Portable Power Stations for Home Backup

What a power station actually is

Strip away the marketing and a portable power station is three parts in one case: a lithium battery pack, an inverter that turns its DC into household AC, and a charger. There is no engine, nothing burns, nothing is emitted. That is why it can sit in a hallway during a storm when a generator cannot come within twenty feet of an open window.

Nearly everything worth buying now uses lithium iron phosphate cells — LiFePO4, often shortened to LFP. The chemistry matters more than the brand. LFP tolerates thousands of full charge cycles where the older lithium-ion chemistry managed several hundred, and it is markedly less prone to thermal runaway. When a spec sheet says 3,000 cycles to 80% capacity, it is claiming the pack still holds four-fifths of its original charge after three thousand full drains. Charged and drained weekly, that is a device with a working life measured in decades rather than years.

The two numbers that decide everything

Watt-hours (Wh) is the tank. It is how much energy the battery holds. A 1,000Wh station running a 100W load lasts roughly ten hours, less conversion losses of typically 10–15%.

Watts (W) is the pipe. It is how much the inverter can deliver at any instant. An 1,800W inverter runs an 1,800W load and refuses a 2,000W one, however full the battery is.

Buyers confuse these constantly, and the failure mode is specific: a large battery with a small inverter will politely decline to start your fridge, because the compressor asks for two or three times its running draw for the first fraction of a second. That spike is called surge, and it is why every honest spec sheet quotes two output figures. Champion's published reference puts a domestic refrigerator at 150–400W running and 800–1,200W starting. A 300W station will never start it. A 1,000W station will.

What capacity is enough

Work it backwards from the outage, not from the price. The EIA's Electric Power Annual for 2024 puts the average U.S. customer at eleven hours without power across the year — but that average hides the shape of the problem. Roughly two of those hours came from ordinary faults, and nearly nine from major events. A typical year is a handful of brief interruptions plus, occasionally, one long one.

So there are really two purchases. The first covers the common case: the router, the phones, a couple of lamps, the fridge cycling for a few hours. Somewhere between 500Wh and 1,100Wh does that comfortably. The second covers the rare case, and it is a different machine — either 2,000Wh-plus with a way to recharge it, or something that burns fuel.

The honest advice most of this category will not give you: if your outages are short, buying 3,000Wh is money spent on a scenario you do not have. Buy for the outage you actually get, and here is how to work out which one that is.

Where a battery genuinely beats a generator

• It works indoors. No carbon monoxide, so no twenty-foot rule and no dead-of-night decision about whether the garage counts as outside.
• It is silent. Roughly 30dB under load on the quieter units — a running fridge is louder.
• It starts. There is no carburetor to gum up over a year in the shed, no fuel to go stale, no pull cord.
• It doubles as a UPS. Models with sub-20ms switchover keep a desktop or a modem alive through the transition without a reboot.

And where it loses: it holds a fixed amount of energy and then it is empty. A generator with a fuel can beside it is effectively unlimited. On day three of an ice storm that difference is the whole argument, which is why the storm scenario page treats them as complements rather than rivals.

What we do not tell you

We have not tested any of these. Nobody here has plugged one in, measured its real inverter efficiency, or left it in a shed for two winters. What we have done is read the published specifications, arrange them so they compare honestly, and compute the numbers the manufacturers leave out — runtime against a real appliance load, and cost per usable kilowatt-hour over the rated cycle life. That method has real limits and we set them out.