Cable is the least interesting part of a solar plant and one of the easier places to save money at tender stage. It is also a decision you cannot revisit – nobody rewires an operating array because the losses turned out higher than modelled.
Resistive loss in a conductor rises with the square of current and linearly with length. On the DC side, strings run at relatively low current but often over long distances across a large roof to a centralised inverter location. Those runs add up.
A typical target is to keep total DC voltage drop under about 1% at maximum power, with a commonly accepted ceiling of 2%. Above that you are giving away energy every hour the plant runs.
The difference between a 1% and a 3% DC loss on a 750 kW plant producing around 900 MWh a year is roughly 18 MWh annually. Over twenty-five years, at commercial tariffs, that is real money – typically far more than the copper would have cost.
Sizing at STC rather than operating conditions. Conductor resistance rises with temperature. Cables in a roof-mounted tray in direct sun run considerably hotter than 25°C – frequently 60-70°C. Copper resistance increases roughly 0.39% per °C, so a cable sized on nominal resistance is optimistic by 15% or more in service.
Ignoring grouping and installation method derating. Cables bundled in a tray or conduit cannot dissipate heat as freely as a single cable in open air. Both IEC 60364-5-52 and the NEC provide derating factors for grouping and ambient temperature; applying them is not optional, though it is sometimes skipped.
Averaging the run length. Strings at the far corner of the roof may be twice the length of the nearest ones. Designing every string to the average means the distant ones exceed the loss target. Either size by longest run or vary conductor size by zone.
Voltage drop between inverter and point of connection is often the larger of the two, particularly where the switchroom is remote from the array. On sites where the inverter sits on the roof and the connection is 200 m away at ground level, this run deserves as much attention as the DC side, and it carries the full plant current.
Cable selection is a calculation. Cable installation is a craft, and it fails in ways calculations do not capture: connectors from mismatched manufacturers mated together, crimps made with the wrong tool, cables left lying in standing water on a flat roof, UV degradation from runs unsupported for years.
Mismatched MC4-type connectors are the classic. They physically mate, they appear fine, and they develop resistance at the junction over several years – producing heat, then arcing, then in a bad case a fire. Specify that all connectors come from a single manufacturer and are crimped with that manufacturer's tool. Then check a sample during construction.
Cable is boring right up until the point it is not.