How does a home battery work? The whole route, from your roof to your kettle

A hundred kilowatt-hours go into your home battery. Eighty-five come out.

Those fifteen did not disappear. Energy cannot disappear; that is the first law of thermodynamics, not an opinion. They are still somewhere in your house, and you can point at exactly where.

The video follows one kilowatt-hour from the roof to the kettle and points at those places. This article does the same thing with the arithmetic attached: the cell voltages you can check yourself, the standard number I deliberately leave out of the video, and the measurement conditions under my own efficiency figure. If you never watch the video, you miss nothing here.

What you will not find here is money. No payback period, no prices, no brand advice. That is a calculation with your own consumption in it, and it lives elsewhere.

Key takeaways

• Your house speaks AC, every battery speaks DC. There is always a translator between the two. Every translation costs heat, and that is where your fifteen percent starts.
• A solar panel that produces AC does not exist. A solar cell is a diode: the built-in field pushes every electron the same way. One direction, always.
• AC beat DC because you can transform it. Loss in a cable scales with the square of the current, so you transport at high voltage and low current. Edison needed a booster station every few kilometres.
• A plug-in battery sits around 48 volts for one reason. Sixteen LFP cells in series come to 58.4 volts when full, and 60 volts DC is the ceiling below which an ordinary person may touch something unprotected. Seventeen cells would break it. A fixed installation put in by an electrician runs at several hundred volts instead — a different class, not a bigger version of the same thing.
• That safety is not free. Lower voltage at the same power means proportionally more current, and copper loss scales quadratically. Halve the voltage and you quadruple the loss in the same wire.
• There is no electricity inside your battery. There is a chemical state. Charging pushes ions to one side of the cell, discharging lets them run back, and that movement is your current.
• On my own units I measure 83 to 88 percent round-trip. That spread has two different causes, and almost everyone conflates them. Below they are pulled apart.

Your house and your battery do not speak the same language

There is one thing you need to know, and it is not complicated.

Your wall socket delivers alternating current. Your battery delivers direct current. No battery on earth stores AC, and no socket in the Netherlands delivers anything but AC.

So there is a translator between the two. Every single time. And that is where the leaking starts.

That is the entire architecture of a home battery, and it explains why there is a cabinet full of electronics around it instead of just a box of cells.

Why a solar cell has no choice

Start at the source. The roof.

A solar cell is a semiconductor with a p-n junction: two layers of silicon doped differently, with a built-in electric field between them. Light knocks electron-hole pairs loose. The field pushes the electrons towards the n-side and the holes towards the p-side, and that asymmetry forces the current through the external circuit in one direction.

A solar cell is therefore a diode that generates current. AC from a panel does not exist; it cannot exist. What comes down from your roof is direct current, and your house can do nothing with it.

Volts, amps and watts, once and properly

Three words that come back throughout this piece.

Volt is the pressure doing the pushing. Ampere is how much passes per second. Pressure times flow is watt: the power, what the thing is doing right now.

The image that goes with this is water in a pipe, and I use it once. It breaks immediately too: water gets consumed, charge does not. Charge comes back. That is exactly why there are always two wires running to everything — out and back — and why a bucket of water works as a picture but a bucket of electrons does not.

AC exists for one reason: you can transform it

Fifty hertz means the direction reverses fifty times per second. Not that the current travels fifty times faster, and not that anything shuttles back and forth fifty times. The direction flips.

Why would you build an entire country on electricity that changes its mind fifty times a second, when DC just goes straight?

Because you can transform AC.

Power loss in a cable scales with the square of the current. To move the same power a hundred kilometres, you raise the voltage; at equal power that comes with a proportionally lower current, and the loss collapses. A transformer can only convert that voltage in a changing field. DC voltage does not step up or down so easily.

That is what broke Edison's DC grid: booster stations every few kilometres and heavy copper, against an AC grid that spanned hundreds of kilometres.

