How-to
How to power a mini PC from a battery safely
Short answer
A mini PC runs from a battery when the pack feeds a DC-DC converter whose output matches the computer's documented DC input voltage, connector and identification scheme. Size the battery from measured input watts divided by converter efficiency — 40 W at 85% is a 47 W battery load — and fuse the battery lead for the battery-side current. A 12 V, 20 Ah battery with 80% usable energy runs that load about four hours.
On this page
- Can any mini PC run from a battery?
- What does the DC input label have to tell you?
- Why can't I size the battery from the adapter's wattage?
- How do I measure what the mini PC actually draws?
- Which DC-DC converter belongs between battery and PC?
- How long will the battery run the mini PC?
- How do I wire and protect the battery circuit?
- What should happen when the battery runs flat?
- Questions asked
Yes, a mini PC runs from a battery, provided the pack feeds a DC-DC converter whose output matches the computer's documented DC input and the battery is sized from measured watts rather than the adapter's rating. The order that works: read the input voltage, polarity and connector on the computer first; measure what it actually draws at its DC input; convert that to battery-side watts with the converter's efficiency (40 W at 85% becomes 47 W); then divide usable watt-hours by that load. A 20 Ah pack at the nominal voltage carries the example load about 2.6 hours on lead-acid and roughly 4 hours on lithium iron phosphate. Fuse the battery lead at the terminal, plan the flat battery as a power outage, and set the firmware to boot the machine again when DC returns.
Can any mini PC run from a battery? #
Yes — but only a machine that accepts an external DC input can run from a battery directly, and the battery's full voltage span, empty to fully charged, must sit inside what that input is documented to accept.
Start at the label beside the DC jack and at the power section of the manual, not at a store listing. Mini PCs that ship with an external adapter take DC directly through a barrel connector, and those are the straightforward case: the battery replaces the adapter, through a converter. A computer with an internal mains supply and an IEC inlet has no DC input at all. It runs from a battery only through an inverter sized for its adapter — one more conversion stage and its losses — or through a manufacturer-documented DC arrangement, if the manual describes one. If the documentation offers neither a DC input nor a documented DC option, the decision for that machine is already made: choose a different computer rather than improvising inside an enclosed supply.
What does the DC input label have to tell you? #
Four entries decide the whole design: the input voltage or range, the polarity symbol beside the jack, the maximum input current or watts, and the connector barrel size — plus any adapter identification the computer checks before it draws power.
Record the input voltage or range first: a single value such as 12 V, or a span the manual prints for wide-range units. Read the polarity symbol beside the jack — center-positive is the common arrangement on small computers, but the symbol is the document that governs, so read it rather than assume it. Convert the supplied adapter's rating into a rail current so you know the machine's class: a 36 W adapter on that rail is up to 3 A, because 36 W ÷ 12 V = 3 A. Then settle identification. Some computers interrogate an identification conductor inside the DC plug before they draw full power, and they refuse to boot or throttle back when the supply cannot answer; a two-wire converter output cannot satisfy that check, and the manual — not the plug shape — tells you whether your machine behaves this way. If any entry is unreadable or the identification question is open, stop and obtain the manufacturer's documentation before wiring a battery to anything.
Why can't I size the battery from the adapter's wattage? #
The adapter's rating is the ceiling the outlet can deliver, not what the computer consumes: a 65 W adapter implies about 5.4 A available on that rail, while the machine's real average only shows up when you measure it.
The multiplication trap looks like this: 5.5 A at the nominal volts gives 66 W, therefore the battery must deliver 66 W. The arithmetic is correct and the basis is not. That product sizes the source's ceiling, and it becomes a load number only if the manual documents 5.5 A as the input's actual maximum draw — which manuals state as an input limit, not as what the machine normally consumes. A 65 W adapter on a machine that can idle near 10 W is not burning 65 W, and a 180 W brick proves even less about its computer. Two numbers follow two different decisions. The converter's continuous rating is sized from the adapter's rating, so it covers the worst case the manufacturer planned for; every watt-hour calculation is sized from the measured average, which is the next step.
