How to size a BMS for your battery pack
Technical review pending — checked against public standards on
- sizing
- selection
Continuous current is a division problem, not a guess. Work out what your pack actually has to deliver, then add the margins that real loads need.
Most BMS sizing mistakes come from one shortcut: picking a board by the number on the label instead of by the current the pack has to carry. The label is a ceiling. The load decides what you actually need.
Step 1 — Continuous current
Continuous current is the highest current your system draws for minutes at a time, not the instant it starts. For a resistive load it is simply power divided by pack voltage.
I_cont = P_load ÷ V_nominal
P_load in watts, V_nominal in volts, result in amps. Use nominal voltage, not fully-charged voltage — nominal is what the pack spends most of its life at.
A 3000 W inverter on a 48 V nominal pack therefore draws about 62 A continuously. On a 12 V pack the same inverter draws about 250 A — which is why the same appliance needs a completely different board depending on pack voltage. This is the single most common sizing error we see.
| Load | 12 V pack | 24 V pack | 48 V pack |
|---|---|---|---|
| 500 W | 42 A | 21 A | 10 A |
| 1000 W | 83 A | 42 A | 21 A |
| 2000 W | 167 A | 83 A | 42 A |
| 3000 W | 250 A | 125 A | 62 A |
| 5000 W | 417 A | 208 A | 104 A |
Step 2 — Peak current, with a duration
Motors, compressors and inverters draw a surge on start-up. A peak rating without a stated duration is not a specification — 300 A for 10 seconds and 300 A for 10 milliseconds are different products.
As a rule of thumb, inductive loads surge to 2–3× running current for 1–3 seconds. If a datasheet gives you a peak number with no time attached, treat it as unavailable rather than as a large continuous rating.
Step 3 — Derating for heat
Every current rating assumes a temperature. MOSFETs conduct worse when hot, so a board rated 100 A at 25 °C cannot deliver 100 A inside a sealed enclosure at 45 °C. Ask what ambient the rating was measured at, and plan for a lower number if the board lives in a box with no airflow.
- Open frame with airflow: use the rating close to face value
- Sealed enclosure: derate by roughly 20–30%
- Ambient above 40 °C: derate further, and expect the board to throttle rather than fail
- Long cable runs: add for voltage drop, which shows up as heat at the terminals, not at the board
Step 4 — Check the pack can supply it
The BMS is a gate, not a source. A 200 A board on a pack built from cells rated 50 A continuous each, in a single parallel string, will be limited by the cells. Calculate pack capability as cell continuous rating multiplied by the number of parallel strings, and take the lower of that and the board rating.
I_pack = I_cell × N_parallel
If this number is below your I_cont from step 1, the pack is the constraint. A bigger BMS will not fix it.
Worked example
A campervan with a 2000 W inverter, a 48 V nominal LiFePO4 pack, and a 4-parallel pack of cells rated 50 A continuous each.
- Continuous current: 2000 W ÷ 48 V ≈ 42 A
- Add 25% headroom for inverter inefficiency and continuous rating honesty: ≈ 52 A
- Peak: inverter surge of about 2× for 2 s → ≈ 84 A for 2 s
- Pack capability: 50 A × 4 = 200 A — not the constraint
- Enclosure is sealed in a van: derate the board rating by 25%
- So look for a board whose continuous rating, after derating, clears 52 A, and whose peak rating is stated with a duration that covers 2 seconds
Notice that the answer is a range, not a model number. Which board you pick then depends on series count, port topology and protocol — each covered in its own note.
