A Starlink terminal draws between 16W and 150W depending on which dish you own: a Starlink Mini sits at roughly 16-25W in steady use on current firmware, a Gen 2 or Gen 3 standard dish runs 50-75W typical with peaks near 100W, and a Flat High Performance terminal pulls 100-150W. Everything else in an off-grid or backup build — battery amp-hours, inverter rating, panel count, charger amps, and a surprising amount of the real cost of the service — follows from that one number and the hours you need to cover.
This guide converts those wattages into the figures you actually buy hardware against, and it is built around the two situations where satellite internet lives or dies on its power supply: a fully off-grid site with no mains at all, and a grid-connected home or business in a load-shedding market where the outage schedule, not the daily average, sets the design. It also does something most off-grid sizing guides skip, which is to tie the power build back to the Starlink plan you pay for every month, because the cheapest way to halve your battery budget is usually a hardware and plan decision rather than an electrical one.
By the Starlink News technical desk. Method, in one line: the wattage figures below are compiled from Starlink’s published support and specification documentation, manufacturer datasheets, and power-meter readings that owners publish in off-grid and satellite internet communities. What we did not do: we did not run a controlled laboratory test, and we do not present one. Where a figure comes from user reports rather than a specification sheet, it is labelled inline as community-sourced, because forum readings vary with firmware version, ambient temperature, obstruction and the accuracy of the meter used. The sizing arithmetic, by contrast, is standard electrical practice and you can reproduce every calculation here yourself — the section below on measuring your own terminal shows how. Electrical work described here is subject to your national wiring rules; see Sourcing, method and revision history.
Starlink Power Draw by Hardware Generation
Power draw is not a single figure per dish. Every terminal has an idle floor, a typical streaming load, and a short peak that appears at boot, during firmware installation, and when the heating element engages. Size the battery on the typical number and size the wiring, fuses and converter on the peak number.
| Hardware | Idle | Typical | Peak | Daily (24h, typical) |
|---|---|---|---|---|
| Gen 1 Round (discontinued, still in service) | ~40W | 80-100W | ~120W | ~2.2 kWh |
| Gen 2 Rectangular | ~20W | 50-75W | 100W (snow melt) | 1.2-1.8 kWh |
| Gen 3 Standard | ~20W | 45-75W | ~100W | 1.1-1.8 kWh |
| Starlink Mini | ~15W | 16-25W (community reports; Starlink’s published figure is 20-40W average) | ~60W | 0.4-0.6 kWh |
| Flat High Performance (mobility/enterprise) | ~50W | 100-150W | ~190W | 2.4-3.6 kWh |
Idle and peak columns are community power-meter reports. Typical ranges for the standard dishes are consistent with Starlink’s own published power figures; the Mini row is the one place where widely reported measurements now sit below the published specification, which is discussed immediately below. Treat every row as a range, not a constant.
The Mini’s draw: what is measured, and what is inferred
Read this one with a hedge attached. Starlink’s published specification for the Mini remains an average in the 20-40W band, and that is the only figure with an official source behind it. Separately, through 2025 and into 2026, a large number of owners posting inline-meter readings in off-grid and satellite internet communities have reported steady-state consumption closer to 17W, with an idle floor around 15W. The community consensus attributes the drop to a firmware revision; SpaceX has not published a changelog entry confirming a power optimisation, and we have not verified one. So: plan on 20-25W if you want a specification-backed margin, and treat 17W as a plausible best case you should confirm with your own meter rather than design around.
The same caution applies to the runtime anecdotes that circulate with it. One frequently repeated community report has a 512 Wh portable station moving from roughly 15 hours on launch firmware to roughly 26 hours on current firmware with the same Mini. That is a single-source claim, not a measurement we produced, and the arithmetic bound is worth knowing: 512 Wh at about 90% delivered efficiency is 460 Wh, which is 27 hours at 17W and 23 hours at 20W. The anecdote is consistent with the arithmetic, which is the most you can say for it. If your build notes predate 2026, re-measure before you buy more battery — and if you measure higher than the forums, believe your meter.
Peak versus idle: what to plan for
- Startup and firmware updates add roughly 20% above typical draw for two to three minutes. Short enough to ignore for energy budgeting, long enough to matter for converter and fuse ratings.
- Snow melt is the largest sustained excursion on standard dishes. Heating elements can hold the terminal near 100W for as long as precipitation continues — an hour of snow melt costs more energy than four hours of ordinary browsing.
- Seasonal variation reported across northern-hemisphere installations runs 15-20% higher in winter than in summer for the same usage pattern, almost entirely from thermal management.
- Idle is not off. A standard dish parked with no traffic still holds the satellite link at around 20W. Twelve idle hours is roughly 240 Wh — a quarter of a 100Ah 12V lithium battery, spent on nothing.
The rest of the numbers in this guide use 75W as the standard-dish planning figure and 20W for the Mini. Both sit at the upper end of typical rather than at the average, which is the correct bias when the consequence of being wrong is a dark router at 2am.
Measure Your Own Terminal in Ten Minutes
Every table in this guide is a substitute for a measurement you can take yourself in an afternoon, and your own number beats any published range because it includes your firmware, your obstruction percentage, your ambient temperature and your usage pattern. This is the part of the process we would rather you trusted than our synthesis of other people’s readings.