How do I measure what the mini PC actually draws? #
Measure at the point the battery will actually feed, with an inline DC watt meter, and record three numbers: idle watts with the normal software up, average watts across the real workload, and the peak at power-on — 40 W average on this input is the example used below.
Put the meter where the battery will sit: in line with the DC feed, so the reading is the load the pack will actually carry. If you measure upstream of the converter instead — on the battery side — the converter's losses are already inside the number, and you must not divide by efficiency a second time.
- Connect an inline DC watt meter between the bench supply, or the existing AC adapter, and the computer's DC input.
- Apply power and record the peak watts during boot; this is the surge the converter has to survive.
- Let the operating system and its normal services settle, then record idle watts with every peripheral that will ship in the final build attached.
- Run the real workload — the application, vision processing, radio traffic, whatever this machine actually does — for at least half an hour, and record both the average and the highest sustained watts.
- Repeat once from a cold start.
Three numbers decide three things: the average sizes the battery in watt-hours, the sustained maximum is checked against the converter's continuous rating, and the boot peak against its surge rating. If the measured average already breaks the energy budget, the fix is a lower draw — less workload or a different computer — because a larger converter never improves runtime; it only adds one more efficiency loss.
Which DC-DC converter belongs between battery and PC? #
When pack and input share a 12 V rating, you need a buck-boost converter, because the pack spans roughly 10.5 V empty to about 14.6 V charging and the output must hold that nominal value through the entire span — a plain buck or a plain boost cannot cover both ends.
A 12 V lead-acid pack rests near 12.7 V full, climbs to roughly 14.4-14.7 V while a charger works, and is effectively empty near 10.5 V. A lithium iron phosphate pack of four cells in series, 12.8 V nominal, runs from about 10.0 V empty to about 14.6 V full. Both spans straddle the nominal output, which is exactly why a converter that only steps down, or only steps up, fails at one end of the charge. The selection rule is mechanical: battery always above the output needs a buck, always below needs a boost, straddling it needs a buck-boost. Check three entries in the converter's documentation: the input range covers the pack's full span including the charger's absorption voltage; the continuous output rating meets or exceeds the supplied adapter's rating; the surge rating covers the measured boot peak. For the example build, a buck-boost rated at 8 A on that bus — 12 V × 8 A = 96 W — clears the 65 W adapter rating and the 47 W battery-side average with room for the boot surge. Take efficiency at your actual load point rather than the headline figure, and treat a claimed efficiency as proven only by measuring watts loss. If the output voltage is adjustable, set it to the documented value and verify it at the plug before the computer is ever connected.
How long will the battery run the mini PC? #
Runtime is usable watt-hours divided by battery-side watts: 40 W at the input and 85% converter efficiency load the battery at 40 ÷ 0.85 ≈ 47 W, so a 20 Ah pack at 12 V (240 Wh) runs about 2.6 hours on lead-acid or about 4 hours on LiFePO4.
Carry the numbers through:
- Battery-side power: 40 W ÷ 0.85 = 47.06, so plan on 47 W from the battery.
- Battery-side current: 47 W ÷ 12 V = 3.9 A.
- Pack energy: 240 Wh, from 20 Ah at the nominal volts.
- Lead-acid, 50% usable depth of discharge: 240 Wh × 0.50 = 120 Wh usable, and 120 Wh ÷ 47 W = 2.55 h.
- LiFePO4, 80% usable: 240 Wh × 0.80 = 192 Wh, and 192 Wh ÷ 47 W = 4.09 h.
Watt-hours convert to amp-hours by dividing by the pack's nominal voltage — Ah = Wh ÷ V — which is why an amp-hour figure alone cannot compare packs at different voltages. The usable fraction comes from the battery's documentation; depth-of-discharge limits, cold temperature and aging all cut it below the nominal label.
For a fixed target, run the arithmetic backwards. An 8-hour shift at 47 W needs 47 W × 8 h = 376 Wh of usable energy: 376 ÷ 0.50 = 752 Wh of lead-acid, about 63 Ah, or 376 ÷ 0.80 = 470 Wh of LiFePO4, about 39 Ah. When the pack that meets the hours becomes the heaviest item on the machine, the better fix is a lower-draw computer such as a fanless industrial mini PC, not more packs in parallel.
