Independent · not affiliated with SpaceX or StarlinkPower data checked · how
Mini & Roam plansReferral link · US only

Methods

How we get our numbers

Where each number comes from, the assumptions we add, and what we can’t test.

    Provenance levels

    Every power number on the site carries one of these tags. Tap a tag to come back here, or to the row on the watts table.

    Owner test Owner test: Measured by the site owner on our bench. No owner test of a Mini exists: the owner doesn’t have one. Owner tests cover batteries, inverters and the home Standard dish on a UPS.

    Official spec Official spec: From a Starlink specification sheet or help article.

    Community report Community report: A dated measurement posted by an owner, cited. We only use reports that state the connection, what the dish was doing and when; each links to the post.

    Vendor claim Vendor claim: A seller’s marketing figure, not tested by us. Shown only beside our own or official numbers, never as the basis for a pick.

    Standard Standard: From a regulator, standards body or reference table.

    Planning assumption Planning assumption: Our stated assumption, used until bench results replace it.

    Starlink figures checked Oct 5, 2026; readings checked Oct 5, 2026. Starlink’s watt figures include the dish, router, power supply and cables, measured at the AC input.

    Planning assumptions

    These are deliberately on the careful side. Bench results will replace the battery figures; we’ll log every change on corrections.

    Where a battery's energy goes on the way to the dishUSB-C PD: of 100 label watt-hours, about 85 are usable and about 77 reach the dish. 12 V DC: of 100 label watt-hours, about 85 are usable and about 77 reach the dish. AC: of 100 label watt-hours, about 85 are usable and about 74 reach the dish.Of 100 Wh on the label (planning assumptions)USB-C PDPD output + cable77 Wh reach the dishcutoff −15 · conversion −912 V DC12 V port + cable77 Wh reach the dishcutoff −15 · conversion −9ACInverter (12 V → 120 V AC)74 Wh reach the dishcutoff −15 · conversion −11
    Planning assumptions, not measurements: usable share of label capacity 85% for a power station; conversion 90% on USB-C PD and 12 V DC, 87% through an inverter. The AC path loses more than this picture shows: the dish's own power supply adds a loss, which Starlink's AC-input watt figures already include, so on AC the dish also draws more (see the community readings on /watts/). Our bench tests (/bench/) will replace these.

    Sun hours

    NASA POWER POWER Climatology API v2.10.0, parameter ALLSKY_SFC_SW_DWN (all-sky surface shortwave downward irradiance on a horizontal surface), 20-year Meteorological and Solar Monthly & Annual Climatologies (January 2001 - December 2020). One reference city per state; local terrain, shade, snow and weather vary. Flat-panel values are the conservative case for portable panels; tilting toward the sun raises winter output. NREL's API host (developer.nrel.gov) did not resolve on 2026-10-05, so NASA POWER is used; a Denver cross-check against NREL's Solar Resource API (via developer.nlr.gov) gave Dec 2.26 vs 2.29 and Jul 7.11 vs 7.00. Retrieved Oct 5, 2026. Data source.

    The battery bench

    The protocol, equipment and limits are on the bench page. In short: a constant 40 W load through a 20 V USB-C PD trigger (or a DC or AC load), logged to cutoff, twice per battery.

    What we can’t test