( NO Smart boards in those days) off a apprentice who misinterpretated what the lecturer was saying about earthing. His quote was "Son, you can call it whatever the hell you want, but when you fall on your backside ( He used another word!!) all earth is DIRTY!!!!!!" At the time rest of us were a little bit frightened to ask any more questions but as my knowledge and career grew, there was a lot of truth in that statement. 

Cheers GTB

Graham, I do not think that we are far apart on this one. Thinking of a MET for the installation within a building is very helpful and I think makes the OP's confusion go away.

If you bring the comma forward a couple of words so that we have, "In each consumer's installation, within a building extraneous-conductive-parts ...", the MET could be elsewhere, but clearly that is not the case.

If the law were always clear, courts and tribunals would not misinterpret it, but the existence of the appellate courts clearly demonstrates the difficulties. (I have personal experience!)

I suggest that 411.3.1.2 would be much clearer if the word, "building's" were inserted before, "main earthing terminal".

I think that part of the confusion stems from the traditional domestic installation which we discussed last week. That probably was the norm when TN-S supplies needed a terminal adjacent to the PILC and fuse boxes did not have provision for bonding conductors, but even the OSG (Section 2.1) now has the MET in the DB. It seems a little unfortunate that it is not labelled as such.

AJJewsbury said:

So these days a completely different approach is taken - rather than try to keep different "earths" apart, have just one and make it a low impedance as possible by having as many parallel paths as possible. Hence all the talk of meshes and rings and stars of meshes in BS 7671 and BS EN 50310. Some of the old language about separate "clean" earths does seem to cling on in some quarters though.

Yes absolutely, as you say trying to maintain a clean earth system is incredibly difficult, and a mesh achieves the same effect - we're just trying to stop any earth conducted noise from "dirty" devices turning into a significant noise p.d. between different equipment earths of sensitive equipment.

Sometimes however it's really difficult to make a thorough mesh, and there is a mix of very sensitive equipment and very noisy equipment and you do end up having to join the earths at just one point. As Jam says, it's not recommended "for general use" in 61000-5-2 (figure 4), but sometimes it is the only way. But that then becomes a real challenge to design and maintain - as 61000-5-2 also says "When properly installed and the topology maintained (their emphasis), this approach has been found satisfactory".

However as far as I'm aware (and from my experience), where "clean" earths are used it doesn't matter from the EMC point of view where the two earths are joined (and for safety they must be joined somewhere, as 61000-5-2 points out), as long as they are only joined at only one place.

I used to come across this with installing sound equipment in TV studios - massive 3 phase lighting dimmers and microphone amplifiers were not comfortable bedfellows! But like Andy says about IT equipment, it was not easy(!) ensuring that the installation was designed such that no-one could accidentally join the two earths at a second unintended place, which would have been electrically safe but potentially extremely noisy.

I am slightly surprised if IT systems these days still need a clean earth, but I'm no expert in the earthing or immunity of IT systems. I'm more used to them being the "dirty" systems that put 'orrible noises all over my nice clean (analogue) audio systems!

It's a really interesting point. And, as everyone has said, full of myths and rumours. But some of the rumours do have a grain of truth.

Cheers,

Andy

Stepping away from the standard words for a second, it may help to realise that from a comms/emc perspective, we usually are considering much higher frequencies than the dc or 50hz needed to blow fuses, and the same piece of metal can be both solidly earthed at low frequencies an yet decidedly live at higher ones. An extreme case could be a vertical antenna mast, earthed solidly enough to be its own lightning conductor, but at the same time 'live' to the transmitter's RF and creating an E- field around it of hundreds or even thousands of volts per metre. So an earth that is clean for one application such as a VSD switching at a few kHz may be entirely inadequate or even act as a resonant antenna for sharp edges for superficially similar switching waveforms but with unfiltered fast edges giving frequency components in the 100s of MHz.

Given that the fields around a wire are created with a wave speed cira 'c' or about 300m per microsecond, a full signal at say 10MHz takes about 30m, and much as you can slam a skipping rope in a door and then set up a standing wave by flapping the free end at a frequency that is a quarter wave, a length of wire 7.5 m long forms a resonator and or antenna  for 10MHz - giving a voltage maximum at that frequency when the other end is earthed. Higher frequencies are even more obvious - in a length of say a few 10s of metres of cable carrying 100Mbit Ethernet, any given bit only starts  coming out of the far end after quite a few of the  subsequent bits have already been clocking in  - in a sense the bits of the signal are queued up inside the cable,

Such delays and phasing effects make for a situation that is very hard to generalize how best to organize high speed systems especially those where the earth also carries signal 'return' currents , beyond, ' keep it short and low impedance or keep it separate' Both approaches do work, but which to employ in any situation is not an easy call.

Good modern systems tend to sidestep this altogether (and wired Ethernet wired to standard with the transformers at both ends is a good example) by providing a balanced transmission line pair, with out and return currents in a controlled geometry ensuring that earth connections are not really part of the signal loop at all and that if the earth 'bounces' the link is unaffected.

Mike.