Why a 6400W inverter can trip a 140A BMS

Two numbers datasheets get away with

An inverter's DC-side current is AC output power divided by DC input voltage divided by conversion efficiency. Every part of that sentence sounds simple, and the two places people get it wrong are exactly the two places manufacturers have an incentive to publish a flattering number instead of a working one.

First, voltage. A "48 V" battery is not 48.0 V under load — it sits around 50.4 V once at rest and charged (this project's vLoaded(48)), because LiFePO4 cells hold a resting voltage well above their nominal rating across most of the usable state of charge. Using the higher, real voltage in the denominator would actually lower the computed current a little — so voltage isn't where the surprise comes from. Efficiency is.

An inverter's published efficiency figure — 90%, 95%, sometimes higher — is measured at whatever load point makes the number look best, typically 50% of rated output, under bench conditions with a resistive load and a fresh unit. Full continuous output, the load an undersized bank actually experiences, sits further down the efficiency curve: switching losses and I²R losses in the output stage both scale up as current rises, and neither scales down the datasheet's headline number to compensate. This project does not try to model that curve panel-by-panel, because that data isn't published for a given inverter — it deals with the uncertainty by capping every published efficiency figure at 85% for the purpose of computing DC-side current (Calc::INVERTER_DC_EFFICIENCY_CAP in src/Calc.php, mirrored as C.INVERTER_DC_EFFICIENCY_CAP in assets/js/engine.js). A 95%-rated inverter is treated, for this one calculation only, as an 85% inverter — never as worse than 85%, since no inverter sold for this use case is that inefficient at any load point, but never credited with more than that either.

It's worth being precise about what this cap is not. It isn't a claim that every inverter's real full-load efficiency is exactly 85% — a well-designed unit under a clean resistive load might do better, and a cheap modified-sine unit driving a motor load might do worse. It's a ceiling applied specifically to the number used for sizing the DC-side conductor and checking it against the BMS, chosen so that the sizing calculation never assumes an efficiency better than what a decent full-load inverter can be expected to deliver. Treat it as a safety margin baked into the formula rather than a measured constant, in the same spirit as this project's fuse-sizing factor of 1.25× rated current — neither number claims to be a physical fact about a specific unit, both exist so a calculation built on a manufacturer's best-case number doesn't quietly inherit that manufacturer's optimism.

The formula

inv_dc_continuous_a = continuous_w / (vLoaded(system_v) × min(efficiency, 0.85))

The same formula with surge_w in place of continuous_w gives the surge draw, which matters against the battery's peak BMS rating rather than its continuous one — a separate, usually much higher, limit that a startup surge or a compressor kick can still blow through even when the continuous math looks fine.

vLoaded(system_v) returns 12.6 V, 25.2 V or 50.4 V for a 12/24/48 V system — the battery's real voltage under a working load, not the round number on the label. Both numbers, the capped efficiency and the loaded voltage, come from this project's own constants; nothing here is an invented industry rule of thumb.

A concrete example

Take a real pairing from this site's catalog: an EG4 WallMount Indoor 280Ah lithium battery (48 V nominal, 51.2 V, 280 Ah, 140 A continuous / 200 A peak BMS) with a Victron MultiPlus-II 48/8000/110-100 inverter (6400 W continuous, 15,000 W surge, 95% published efficiency).

The continuous draw alone overshoots the BMS's 140 A rating by 9.4 A — 6.7% over, enough to trip the BMS at full inverter load even though the naive calculation showed 8.4 A of headroom in the other direction. The surge figure is worse: 350.1 A against a 200 A peak rating, 75% over. A dishwasher's or air compressor's startup surge on this inverter would trip the BMS's peak protection well before it trips on sustained draw. Nothing about the pairing looks wrong from the box labels; the battery has more energy capacity than the inverter could use in two hours, and the naive current check clears the BMS's continuous rating with room to spare. The actual load on the DC bus is a different number from either of those, and it's the one the BMS reacts to.