Instrumentation, cheapest first
- Plug-in AC energy meter (Kill A Watt P4400 class, or any meter with a cumulative kWh register). Put it between the wall and the stock power brick, reset it, and read kilowatt-hours after exactly 24 hours. One reading, no maths, and it captures snow melt, updates and idle periods you would otherwise miss. This is the single most useful measurement in the whole exercise.
- Inline DC watt meter or shunt (Victron SmartShunt, Renogy or a generic 150A DC meter) on the battery side of a 12V system. Critically, this measures what leaves the battery, which includes converter or inverter losses — that is the number you size amp-hours against, not the terminal’s own consumption.
- True-RMS clamp meter with DC capability for peaks. Peak events are short; an averaging meter will smooth them away and you will under-specify the converter.
- Smart plugs are a last resort. Most sample once per second or slower and under-report short peaks, and many report power factor incorrectly on switch-mode supplies. They are fine for tracking daily kWh trends, unreliable for sizing fuses.
A protocol that produces a comparable number
- Record the firmware version from the Starlink app (Settings, then debug data) and the obstruction percentage. Both change consumption; a report without them is not comparable to anyone else’s.
- Log a full 24 hours, not a spot reading, and note minimum and maximum ambient temperature over the period.
- Log a separate overnight block with no traffic to establish your idle floor. Idle is usually a third of a backup battery’s job and almost nobody measures it.
- Note whether snow melt or any heating cycle engaged. In winter this can double a daily figure and it is the most common reason two owners of identical hardware report wildly different numbers.
- Report the result in watt-hours per day, with the average watts derived from it. Watts alone, quoted from a live display, is the least useful number anyone can share.
If you run that protocol, send the log through the site contact page with hardware generation, firmware, 24-hour Wh, idle Wh, ambient range and the meter you used. Submissions that include all six fields are added to the reference ranges above with attribution at the next revision, and dated in the revision history at the foot of this page. That is how the numbers here improve: reader meters, not our assertions.
The One Formula, the Cheat Sheet and the Printable Table
There is exactly one equation in off-grid sizing, and everything else in this guide is that equation with a correction factor bolted on. It appears here, once, and later sections refer back to it rather than re-deriving it.
(Watts × Hours) ÷ (Battery Voltage × Depth of Discharge) = Required Amp-Hours
Worked once, for a standard dish over an eight-hour outage: 75W × 8h = 600 Wh. Divided by (12V × 0.8 depth of discharge) = 62.5Ah minimum from a lithium bank. Round up to the nearest commercial size and you are buying a 100Ah LiFePO4 battery. Three corrections turn that clean figure into a real one — chemistry, inverter losses and temperature — and they are handled in the next section.
Three rules of thumb worth memorising
These are the formula pre-solved for a 12V lithium bank at 80% depth of discharge, then rounded down on purpose so that memory never over-promises:
- Starlink Mini (20W): 1Ah of 12V lithium ≈ 30 minutes of runtime.
- Standard dish (75W): 1Ah of 12V lithium ≈ 8 minutes of runtime.
- Dish + router + laptop (135W): 1Ah of 12V lithium ≈ 4.5 minutes of runtime.
Applied to a 100Ah LiFePO4 battery, the rules give roughly 50 hours, 13.5 hours and 7.5 hours. The exact arithmetic gives 51.2, 13.6 and 7.6 hours. Where this guide quotes a rule-of-thumb figure and a table figure that differ by a few percent, the rule of thumb is the deliberately conservative one. Knock a further 10-15% off all three if you power the terminal through an inverter rather than a DC-DC converter, and a further 5-8% if you want to account strictly for converter efficiency.
Printable sizing table: battery amp-hours by runtime and hardware
Figures are 12V lithium (LiFePO4) at 80% depth of discharge, powered by DC-DC conversion. Add 15% for an inverter chain, and double the whole table if you are using lead-acid.
| Runtime needed | Mini (20W) | Standard dish (75W) | Dish + router + laptop (135W) | Flat High Performance (150W) |
|---|---|---|---|---|
| 2 hours | 4Ah | 16Ah | 28Ah | 31Ah |
| 4 hours | 8Ah | 31Ah | 56Ah | 63Ah |
| 6 hours | 13Ah | 47Ah | 84Ah | 94Ah |
| 8 hours | 17Ah | 63Ah | 113Ah | 125Ah |
| 12 hours | 25Ah | 94Ah | 169Ah | 188Ah |
| 24 hours | 50Ah | 188Ah | 338Ah | 375Ah |
Round up to the next commercial size — batteries are sold at 50, 100, 200, 300 and 400Ah, and you should never plan to land exactly on the calculated figure.
Copy this into a spreadsheet
Four cells do the entire job, including the two calculations later sections depend on. Put your load in watts in A1, hours needed in B1, battery voltage in C1 (12, 24 or 48), depth of discharge in D1 (0.8 lithium, 0.5 lead-acid), winter peak sun hours in E1, and hours of grid between outages in F1:
- DC-powered (converter):
=(A1*B1)/(C1*D1)→ amp-hours required - AC-powered (inverter):
=(A1*B1)/(C1*D1*0.87)→ amp-hours required - Solar array watts:
=((A1*24)/E1)*1.3 - Charger amps to recover between outages:
=((A1*B1)/C1)/F1*1.3
Print the table above, tape it inside the battery box lid, and you will not need to re-derive anything when you expand the system.
