Easy Discrimination Quiz

[mfarrow]

Out of interest, what make was the MCB?

 

[ban-all-sheds]

Put your hand down, that boy.

 

[ban-all-sheds]

Put your hand down, that boy.

A wonderful example of the importance of correct punctuation.

 

[echoes]

Put your hand down, that boy.

A wonderful example of the importance of correct punctuation.

Good job I wasn't asking to put his hand up...

 

[echoes]

Out of interest, what make was the MCB?

Wylex (NHX B20)

 

[JohnW2]

Almost 'on cue' (for this thread, my neighbour) sliced through the 2-core cable of his (Class II) hedge trimmer this afternoon, and it was a case of 'none of the above'.

A 5A 1362 in the appliance plug, a 13A 1362 in the plug of the extension reel he was using, a B16 RCBO and the cutout fuse all failed to operate. Fortunately he did not touch the live cut end of the cable.

I thought I ought to check (as well as mending the lead for him!) ... L-N loop impedance measured at the end of his (3-core) extension reel was about 1.2Ω

Kind Regards, John

 

[mfarrow]

Yes John, an exact replica of my neighbour's sliced mower lead - just a big bang but nothing tripped.

Wylex (NHX B20)

I can't seem to find a curve for the 5A BS 1362 fuse, but it looks probable that the fault current was in excess of what was needed to blow the fuse within 0.01s seconds. From what I can deduce by Wylex's literature, their MCBs, in common with many others, take minimum that long to trip anyway, so the fuse was always likely to blow. Whether a BS 1361 or 88-3 (whatever they are now) would have avoided this, I could not say, but as they are not limited by a physical mechanism, it is likely.

 

[mfarrow]

Deleted.

 

[JohnW2]

I can't seem to find a curve for the 5A BS 1362 fuse, but it looks probable that the fault current was in excess of what was needed to blow the fuse within 0.01s seconds. From what I can deduce by Wylex's literature, their MCBs, in common with many others, take minimum that long to trip anyway, so the fuse was always likely to blow. Whether a BS 1361 or 88-3 (whatever they are now) would have avoided this, I could not say, but as they are not limited by a physical mechanism, it is likely.

When one is dealing with very high currents and very short disconnection times, I suspect that this is not really quite as much a ‘science’ as we might like to think – which is perhaps the reason for some of the ‘unexpected’ behaviour one sees in these various ‘discrimination (or non-discrimination) stories’.

Finding adequate information (‘curves’) for the operational characteristics of OPDs is not easy. The curves in BS7671, for both fuses and MCBs, ‘stop’ (i.e. don’t go below) at 100 ms. Curves for MCBs invariably show the magnetic operation as a ‘vertical line’ – from the shortest operation time displayed on the curve (100 mS for the BS7671 tables) up to 5 seconds. Even this is a bit misleading – to reflect reality, there really ought to be a gap in the curve between 5 seconds and a very much lower current (maybe ~10ms), since the implication of the curve that magnetic operation can take anything up to 5 seconds clearly is not correct. Indeed, one wouldn’t really expect operation time for an MCB to be appreciably influenced by current magnitude – once a current equal to or above the ‘threshold’ is reached, the actually tripping is yes/no, after which the operation time is just down to how fast the mechanical mechanism operates – so I would expect that operation time to be essentially the same (any very short) regardless of the magnitude of the fault current.

The Wylex MCB data I’ve seen is also a little misleading in another respect, because they show the ‘vertical’ magnetic part of the curves right down to 1 ms, and I don’t believe that they ever operate that rapidly. The text of the Wylex technical documentation indicates that their standard MCBs generally operate in 3.5 to 5.0 ms. However, as above, this is going to be primarily the amount of time is takes for the physical mechanism to operate (i.e. for the contacts to move) after the mechanism has been tripped, but we’re never told how rapidly that ‘tripping’ occurs. With an operation “within 3.5-5.0 ms” one might guess that the ‘tripping’ (after which operation is inevitable) probably doesn’t take more than about 1ms – in which case a fuse would have to disconnect the supply in less than 1 ms to prevent the MCB getting to ‘the point of no return’.

Another complication in terms of discrimination between fuses and MCBs is that their operation presumably depends on different things. An MCB presumably simply ‘trips’ when a certain absolute level of current is exceeded. With a fuse, it is the amount of energy put into the fuse link over a period of time which matters. If (per above), we’re talking about ‘tripping’ (not operation) times of the order of 1 ms for an MCB (i.e. 0.05 cycles at 50Hz), then the relationship between operating times (and, indeed, the operating times themselves) is likely to be at least somewhat dependent on the point in the cycle at which the fault commences.

... and then, of course, operation of a fuse has got to have some variability to it – melting and/or vaporisation of a fuse link is never going to be totally reproducible.

So, all in all, whilst the operating characteristics of OPDs obviously provide a reasonable guide to likely discrimination between devices, I think there are always going to be appreciable uncertainties and unpredictabilities (not to mention difficulties in sourcing adequate information about the characteristics) regarding discrimination, unless, one is talking about devices of very different In.