How do I wire and protect the battery circuit? #
Fuse the positive battery lead within a few centimeters of the terminal at the next standard value above the converter's documented maximum input current — with a converter rated 8 A, that is a 10 A fuse on conductors rated for at least the fuse value.
Work in this order, and keep the converter output unplugged until the input side is proven:
- Connect battery negative to the converter's input minus terminal.
- Connect the fused positive lead: fuse first at the battery terminal, then on to the converter's input plus.
- Set an adjustable converter's output to the documented voltage and verify it at the plug with a meter before the computer is connected.
- Connect the output plug to the computer, polarity matching the label symbol; a plug that fits loosely causes brownouts and reboots that look exactly like software faults.
- Power up and re-measure on the battery side to confirm the watts match the prediction.
Size the conductors for the fuse value, not for the average current: the fuse protects the wire, so the wire has to survive everything the fuse lets through. The 3.9 A average on this build sits far below the 10 A fuse, but the fuse exists for the short circuit, not for the schedule. Batteries store enough energy to weld a dropped wrench across a terminal, flooded lead-acid gives off hydrogen while charging and needs a ventilated compartment, and a lithium pack's battery management system stays in circuit — it is the piece that keeps the cells inside their limits.
What should happen when the battery runs flat? #
Plan the flat battery as a normal power outage: the low-voltage cutoff opens the feed, the PC drops dead, and the firmware option that boots the machine whenever DC input returns brings it back on its own — set that option, because many machines default to staying off.
Give the pack a low-voltage cutoff set to the battery documentation's end-of-discharge voltage, so the cells cannot be dragged below their limit while the machine sits unattended. Better still, shut the computer down in software one step above that hardware cutoff: a PC that loses power mid-write can lose data, and a monitored shutdown at a defined state of charge always comes back clean. Then set the firmware power option that applies power whenever DC input returns — the behaviour usually labelled as restoring power after a loss — because without it the machine waits for a button press nobody is there to press. With that set, the flat battery stops being an incident: the charger or the sun refills the pack, DC returns, and the box boots straight into its application. A battery-backed edge box living that cycle wants a light startup path; on Linux, installing Node-RED natively on the edge computer puts the collection job directly on the machine, with no container layer to start first.
Questions asked
Can the battery charge while the mini PC is running?
Yes, and that is the normal arrangement on a vehicle or a solar cabinet, provided the charger's output voltage stays inside the converter's documented input range through the whole charge cycle. A lead-acid charger on a nominal 12 V bus can push it to about 14.4 V at absorption, so the converter must accept that on its input while still holding the computer's output steady. Feed the computer from the converter, never directly from the charger output.
Can I power the mini PC from a vehicle 12 V socket instead of a converter?
Only when the computer's documented input range covers everything the socket delivers, because a vehicle bus sits near 12.6 V parked, near 14.2 V with the engine running, and collapses while the starter cranks, and the same wiring carries load dump spikes well above nominal during alternator faults. Converters with an automotive surge rating exist for exactly this reason; check the input range and the surge rating before trusting a socket feed.
Why not use a power bank with an AC outlet to run the normal adapter?
You can, but the efficiency arithmetic argues against it, because the losses multiply: at 90% inverter efficiency and 85% adapter efficiency, 0.90 × 0.85 = 0.765, so roughly a quarter of the stored energy heats the air before it reaches the computer. One DC-DC converter from the power bank's DC output does the same job with a single loss stage, and the runtime math stays a single division.
Does sleep mode change battery runtime planning?
Yes. A sleeping PC still draws watts keeping memory and the network interface alive, it may not wake cleanly when no display is attached, and it does nothing useful for a plant application in the meantime. Set the power plan so the machine never sleeps, then re-measure idle with that setting active; otherwise the idle number you measured hides a state the machine enters between shifts, and the runtime math splits in two.