Cluttered!

[JohnW2]

Oh fair enough maybe I misread which circuits it applied to, or maybe that was for RCD protection it varies.

As far as I am aware, the only things which are required to have RCD protection are new sockets (up to 32A) and circuits supplying a bathroom. In addition, of course, new buried cables will also usually require RCD protection, regardless of the nature of the circuit.

even the lighting circuits should stay below around 1.2ohms

I would imagine that, in any remotely 'normal' house, one would probably struggle to exceed the maximum permitted Zs on a lighting circuit!

Kind Regards, John

 

[EFLImpudence]

"NOTE: Disconnection is not required for protection against electric shock but may be required for other reasons, such as protection against thermal effects". Any thoughts about that?

No.

Especially as Chapter 41 is titled "Protection Against Electric Shock" and reading 411.3.2.6 which also seems inconsitent if the ADS is not for shock prevention.

 

[JohnW2]

Especially as Chapter 41 is titled "Protection Against Electric Shock" and reading 411.3.2.6 which also seems inconsitent if the ADS is not for shock prevention.

Exactly my point.

I'm even struggling to think what that footnote might be trying to say (badly!), given that it seems fairly clear that it was not intended to say "what it actually says"! Any ideas?

Kind Regards, John

 

[ericmark]

It was 16th Edition time I was talking about, today with RCD protection having a MCB tripping on the magnetic part is not so critical. The problem is not if it trips in time or not, but can one as the person doing the testing and inspecting accept any reading out of spec? even if only 0.02Ω out, one simply does not know if the trip time is going to be 0.01 second or 5 seconds once you exceed the pass limit.

I was worried at one time about inspection and testing, with the idea if you did not follow the limits to the letter how some one some 9 years latter could re-test and show your installation was not good enough and want it to be corrected FOC. My worry was more about volt drop than tripping times. However when I came to look at the results and work out the testing tolerance I realised it was near impossible to measure 0.96Ω accurately enough to prove it was wrong from day one. Test three times and it is unlikely that two readings will match even using the same meter.

So in real terms 1.50Ω was likely good enough, and no one was really likely to have a come back because of the 0.06Ω over the limit, it was more down to the electrical firm being annoyed that the labour time required was increased without them being able to charge for the extra time required.

In real terms I know realise although I did not realise at the time, even 1.44Ω was far too high in most cases to drop within the volt drop limits, I am sure if we really measured the volt drop many of the older installations would have failed. Only when volt drop and ELI became separate issues with the RCD covering the ELI requirement did I really consider the line - neutral loop impedance and how that would effect both volt drop and speed of disconnection with a short circuit. Yes I always knew you should test earth and neutral impedance to line and use the highest figure, but with twin and earth the ELI was nearly always the highest figure.

Since it was around 2003 I can't remember what the incoming supply was so can't really say if the volt drop was exceeded or not, but I would guess it was. When BS7671:2001 came out, I decided to re-take some of my qualifications. I did my apprenticeship in a time when if you asked to use the loop impedance meter, or to read the regulations book, one was subjected to a line of questions which ensured you did not ask again. It was 1980 when working in Algeria when I first even considered testing an earth rod, before that I just banged them in and connected up, I never tested them.

The first 10 years as an electrician I now realise I did a lot of sub-standard work, I knew what happened when the dash pot oil dried up, and how to use a micrometer to measure fuse wire thickness, but ELI was some thing only done in collage, it was not done in the real world that I worked in. Maybe just the firm I worked for, the local authority, when I went abroad in 1980 it was a steep learning curve, and when I returned to UK in 1989 again it was a very different world.

 

[JohnW2]

It was 16th Edition time I was talking about, today with RCD protection having a MCB tripping on the magnetic part is not so critical.

Maybe - but it's as 'critical' as it ever was on a non-RCD-protected circuit (of which I'm sure there are still millions around).

The problem is not if it trips in time or not, but can one as the person doing the testing and inspecting accept any reading out of spec? even if only 0.02Ω out ...

That's not really "a problem", or a difficult question to answer, is it? The requirements of BS7671 are explicit enough. If the Zs exceeds the maximum permitted by BS7671, then that's a 'fail', with no wriggle room, isn't it? How (in the context of an EICR) one 'codes' a very small deviation from the requirements is, of course, a different matter - and no doubt open to debate.

In real terms I know realise although I did not realise at the time, even 1.44Ω was far too high in most cases to drop within the volt drop limits, I am sure if we really measured the volt drop many of the older installations would have failed.

I don't know about the pre-17th days, but throughout the period of the 17th ed. it has very rarely been possible to 'fail' a circuit in relation to VD, since BS7671:2008 has never included any explicit VD requirements.

Yes I always knew you should test earth and neutral impedance to line and use the highest figure, but with twin and earth the ELI was nearly always the highest figure.

If you're talking about L-E and L-N loop impedances, then I can't really see how it could ever be otherwise with T+E, since the CPC is always of lower CSA than the neutral conductor.