Remember that law, because it comes back on a scale of about a metre and a half. Same law, twice, at two scales: on the transmission grid and in the cable to your own cabinet.

A kilowatt is not a kilowatt-hour

This is where it goes wrong most often, and it costs people money.

Kilowatt is how far the tap can open. Kilowatt-hour is how much is in the bucket.

A ten kilowatt-hour battery that only delivers 800 watts will not run your kettle. The bucket is plenty big, the tap is too small. It works the other way round too: plenty of power on a small bucket is empty within half an hour.

And this bucket leaks. A home battery draws 5 to 8 watts at rest, day and night, whether you use it or not. That is the point where the bucket gives up as an image, and it is no accident that it is also loss number four further down this piece.

Why a plug-in battery sits at 48 volts — and a fixed installation does not

This is the nicest number in the whole video, and you can check it on one line.

An LFP cell — lithium iron phosphate, the chemistry nearly every home battery uses — has a nominal voltage of 3.2 volts. Fully charged that is 3.65 volts. Empty, 2.5 volts is the floor you do not want to go below.

Sixteen times 3.2 is 51.2 volts. That is what the box calls "48 volts". Full: sixteen times 3.65 is 58.4 volts.

Why sixteen? Why not seventeen?

Because 60 volts DC is the ceiling below which an ordinary person may touch something without protection. Below it, the thing is allowed to be a consumer product: something you unbox and install yourself. Seventeen cells would sit at 62 volts when full, over the line.

58.4 against 60. A little over one and a half volts of headroom. The entire existence of the plug-in battery as a category hangs on that margin.

48 volts is a class, not a law

This is where the comments reliably go wrong, so let me be explicit: 48 volts applies to plug-in batteries and LV stacks, not to every home battery. The moment an electrician installs the system, the reason to stay under 60 volts disappears — nobody needs to be able to put their hands on it. Those systems therefore run at several hundred volts DC.

That second reason is not a detail: loss in a cable is I²R, so current counts twice. Four times the voltage at the same power is four times less current and sixteen times less loss in that same copper. That is exactly why an installed system accepts the insulation monitoring, the arc-fault detection and the electrician that come with it — it gets efficiency and thinner cables in return. The plug-in battery trades all of that away for the right to stand on your floor without a fitter.

How an HV system reaches those hundreds of volts

There are three layers to this, and they get conflated constantly — including by me, until I put the datasheets side by side.

Layer 1 — cells in series. This is the only layer that makes voltage. An LFP cell is 3.2 volts and that never changes. Hundreds of volts exists only when hundreds of cells sit end to end. Sixteen cells is 51.2 volts, thirty-two is 102.4, eighty is 256.

Layer 2 — where that series chain ends differs per manufacturer. This is where most explanations go wrong. There is no standard answer:

With BYD the high voltage arises outside the module, with AlphaESS inside it. Both are HV, both land at hundreds of volts, and the build is entirely different.

Layer 3 — expansion goes in parallel. Almost always. That is not an arbitrary choice but the weakest-link problem. In a series chain the same current flows through every element, so the weakest one decides what the whole chain can do. Add a new module in series to a four-year-old set and the new one drops to the oldest one's window. Parallel does not have that: each pack keeps its own voltage and delivers what it can.

Which is why Pylontechs run up to sixteen in parallel, AlphaESS puts its 256 V modules 1 to 6 in parallel, and BYD parallels whole towers rather than stacking a sixth module — even though the modules inside that tower do sit in series.

The newest answer to the same problem is in the Sigenergy SigenStor: every pack gets its own DC-DC optimizer. With power electronics per pack, mismatched packs can hang side by side without the weakest dictating terms to the rest. It is the same trick as a per-panel solar optimizer, applied one layer down.

So what survives of the plug-in battery rule? Exactly the same rule. There too you expand in parallel: voltage stays at 51.2 volts, amp-hours are added. The difference with HV is not the direction of expansion, but how tall the series chain was before you started.

