Five myths about charging lithium-ion batteries, and what actually wears them out

Reviewed: September 2026

The short answer: what actually wears out a lithium-ion battery

Heat, not amperage, is what makes fast charging damage battery capacity, and inside a phone that stays near room temperature the loss is small enough to disappear into ordinary ageing. The things that genuinely shorten a cell’s life are duller than the folklore: hours spent sitting at a high state of charge, hours spent hot, occasional deep discharges, and the plain passage of time.

A lithium-ion cell is a rechargeable electrochemical cell that stores energy by shuttling lithium ions between a graphite anode and a metal-oxide cathode. Nothing about that process is reversible with perfect efficiency. Every charge leaves a little lithium locked up in places it cannot come back from, and the rate at which that happens is set mostly by temperature and voltage, not by how many amps the charger claims on the box.

Two clocks run at once. One counts cycles: each full charge and discharge moves lithium in and out of the electrodes and costs a small amount of usable capacity. The other counts days. A cell parked in a drawer at 90% charge loses capacity without doing any work at all, which is why a lightly used two-year-old phone can show worse battery health than a heavily used one-year-old phone. Most articles about charging habits ignore the second clock entirely, and it explains a large share of the “why is it worse than it was” question.

Here is the whole article in one table.

Myth	Verdict	What actually happens	Named source
1. Fast charging destroys your battery	True only when the cell is cold or already hot	At moderate temperature the charging circuit tapers current before damage occurs; at 0 °C a high charge rate plates metallic lithium onto the anode	Yang et al., PNAS 2018; Gao et al., Frontiers in Energy Research 2022
2. You should always charge to 100 percent	False	Higher upper voltage buys capacity today and costs cycle life; every major phone maker now ships a way to stop short of 100%	Battery University BU-808; Apple; Samsung; Google
3. A new phone needs three full charge cycles to calibrate	False	The idea is a nickel-cadmium habit; lithium-ion has no memory effect, and modern gauges learn from partial cycles	Battery University BU-807 and BU-415; Texas Instruments SLUA364B
4. Leaving it plugged in overnight damages the battery	True only if the phone stays hot or sits at full charge for many hours	The charger terminates and switches off, then restarts only after the cell drops roughly 100 mV; the real cost is time spent at high voltage, not “overcharging”	Texas Instruments BQ25170 datasheet; Google Pixel Help
5. Draining to zero is good for the battery	False	Deep discharge is the single worst routine habit on this list; below about 2.50 V per cell the protection circuit latches off	Battery University BU-415 and BU-802c

How a lithium-ion cell degrades

Two mechanisms account for most of the capacity a phone battery loses: a passivation film that thickens on the anode over time, and metallic lithium that deposits on the anode surface when charging is pushed harder than the chemistry can absorb. Both consume lithium that would otherwise be available to store your charge, and both speed up with temperature.

A third mechanism, rising internal resistance, does not remove capacity but changes how the phone behaves. Internal resistance is the opposition a cell presents to current flow, measured in milliohms. As it climbs, voltage sags harder under load, and a phone can shut down at an indicated 15% because the processor asked for a burst of current the aged cell could not supply at a usable voltage.

SEI layer growth and what feeds it

The solid electrolyte interphase, or SEI, is a thin film that forms on the graphite anode during the very first charges and keeps growing slowly for the rest of the cell’s life. It is not a defect. Without it the electrolyte would decompose continuously against the anode and the cell would be dead in weeks. The problem is that the film is built out of lithium taken from the working inventory, so every nanometre of growth is capacity that never comes back.

Growth is diffusion-limited, which in practice means it slows as the film thickens. Amelie Krupp and colleagues, publishing in the Journal of Energy Storage in 2021, stored commercial 64 Ah NMC graphite pouch cells for up to 422 days and fitted the fade to a power law. Over the first 220 days the exponent came out at 0.581, close to the square-root-of-time behaviour expected from diffusion-limited SEI growth; across the full 422 days it drifted to 0.789. So the textbook square-root law is a good first approximation and not a law.

What feeds SEI growth is temperature and anode potential, which is to say heat and high state of charge. In the same study, cells held at 23 °C and 50% state of charge lost roughly 0.1-0.3% of capacity over more than 400 days. Cells at 40 °C and 90% state of charge reached about 1% loss by day 200 and then curved upward. Those are pouch cells in a laboratory, not a phone in a pocket, and the absolute numbers should not be carried across. The ranking is what transfers: warm and full is worse than cool and half empty, every time.