Kind Regards, John

 

[JohnW2]

Yes John, an exact replica of my neighbour's sliced mower lead - just a big bang but nothing tripped.

He said that there wasn't even a bang - just 'silence' as the trimmer stopped. I suppose there is really 'no telling' exactly what is going to happen when one chops through a cable in this sort of way. There is certainly the possibility that the duration of the L-N contact could be extremely short - too brief for any OPD to operate. It is even just about possible that it could happen without there ever being any electrical connection between L and N - one conductor could be sliced and 'pushed out of the way' before the other conductor came into contact with a blade. ... and, of course, if it were all over incredibly quickly (say <1 ms), whilst the supply waveform happened to be around its zero-crossing point, there might not actually be much of a fault current.

... lots of unknowns/uncertainties/unpredictabilities!

Kind Regards, John

 

[mfarrow]

Finding adequate information (‘curves’) for the operational characteristics of OPDs is not easy. The curves in BS7671, for both fuses and MCBs, ‘stop’ (i.e. don’t go below) at 100 ms.

And for a type D they're downright confusing.

Even this is a bit misleading – to reflect reality, there really ought to be a gap in the curve between 5 seconds and a very much lower current (maybe ~10ms), since the implication of the curve that magnetic operation can take anything up to 5 seconds clearly is not correct.

I've often wondered why this is the case - I can see how it's technically correct, but it takes some getting used to.

Indeed, one wouldn’t really expect operation time for an MCB to be appreciably influenced by current magnitude – once a current equal to or above the ‘threshold’ is reached, the actually tripping is yes/no, after which the operation time is just down to how fast the mechanical mechanism operates.

Which is the achilles heal of the MCB - it looks great on paper where all fault start at 0A on the sinusoidal waveform, but as soon as you have a fault occurring anywhere else on the curve, you're going to see a large let-through current before the MCB will trip. And, in the time it's taken to unlatch itself and start breaking, the downstream fuse will have removed the fault.

The Wylex MCB data I’ve seen is also a little misleading in another respect, because they show the ‘vertical’ magnetic part of the curves right down to 1 ms, and I don’t believe that they ever operate that rapidly. The text of the Wylex technical documentation indicates that their standard MCBs generally operate in 3.5 to 5.0 ms.

That is a typical time from BS EN 60898, however the let-through energy graph from the same literature says "HW=0.01s". I've assumed this is the time they've used to calculate what else is on the graphs, but I can't find where HW comes from.

 

[JohnW2]

Even this is a bit misleading – to reflect reality, there really ought to be a gap in the curve between 5 seconds and a very much lower current (maybe ~10ms), since the implication of the curve that magnetic operation can take anything up to 5 seconds clearly is not correct.

I've often wondered why this is the case - I can see how it's technically correct, but it takes some getting used to.

I would not say that it is even technically correct. By presenting it as a continuous curve, it implies that it is possible to get operation times between (for the BS7671 curves) between 100 ms and 5 seconds with a current at the minimum level which should result in magnetic tripping (e.g. 5*In for a Type B MCB), and I don’t think that’s correct or possible. The Wylex curves are even more incorrect, in that the continuous curve implies that it’s possible to get any operation time between 1 ms and 5 seconds at that current. As I said, there should be a gap in the curve between 5 seconds and the longest possible operation time at that current (say, about 10 ms).

Which is the achilles heal of the MCB - it looks great on paper where all fault start at 0A on the sinusoidal waveform, but as soon as you have a fault occurring anywhere else on the curve, you're going to see a large let-through current before the MCB will trip. And, in the time it's taken to unlatch itself and start breaking, the downstream fuse will have removed the fault.

Given that we seem to be talking about operation times which will usually be less than half a cycle (possibly, per Wylex, no more than quarter of a cycle), I don’t even know at what current they are meant to ‘trip’ (‘unlatch’). When we say that, for example, a B20 should result in the magnetic trip at “100A” what does this actually mean? The calculations (e.g. of PFC etc.) we do are in terms of RMS voltage and current. So, do we expect the B20 to trip at an instantaneous current of 100A ('RMS') or the corresponding peak current of about 141A, or what?

Whatever ‘instantaneous trigger current’ we are talking about, if I understand you correctly, I would have thought that you comment is ‘the wrong way around’. I would have said that the ‘worst case’ (for the MCB) would be a fault starting at the zero-crossing point of the curve, since that would represent the situation in which it would take the longest for the current to reach the MCB’s ‘trigger threshold’. Conversely, if the fault started at any point on the curve corresponding to a current above the trigger threshold, the MCB would trip (‘unlatch’) ‘immediately’, without having to wait for the current to rise to the threshold ... or did I misunderstand what you were saying?

However, with a B20, if I’ve got my sums right, even the ‘worst case’ wouldn’t be too bad. If one hypothesises a fault current of 400A (RMS, hence about 566A peak) starting at the zero-crossing point of the waveform, it would only take about 0.57 ms (about 10.2° of cycle) for the current to rise to a 100A trip threshold (or about 0.8 ms, about 14.4° of cycle if the threshold were 141A). These times are pretty small because, of course, the voltage/current are rising at their fastest around the point of zero-crossing.

Kind Regards, John