The Three Corrections: Chemistry, Inverters and Cold
The formula gives you a clean amp-hour figure. These three factors are why the clean figure is optimistic, and they are the reason most first builds come up short.
Chemistry sets usable capacity
Lithium iron phosphate delivers about 80% of nameplate capacity without shortening life meaningfully. Flooded or AGM lead-acid delivers about 50% before cycle life falls off a cliff. A 100Ah lead-acid battery is therefore a 50Ah battery for planning purposes — it will carry a standard dish for roughly six and a half hours, not thirteen. Lead-acid also sags under load in a way lithium does not, which is the single most common cause of the voltage-triggered reboots covered further down. Both chemistries are treated in detail in Victron Energy’s technical information library, the most accessible free reference for depth-of-discharge and cycle-life curves.
Inverter losses
If you keep the stock AC power supply, the energy goes battery → inverter → AC → the terminal’s internal supply. Pure sine wave inverters run 85-90% efficient at reasonable load, worse at very light load, and they draw 5-15W just being switched on. That idle penalty is brutal on a Mini: a 15W inverter overhead on a 17W load nearly doubles your consumption, which is why the Mini’s native DC input is worth more than its headline wattage suggests. On a standard dish, add 10-15% to the amp-hour figure: the 600 Wh worked example becomes roughly 690 Wh, and 62.5Ah becomes about 72Ah.
Temperature derating
Below freezing, battery capacity falls 20-30% for both chemistries. LiFePO4 cells additionally must not be charged below 0°C without an internal heater or a charger that blocks sub-zero charging — a rule that catches out first-winter builders in Canada, Scandinavia and highland regions every year. If your battery lives in an unheated shed, buy a heated model or size for the cold-weather capacity, not the datasheet capacity.
Runtime reference table
| Battery (12V lithium) | Usable energy | Mini only (20W) | Standard dish (75W) | Dish + router + laptop (135W) |
|---|---|---|---|---|
| 50Ah | ~512 Wh | ~25h | ~6.8h | ~3.8h |
| 100Ah | ~1,024 Wh | ~51h | ~13.6h | ~7.6h |
| 200Ah | ~2,048 Wh | ~102h | ~27h | ~15h |
| 400Ah | ~4,096 Wh | ~205h | ~55h | ~30h |
Figures assume 80% depth of discharge, energy measured at the battery terminals with a direct DC supply. Subtract 10-15% throughout if you run through an inverter; these are the exact numbers the rules of thumb round down from.
DC Conversion: Eliminating Inverter Losses
Starlink’s standard terminals are powered over Ethernet at 48V. The stock supply accepts 100-240V AC at 50-60Hz and outputs that 48V down the proprietary cable. On a battery system, running 12V DC up to 230V AC and straight back down to 48V DC is two conversions you did not need.
A DC-DC step-up (boost) converter takes 12V or 24V nominal battery voltage directly to 48V and feeds the PoE injector or, on some builds, the terminal cable. Measured against an inverter chain, the saving is 10-20% of total energy drawn — on a 100Ah battery carrying a standard dish, roughly an extra one to two and a half hours of runtime, for a component costing less than a spare battery terminal.
Specifying the converter
- Rate for peak, not typical. A converter sized at 100W will brown out during snow melt. 150W continuous is the sensible floor for a standard dish; 250W if you are in a snow region or running Flat High Performance.
- Check the input range. A 12V LiFePO4 bank swings from about 14.6V at full charge to 10.5V at cut-off. The converter must hold 48V across that whole window, not just at 12.8V.
- Fuse both sides and keep the low-voltage run short. At 12V input, a 150W load is over 12A — the current, and therefore the voltage drop, lives on the battery side of the converter.
- The Mini is the exception. It accepts 12-48V DC on its own barrel input, so no step-up is needed at all. This is the main reason it has become the default choice for vans, boats and small cabins.
Two cautions. First, a non-standard power supply gives SpaceX a reasonable argument in any hardware warranty claim, so keep the stock brick and be prepared to reinstate it before you contact support. Second — and this is a clearly-labelled community observation rather than bench testing — owners posting failure reports in off-grid and satellite internet forums since 2022 have consistently blamed cheap unbranded boost converters run close to their rated ceiling. The pattern is anecdotal, but it is consistent enough to act on: buy a converter with a documented thermal shutdown from a manufacturer that publishes a datasheet, and derate it by a third.
Solar Panel Sizing for Continuous Operation
Battery capacity buys you hours. Solar buys you days. Using the array formula from the cheat sheet: daily energy divided by local winter peak sun hours, multiplied by a weather margin.
Standard dish, worked: 75W × 24h = 1.8 kWh per day. At 4 peak sun hours, 1,800 Wh ÷ 4 = 450W of array as the bare minimum with no margin at all. The widely used rule of thumb is to size for around three times daily consumption in usable generation, so the bank still recovers through cloud and short winter days.
