12.2 Electromagnetic Interference and DC Surge Protection

Why loop area belongs in a wiring lecture, not only in an SPD catalogue

Lightning does not have to strike the array to damage an inverter. A stroke to a nearby chimney, tree, or air termination dumps a very large, very fast current into a conductor. That current produces a rapidly changing magnetic field. If your positive and negative DC cables enclose a large loop, that field cuts a large area, and a large induced voltage appears between the poles — or between the DC circuit and earth. The inverter’s DC input, module bypass diodes, and insulation all see that transient.

The same loop is also a better antenna for everyday electromagnetic interference (EMI): variable-speed drives in a workshop, amateur radio, or the inverter’s own high-frequency switching. Handbook 4.2.7 therefore starts with cable geometry, not with a part number. Independent OpenExamPrep teaching for City & Guilds 2922-34 candidates uses that geometry plus the exact 712.443.101 sentence in handbook v1.3.

Run positive and negative side by side — and do not build loops

Requirement 4.2.7(a) is blunt: run positive and negative cables side by side and avoid “loops” to reduce susceptibility to surges and electromagnetic interference.

Side-by-side (clipped as a pair, or a purpose-made two-core PV cable) keeps the area between the conductors down to a few millimetres along the run. Equal-and-opposite currents then produce magnetic fields that largely cancel (less EMI emitted by the DC circuit) and the pair picks up less flux from an external field (less EMI and surge received).

A loop is any path where + goes one way around an obstacle and − goes the other way, so the two conductors and the inverter terminals enclose a polygon of roof. Classic loops on 2922 jobs:

• • along the east eaves and − along the west eaves, meeting at an inverter on the north gable.
• • and − split around a chimney, dormer, or soil stack “to keep the cable tidy.”
• A single-core run down one rafter bay and the return down a bay three metres away.

The loop area is what the examiner is pointing at. Doubling the enclosed area roughly doubles the flux linkage for the same magnetic field, so the induced voltage rises with area. You cannot “make it up” with 6 mm² cable. Cross-sectional area fixes voltage drop and current rating; it does not cancel a five-metre-by-four-metre roof loop.

Keep cable-runs short

Handbook 4.2.7(b) asks you to consider keeping cable-runs short. Short DC cables do four useful jobs at once:

• Less length means less chance to wander into a loop.
• Less inductance in the DC circuit, so switching edges and surge fronts are tamer.
• Less “antenna” for EMI.
• Less voltage drop — which 4.2.5 already cares about for performance.

That is one reason the next chapter will push you to site the inverter so the DC route is short, even if the AC route is a little longer. A 20 m paired run along a ridge is still better than a 12 m run that splits around the chimney, but the shortest paired route is the design you should be able to justify.

Induced surge versus a direct strike

Two different physical events get lumped together in site chat as “lightning.” Keep them separate when you write an answer:

• Induced / conducted surge — the magnetic pulse from a nearby stroke, or a surge arriving on the AC supply. This is the everyday UK risk on a house without a lightning-protection system (LPS). A Type 2 SPD is the usual device class for that wave shape.
• Direct or partial lightning current — the building has an external LPS, or the array is in a zone where a strike to the structure can inject lightning current into the DC conductors. That is a Type 1 (or combined Type 1+2) conversation, not a “fit any metal-oxide varistor and tick the box” conversation.

Loop control reduces how much voltage the magnetic pulse can induce before the SPD has to work. The SPD is the clamp. You still need both ideas: a small loop and, when 443 says so, a real DC SPD.

BS 7671 regulation 712.443.101

Handbook 4.2.7(c) quotes the regulation: where surge protection is required by section 443 it shall also be applied to the DC side of the installation.

Read the “where” clause. Section 443 is the general-installation rule for transient overvoltage. Amendment 2 of BS 7671:2018 simplified the old calculated-risk map: protection against transient overvoltage is required where the consequence of overvoltage could include serious injury or loss of human life, failure of a safety service, or significant financial or data loss. A great deal of new dwelling work therefore carries SPD at the origin. If that Section 443 duty is in play for the installation, 712.443.101 extends it to the PV DC side. If a documented 443 decision means SPD is not required on that installation, 712.443.101’s extra DC duty is not triggered — but paired short cables still are. Never skip 4.2.7(a)–(b) because someone skipped an SPD.

Type 2 in the inverter — until the manufacturer does not cater for it

The handbook’s own next sentence is the one to recite: this DC-side protection is usually achieved by a Type 2 SPD in the inverter, however some manufacturers do not cater for this.

So the exam algorithm is:

1. Does Section 443 require surge protection on this installation? If no, you still pair the cables; you do not invent a 712.443.101 duty that the “where” clause did not switch on.
2. If yes, the DC side needs surge protection too.
3. Open the inverter datasheet and installation manual. If the manufacturer specifies an integral device as the DC SPD (Type 2 is the usual class on a house without LPS), that can satisfy 712.443.101.
4. If the manufacturer does not provide a DC SPD — unmarked varistors on a control board are not a declared SPD — fit an external DC SPD.

BS 7671’s note around the same regulation (installer knowledge you should carry even when the 2922 handbook is shorter) is that an inverter “SPD” only counts if the manufacturer specifies it for the DC side, and that varistors included in the inverter are not considered an SPD. DC SPDs are PV-specific products, typically to BS EN 61643-31, with + and − terminals, not a borrowed AC Type 2 with L and N labels. Generally Type 2 on the DC side; Type 1 (or 1+2) where the building has an external LPS.

Mount an external DC SPD close to the inverter so the connecting tails do not recreate a loop. If the DC cable on the roof is very long, manufacturers and BS 7671 guidance may also want protection near the array — that is a coordination exercise, not a reason to skip the inverter-end device when 443 has already required SPD.

Scenario: + and − on opposite eaves around a chimney

A ten-module landscape array is split by a chimney stack. The installer sends the positive down the east eaves, under the gutter clips, around the stack, and the negative down the west eaves, “so each colour has its own tidy route.” They meet at an inverter in the airing cupboard. The enclosed loop is essentially the roof plan around the chimney — several metres by several metres.

That layout increases surge and EMI susceptibility even if every cable is PV-rated, even if voltage drop still passes, and even if a Type 2 SPD sits in the inverter. The magnetic field from a stroke to the chimney or a neighbouring air termination cuts that huge loop. The correct construction brings + and − together at the array, clips them side by side down one designated route (avoiding the hot rear of the modules where possible, on a 70 °C-rated cable as the previous chapter required), and keeps that paired run short to the inverter. If 443 requires SPD and this inverter’s manual does not offer a DC Type 2, add the external DC SPD at the inverter rather than hoping the chimney loop was “only aesthetic.”

If a knowledge-test stem shows two colours taking opposite routes around a chimney, mark the loop, not the cable size.