Kind Regards, John

 

[ericmark]

Every measurement must have a tolerance, some where there must be a +/- some figure. It is impossible in most cases to be exact. With a DC supply with no load one may be able to get a reasonably accurate result for the resistance, but with an AC supply and some load to measure the loop impedance is always a little hit and miss.

The only way would be to have a 50 Hz generator supplying a test supply with no load and a very stable origin loop impedance, without this supply the results are never going to be very accurate.

The meter in essence reads a voltage, then puts a load on the circuit and reads the voltage again and calculates the impedance from the volt drop on a set load. If the supply volts vary during the measurement then there will be error. Since we have no control over the supply volts the only option is to measure many times, discard any measurements which are well out, and take an average of the rest.

So if we want 1.44 and we get 3.44 then there is clearly something wrong and it clearly fails. But if we want 1.44 and get 1.45 then to fail it would be hard, as we don't really know if the out of tolerance is due to measurement errors or installation faults.

The idea of doing repeated tests until one test passes is clearly flawed, also the idea of one pass limit for original installation and another for periodic inspections after the originally installation. One would hope the design and installation did not result in reading right on the limit, however it does happen.

Of the 1.44 the reading is in fact split into two parts, the supply impedance and the installation impedance, we are told the supply should be better than 0.35Ω with a 100A DNO fuse. However often the supply is below that figure, this raises another question, if the supply is 0.25Ω can the installer use the extra 0.1Ω? or should he take the 1.37Ω - 0.35Ω and say the circuit needs to be no more than 1.02Ω.

The problem is with other than a 100A supply, with 100A we can consider if the supply is 230 volt and the DNO is permitted a 6% volt drop or 13.8 volt then the impedance can be 0.138Ω but it does not really work out like that, as in the main there are many tapping all taking 100A so there is no real way to determine exactly what impedance the supply needs to be at. And we have all seen how 60A fuses are not always replaced with 60A and how supplies are modified from time to time, be it new cable or new transformer.

So the question, if the loop impedance is higher than required what happens then?
1) Complain to DNO that the supply impedance has dropped.
2) Complain to installer that the impedance is too high even when it was not too high when installed.
3) Swap the MCB for a smaller one.
4) Replace part of the circuit with larger cable.
The list could go on and on, but simply recording fail is not really an option. It could simply be that on the day of testing the DNO was using a temporary supply generator while it was doing some work on the network. Maybe we should return to using the humble fuse?

 

[bernardgreen]

It is possible that the prescribed values take account of all possible errrors due to the method used to measure the impedance.

For example the required impedance as measured is stipulated as 0.35 Ω or lower.

With very in-accurate measurements an actual impedance of say 1.0 Ω could give a false reading of 0.35 Ω

Assume an impedance of 1.0 Ω is acceptable for safe operation (*) but a reading of 0.35 Ω is required to ensure that the most inaccurate measuring techniques will not "pass" the installation when the actual impedance is 1.1 Ω but the measurement gives a reading of 0.36 Ω

(*) 1.0 Ω is probably not acceptable but chosen for clarity in the explanation

 

[JohnW2]

It is possible that the prescribed values take account of all possible errrors due to the method used to measure the impedance.

Indeed. In fact, given that the regs are silent as regards the matter of measurement error, I cannot see how it could be otherwise.

"All possible errors" is perhaps going a bit too far (since that could be taken to encompass grossly malfunctioning meters!), but I would expect the specified required limits (maxima or minima, as the case might be) would take into account the anticipated accuracy of commonly used (and 'in-spec') measuring equipment - and I presume that's what you mean.

Kind Regards, John

 

[EFLImpudence]

How can measurement errors be taken in to account when the values are calculations of the voltage and tripping characteristics?

It doesn't make any difference to the requirements and complying with them but -
Type B MCBs shall have an Ia of between 3 and 5 In - so 5 x is the worst case.
Does anyone know what the general figure for B MCBs is?

 

[JohnW2]

Every measurement must have a tolerance .... if we want 1.44 and get 1.45 then to fail it would be hard, as we don't really know if the out of tolerance is due to measurement errors or installation faults.

As bernard has said, and given that the regs are silent about measurement error, one has to assume that the stated requirements are absolute, and take into account the anticipated measurement error of acceptable and in-spec measurement equipment. Hence, if the limit is 1.44Ω, then a measurement of 1.45Ω is, indeed, a 'fail'.

Of the 1.44 the reading is in fact split into two parts, the supply impedance and the installation impedance, we are told the supply should be better than 0.35Ω with a 100A DNO fuse. However often the supply is below that figure, this raises another question, if the supply is 0.25Ω can the installer use the extra 0.1Ω? or should he take the 1.37Ω - 0.35Ω and say the circuit needs to be no more than 1.02Ω.

The answer is surely yes. Both the regs and electrical common sense say that all that matters is the total loop impedance, regardless of how that is divided between 'internal' and external components. The requirements of the regs are therefore stated in terms of Zs, regardless of what proportion of that consists of the R1+R2 of the circuit within the installation.

The problem is with other than a 100A supply, with 100A we can consider if the supply is 230 volt and the DNO is permitted a 6% volt drop or 13.8 volt then the impedance can be 0.138Ω but it does not really work out like that ....