Array size by latitude band
| Location profile | Winter peak sun hours | Mini (0.5 kWh/day) | Standard dish (1.8 kWh/day) |
|---|---|---|---|
| Equatorial / sunbelt (0-25°) | 5-6 | 150-200W | 400-600W |
| Subtropical (25-40°) | 3.5-5 | 200-300W | 600-800W |
| Temperate (40-50°) | 1.5-3 | 300-450W | 900-1,200W |
| High latitude (50°+) | 0.5-1.5 | 600W+ or generator | 1,500W+ or generator |
The high-latitude column is where most undersizing happens. A recurring pattern in European winter reports: a 200W array that ran a standard dish comfortably from April to September goes three consecutive days without recharging the bank in December, and the owner discovers the problem at the point the battery is already flat. If you are north of 50 degrees and need year-round uptime, budget for a generator or accept scheduled downtime rather than buying your way out with panels alone.
Controller and mounting details that change the answer
- Use MPPT, not PWM. A maximum power point tracking controller harvests 15-30% more from the same panels, and the advantage widens in cold, bright and low-light conditions — exactly the conditions in which you are short of energy.
- Tilt matters more than watts at high latitude. A winter tilt close to latitude plus 15 degrees can outperform a 30% larger flat array. Adjustable mounts are cheap compared with extra panels.
- Azimuth tolerance is wider than people think — within 30 degrees of true south (or north, below the equator) costs only a few percent — but shading is not tolerant at all. One shaded cell string can halve a panel’s output.
- Do not mount panels where they shadow the dish. Satellite internet needs a clear field of view; obstruction losses will cost more throughput than the panels return in uptime.
Victron’s sizing documentation and the IEC’s standards for renewable energy systems are the two reference sets worth reading before you commit to an array layout; both treat weather margin as a design input rather than an afterthought.
Sizing to a Load-Shedding Schedule, Not to a Day
This is where most generic off-grid guides mislead the readers who need them most. If you live with rotational power cuts — South Africa, Nigeria, Pakistan, Lebanon, Ecuador, Cuba — you are not sizing for 24 hours of autonomy. You are sizing for the longest single block, and then for the charge rate that refills the bank before the next one. Get the second half wrong and a perfectly adequate battery still dies on day two.
Take the standard household load used later in this guide: dish, router and laptop at 135W. Against Eskom’s published stage definitions and the equivalent rotational schedules used elsewhere, the design numbers look like this:
| Stage / severity | Outage hours per day | Longest single block | Energy per block (135W) | Energy per day | Sensible 12V lithium bank | Charger to recover between blocks |
|---|---|---|---|---|---|---|
| Stage 1-2 (mild) | 2-4h | 2.5h | ~340 Wh | 270-540 Wh | 50Ah | 10A |
| Stage 3-4 (moderate) | 6-8h | 2.5h | ~340 Wh | 0.8-1.1 kWh | 100Ah | 15A |
| Stage 5-6 (severe) | 10-12h | 4h | ~540 Wh | 1.4-1.6 kWh | 200Ah | 20A |
| Stage 7-8 (extreme) | 14-16h | 4h | ~540 Wh | 1.9-2.2 kWh | 300-400Ah | 30-40A or solar |
Two things fall out of that table that a daily-average calculation never shows you.
The battery is usually the easy part. At Stage 4, one block costs about 340 Wh — a 50Ah lithium battery covers it twice over. The reason people still run out is that at Stage 4 the grid window between blocks can be as short as three hours, and a 5A trickle charger only returns about 180 Wh in that time. You then start each successive block lower than the last. The charger amp figure in the final column, not the battery size, is the number that decides whether you make it through a bad week.
Charging is not 100% efficient either. Lithium takes bulk charge quickly but tapers in absorption, which is why the charger formula in the cheat sheet carries a 30% margin. If your grid window is under two hours, stop trying to solve it with mains charging and add 300-400W of solar; panels charge during the outage, mains charging by definition cannot.
The duty-cycling trick, and its limits
Because a standard dish idles at around 20W, some owners put the terminal on a smart plug or timer and cut it outside working hours, saving roughly 240 Wh over a twelve-hour overnight. It works, with three caveats: the terminal takes one to two minutes to reacquire and will occasionally take longer after a firmware update; scheduled downtime means missed alerts, VoIP calls and remote-access sessions; and repeated cold starts each cost a short 20% current spike. Duty-cycling is worth it on a Gen 2 dish in a severe-outage market. It is barely worth it on a Mini, where the idle floor is around 15W and the whole night costs less than 200 Wh.
What the Power Build Costs Next to the Starlink Price You Already Pay
Satellite internet is sold as a monthly subscription plus a hardware kit, and almost every buyer budgets for those two lines and nothing else. In an unreliable-grid or off-grid market the power system is a third line item of comparable size, and it is determined by which plan and terminal you chose.
| Plan tier | Typical terminal | Typical draw | Daily energy | Bank for 8h backup | Array for 24/7 (sunbelt) |
|---|---|---|---|---|---|
| Residential Lite / Residential | Gen 3 standard dish | 45-75W | 1.1-1.8 kWh | 100Ah | 400-600W |
| Roam (mobile, regional) | Starlink Mini | 16-25W | 0.4-0.6 kWh | 30-50Ah | 150-200W |
| Roam Unlimited / in-motion | Standard dish or Mini | 25-75W | 0.6-1.8 kWh | 50-100Ah | 200-600W |
| Priority / business | Flat High Performance | 100-150W | 2.4-3.6 kWh | 200Ah+ | 900-1,200W |
The decision that actually moves money is the one at the top of the table. Choosing a Mini over a standard dish removes roughly two-thirds of the battery and array from the build — in practice USD 400-900 of hardware for a typical home backup, more for a 24/7 site — on top of usually sitting on a cheaper monthly tier. Against that you trade peak throughput and field of view, a trade examined in detail in Starlink Mini vs the standard dish. Current subscription figures for each tier are tracked in our breakdown of Starlink price by plan tier, and Starlink’s own site remains the authority for what your address is charged, since pricing, promotions and hardware discounts vary by market and change several times a year.
