How our numbers are calculated

Usable watt-hours

Manufacturers advertise a rated watt-hour figure, but you never get all of it. Two losses stand between the rated number and what actually reaches your devices: depth-of-discharge limits and inverter conversion losses.

We calculate usable Wh as:

usable_wh = rated_wh × depth-of-discharge × inverter efficiency

Depth-of-discharge (DoD) is the proportion of the battery a chemistry can safely cycle without accelerating degradation. Inverter efficiency is the proportion of stored energy that survives the DC-to-AC conversion as heat is lost. Multiplying rated Wh by both figures gives a realistic estimate of what a power station delivers through its AC output, rather than the marketing number on the box.

For the fuller explanation, including why this catches people out, see You only get half the battery.

Chemistry defaults

Where a product's battery chemistry is known but a manufacturer has not published DoD or inverter efficiency figures, we apply these class defaults:

Where a manufacturer's spec sheet documents better figures for a specific model, we override these defaults for that model and record the source.

Price data

Prices come from the Amazon Product Advertising / Creators API and are refreshed twice daily. Every displayed price shows its retrieval time, so you can see how current it is at a glance.

We do not store price history. Prices older than 24 hours are not displayed. Because the site is built statically, the timestamp you see between builds reflects the most recent data fetch at build time, not a live lookup.

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply to the purchase of this product.

Runtime estimates

An inverter draws power just being switched on, so small loads run for far less time than a naive watt-hours divided by watts calculation suggests. This is the model behind the runtime figures shown wherever PowerLasts displays them, rather than dividing usable Wh by device wattage alone.

For AC output, runtime is calculated as:

runtime = rated Wh × depth-of-discharge ÷ (device W × duty cycle ÷ inverter efficiency + inverter idle draw)

Duty cycle is applied before inverter losses, since it reflects how long the device actually draws power (a compressor cycling on 35% of the time, for example). Inverter idle draw is then added to the converted load, and the battery's energy is divided by that total.

Where no published or community-measured idle draw figure exists for a model, we use a conservative class default based on rated capacity. Where a page shows an estimated runtime figure, it is flagged "est." in the UI:

A 10W router on a 1024Wh station runs roughly 52 hours, not the 100 hours a naive watt-hour division suggests, because the inverter draws about 8W just being switched on. This is why our runtime figures are lower than most sites' and closer to what owners measure.

For DC output (12V ports, USB-C), there is no inverter to switch on, so idle draw does not apply. We instead apply a 0.95 regulator efficiency factor to account for voltage conversion losses:

runtime = rated Wh × depth-of-discharge × 0.95 ÷ (device W × duty cycle)

This is why DC-capable devices, such as a CPAP machine or a laptop that accepts USB-C power delivery, often run measurably longer per watt-hour than the same load run through the AC inverter.

UPS runtime figures

UPS runtime is the one number on this site we do not calculate. Everything above describes a model we apply to portable power stations; for a UPS we transcribe what the manufacturer publishes and show the source next to it.

The reason is the battery. A sealed lead-acid battery delivers less of its nominal energy as the discharge current rises, so the same cell that supports a light network load for hours gives only minutes near the unit's output limit. The effect is steep and specific to a given battery and inverter pairing, which means no watt-hour figure and no VA label can be turned into minutes reliably. Manufacturers measure each model instead of deriving it, and so do we.

We record published load-and-runtime pairs for the 11 active UPS models we list, taken from manufacturer product pages, runtime graph tools and brochures. Those figures were last read on September 11, 2026, and every row on the UPS runtime comparison links to the page it came from. Manufacturers revise these curves and change battery suppliers without renaming the model, so the linked source is the authority and our table is a dated snapshot of it.

Three limits are worth stating plainly:

• The figures describe a new battery. Published curves assume a fully charged, healthy cell under the manufacturer's own test conditions. Lead-acid capacity falls with age and heat, so a unit three years into service will not reach its published minutes.
• Do not interpolate between points. The curve is not a straight line. Averaging the APC SMC1500C's published 90 minutes at 100W and 32.5 minutes at 300W predicts 61 minutes at 200W, where the manufacturer in fact publishes 49.2, about 20% less than the straight-line guess.
• A point below your load is not a promise. Runtime falls faster than load rises, so read the nearest published point at or above the load you measured, never the one below it.

Where a manufacturer publishes only half-load and full-load figures, those two points are all we list. A short list is not a gap in our research: it is the whole of what the manufacturer states, and inventing intermediate points would turn a sourced figure into a guess. For how to use these curves to size a UPS, see How long will a UPS run?