Usable capacity: why a 1,000 Wh station doesn't give 1,000 Wh

On paper the DJI Power 1000 V2 holds 1,024 Wh. Run a 120 W TV from its AC outlet and about 756 Wh of that reaches the TV, roughly 74%. Run a 10 W router instead and only 42% arrives, because the inverter's own draw now rivals the load.

Three things sit between the label and your device

The rating on the box is the energy stored in the cells. Before any of it reaches a plug, three things take a cut, and they stack:

• Depth of discharge. The battery management system shuts the station off before the cells are fully empty, to protect them. BLUETTI puts this at 90% to 95% for its LiFePO4 packs and uses it as a term in its own runtime formula (BLUETTI, What is inverter efficiency).
• Inverter loss. Turning battery DC into 120 V AC turns some energy into heat. The same BLUETTI article calls 85% or better the benchmark for portable stations. Anker measured its SOLIX S2000 at 86% with a 100 W load against an industry average of 76% for mainstream 2 kWh stations tested the same way (Anker SOLIX S2000 page, footnotes).
• Idle draw. With the AC output switched on, the inverter consumes power even with nothing plugged in. Anker's same footnotes give 10 W for the S2000 and an industry average of 17 W.

None of the stations in our dataset publish their own figures for these three, so KiloHour fills them with labelled defaults: 90% usable, 88% inverter efficiency and 10 W idle on AC, 95% and 2 W on DC. The reasoning behind each one is on the methodology page. Note that our idle default is Anker's best case, not the 17 W average, so real losses on light loads can be larger than shown here.

Worked example 1: peeling the losses off one TV session

Inputs: the DJI Power 1000 V2 at 1,024 Wh, a 120 W flat screen TV on the AC outlet. Each row adds one loss to the row above it, and the engine computes every line at render time. "Reaches the TV" is runtime times 120 W.

At this load the inverter's conversion loss is the biggest single cut, and idle draw is the smallest because 10 W is small next to 120 W. Swap the TV for the router and the order changes: the same station then delivers 431 Wh over 43 h 8 min, and the battery is spending 21.4 W to supply 10 W.

Worked example 2: a small load, AC vs DC, new vs aged battery

Inputs: the EcoFlow DELTA 3 at 1,024 Wh, with a 126 W rated 12 V port, powering a 10 W Wi-Fi router around the clock. The aged rows set battery health to 80%, a hypothetical older pack, not a figure EcoFlow publishes.

On AC, idle draw is 47% of what leaves the battery, so moving the router to the 12 V port stretches runtime about 1.7 times. Battery health works differently: it shrinks the usable energy, so it cuts runtime by the same 20% on either path. A worn pack and a wasteful output path compound each other.

How much reaches the device on 1 kWh class stations

Rated capacity is from each maker's published spec. The other columns use the same defaults for every model, since none of these makers publish usable share, inverter efficiency or idle draw, so the percentage of the rating delivered is the same down each column. What differs is the absolute energy, and whether a rated 12 V port exists to skip the inverter.

The Power 1000 V2 and the DELTA 3 share a capacity, so they land on identical numbers here. Between stations of the same size, output and ports are what separate them; see the Power 1000 V2 and DELTA 3 Plus and SOLIX C1000 and AC180 comparisons.

Common mistakes

• Dividing the box rating by the device's watts. 1,024 Wh ÷ 120 W says 8 h 32 min; the engine's estimate is 6 h 17 min.
• Applying one fixed percentage to every load. Idle draw is a flat cost per hour, so the share it takes grows as the load shrinks. A rule of thumb that holds for a TV is far too generous for a router or a phone charger left on all night.
• Leaving the AC output on with nothing plugged in. At 10 W that is 240 Wh a day, and at Anker's 17 W industry average it would be 408 Wh.
• Planning with a new battery's numbers on an old unit. Our runtimes assume a new pack. If the station has seen years of use, scale the result down by the capacity you think it has lost, as in the aged rows above. We don't apply a fixed aging curve because makers don't publish one per model.
• Trusting a maker's runtime claim without its conditions. Check the load, output path and settings it was measured at before comparing it with your own setup.