The cost-per-hour framing, with its assumptions on the table
A quality 100Ah lithium backup build — battery, inverter/charger, fusing, transfer switch, cable — lands around USD 800-1,500 in a market without punitive import duties. The commonly quoted conclusion is that such a build costs about the same per hour as the subscription it protects. That is true under a specific set of assumptions, and it is worth showing them rather than asserting the answer.
On the five-year life assumption. A 100Ah LiFePO4 cell is typically rated for 3,000-5,000 cycles at 80% depth of discharge. A load-shedding build running 540 Wh out of a 1,280 Wh nameplate battery is roughly 0.4 of a cycle per day, or about 150 equivalent full cycles a year — on cycle count alone the cells would last decades. In practice, calendar ageing (commonly 10-15 years) and the surrounding electronics govern: inverters, chargers and MPPT controllers in continuous duty typically need replacement or repair somewhere between five and ten years. Five years is therefore a conservative figure for the battery and a realistic one for the system as a whole. Here is the same build across three lifetimes and three outage regimes:
| Outage hours protected per day | 3-year system life | 5-year system life | 8-year system life |
|---|---|---|---|
| 2 hours | USD 0.46/h | USD 0.27/h | USD 0.17/h |
| 4 hours | USD 0.23/h | USD 0.14/h | USD 0.09/h |
| 8 hours | USD 0.11/h | USD 0.07/h | USD 0.04/h |
On the subscription side. A USD 120/month residential plan, billed continuously, is about USD 0.16 per hour — but that is a high-income-market figure, and this guide is read mostly in markets where it does not apply. Monthly pricing varies from roughly USD 30 to USD 140 depending on country, tier and promotion:
| Monthly plan price | Cost per hour of service | Verdict against a USD 1,000 build protecting 4h/day for 5 years (USD 0.14/h) |
|---|---|---|
| USD 30 | USD 0.04 | Power build costs ~3.4× the subscription per hour — go Mini and portable station |
| USD 50 | USD 0.07 | Power build costs ~2× — keep the build lean |
| USD 80 | USD 0.11 | Roughly comparable |
| USD 120 | USD 0.16 | Backup is cheaper per hour than the service it protects |
The point is not the headline number, it is the structure: in expensive markets the battery is easy to justify because you are already paying for hours you cannot use, while in cheap markets the power build can cost several times the subscription per protected hour, and the right response is a smaller terminal rather than a bigger bank. Regional pricing moves the hardware side too — the same specification is routinely 30-50% more expensive where import tariffs on lithium batteries are high, and in several countries the cheapest viable route is a sealed portable power station rather than a component build, because the station ships as a certified appliance rather than as loose cells.
Availability: Power Is the Real Uptime Constraint for Satellite Internet in Remote Areas
There are three different things people mean by availability, and only one of them is about satellites.
- Regulatory availability. Service must be licensed in your country. Several of the heaviest load-shedding markets are not officially served — South Africa being the most-discussed example — and users there rely on regional roaming kits whose legal standing and terms change. Confirm at the official availability map before you spend a cent on batteries; our country-by-country tracker on Starlink availability by country follows the regulatory picture as it moves.
- Cell capacity availability. Even inside a licensed country, individual cells sell out or go waitlisted, and congested cells deliver lower speeds at peak. This is the variable that determines whether a Mini’s smaller phased array is a sensible economy or a bottleneck.
- Effective availability — your uptime. This is the one you control. A terminal in a licensed, uncongested cell with a clear sky view is still down four to twelve hours a day if the grid is down and there is no battery behind it.
That third definition is why this guide exists. In the places where satellite internet is most transformative — rural clinics, farms, remote schools, small businesses outside the fibre footprint — the practical difference between 60% and 99% availability is a battery bank and a charge controller, not anything SpaceX does in orbit. It is also why the power chain deserves the same redundancy thinking as the link itself, covered in the business section below.
Sample Home Setup: Load-Shedding Backup
This is the most common build worldwide — South Africa, Nigeria, Pakistan, Lebanon, Puerto Rico, rural Australia — and it targets a four to eight hour outage rather than full independence.