Of course it doesn't. It all depends upon what the voltage is at the secondary of the transformer. If it is 253V, then the 'permitted voltage drop' for the most distant consumer is 36.8V, not 13.8V.

So the question, if the loop impedance is higher than required what happens then?
1) Complain to DNO that the supply impedance has dropped.

Is there a typo there? If not, I don't understand.

2) Complain to installer that the impedance is too high even when it was not too high when installed.

That would obviously be ridiculous.

3) Swap the MCB for a smaller one.
4) Replace part of the circuit with larger cable.

As you go on to imply, if (I would think very unusually, in the absence of network faults) the Zs of circuits had become too high because of a rise in Ze since installation, one would need to discuss the matter with the DNO and determine whether the new (higher) Ze was going to persist. In the (I would think unlikely) even that they did say that it was going to persist then, yes, the only way to bring the circuits(s) into compliance would be to change the cable and/or OPD. However, I very much doubt that that situation is going to arise in practice, other than in very exceptional circumstances. Other than in the presence of cable faults, I find it hard to believe that the DNO would make changes which resulted in the Ze seen by a particular property to increase - not the least because of the potential problem you mention.

Kind Regards, John

 

[JohnW2]

How can measurement errors be taken in to account when the values are calculations of the voltage and tripping characteristics? It doesn't make any difference to the requirements and complying with them but -
Type B MCBs shall have an Ia of between 3 and 5 In - so 5 x is the worst case.

I think the assumption would have to be that all of the figures we work with (including the Ia of MCBs) include a consideration of measurement errors.

If that were not the case then we would have the ridiculous situation in which, other than in a fictional utopian world of 100% accurate measuring equipment, an apparently 'compliant and safe' installation could actually be unsafe.

Perhaps someone (stillp?) could tell us exactly what the Standard requires of the Ia of an MCB - and what, if anything, it syas about considerations of measurement errors?

Does anyone know what the general figure for B MCBs is?

I don't quite understand that question. Are you perhaps asking the same as I have just asked?

Kind Regards, John

 

[EFLImpudence]

I think the assumption would have to be that all of the figures we work with (including the Ia of MCBs) include a consideration of measurement errors.

How can it?
For 32A the requirement is simply (230/(32 x 5)) x 0.95 (with correction factors for temperature), i.e. 1.37Ω.
If your meter reads 1.30 when it is actually 1.40, no one knows. How can that be a consideration?

If that were not the case then we would have the ridiculous situation in which, other than in a fictional utopian world of 100% accurate measuring equipment, an apparently 'compliant and safe' installation could actually be unsafe.

So, what is this consideration of errors of which you speak?

Perhaps someone (stillp?) could tell us exactly what the Standard requires of the Ia of an MCB - and what, if anything, it syas about considerations of measurement errors?

It is, for Type B, that they trip (instantaneously) at between 3 and 5 times their In.
We calculate the worst case at 5x but I do not know what they actually are - maybe 4x.

I don't quite understand that question. Are you perhaps asking the same as I have just asked?

Similar, although you don't seem to know about the 3 to 5 times requirement.

 

[slippyr4]

3 pages of arguing about the regs and no one has complimented the OP on his fairy lights!!

 

[JohnW2]

It is, for Type B, that they trip (instantaneously) at between 3 and 5 times their In. We calculate the worst case at 5x but I do not know what they actually are - maybe 4x. .... Similar, although you don't seem to know about the 3 to 5 times requirement.

You seem to be assuming an answer to the question I'm asking, and to which I don't know the answer for certain.

Yes, we talk about a Type B MCB tripping magnetically between 3In and 5In, but what does the Standard actually say? If it really does say that, in the 'worst case' scenario, the device has to be guaranteed to trip magnetically at exactly 5In, but not necessarily at a current fractionally below that, then this would mean (assuming symmetrical measurement errors) that 50% of cases in which measurement + calculation indicated a fault current of exactly 5In would actually be 'unsafe'.

Maybe that is the case, but, if so, it would seem rather unsatisfactory. I'm therefore speculating that when we say that the MCB has to trip at 5In, the Standard might actually require that it trips at 5In - x%, to make allowance for measurement error. If it doesn't, then BS7671 really ought to drop its 'maximum Zs' figures a bit to 'keep things safe' even when loop impedance measurement was done by a slightly 'under-reading' meter.

Returning to your initial comments, I wonder if, in practice, even if there is a requirement for magnetic tripping at "3In-5In" that, in practice, virtually all of them trip at a current closer to 3In than 5In? Unfortunately, very few of us have the ability to determine that by experimentation ourselves!

Kind Regards, John

 

[bernardgreen]

It is, for Type B, that they trip (instantaneously) at between 3 and 5 times their In.

It is very difficult to measure something that happens "instantaneously" at some point whenm the fault current is rapidly increasing. A digital meter may not sample and hold at the instant the disconnect happens. The inertia of an analogue meter's mechanism will not allow the needle to follow the current waveform. A storage oscilloscope with a fast sample rate will be the best option.

For laboratory testing of an MCB other methods can be used.