Cellmetry research
You’re wasting electricity by charging slower
With AC charging, the slower you charge, the more you lose. Measured on 19 Teslas.

To me this was somewhat obvious. I just didn’t know how bad it was until I measured it. While a Tesla charges, Cellmetry shows two numbers side by side: the power arriving from the charger, and the power actually going into the battery. The gap between them is the charging loss.
I normally charge at 11 kW, at home or at work. This summer I plugged into our summer cabin’s ordinary Schuko outlet instead, and to play it safe I kept the power down. I was charging at 1.8 kW, but only 1.5 kW was reaching the battery. That’s 17 % of the energy I paid for, gone.
Where it goes
The obvious suspect is the on-board charger, the part of the car that turns AC from the wall into DC for the battery. It does lose some: about 3.2 % of whatever comes in. But that is not where most of a slow charge disappears.
The biggest factor is what’s normally the smallest: the car being awake. It can’t sleep while it charges, and across the 19 Teslas in this analysis there is a fixed cost of about 199 W that no charge rate avoids. Most of it is the car keeping its pumps, its electronics and the computer that looks after the battery running. Cars may differ a little, but the car reports charging power in 100 W steps, and the differences between cars are no larger than that, so this data cannot tell them apart. If you were charging at 1 kW, that alone would be 20 % gone.
So an 11 kW wallbox puts 95.0 % of the energy into the battery, and a household socket 82.6 %.
Lower watts, more hours
When I shared this on LinkedIn, the objection I heard most, several times from engineers, was this: at 1.4 kW the loss is 244 W, at 11 kW it is 554 W, so isn’t the slow charge losing less? Per hour, yes. But you don’t pay per hour, you pay per kWh, and the cost of being awake runs for as long as the charge lasts.
Put 10 kWh into the battery. From a socket at 1.4 kW that takes about 8.6 hours, and 244 W for 8.6 hours is 2.1 kWh lost. On an 11 kW wallbox it takes under an hour and loses 0.5 kWh. The same energy in the battery, about four times the waste.
What it costs in a year
Say you drive 20,000 km a year at 16.1 kWh/100 km. That is 3,220 kWh into the battery. At 15 c/kWh:
| Charging at | From the wall | Per year |
|---|---|---|
| 1.4 kWHousehold socket | 3,899 kWh | 585 € |
| 3.7 kWWallbox, single-phase | 3,523 kWh | 528 € |
| 11 kWWallbox, three-phase | 3,391 kWh | 509 € |
About 76 € a year between a socket and a three-phase wallbox. Not a fortune, but not nothing. And on a spot-price contract, a charge that takes under an hour instead of eight and a half also fits into the cheapest hours of the night.
What this doesn’t tell you
- Whether all of the fixed part is the car being awake. Some of it could be the on-board charger’s own standing loss, and a charging session alone cannot separate the two. What points to the car: parked and awake with no charger connected, the median car here draws 193 W from its own battery, almost exactly the fixed part. And Sentry Mode, a load of the car and not of the charger, adds 45 W on top of it.
- Where the best rate is. Above about 7 kW the curve is nearly flat: 7.4 kW and 11 kW are 0.9 percentage points apart. The fit is a straight line in watts, so it cannot show whether there is a sweet spot in between, which rising resistive losses would eventually create. The readings at the highest rates do sit slightly above the line (see the data at the end). The big step is the one from a socket to a wallbox.
- Whether slow charging is better for the battery. This measures energy, not battery health, so here I lean on other people’s work. The closest evidence is a study by Idaho National Laboratory that compared Nissan Leaf batteries charged at 3.3 kW AC with ones charged at 50 kW DC, about fifteen times faster. Whole packs charged fast did lose more capacity, 28.1 % against 23.1 %, which the authors link to the fast-charged pack running hotter. With the temperature held equal, individual cells differed by less than 1 %. Recurrent, looking at 13,000 Teslas in the US, found no statistically significant difference in range loss between cars that fast-charged more than 70 % of the time and cars that did so less than 30 %, though its measure is the car’s own range estimate rather than a capacity test. And in a fleet analysis of 10,000 EVs, Geotab found that cars charged at Level 1 and at Level 2 differed, but not by a statistically significant amount.No study I found compares 1.4 kW with 11 kW directly. But if fifteen times the charging rate barely matters once temperature is equal, the step from a socket to a wallbox is unlikely to. Temperature and time spent at a high charge level are the better places to look.
- Winter. This is summer data, with cabin climate and battery heating left out. In the cold the car also heats its battery while it charges, so a slow charge is likely to lose more. I’ll measure it when it gets cold.
So
At AC rates, charging slower mainly costs you energy. Charge at the fastest rate you have. A socket is fine when it is the only option; just know what it costs. What slow charging really affects is your wallet.
Appendix
Every rate
| Charge rate | Lost | Into the battery | For 10 kWh in |
|---|---|---|---|
| 1.4 kWHousehold socket, 6 A | 244 W | 82.6 % | 12.1 kWh |
| 2.3 kWHousehold socket, 10 A | 273 W | 88.1 % | 11.3 kWh |
| 3.7 kWWallbox, 16 A single-phase | 318 W | 91.4 % | 10.9 kWh |
| 7.4 kWWallbox, 32 A single-phase | 438 W | 94.1 % | 10.6 kWh |
| 11 kWWallbox, 16 A three-phase | 554 W | 95.0 % | 10.5 kWh |
Loss = 199 W + 3.2 % of the charge power.
The data
How it was measured
- It is a subtraction. The car reports the power arriving from the charger, and separately the power going into the battery. The loss is the difference.
- Only quiet, steady moments count. Car parked, charge rate steady, cabin climate and battery heating off; moments with any of those running were left out. Sentry Mode was tracked separately. 138,007 such readings from 19 cars, 10 July 2026 to 10 September 2026.
- Temperature and charge level made no measurable difference. Not in this summer data. Winter may change that.