Load budget
- Starlink standard dish: 75W
- Household router / mesh node: 10W
- Laptop: 50W
- Total: 135W → 1.08 kWh over 8 hours
Bill of materials
- 100Ah 12.8V LiFePO4 battery with integrated BMS rated 100A continuous and a low-temperature charge cut-off — 1,280 Wh nameplate, 1,024 Wh usable at 80% depth of discharge
- 600W pure sine wave inverter, or better, a 150W DC-DC converter for the dish plus a small inverter for the laptop only
- 30A MPPT charge controller if solar is present, or a 15-20A mains charger if you recharge from the grid between outages (match the amp figure to your stage in the load-shedding table)
- Automatic transfer switch or a UPS-style inverter/charger so the satellite link never drops during changeover
- Class-T or ANL fuse within a few inches of the positive terminal, 12AWG minimum on DC runs under 10ft, heavier beyond that
Using the runtime table: 1,024 Wh usable covers about 7.6 hours of the full 135W load on DC conversion, about 6.6 hours through an inverter chain, or roughly 12 hours with the laptop unplugged. If your longest block is a true eight hours and you want the laptop on throughout, step up to 150Ah. Component cost lands in the region of USD 800-1,500 for quality parts, subject to the regional pricing caveats above.
Adding it to an existing home solar system
Two routes. AC coupling leaves your existing hybrid inverter in charge and simply puts the terminal on a backed-up circuit; it is the simpler option and the one an installer will sign off. A dedicated circuit — small separate battery and converter feeding only the dish and router — costs more in duplicated hardware but keeps connectivity alive even when the main system is being serviced or has tripped. For households that also travel, a single battery station that handles both duties is worth considering alongside portable Starlink setups for mobile use, since a Mini’s native DC input suits both the car and the house.
Sample Small Business Setup: Multi-Day Resilience
Commercial users have a different failure cost. A clinic, a site office or an emergency operations post is not sizing for one evening; it is sizing for a week with no assumption of sunshine, and it usually cannot tolerate a single point of failure.
Reference configuration
- Storage: 400Ah lithium at 12V, or 200Ah at 24V, which is the better choice — half the current, half the cable cost — giving roughly 4 kWh usable
- Conversion: 2,000W inverter/charger with pass-through and automatic transfer
- Generation: 800W solar minimum, 1,200W in temperate winters
- Connectivity redundancy: two terminals on separate DC circuits and separate fuses, joined by a dual-WAN router with automatic failover
- Generator: auto-start on low state of charge, sized to return the bank to 80% in about two hours of runtime
- Monitoring: Victron, Renogy or equivalent with remote alerting on state of charge, converter temperature and terminal uptime
Two standard dishes plus networking is roughly 170W continuous, or 4.1 kWh per day. The 400Ah bank alone is therefore about one day of autonomy; the solar array is what turns it into a week. This is why a generator is in the specification rather than more panels — beyond about 1,200W, each additional 100W of array buys progressively less winter energy per unit of cost, while a generator delivers on demand regardless of weather.
Redundancy deserves a note on placement. Two terminals fed from the same converter and the same fuse are not redundant; they share a failure mode. Split them across separate converters, separate fuses and, where possible, separate battery banks. Published accounts from remote construction sites, mobile clinics and disaster response deployments repeatedly describe power-chain faults — a failed converter, a blown fuse, a flat bank — taking satellite internet down far more often than the link itself fails.
Regulatory and Safety Considerations
Electrical rules are national, but a short list of requirements applies almost everywhere a Starlink power build is installed.
Transport and import
UN38.3 certification is mandatory for lithium batteries moved by air under the IATA Dangerous Goods Regulations, and carriers normally require cells to be shipped at 30% state of charge or below. Batteries above 100 Wh cannot travel in checked luggage, and units above 160 Wh are not permitted in passenger baggage at all — relevant if you were planning to fly a portable power station to a remote site. Several jurisdictions additionally require import permits or local certification marks for battery energy storage, so check before ordering across a border.
Installation standards
Fixed battery installations fall under national wiring rules: NFPA 70 (NEC) Articles 480 and 706 in the United States, CSA C22.1 in Canada, BS 7671 with the IET Code of Practice in the United Kingdom, AS/NZS 5139 in Australia and New Zealand, SANS 10142-1 in South Africa, and IEC 60364 as the parent framework much of the rest of the world harmonises with. In most of these, a portable plug-in power station is unregulated while a permanently wired battery connected to the building’s electrical installation requires a licensed electrician and, frequently, a certificate of compliance. Nothing in this guide substitutes for that sign-off.
Fire, ventilation and earthing
- Ventilation: lead-acid vents hydrogen and needs passive or forced ventilation to outside air. LiFePO4 does not vent in normal operation but still needs airflow for thermal management and clearance from combustible materials.
- Enclosure: keep batteries out of living and sleeping spaces where local rules require it, and fit a correctly rated fuse or breaker within inches of the positive terminal.
- Earthing and surge protection: outdoor solar arrays and a roof-mounted dish are both lightning exposure paths. Bond the array frame, fit DC surge protection on the array and Ethernet surge protection on the terminal cable — on exposed rural sites this is the cheapest insurance in the build.
- Insurance: tell your insurer. Undeclared battery storage has been used to decline claims, and most insurers simply want confirmation the work was done by a certified installer.
Troubleshooting Power-Related Performance Issues
Power problems rarely announce themselves as power problems. They look like a slow internet connection.
The classic symptom set
- Reboot loops as the battery discharges. The terminal drops out when PoE voltage sags too far — community logs cluster the failure in the mid-40s of volts, though no official threshold is published. Most often the cause is sag under load rather than a flat battery: a lead-acid bank near end of life, an undersized boost converter, or thin cable.
