1  Combination: array field + substation

2  Laying out the field grid

Earthing a PV field is not a resistance problem; it is an equipotential and coverage problem. DC-side fault current is limited by the inverters to 1.1–1.5 times rated current – tens to a hundred-odd amperes – which any conductor size carries. What has to be achieved is that every structure row, combiner box and inverter sits on one equipotential body, with no floating metal when lightning strikes.

• Rings per block: divide the field into blocks (usually one inverter or one sub-array each) and run a ring electrode around the outside of each block, buried ≥ 0.6 m (≥ 0.8 m in farmland, below the plough layer).
• Mains into a mesh: tie the block rings together with main conductors along cable trenches or roads, bonded to trench supports, fences and kiosk enclosures as they pass; the mesh size equals the block size (tens of metres) – there is no need for the 5–10 m densification of a substation.
• Structure bonding: every row bonded to the ring or branch at two points at least, every 20–30 m on long rows; the posts themselves are natural electrodes – driven piles go 1.5–2.5 m into the ground and there are tens of thousands of them, and using them beats driving extra rods by a wide margin.
• Combiner boxes, inverters, kiosks: one or two Ø 14–16 × 2.4 m rods at each foundation, tied to the branch – not for resistance but to give lightning current a local path to earth.
• Air termination: the structures themselves act as air terminals; no separate masts (they shade modules). The posts are the down-conductors, so the post-to-ring joint is a lightning path: exothermic weld or purpose-made clamp, never a bolted splice.

3  Targets: resistance and grading

Chinese design practice under GB 50797 takes the field earth resistance at ≤ 4 Ω; on a grid of hundreds of hectares this is met almost automatically – by the IEEE 80 estimate R ≈ ρ/(4r), a 200 ha field (radius about 800 m) on 1000 Ω·m desert comes to 0.3 Ω. The field never needs resistance-reducing measures; anyone selling "resistance-reducing products" for a PV field is selling something useless.

Two things do need checking: ① step and touch voltage at the step-up substation (section 6); ② step voltage at the field perimeter and roads – once field and substation grids are tied together, the substation's ground potential rise reaches the field edge, and the potential difference where a passer-by stands outside the fence has to be checked per IEEE 80 clause 16.5, usually settled by a grading ring 1 m outside the fence.

4  Soil, corrosion and theft

PV plants are built on land nobody else wants: desert, salt flats, tidal flats, mining subsidence, fish ponds. These are exactly the "high corrosivity" soils of DL/T 2650-2023 Table 1. By its material ladder:

Three lessons:

• The soil–air line corrodes first. The post-to-electrode joint and the 30 cm of down-lead above ground see the greatest oxygen-concentration difference; galvanised steel rusts through there in 3–5 years. Either weld copper-clad steel straight to the post and coat the joint, or use prefabricated tinned copper-clad steel to galvanised-tape transition pieces.
• Nobody walks a PV plant. Operation relies on remote monitoring, which cannot see earthing corrosion; a retrofit means trenching hundreds of hectares under the modules, and the labour plus module damage cost an order of magnitude more than the material saved. That is the core reason a PV plant takes 0.6/0.8 mm rather than 0.25 mm.
• Theft. Bare copper tape and cable in open country are the first things to go; copper-clad steel cannot be stripped and has negligible scrap value – the direct answer to field theft.

5  Quantities and bill of materials

A 100 MW-class plant covers about 200 ha, fifteen times a 2 × 350 MW thermal plant, with ten times the connection points – PV is the largest segment by kilometres, with small sections. The actual bill of materials from one 100 MW-class plant we supplied:

Rule of thumb: about 350–400 m of horizontal conductor, 1.5 rods and 70 exothermic joints per MW; add 20–30 % on hill sites and scattered plots. An exothermic joint takes 3–5 minutes; a three-man crew makes 60–80 a day, so the earthing of a 100 MW plant is done in 2–3 weeks.

6  Tying the field to the substation

Walk the step-up substation through the substation page: the 110 / 220 kV single-phase fault current sets the grid section (usually Ø 14–16, 0.8 mm), and step and touch voltages are checked per IEEE 80. The field grid and the substation grid must be one grid – left separate, a fault produces a dangerous potential difference between them, and the collector-cable screens carry it into the kiosks. Tied together, the substation resistance becomes very low (the field is large) but the GPR spreads over the whole field, which is why section 3 checks perimeter step voltage. We issue the substation calculation to IEEE 80 for the design institute to check.

7  Field-grid calculator

R ≈ ρ/(4r) + ρ/L, r = √(A/π). For showing that the field needs no resistance reduction; for the substation's R_g, GPR, step and touch voltages use the calculator on the substation page.

8  What to send, common mistakes

Send: capacity and plot coordinates (we estimate the soil type from them), step-up voltage and fault current, structure type (fixed / tracker, piles / concrete), design life, whether theft resistance is a requirement.

Common mistakes: ① measuring resistivity at one point by the substation – 200 ha needs 5–10 points, salt flat and dune can differ a hundredfold; ② buying "resistance-reducing products" for the field (useless, and corrosive); ③ bolted splices as lightning paths on structures (arc once loose); ④ field and substation grids not tied together; ⑤ 0.25 mm plated wire as the horizontal electrode on a salt flat – it belongs in dry, low-corrosivity, maintainable sites, not here (see position 1).