Quick answer
Off grid, a 9,000–12,000 BTU inverter mini split typically draws about 200–600W at mild load, up to roughly 1,000–1,500W at full output on a hot day — a 12,000 BTU (1-ton) unit in the SEER 17–22 class averages about 400–700W in cooling. That’s roughly 4kWh for an 8-hour cooling night, needing about 4,700Wh of usable battery (~100Ah at 48V or ~400Ah at 12V) and about 1,000–1,300W of panels to refill it on a typical sunny day. Bottom line: a mini split is a big off-grid load — realistic for a large DIY battery bank, not for a small portable power station.
Key takeaways
- Inverter mini splits modulate: ~200–600W at mild load, ~1,000–1,500W only at full output on a hot day.
- Startup surge is modest (~2× running) — nothing like the 3–5× spike of old single-speed compressors.
- A 12,000 BTU unit at SEER 17–22 averages ~400–700W in cooling — about 3–5.6kWh for an 8-hour night.
- Measure first: a 24-hour plug-in wattmeter beats nameplate math; the NEEP cold-climate heat pump list covers published heating watts.
- Battery: ~4kWh/night ÷ 0.85 usable ≈ 4,700Wh (~100Ah@48V or ~400Ah@12V) before adding autonomy days. Why 48V: our explainer.
- Refill: ~1,000–1,300W of panels at 4 sun hours with a 0.75 derate — a real array, not a couple of portable panels.
- Honest verdict: mini splits suit large DIY banks or generator-fed systems; small portable stations and smaller loads (fans, DC evaporative) are the realistic alternatives.
Section 1: Measure YOUR draw — meter first
The nameplate figure is the maximum rated input at full output — not your typical draw. Inverter mini splits run at partial capacity most of the time, so the average sits far below the maximum.
Step 1: plug in a wattmeter (Kill A Watt style) between the wall and a 120V mini split for a full 24 hours, including a warm afternoon and your usual set-point. You get: running watts at your set-point, max watts on the hot day, and total kWh over 24h (divide by run hours for your average draw).
Step 2 (heating): NEEP cold-climate listings. For heating, published rated watts at low outdoor temperatures beat guesses. NEEP’s Cold Climate Air Source Heat Pumps product list (ashp.neep.org) lists models with rated input watts at low-temperature conditions — use those at your design outdoor temperature, not cooling watts. The database moved to AHRI reporting in 2025; re-verify before you buy.
| Load condition (9,000–12,000 BTU inverter unit) | Typical draw | What it means |
|---|---|---|
| Running, mild load (set-point mostly met) | ~200–600W | Modulating compressor at partial capacity — most of its runtime |
| Full output, hot day | ~1,000–1,500W | Unit pinned at max cooling |
| Average in cooling, SEER 17–22 (12k BTU / 1 ton) | ~400–700W | The number to use for nightly energy math |
| Startup surge | ~2× running | Inverter ramp, not the 3–5× of old compressors |
Section 2: Energy budget — 8 hours × 500W ≈ 4kWh
The working case: 0.5kW average × 8h = 4kWh per night. That’s “worst-case-ish” — 500W sits mid-range of a SEER 17–22 unit’s 400–700W. A mild night lands closer to 3kWh; a long, hot one can run 5kWh+.
| Scenario (12k BTU cooling) | Average draw | Hours | Nightly energy |
|---|---|---|---|
| Mild night | ~400W | 8 | ~3.2kWh |
| Hot night (our working case) | ~500W | 8 | ~4.0kWh |
| Worst-ish continuous | ~700W | 8 | ~5.6kWh |
Heating changes the budget. A heat pump at low outdoor temperatures draws more power and delivers less capacity per watt — the SEER averages above are cooling numbers. Budget heating from the unit’s published low-temperature input watts (NEEP listing or manufacturer spec) for the coldest night you expect. A common rule of thumb is 1.2–1.5× the cooling budget, but verify with your model’s rated watts.
Section 3: Battery sizing — 4kWh ÷ 0.85 ≈ 4,700Wh usable
Batteries are rated in total (nominal) watt-hours, but you can only use a fraction. For LiFePO4, usable depth-of-discharge is ~80–90%; we use 0.85 — the same factor the site’s fridge article uses, so all math stays consistent.