Lithium plating and when it happens

Lithium plating is the deposition of metallic lithium on the anode surface instead of intercalation into the graphite, and it happens when lithium ions arrive at the anode faster than the graphite can absorb them. Push enough current, or chill the cell enough that diffusion slows down, and the anode potential drops below 0 V versus Li/Li+. At that point plating becomes thermodynamically favourable and metal starts building on the surface.

Some of that metal reconnects on the next discharge. Some of it does not, and what stays behind is doubly expensive: it is lithium removed from inventory, and it is fresh reactive surface area that grows new SEI. In severe cases the deposits form dendrites that can eventually reach the separator, which is the failure mode behind the safety warnings on cold-weather charging.

The numbers here are brutal and specific. Xiao-Guang Yang and colleagues at Penn State, writing in PNAS in 2018, cycled conventional commercial cells at a 3.5 C charge rate at 0 °C. The baseline cells lost 20% of capacity in 50 cycles. A cell design that self-heats before accepting the charge ran 4,500 cycles under the same nominal conditions with less than 20% loss. Same charge rate, same ambient temperature, two orders of magnitude difference in life, and the only variable that mattered was the temperature of the electrodes at the moment the current went in.

Gao and colleagues, modelling an NCA graphite cell in Frontiers in Energy Research in 2022, put a figure on the short-term cost: charging at 3-5 C at 0 °C cost nearly 10% of available capacity to plating alone in the early cycles. Their model also showed plating dominating the first handful of cycles and SEI thickening taking over afterwards. At 25 °C and 1-2 C, their simulation showed minimal plating.

Why heat is the common factor behind both

Heat accelerates SEI growth directly, and heat is also the thing that decides whether a given charge rate causes plating. Both mechanisms follow Arrhenius kinetics, which means their rates rise roughly exponentially with temperature.

The awkward part is that the two effects point in opposite directions for cold weather. Cold slows SEI growth, which is why storing cells cold is good advice. Cold also slows lithium diffusion into graphite, which is exactly what causes plating during charging. A cell at 0 °C is ageing slowly and is simultaneously in the worst possible state to accept a fast charge. Battery University treats anything above 30 °C as elevated temperature for a lithium-ion cell, and Apple’s own guidance puts the safe ambient operating window for iPhone at 0-35 °C, warning that above 35 °C battery capacity can be permanently damaged.

Named limitation on all of the above: none of these figures were measured on a phone. They come from cell-level and pack-level testing where researchers control temperature to a degree that a device in a pocket never achieves.

Myth 1. Fast charging destroys your battery

True only when the cell is cold or already hot. At a comfortable cell temperature, a charging system that is doing its job will not let you damage the battery with a fast charger, because the system decides the current, not the charger.

This is the point that gets lost. A charger advertising 65 W is offering an upper bound. The phone’s battery management system, a controller that monitors cell voltage, current and temperature and sets the charging profile accordingly, negotiates what it will actually take. Every mainstream fast-charging implementation front-loads the high current into the low-state-of-charge region, where the anode has plenty of empty sites and plating risk is lowest, then tapers hard above roughly 60-80%. The advertised wattage describes the first ten minutes, not the whole charge.

What fast charging actually does to cell temperature

Two heat sources get confused constantly, and separating them is the most useful thing in this article.

The first is ohmic and entropic heating inside the cell, driven by the current flowing through internal resistance. This is the heat that fast charging genuinely creates, it scales with the square of the current, and it is highest early in the charge when current is highest.

The second is everything else: ambient temperature, direct sun, a thick case, a wireless charging pad’s coil losses, and the phone’s own processor if you are gaming while plugged in. This second category routinely dominates. Google’s Pixel documentation states plainly that a phone warms up while plugged in and advises against leaving it charging longer than needed, and Apple warns that charging inside certain styles of cases generates excess heat that can affect battery capacity.

There is no publicly available primary measurement of phone cell temperature against charging power that we could locate, which means the exact crossover point where a faster charger meaningfully raises cell temperature in a specific phone is not confirmed. What is confirmed is the protection layer. Charger controllers monitor cell temperature through a thermistor and cut charging outside a window, typically 0-45 °C in the Texas Instruments BQ25170 reference design, and they reduce charge current when the die reaches its thermal regulation limit. Google documents this behaviour at the product level too: a Pixel charging continuously for hours at high temperature limits itself to roughly 70-80%.

When it does cause damage

Cold is the real danger case. Charging a cell below about 5 °C at a high rate is where plating stops being a modelling curiosity. In the PNAS 2018 work a conventional 9.5 Ah cell charged at 3.5 C at 0 °C lost 20% of its capacity in 50 cycles. Those are EV-cell conditions, not phone conditions: a phone charges at roughly 1-2 C and its controller refuses the charge outright when the cell is that cold. The direction transfers, the magnitude does not. In a phone this is mostly a car-in-winter problem: the device has been outside for hours, you plug it into a car charger, and the cell is still near ambient. Most modern phones detect this and refuse or throttle the charge, which is the feature working, not the phone misbehaving.

