When operating normally, SPDs only handle brief transients - the sort you get from e.g. a lighting strike near the distibution network somewhere. So although a significant current may flow through the SPD, the total energy passed is small and the cable won't overheat. If the transient is too long, the fusible element built into the SPD will open, again before the wire has overheated. Finally, the fact the OCPD in the SPD is after the cable is probably covered by [433.2.2] and/or [434.2.1]


(All the above is speculation on my part.)

Humm 2.5mm² does seem a bit smaller than I would have expected - hardly in the spirit of 534.4.10 (although presumably these will be covered by the product standard rather than BS 7671). Maybe it says something about the size of surge than the manufacturers think it'll be able t cope with.  It even seems a bit thin for ordinary fault protection given that the upstream protective device is likely to be the DNO's fuse at anything up to 100A.

   - Andy.

there is no thermal reason for a bus bar.

Take rather more than the highest rated current, let's say 40kA, and use an approximation that over-estimates the dual exponential (8us rise 20us fall) of the official
spec for the unit   by a simple rectangular current pulse duration of perhaps 20 microseconds at this higher level,  and bung it back into the adiabatic forumula, to estimate a conservative cable size.

K²A²   >   I²t  (units joules per ohm)


re-arranges as to give the area A in mm2 as


A²   >   I²t/K²


(K is a constant set by the heat capacity of copper- temp rise per joule), the electrical resistance of a unit cube of copper, and the maximum temperature rise we are happy with,)

for copper under pvc with a cold start, K  is perhaps 120 ish, could be up to 140.  so K²  is in the region of 10000 to perhaps 20000 ish

So I²  might be 1600 million, t might be 20 millionths  of a second so the top of the fraction is perhaps 32000

divide that by a K² of at least 10000 and A becomes square root of  3 and a bit,  or a touch under 2mm2


In reality, just because it does not have time to get hot at 40,000 A does not mean you should use the bare minimum - we need to keep inductance down, and avoid damage by magnetoconvulsion - so a bit of extra 'meat' is desirable. Personally I prefer flat braid. I imagine the Wylex accountant prefers plain round wire however.


regards Mike

Thanks for taking the time to go through it all mike, much appreciated! As you say I guess like most things these days it comes down to cost and manufactures getting away with the bare minimum to maximise profit.

I have 2 Wylex SPD REC 2 with SPDs and both came with bus bar links.

That does look  bit cheap - however, as the calculation above shows, it will still pass the test spec of the 8us/20us  at 20kA, probably with a bit in hand. I imagine there has been a design review and a decision to do some 'value engineering' and save a few pence.

In terms of SPD function it would be better if the links were short and fat, and routed together in a way that their magnetic fields cancelled (*), as this would give the lowest overshoot voltage during the surge. (well not quite, the lowest would be if the mains came in as a piar to the SPD, and then out again from the SPD, but a short 'T branch' is unavoidable in most designs.)

Looking at that, is the SPD on the incoming or outgoing side of the switch ? - I'd expect out going and I presume this is in a box on its own, not in a CU where the bus bar is at the bottom.


No prob with the calculation - it highlights something I first realised when looking at a RADAR designs, that the short pulse ratings of components  can be thousands of times the steady state rating - when limited by thermal considerations, and shorter more peaky pulses are permitted, until we become limited by voltage breakdown, or some other 'near instant' effect, then this scaling up can no longer apply.


Regards Mike


* Aside. If you use a clamp meter as a measure of the  magnetic field around a wire,. you can compare the effect of looking at the live only on a wire feeding a load, or the neutral wire only, or clamping around the whole cable , where suddenly you only see the L-N difference current, as the two fields essentially cancel.out  - energy wasted in filling the space with magnetic fields is the inductive part of the voltage drop, which we can reduce quite noticeably by keeping flow and return currents together in this way.

At 50Hz it only really matters on kA mains wired in singles.

When folk use scope probes to measure microsecond events and find that there is high frequency ringing that changes with how close the earthing wire is brought to the place the probe tip connects, this is the same effect at work at smaller scale.

Where it really matters a lot, ultra low inductance lines can be made of  either coaxial cables with very chunky centres , or if we need extra capacitance, as we might on power supplies to heavy RF devices, then interleaved flat bars of alternating polarity, or on a PCB alternate planes of copper.

Yes, it is on the outgoing side and is a separate unit on its own outside of the consumer unit. 


Thanks for all all the information regarding SPDs, it's something I'm fairly new to  !

That’s a consumer unit main switch, not a REC main switch ?