A second example: enough headroom to survive both checks

Downsize the inverter on the same battery and the picture changes completely. A Victron MultiPlus 48/3000/35 (2400 W continuous, 6000 W surge, 95% published efficiency) on the same EG4 280Ah bank: real continuous draw is 56.0 A (56.0 A of the bank's 140 A limit used, 84.0 A of margin — a 60% cushion) and real surge draw is 140.1 A against a 200 A peak rating (a 30% cushion). Both checks clear with margin large enough to absorb a compressor start or a second, unmodeled load on the same circuit.

The lesson generalizes past this one battery: the number that matters is not the inverter's power rating relative to the battery's Ah or Wh capacity — those describe how long the battery can run something, not how hard the inverter pulls at any given instant. It's the inverter's continuous_w and surge_w relative to the battery's BMS continuous and peak current ratings, run through the real loaded voltage and a realistic, not a flattering, efficiency number.

Why continuous and peak get checked separately

A BMS enforces two different limits because it protects against two different failure modes. The continuous rating protects the cells and the internal wiring from sustained heating — exceed it for more than a few seconds and the pack heats past what its thermal design tolerates over time. The peak (surge) rating is a much shorter-duration allowance, usually a few seconds, sized around motor starts and other transient loads; it exists precisely because a compressor, pump or power tool can draw several times its running current for a fraction of a second without doing any damage, as long as it's brief. A pairing can clear one check and fail the other — the second example above clears both, the first example fails both, but plenty of real systems fail only the surge check with a continuous draw that looks perfectly safe. Checking only one number is checking half the problem.

This project's rule set treats them accordingly: inverter-exceeds-bms fires as an error when continuous draw exceeds the BMS's continuous rating, a separate inverter-bms-margin warning fires below that when the draw is already past 85% of the continuous limit (headroom under 15%, before anything actually trips), and surge-exceeds-bms-peak checks the surge number against the peak rating independently. None of the three substitutes for the others.

The margin warning exists because a bank sitting at, say, 90 A of draw against a 100 A continuous rating is not actually safe in practice even though the error rule stays silent — a single degraded cell group, a slightly cold pack, or a BMS firmware update that trims its own limit for safety can turn a 10 A cushion into an overage without anything about the inverter or the wiring changing. Warning at 85% of the limit rather than at 100% gives a builder room to notice a thin margin before a firmware update or a bad day turns it into a nuisance trip in the field, rather than finding out the first time the BMS actually opens under load.

Why energy balance uses a different efficiency number

This project has two separate constants that both look like "inverter efficiency," and mixing them up produces confusing results. DEFAULT_INVERTER_EFFICIENCY (88%) is used when converting AC appliance loads into daily watt-hours drawn from the battery — an energy-balance calculation, run once per day's usage rather than at any single instant, where the difference between 85% and 88% shifts a day's total load by a few percent and doesn't change which appliances the system can run at all. INVERTER_DC_EFFICIENCY_CAP (85%) is used only for the instantaneous DC-side current calculation this guide is about — the number that decides whether a BMS trips right now, at full output, where the same few percentage points are the entire difference between passing and failing the check. The two figures aren't inconsistent; they answer different questions at different time scales, and a system model that used 88% for the current check would understate the exact failure mode this guide exists to catch.

Try it with your own inverter and battery

Enter an inverter's continuous power and efficiency from its datasheet, the system voltage, and the battery bank's continuous BMS rating.

What this means for pairing an inverter with a battery

Three practical consequences follow directly from the formula:

1. Never size an inverter against Ah or Wh capacity alone. A battery with plenty of stored energy can still have a BMS that opens under a load the inverter is fully capable of drawing for hours — capacity and current rating are independent numbers, and only one of them is checked when someone eyeballs "how long will this run."
2. Use the manufacturer's efficiency only as a ceiling, not the real number. Capping it at 85% for the DC-current calculation, as this project does, is not pessimism for its own sake — it is the difference between a pairing that looks like it has 8 A of margin and one that actually has 9 A of overage, on the same hardware.
3. Check surge against peak separately from continuous against continuous. A pairing that clears the continuous check by a wide margin can still trip the BMS's peak protection on the first motor start, because the surge multiplier on inverter output and the peak multiplier on BMS rating are set independently by their respective manufacturers and don't move together.

The inverter catalog runs this exact inverterDc calculation against every battery bank size in the database, so a pairing can be checked before it's bought rather than after the BMS starts tripping.