The second damage case is stacked heat. Fast charge, in a case, in sunlight, while running navigation. Each of those alone is survivable. Together they push cell temperature into the range where SEI growth accelerates measurably.

The honest limitation on the “fast charging is fine” advice: it assumes the phone’s thermal management and charging circuit are working correctly and that the cable and adapter negotiate a standard protocol properly. A damaged cable or a non-compliant adapter that ignores negotiation is a different situation, and one this article cannot assess for any specific product. Counterfeit adapters are their own failure mode, and we cover the physical tells in how to spot a fake laptop power supply.

Myth 2. You should always charge to 100 percent

False. Charging to 100% is convenient and gives you the full capacity you paid for, but it is measurably harder on the cell than stopping short, and every major phone manufacturer now ships a setting that stops short on purpose.

The mechanism is voltage, not percentage. A phone showing 100% is holding its cell at or near its maximum design voltage, commonly 4.20 V or 4.35 V per cell depending on the chemistry. High anode potential accelerates the side reactions that grow SEI, and it does so for as long as the cell sits there. Jeff Dahn, quoted by Battery University, puts it directly: a voltage above 4.10 V per cell at elevated temperature causes a demise that can be more harmful than cycling a battery.

What upper voltage limit does to cycle life

One or two sentences cannot carry this; the numbers can. Note carefully that the sources below define cycle life differently and were measured under different conditions, which is exactly why the absolute numbers disagree while the direction does not.

Upper charge limit	Cycles reported	End-of-life definition	Stated test conditions	Source
4.20 V/cell (100% of stored energy)	300-500	not stated	not stated	Battery University BU-808, Table 4
4.13 V/cell (90% of stored energy)	400-700	not stated	not stated	Battery University BU-808
4.06 V/cell (81% of stored energy)	600-1,000	not stated	not stated	Battery University BU-808
4.00 V/cell (73% of stored energy)	850-1,500	not stated	not stated	Battery University BU-808
3.92 V/cell (65% of stored energy)	1,200-2,000	not stated	not stated	Battery University BU-808
3.85 V/cell (60% of stored energy)	2,400-4,000	not stated	not stated	Battery University BU-808
4.15 V/cell, single 18650 cell	1,500	above 81% of initial capacity	0.3 C charge, 1 C discharge, 25 °C	Panasonic NCR18650BD-Improved datasheet
Device level, iPhone 14 and earlier	500	80% of original capacity	“under ideal conditions”, not further specified	Apple
Device level, iPhone 15	1,000	80% of original capacity	“under ideal conditions”, not further specified	Apple
Device level, iPad, MacBook, Apple Watch	1,000	up to 80% of original capacity	not specified	Apple

The Battery University rows carry a rule of thumb worth remembering and a caveat worth remembering harder. The rule: every 0.10 V drop below 4.20 V per cell roughly doubles cycle count while holding less capacity. The caveat: that table is published without test conditions, no temperature, no discharge rate, no depth of discharge, so it shows a shape rather than a set of numbers you can plan around. The Panasonic datasheet row is the opposite case, a fully specified measurement of one cell type that tells you nothing about the cell in your phone.

Depth of discharge moves the same lever from the other end. Battery University’s figures for NMC chemistry give roughly 300 cycles to 70% capacity at 100% depth of discharge, about 600 at 60% depth, about 1,000 at 40%, and around 2,000 at 20%. Shallow cycles are cheap. That is the actual argument for the 20-80% habit, and it is an argument about total energy throughput, not about a magic number.

Why manufacturers added charge limiting

Three companies, three implementations, same physics.

Apple lets you set a charge limit anywhere between 80% and 100% in 5% increments, and separately runs Optimized Battery Charging, which learns your routine over about 14 days and delays charging past 80% until shortly before you normally unplug. Apple’s stated reason is reducing the time the iPhone spends fully charged. One detail matters for what comes later in this article: if you set a limit below 100%, Apple says the iPhone will occasionally charge to 100% anyway to keep its state-of-charge estimate accurate.

Samsung’s One UI Battery protection has three modes. Basic stops at 100% and does not resume until the level falls to 95%. Maximum stops at 80%. Adaptive uses Maximum while you sleep and switches to Basic before you wake. Samsung’s stated reason is that keeping the battery at a full 100% charge for a long time reduces its lifespan.

Google offers an 80% charging limit on Pixel 6a and later, plus Adaptive Charging that holds at a lower level overnight and finishes to 100% about an hour before you normally unplug.