- Intermittent disconnections at otherwise normal state of charge. Check converter temperature. Boost converters running close to their rating throttle or shut down when hot, then restart, producing a drop every few minutes.
- Speeds that fall away after 20-30 minutes. Thermal throttling in the terminal, or a power supply that cannot hold voltage once warm. Compare a cold-start speed test with one taken after half an hour.
- Everything fine on grid, poor on battery. Almost always the inverter: a modified sine wave unit, or a pure sine unit loaded so lightly it is running in its inefficient band.
A diagnostic sequence that works
- Open the Starlink app and read uptime. If uptime keeps resetting, you have a power fault, not a network fault.
- Put an inline DC meter on the battery side and a plug-in meter on the AC side. Compare measured draw with the tables above; a standard dish pulling 110W continuously is heating, not browsing.
- Measure voltage at the PoE injector while the dish is under load, not at rest. A 2-3V difference between battery and injector means the cable or converter is the problem.
- Upsize the cable. 12AWG is the minimum for a 10ft DC run at these currents; 10AWG or 8AWG for longer runs. Voltage drop scales with length and current, and it is the single most common defect in home-built installations.
- Re-test after a firmware update. Updates draw temporarily higher current and install without warning, so leave headroom in both the converter rating and the battery reserve.
On whether undervoltage damages the dish: there is no evidence in published reports that sustained brownouts destroy terminals, and the hardware protects itself by restarting. The real cost is to the battery — repeatedly running lead-acid flat, or holding lithium at the low-voltage cut-off, shortens the bank’s life far faster than it affects the Starlink hardware. Set the inverter or converter low-voltage cut-off above the battery’s own protection threshold so the terminal shuts down before the cells do.
Putting the Numbers Together
Four figures decide the whole build.
- Daily energy: 0.4-0.6 kWh for a Mini, 1.2-1.8 kWh for a standard dish, up to 3.6 kWh for Flat High Performance — ideally measured on your own meter rather than taken from a table.
- Usable battery capacity: 80% of nameplate for lithium, 50% for lead-acid, less 10-15% again if you keep an inverter in the chain, less 20-30% again below freezing.
- Array size: daily energy divided by local winter peak sun hours, multiplied by a weather margin that grows with latitude.
- Charge rate: block energy divided by the hours of grid or sun you get between outages, plus 30%. This is the number generic guides omit and the one that decides whether your system survives a bad week.
Get those four right and the rest — cable gauge, fuse rating, controller type — follows from standard electrical practice in your jurisdiction.
The trend line is in the reader’s favour. Terminal efficiency has improved faster over the past two years than battery or panel prices have fallen, which has quietly made the smallest, cheapest terminal the most practical one for anyone running on sun, and has put reliable satellite internet inside the budget of sites that could not have carried a standard dish at all. Before you price a single battery, settle the hardware question — and if you want the plan and coverage side of the decision, start with our full library of Starlink guides.
Sourcing, method and revision history
Published 19 September 2026 by the Starlink News technical desk. Method: power figures are compiled from Starlink’s official support documentation on power requirements and terminal specifications, manufacturer datasheets (Victron, Renogy, Battle Born), and power-meter readings published by owners in off-grid and satellite internet communities. Community-reported figures are labelled as such at the point of use and should be treated as indicative ranges, not measurements produced under controlled conditions by this desk. Two claims in particular are single-source or forum-sourced and are hedged in the body: the reduction in Starlink Mini steady-state draw attributed to a 2026 firmware revision, and the voltage at which terminals reboot under sag. Sizing methodology follows Victron Energy’s technical papers on off-grid system design and IEC standards for renewable energy systems; every calculation in this guide is reproducible from the single formula and the four spreadsheet cells given above. Transport rules follow the IATA Dangerous Goods Regulations; installation references are to the current published editions of NFPA 70, CSA C22.1, BS 7671, AS/NZS 5139, SANS 10142-1 and IEC 60364. Pricing is indicative, varies by market and promotion, and should be checked against Starlink’s own site for your address.
What has changed in this guide. This page is maintained rather than archived, and each revision is tied to a change in a source above. The DC conversion section was added in March 2025 after low-cost 48V boost converters became widely stocked. Gen 2 peak draw was revised from 80W to 100W in January 2025 to reflect sustained snow-melt readings reported by owners in cold-climate communities. Mini specifications were updated in October 2025 and again in September 2026, the latter to record widely reported lower steady-state readings while retaining Starlink’s published 20-40W figure as the specification of record. National electrical code references were expanded in September 2026 to cover IEC, AS/NZS, SANS, CSA and BS frameworks alongside the NEC. Also added in September 2026: load-shedding schedule sizing, the plan-price comparison with a lifetime and regional sensitivity analysis replacing the earlier single point estimate, the quick-reference cheat sheet, and the reader measurement protocol. Duplicated worked examples were consolidated into a single canonical calculation in the same revision. Corrections and meter logs are welcome via the site contact page and will be credited by name at the next revision.
Frequently Asked Questions
How many watts does Starlink actually use per hour?
