How to Know If Your Tesla Is Balancing Its Battery

BatterMachine, August 2026.

Below are charging sessions from two Teslas, recorded in Dr.EV. Both cars charge to nearly full. The first graph is smooth. In the second, the highest cell voltage line turns into the teeth of a saw. That sawtooth is the sign to look for: it means the car is balancing its cells, and you can watch it happen live in the graph.

A balanced pack charging. The highest and lowest cell voltages rise together and finish smoothly, with no teeth.

An unbalanced pack charging. Near full, the highest cell voltage line turns into a sawtooth, and the teeth continue for hours.

This article has two parts. Part 1 is for drivers and takes a minute to read. Part 2 is for engineers who want to know where the sawtooth comes from; drivers do not need it at all.

Part 1. For drivers

The sawtooth is cell balancing. The car has found some cells holding more charge than the rest and is bringing them back in line with the others. It is not damage, and it is not a fault in the charger.

Many drivers believe the Supercharger balances the cells, since it is the most powerful charger the car ever sees. The truth is the opposite. Balancing is very slow and needs many hours to finish, and a Supercharger session ends after about half an hour, long before the work is done. A car that only ever fast charges starts balancing at the end of every session and never gets to finish, so the sawtooth comes back larger in every charging graph. Slow charging at home is what gives the car the hours it needs.

So if your charging graph shows the sawtooth, charge at home or on any slow AC charger, let the charge run all the way to your set limit, and then leave the car plugged in for several hours more. Overnight is the easiest way to do all three at once. Do this regularly, not once: a single slow charge removes only part of a large imbalance, and it is the habit that keeps the cells together. Fast charging on the road stops mattering once the pack gets its slow hours at home.

That is everything a driver needs. The rest of the article is for engineers, and you can stop reading here.

Part 2. For engineers

What the two voltage lines measure

A Tesla pack is built from groups of cells welded together in parallel, and these groups are then connected in series. The battery management system does not measure individual cells. It measures the voltage of each group, because the cells inside a group are bonded together and share one voltage. The two lines in the charging graph are the highest and the lowest of these group voltages. When all groups hold the same amount of charge, the two lines sit close together and rise as one.

Groups drift apart slowly over time, because no two groups have exactly the same self-discharge or the same temperature history. Charging cannot correct the drift: the groups are in series, so every group receives exactly the same current. The only way to level them is to drain a little charge out of the fullest groups, and that is what the balancing circuit does.

The balancing circuit

Each group has a discharge resistor and a switch across it. Closing the switch lets a small current bleed out of that group alone, through the resistor, as heat. This is called passive balancing, and it is standard in production battery monitor chips.

We do not know which chip Tesla uses. As a documented example of how the industry builds this circuit, take the Texas Instruments BQ79616, an automotive 16-channel battery monitor with the balancing switches built into the chip. The figure below is TI's own schematic of the circuit.

The balancing circuit of the TI BQ79616. The dashed loop is the balancing current: out of the cell tap, through the external resistor RCB into the CB pin, through the switch QCB inside the chip, and back to the next tap. The voltage measurement enters separately, through the RVC filter into the VC pin. Source: Texas Instruments, BQ79616 datasheet SLUSF21A, Figure 6-13.

Reading the schematic from left to right: each cell group connects to the chip twice. One path goes through the filter resistor RVC to the VC pin, and this is the measurement path that feeds the voltage ADC. The other goes through the balancing resistor RCB to the CB pin, and behind each CB pin sits a transistor switch QCB inside the chip. When the chip closes QCB, the balancing current, drawn as the dashed loop, flows out of the group's positive tap, through RCB, through the switch, and back into the tap below. The group discharges through the two resistors, and everything the current passes through, the resistors and the switch, turns that charge into heat. The right-hand half of the figure shows the case where two adjacent switches are on at once, which changes the current path and is why the chip manages adjacent channels carefully.

The datasheet (Texas Instruments, BQ79616 16-Series Battery Monitor, Balancer, and Integrated Hardware Protector, SLUSF21A, June 2023, revised June 2026) specifies how the circuit is operated:

• a balancing current of 240 mA at 75 °C ambient, with higher current allowed when the board is cooler;

• balancing that runs on its own once started, switching between odd and even channels on a configurable duty cycle, because two adjacent switches share a connection and cannot run at once;

• a stop rule based on voltage: balancing on a channel ends when that group's voltage falls below a programmed threshold;

• thermal protection: balancing pauses automatically when the chip die exceeds a nominal 105 °C and resumes after it cools, since the switches and resistors dissipate real heat.

The picture to keep is a quarter of an ampere, switched on and off by rules like these, against a group that a Supercharger feeds with several hundred amperes. The balancing current is roughly two thousand times smaller than the charging current.

Why the switching draws teeth

While a group's switch is closed, the measured voltage of that group sags by a few tens of millivolts. When the switch opens, the reading springs back, and charging lifts the group further. Each close-and-open cycle draws one tooth: the falling edge is the drain switching on, the rising edge is the drain switching off. The switching is fast, but the draining is slow, so the teeth run on for hours and stop only when the fullest group has been drained level with the rest. The balanced car shows no teeth because no group stands above the rest, so its switches stay open.

The sawtooth up close. Each tooth is one cycle of the drain switch: the voltage sags while the drain is on, and springs back when it switches off.

The size of the tooth deserves a closer look, because the cells themselves cannot explain it. A group of tens of parallel cells has a resistance well below a milliohm, so 240 mA flowing out of it moves the true group voltage by well under a millivolt. What moves the reading is the wiring: the balancing loop and the measurement path hang off the same tap wire, so while the drain is on, the drained group's reading drops by the voltage lost in that shared wiring. That drop is what makes the line dip at all. But the line in the graph is the maximum over all the groups, so it can only fall as far as the second-highest group before it starts tracking that group instead. The depth of a tooth therefore measures how far the runner-up group sits below the top one. Our fleet data confirms this: teeth deepen as the pack approaches full, because the charge curve steepens there and a fixed surplus of charge turns into a wider voltage gap, and they shrink and fade out as balancing drains the top group level with the rest. The fading is worth knowing: teeth that get shallower and then disappear are balancing finishing its job.

Two checks in our fleet data rule out the other explanation drivers sometimes reach for, that the charger is cutting its current. The pack current trace shows no oscillation matching the teeth. And even if it did, charging current flows through the group's sub-milliohm resistance, so swings of tens of amperes would move the reading by a few millivolts, not the 30 to 50 mV the teeth actually span.

The arithmetic of why fast charging cannot balance

Take a long-range pack as an illustration: 46 cells in parallel per group, roughly 220 Ah of group capacity. If one group holds just 1% more charge than the rest, the surplus is about 2.2 Ah. A balancing current in the 240 mA class needs around nine hours of switch-on time to drain it. A half-hour Supercharger session, even if balancing ran throughout, removes about 0.12 Ah, a twentieth of the job. The session ends, the car drives off, and the surplus is still there, which is why the sawtooth comes back larger in every charging graph of a car that only ever fast charges. An overnight AC charge is the only common situation where the pack gets the uninterrupted hours the arithmetic demands.

Reading the graph

In our fleet data the teeth behave exactly as this mechanism predicts. They appear when one group stands above the rest, run for hours of slow charging, and stop when the groups are level. A brief jump in the cell voltage difference number during charging means little on its own; isolated single samples come and go in normal data. What identifies balancing is the shape, teeth repeating for hours, and what identifies a healthy outcome is that after enough slow charging the teeth are gone.