The price of all these models is the BMS. Each module watches its own cells, with a controller above them watching the modules against each other — a master-slave arrangement instead of the single BMS that does everything inside a plug-in battery.

The standard that applies, and the standard that does not

The video names no standard on purpose: the moment a number like that drops, the conversation becomes about the number. Here it belongs, because it will absolutely come up in the comments — and the standard most people cite is the wrong one.

The 60 volts does not come from IEC 61140. That standard puts extra-low voltage at 50 volts AC or 120 volts DC. So anyone waving IEC 61140 in the comments is right that 60 volts is not in it.

Where the 60 volts does come from is IEC 62368-1, the safety standard for consumer electronics. It sets the boundary of class ES1 — an energy source an ordinary person may touch unprotected — at frequencies below 1 kHz at 30 volts RMS, 42.4 volts peak and 60 volts DC. That is the ceiling a device has to stay under to be allowed to be a consumer product, and it lines up exactly with the 58.4 volts above.

That third row explains why the number is so persistent: two different routes land on the same figure. But the route that runs through your home battery is ES1.

What happens if you want to go higher

Above that ceiling the product changes category. An installer system at a few hundred volts gets insulation monitoring, arc-fault detection and a technician along with it. Not out of manufacturer caution, but because the voltage itself triggers a different safety regime.

Which brings up a misconception that comes up often: if 60 volts is the ceiling, why not just stack more packs for more storage?

You do. Just not in series.

Two packs in series would give 116.8 volts and break the ceiling. Two packs in parallel keep the same voltage and double the amp-hours, and that is exactly what an expansion module does. No separate source needed for that; it follows from the arithmetic above.

What low voltage costs you

Back to that law: the square of the current.

A plug-in battery sits at 51 volts. An installer's fixed system often sits at a few hundred. At the same power, lower voltage means proportionally more current, and loss in copper scales with I²R.

Halve the voltage and you double the current, which quadruples the loss in the same cable.

So that safety is not free. You pay for it in heat. It is not a design flaw and not a bad manufacturer: it is the price of the category, and it follows directly from the ceiling in the previous section.

The image that does hold here is friction: the same litres through a narrower hose create more friction, and that also scales roughly quadratically with flow. This is the one place in this piece where the water analogy stays exactly right instead of approximately right.

There is no electricity inside your battery

The pack itself is the least understood part of the whole device.

You are not storing electricity. You are storing a chemical state.

Charging pushes lithium ions to one side of the cell. Discharging lets them run back, and that movement is your current. There are no electrons sitting in a bucket waiting. There is a chemical equilibrium you shift and let fall back, thousands of times.

That also explains why a battery's voltage changes as it empties: that voltage is a chemical state you are reading, not a level in a tank.

A BMS is not a doorman

Next to the cells sits a BMS, a battery management system. The image everyone uses is the doorman: it measures every cell separately — voltage, current, temperature — and pulls the plug the moment something goes outside the lines. Overvoltage, undervoltage, overcurrent, short circuit, an unsafe temperature.

But guarding is not all it does, and that is where the image breaks. A doorman does not balance cells. A BMS does.

Cells drift out of step over their lifetime. Without balancing, the weakest cell decides when the whole pack has to stop: it hits its upper or lower limit first, and the rest is left holding usable capacity. Balancing pulls them back into line.

That is not a footnote. That is a tenth to a fifth of your storage you simply cannot use without a BMS.

Back outside, in step with the grid

Evening. The sun is gone, the kettle goes on.

Third translation: DC out of the pack, AC into your house. And it has to run exactly in step with the grid.

The Dutch low-voltage grid runs at 50 hertz, and NEN-EN 50160 requires the frequency to stay between 49.5 and 50.5 hertz for 99.5 percent of the measured time. If your inverter falls out of step, it starts fighting the grid instead of joining it. So it watches the outside continuously and lays its own rhythm alongside.

Voltage has a comparable band, and it is less symmetrical than people assume.