Worked example:
- Daily need: 4,000Wh (8h × 500W)
- ÷ 0.85 usable = ~4,706Wh ≈ 4,700Wh usable
- At 48V: 4,700 ÷ 48 ≈ 98Ah → 100Ah@48V
- At 12V: 4,700 ÷ 12 ≈ 392Ah → 400Ah@12V
| Nightly energy | ÷ 0.85 usable = | ≈ @48V | ≈ @12V |
|---|---|---|---|
| 3.2 kWh (mild) | ~3,765 Wh | ~80Ah | ~315Ah |
| 4.0 kWh (our case) | ~4,700 Wh | ~100Ah | ~400Ah |
| 5.6 kWh (worst-ish) | ~6,590 Wh | ~138Ah | ~550Ah |
That’s one night for one mini split before any other loads — no fridge, lights, or pump. Add autonomy days: 2 days × ~4,700Wh = ~9,400Wh usable (~200Ah@48V). The site’s battery capacity calculator does this for you.
Why 48V and not 12V? A 4,700Wh battery at 12V means ~390A at full draw — wiring, fuses, and inverter connections get heavy and lossy. At 48V currents drop 4×, which is why large banks practically need 48V. Full trade-off: 12V vs 24V vs 48V; inverter and battery-draw sizing: solar inverter sizing.
Section 4: Solar array to refill — 4kWh/day ≈ 1,000–1,300W of panels
The panels must put the 4kWh back on a good solar day:
Panel watts ≈ daily Wh ÷ (sun hours × 0.75)
Worked:
- 4,000 ÷ (4 × 0.75) = 4,000 ÷ 3 ≈ 1,333W → ~1,300W
- At 5 sun hours: 4,000 ÷ (5 × 0.75) ≈ ~1,067W → ~1,100W
- At 3 sun hours: 4,000 ÷ (3 × 0.75) ≈ ~1,780W
So the honest planning range at typical US locations (3–5 good sun hours) is ~1,000–1,300W of panels for a 4kWh/night cooling budget — push toward the higher end in summer, when nights are cooling-heavy and usable sun shrinks. The 0.75 derate covers panel angle, temperature, controller losses, and cloudy gaps — full method in solar panel output.
| Sun hours | Panel watts for 4kWh/day (0.75 derate) | Notes |
|---|---|---|
| 5 | ~1,100W | Sunny Southwest locations |
| 4 | ~1,300W | Typical US planning day |
| 3 | ~1,780W | Shaded/northern summers |
Cloudy-day honesty: an overcast day can deliver 10–20% of rated output (see solar system sizing). The battery is the buffer — size for the worst stretch, treat solar as the recharge.
Section 5: Honest verdict — big load, big bank
A mini split is comfortable, quiet, and efficient — but one of the biggest single loads you can put off grid. A modern fridge runs on ~1.4kWh/day; a mini split cooling night runs ~4kWh — nearly 3× the energy, before heating season.
- Works well: large DIY battery banks (10kWh+), cabins with real arrays (1.5kW+), whole-house systems sized with the solar system sizing planner.
- Struggles: a “solar generator” with 1–2kWh of battery runs a mini split for a couple of hours at mild load before draining — not a nightly solution. A 100W panel can’t refill what the unit uses; you’d need ~10× that.
- Alternatives for small systems: 12V DC fans, DC evaporative coolers in dry climates (point-cooling, not whole-room), and spot-cooling an occupied chair or bed — all happy on modest batteries and small panels.
Common mistakes
- Sizing from the nameplate maximum. A 12,000 BTU unit labeled ~1,100–1,500W max doesn’t run at that all night — it modulates. Nameplate math over-sizes the system 2–3×.
- Mixing average and max. The 400–700W average sets battery kWh; the 1,000–1,500W max sets inverter continuous rating. Confusing the two doubles battery cost or trips the inverter.
- Treating surge like an old fridge. Inverter mini splits start at ~2× running — no 5× surge buffer needed, but check the ramp in solar inverter sizing.
- Spending the whole battery budget on cooling. The fridge, lights, and water pump still eat kWh/day — subtract them first.
- Assuming heating obeys the cooling average. Low outdoor temps mean higher draw and lower capacity; plan from published low-temperature watts, not the SEER cooling figure.
- Skipping the meter. Spec tables are averages across models, seasons, and set-points; 24 hours of your actual unit at your actual set-point beats them all.