The distance between the roof and the consumer unit is not free

Resistance turns a fraction of the output into warm cable

A conductor is not a perfect path. It has resistance, that resistance is proportional to length and inversely proportional to cross-sectional area, and the power lost in it rises with the square of the current flowing through. Double the current and the loss quadruples. That single relationship governs almost every cabling decision in a solar installation.

In a well-designed domestic system the total loss in the cabling is a modest percentage, small enough that nobody would notice it in a month’s figures. But it is permanent. It applies to every unit the array produces for as long as the system exists, and unlike shading or soiling there is no weather in which it goes away.

It is also the cheapest thing in the project to fix, provided it is fixed on the day. Copper of a larger cross-section costs a little more per metre and nothing at all in labour if it goes in with everything else. Replacing an undersized run later means opening the same route again, which is where the real expense sits.

High voltage and low current is why the long run is on the direct-current side

Panels wired in series produce a string voltage in the hundreds of volts at a current of only a few amps. Because loss scales with the square of current, that combination is remarkably tolerant of distance, which is precisely why the cable from the roof to the inverter can be long without much penalty.

The alternating-current side is the opposite. It sits at the supply voltage, which is much lower than a string voltage, so delivering the same power requires considerably more current and the losses grow accordingly. The design consequence is straightforward: the run between the inverter and the consumer unit wants to be short, or generously sized, or both.

That’s the real reason an inverter is often mounted near the board rather than near the array, and it is a genuine trade rather than a rule. Somewhere in the middle of a long building, both runs are mediocre. At one end or the other, one of them is short and the other is carrying the form of current that copes with length better.

On export the cable pushes voltage up rather than down

This is the part that surprises people, and it is the part that actually costs generation. To push current out of the house, an inverter must hold its output slightly above the voltage of the grid at the connection point. The size of that difference is the drop across the cable between them — except that on export it appears as a rise at the inverter end.

Supply voltage is permitted to vary within a statutory band, and an inverter is required to disconnect if the voltage at its terminals leaves that band. Now stack the effects. The street voltage is already towards the upper end, neighbours with arrays are exporting on the same feeder, your own cable adds a further rise, and the inverter finds itself outside its limits.

The symptom is unmistakable once you know it: generation flattens off or drops out entirely on the brightest, sunniest days, precisely when it should be highest, and returns as soon as output falls. Some inverters throttle gently instead of disconnecting, which hides the problem while still costing units. A generously sized alternating-current cable is one of the few parts of the fix within the installer’s control.

The route matters for reasons that have nothing to do with energy

Cable does not simply need to be thick enough; it needs to be somewhere sensible. Direct-current cabling from a live array is energised whenever there is light, so where it passes through the building, how it is contained, and how clearly it is identified are all matters that wiring rules address in detail rather than leaving to judgement.

Physical conditions change the sizing too. A cable in a hot loft carries less current safely than the same cable in a cool cupboard, and cables bunched together in a conduit each derate the others by warming the group. A calculation that ignores the installation method produces an answer that is right on paper and wrong on the roof.

Rodents deserve a mention too, since cable in a loft void is exactly the sort of thing that gets chewed, and damaged insulation on a direct-current conductor is a more serious matter than a chewed aerial lead. Containment costs little. Nobody regrets it.

Where the inverter goes is the same decision in disguise

Inverter placement is usually discussed in terms of heat and access, and those matter: electronics in a loft that reaches oven temperatures in August will not last as long as the same box in a garage, and a unit mounted somewhere awkward will be inspected less often and replaced at greater cost. But placement is also, quietly, a cabling decision.

Moving the inverter towards the array shortens the direct-current run and lengthens the alternating-current one, which is the wrong direction for losses and for voltage rise alike. Moving it towards the board does the reverse. Given the choice, and given that the direct-current side tolerates distance better, near the board is usually the better compromise — subject to the space being cool, dry and reachable.

None of which justifies obsessing over a fraction of a per cent. Cable losses are a minor line in the energy budget of a domestic system, and a design that sacrifices sensible inverter placement to chase them has optimised the wrong thing. Voltage rise on export is the item worth caring about, and its limits are set by local wiring rules.