In the video I only say "230 volts, and in practice that swings by a few percent", and that is deliberate. "Plus or minus ten percent" is not quite right: the lower bound is wider than the upper one. Anyone with an inverter probably knows the top of that band from their shutdown notifications, and that is a completely different subject from this one.

Two routes, and why the count does not add up if you are counting along

The journey does not run the same way for everyone.

If you were counting along: that is two or three conversions, but in both cases two of them are on your battery's tab. On the first path, the step from panel to house happens anyway, battery or no battery. That inverter runs regardless, so its loss does not belong to the battery.

Look at the second row. Charging from the grid during the cheapest hours is precisely the whole point for a lot of people, and then there is no panel step at all. I deliberately say "cheapest hours" and not "overnight": with the amount of solar on the grid today, the cheapest hour is often in the middle of the day, and a windless winter night can be expensive. Which hours those are shifts by season and by day.

Finding the fifteen kilowatt-hours

Now the settlement. Four places, and two of them are in the same box.

For me that adds up to between 83 and 88 percent round-trip.

Call it fifteen percent lost, somewhere between twelve and seventeen. The hundred-against-eighty-five in the hook is a rounding, and this is the real band.

The spread is not sloppiness — and it has two causes you must keep apart

This is where nearly every explanation of efficiency goes wrong, including my own earlier version.

If I read the efficiency out while the battery is halfway through discharging, I see 83 percent. If I let it run down to ten percent state of charge, that same figure climbs to 88.

Cause one is a measurement artefact, and it works within a single cycle. The charge is already fully booked, the discharge is still in progress. The denominator is done, the numerator is still climbing. Read it out halfway and you are measuring half a discharge against a whole charge. The number is simply lagging reality.

Cause two is real, and it works over the lifetime. Standby draw is a fixed number of watts. The more kilowatt-hours you push through, the less it costs you in percentage terms. A battery that works deep and often is, over months, more efficient than the same device sitting idle. Same device, different number.

Those two get thrown on one pile constantly, and it cannot be right: a lifetime counter does not jump five points inside a single discharge because of 5 to 8 watts of standby draw. The denominator is far too large for that. The first explains why the number is low right now. The second explains why deep cycling really is better in the long run.

So those fifteen kilowatt-hours are not gone. They are hanging as heat in a few appliances and a few metres of copper. You paid for them. You just did not get them back as electricity.

What is deliberately not in here

This piece is about the mechanism. A few things will certainly come up in the comments, and they belong elsewhere:

• Whether it pays off. Payback period, prices, what size you need. That is a calculation with your consumption, your contract and your roof in it. Start with Is a home battery worth it in 2026? and Your home battery is too big.
• The 800 watt limit, registration rules and net metering. Regulation, not physics. See The end of the Dutch net metering scheme.
• Plug-in versus a fixed system, and AC versus DC coupling. Deserves its own treatment; only the correct core of it is above.
• Backup power and island mode. Grid-forming versus grid-following is a different story with different conditions.
• Sodium and LMFP. Different chemistry, different piece.

Frequently asked questions

The video

The accessible version of this story is on Jay's Desk: eleven minutes, one kilowatt-hour followed from the roof to the kettle, with the cabinet itself on screen. No formula in sight.

This article is the complete version: the cell voltages, the standards, the measurement conditions and the sources.

Transparency

No sponsorship, no affiliate links, no brand preference. Nothing was lent, given or paid for this video or this article.

The efficiency figures come from plug-in batteries I use myself. They are Zendure units I received as review models during my time at Thuisbatterij Nederland; there is no arrangement of any kind left, and Zendure does not know this piece exists. They are here as measurement equipment, not as a recommendation — this article reviews no product.

Sources

All sources accessed on 12 August 2026. The efficiency figures are my own measurements and therefore not externally verifiable; the measurement conditions are stated explicitly above so you can check them against your own setup. If you land on something different, or you find an error — let me know and this article gets updated.