Watts are an instantaneous measure, so the number you plan hardware around is watt-hours per day. A Gen 2 or Gen 3 standard dish averaging 50-75W consumes roughly 1.2-1.8 kWh per day running continuously; a Starlink Mini averaging 17-25W consumes about 400-600 Wh per day. Idle draw on a standard dish is around 20W because the terminal holds the satellite link even with no traffic, so switching it off overnight saves less than most people expect. If you want your own figure rather than a range, put a plug-in energy meter on the stock power brick and read kilowatt-hours after a full 24 hours — that single reading replaces every estimate in this guide for your specific unit, firmware and climate.
Can I run Starlink directly off a 12V battery without an inverter?
Not directly. The standard dish needs 48V delivered over its Ethernet cable, and the Mini needs 12-48V DC through its barrel jack. For a standard dish you fit a 12V-to-48V DC-DC step-up converter rated for at least 150W continuous and feed the PoE injector or the terminal cable from it. This avoids two conversion stages and typically recovers 10-20% of your battery capacity compared with an inverter chain. Two practical cautions: a non-standard supply gives SpaceX an argument in a warranty claim, so keep the stock brick and reinstate it before contacting support; and buy a converter with a published datasheet and documented thermal shutdown, then derate it by a third, because cheap unbranded boost converters run near their ceiling are the most frequently blamed component in community failure reports.
What should I actually buy for 8 hours of Starlink backup during load shedding?
The arithmetic is simple — a standard dish at 75W needs about 63Ah of 12V lithium for 8 hours, so you buy a 100Ah LiFePO4 battery — but the buying decisions that matter are the ones around it. Specify a battery with a 100A continuous BMS and low-temperature charge cut-off, a Class-T or ANL fuse within a few inches of the positive terminal, and a mains charger sized to your outage schedule rather than the smallest one that fits (a 5A trickle charger cannot refill a bank inside a three-hour grid window). If you cannot get certified loose cells imported without punitive duty — common across parts of Africa, South Asia and the Caribbean — a sealed 1,000-1,500 Wh portable power station with pass-through charging is usually cheaper landed than a component build, because it clears customs as a finished appliance. Whatever you buy, size the changeover device too: a UPS-style inverter/charger or an automatic transfer switch keeps the terminal from cold-booting every time the grid drops.
How many solar panels do I need to run Starlink 24/7 off-grid?
Start from daily energy, divide by your local winter peak sun hours, then add a weather margin. A standard dish at 1.8 kWh per day in a sunbelt location with 5 peak sun hours needs a 450-600W array, which is two 300W panels or a pair of 400W residential modules; in temperate Europe or the northern US, plan 800-1,200W, or three to four modules; above 50 degrees latitude in midwinter, 1,500W plus a generator is realistic, and tilt matters more than adding panels. A Starlink Mini cuts all of those numbers by roughly two-thirds — a single 200W panel and a 100Ah battery is a complete year-round satellite internet supply in the tropics.
Why does my Starlink keep rebooting when running on battery power?
Owners consistently report terminals dropping out when supply voltage sags, with community meter logs clustering the failure point in the mid-40s of volts on the PoE line; SpaceX does not publish an undervoltage cut-off, so treat any exact threshold you read online as indicative. The cause is almost always voltage sag rather than a flat battery: an undersized DC-DC converter, thin cable on the 12V side, or a lead-acid bank that collapses under load. Fit 12AWG or heavier for a 10ft DC run, size the converter to peak draw rather than average, and measure voltage at the injector while the dish is under load rather than at rest.
Is the Starlink Mini worth it for solar-only setups?
For power-limited installations, yes. At roughly 16-25W typical draw the Mini needs about a third of the battery and array of a standard dish, and it accepts 12-48V DC natively so no step-up converter is required. Choosing the Mini typically removes USD 400-900 of battery and panel from the build, and it removes a failure point as well as a cost. The trade-off is a smaller phased array with lower peak throughput and a narrower field of view, which matters more on a congested suburban cell than on a remote off-grid site where you are the only subscriber for miles.
How much does the power system add to the true monthly cost of satellite internet?
Amortise it. A USD 1,000 backup build over five years is about USD 17 per month; over three years it is USD 28, and over eight it is USD 10. Against a USD 120 residential plan that is a 8-23% uplift on the true cost of service, which most households accept without argument. The picture inverts in markets where the subscription itself is cheap: at USD 30 per month the same build adds 33-93% to what you pay, which is why buyers in low-price markets are usually better served by a Mini and a single portable power station than by a component build sized for a standard dish. Work out the uplift for your own market before you decide which terminal to order, because the terminal choice sets the power budget for the life of the installation.
Is satellite internet realistic somewhere with no grid power at all?
Yes, and it is one of the most common Starlink deployments. A Starlink Mini on a 100Ah lithium battery with a 200-300W array runs continuously in a sunbelt location; a standard dish needs roughly 200Ah and 600-800W in the same climate. The limiting factor in genuinely off-grid sites is almost never the satellite link — it is winter solar yield and the size of the battery bank. Above about 50 degrees latitude, plan a hybrid: panels for nine months, a small inverter generator with auto-start on low state of charge for the three months when the array cannot keep up. Adding a generator is nearly always cheaper than the array size that would make one unnecessary.
Follow Starlink News on Google. Make us a preferred source to see more of our reporting in Google search results.
Comments 1