This isn't that easy to answer, but please bear with me whilst I try.

The parts of BS 7671 you refer to are aligned with IEC 60664-1. Whilst IEC 60664-1 provides some common rules, it's not always possible to do a direct read-across between the "generic" information, and product standards which have more specific details for the requirements for a particular use-case. So, it's the product standard we need to look at.

Before we look at the product standard, it's worth noting that according to IEC 60664-1, the product standard may require either an electric strength test (under specified conditions), and/or an impulse withstand test (again under specified conditions). As well as electronic equipment, switch and isolator gaps, in particular should be rated for a particular impulse withstand test (to see whether the isolation gap is breached under impulse conditions for example) whereas tests for the performance of insulation alone may be more suited to an electric strength test.

The relevant product standards (as quoted by the popular manufacturer I think we are both looking at ... they certainly say the same things) are BS EN 60998-1 and BS EN 60998-2-2. These standards require (Section 13 of BS EN 60998-1) an electric strength test, which is conducted at 2.5 kV for rated insulation voltage up to 450 V AC. There is no requirement for an impulse withstand test in the product standard.

So, I think in this case, the manufacturer is probably advising you that they have complied with the relevant standard, and may have "read-across" to the 4 kV "surge voltage" (which I take to be the impulse voltage U_W) ... which, other than "read across" has absolutely nothing to do with the "electric strength test" value (applied for 1 minute) according to the product standard.

In summary, the "electric strength test voltage" of 2.5 kV is NOT the same as a U_W of 2.5 kV in Table 443.2 ... and for these reasons, I don't think there's anything for you to worry about, particularly where the products comply with the relevant standard.

Thank you very much for taking the time to prepare such a detailed response Graham. I believe we are both looking at data from the same manufacturer and I am not surprised to learn that my confusion is a result my misinterpretation of the requirements.

I think that the learning point for me is not to take the guidance in section 443.6.2 too literally and that the specific product standards will take precedence.

Thank you very much for taking the time to prepare such a detailed response Graham. I believe we are both looking at data from the same manufacturer and I am not surprised to learn that my confusion is a result my misinterpretation of the requirements.

I think that the learning point for me is not to take the guidance in section 443.6.2 too literally and that the specific product standards will take precedence.

With single terminals like wago connectors flapping in a box, the real sideflash rating is an uncontrolled quantity. Clearly within the same connector all the metal is at one voltage, so there is nothing to induce a breakdown. The question is how close is the neighbour.

It may amuse you that we use some 2 pole 222 style Wagos in a small 150kV test rig that I am DA for. But the ones at the highest voltages are in effect in free space with at least about a foot of dry air to nearest near-earth potential. They do corona a bit with the rig at full chat ;-) ...

The situation changes totally when they are rigidly mounted.

but now the voltage to initiate breakdown between adjacent nets is still not a function of the connector per se, but of its supports, and how well that supports or interrupts either air discharge or  surface tracking through dust or condensation - and in turn if the environment actually has that.

One solution for situations with higher risk to increase the tracking  rating between terminals is then to use a 'missing man' contact formation - actually years ago we used to knock out every other terminal in choc-block to achieve a similar end.

I suspect that by choosing the geometry for the test method almost any result is possible, and there is no substitute for actually inspecting the particular installation.

Mike.