Tare loss is the power an inverter draws from the grid when it is not exporting. For a single modern micro-inverter it is tiny, typically 15 mW to 50 mW. The issue is multiplication and duration. A UK array has more than 4,000 non-generating night-time hours a year, and every one of its micro-inverters keeps drawing standby power through all of them.
The cause is topology. A string inverter consolidates DC-to-AC conversion into one enclosure and one isolation layer, so at dusk a single control board drops into sleep. A micro-inverter array spreads that conversion across dozens of independent printed circuit boards (PCBs). Each keeps its own grid-sync, monitoring and communications circuitry partly powered, and each adds a component that can fail and a path for residual leakage.
| Metric | Micro-inverter array (15–30 units) | Centralised string inverter |
|---|---|---|
| Conversion points | 15–30 independent PCBs | 1 consolidated stage |
| Per-unit standby draw | 15–50 mW | Single sleep-mode load, typically under 1 W |
| Aggregate night draw | 0.23–1.5 W | Under 1 W |
| Annual tare (~4,400 h) | ~1–6.6 kWh | Up to ~4 kWh |
| December night (16 h) | ~4–24 Wh | Up to ~16 Wh |
| Grid-tied components at night | Every module | One |
Set against a typical 4 kWp array producing about 3,500 kWh a year, annual tare is well under 0.2% of output. Seasonality changes the picture. On an overcast December day the same array may only yield 1–2 kWh, so a 30-unit stack's 24 Wh night draw cuts net output by roughly 1–2%. That is several times the annual average, and it lands in the months when every kilowatt-hour is worth most.
For auditors and specifiers, tare loss is a measurable but secondary line item, not a reason to reject micro-inverters. Model it alongside shading gains and component count, check datasheet night-consumption figures, and compare aggregate draw rather than per-unit marketing values. The same standby-loss logic applies to hot water cylinders on smart tariffs.