Named limitation: capping at 80% costs you 20% of your usable runtime every single day, permanently, in exchange for a slower rate of capacity loss that you will not be able to observe for a year or more. For someone who ends most days at 40%, that trade is free. For someone who runs out at 6 p.m., it is a real daily cost paid for a benefit that arrives late.

Myth 3. A new phone needs three full charge cycles to calibrate

False, and it has been false for about twenty-five years. This advice is a nickel-cadmium habit that outlived the chemistry it belonged to.

Where this came from and why it no longer applies

Nickel-cadmium batteries had a genuine problem that looked like forgetfulness. Repeated shallow discharges allowed crystals in the electrode to grow from around 1 micron to 50-100 microns across, and those enlarged crystals hid active material from the electrolyte. Battery University describes the resulting behaviour as the cell remembering how much energy was drawn on previous discharges and refusing to deliver more, with the voltage dropping abruptly as if in protest. The standard cure was a deliberate deep discharge to 1 V per cell every one to three months, performed as maintenance and known in the trade as exercising the pack. NiMH inherited a milder version of the same problem when it arrived in the early 1990s.

Then lithium-ion arrived and the problem did not come with it. Battery University’s charging guidance states flatly that lithium-ion has no memory and that partial charge causes no harm. There is no crystal growth mechanism to reverse, so there is nothing for three initiation cycles to fix.

What survives from the old advice, in a much narrower form, is fuel gauge calibration, and it is not the same thing as conditioning the chemistry. The fuel gauge is the circuit that estimates how much charge remains; it needs reference points, and it gets them from full charges and from long rest periods. Google asks Pixel owners to fully charge every tenth cycle and leave the phone plugged in for at least 30 minutes after the screen reads 100%, specifically for gauge accuracy. Apple takes the automatic route and has the iPhone top up to 100% by itself now and then when a charge limit is set.

Those are software housekeeping tasks. They improve the accuracy of the number on your screen. They do not add capacity, they do not condition the cell, and doing them three times on a new phone accomplishes nothing at all.

Myth 4. Leaving it plugged in overnight damages the battery

True only if the phone stays hot or sits at 100% for many hours, and false in the way most people mean it, which is that the charger keeps forcing energy into a full battery.

What the charging circuit does once the cell is full

Charging runs in two stages plus an exit. Stage one is constant current: the charger pushes a fixed current and the cell voltage climbs. Stage two is constant voltage: once the cell reaches its regulation voltage, the charger holds that voltage steady and the current falls away on its own as the cell fills.

Then it stops. In the Texas Instruments BQ25170, a representative single-cell charger, charging terminates when the current drops below 10% of the programmed fast-charge current, and the datasheet states that the linear regulator turns off and the device enters a standby state. Charging restarts only when the cell voltage has fallen below a recharge threshold, specified as 75-125 mV below the regulation voltage with a typical value of 100 mV for lithium-ion. Nothing is being forced anywhere. There is no trickle charge in the nickel-chemistry sense, and lithium-ion cannot absorb a continuous overcharge the way a nickel cell can dissipate one.

So the phone plugged in at 3 a.m. is not being overcharged. It is doing something slower and duller: sitting at high voltage, warm, for six or seven hours, and repeating that every night. Refer back to the Krupp storage data, where the difference between 50% and 90% state of charge showed up clearly in cells that were doing nothing else. That is the actual cost of the overnight habit, and it is the reason Optimized Battery Charging and Adaptive Charging exist. Both features exist to shrink the hours at 100%, not to prevent an overcharge that was never happening.

Heat during overnight charging deserves its own mention, because it is the variable you control. A phone charging under a pillow or on a bed cannot shed heat, and Google’s guidance explicitly connects heat to battery drain that can damage the battery. Hard surface, not a soft one. Out of the blankets.

Named limitation: on a phone with no charge limiting feature and no scheduling, “do not leave it plugged in overnight” is not bad advice, it is just aimed at the wrong mechanism. The fix is the time at full charge, not the plug.

Myth 5. Draining to zero is good for the battery

False, and this is the single worst thing on the list you can do to a cell on purpose. Deep discharge damages lithium-ion cells and has no compensating benefit whatsoever.

Deep discharge and why it is the opposite of helpful

Battery University’s charging guidance is blunt: deep discharge wears the battery down, and a partial discharge is better than a full discharge. The depth-of-discharge figures quoted earlier make the point quantitatively. Roughly 300 cycles at 100% depth of discharge against roughly 2,000 at 20% depth, for the same NMC chemistry, means the same total energy delivered costs the cell several times more when you take it in full swings.

Below the 0% mark there is a second, harder failure. What a phone displays as 0% is not an empty cell; the software reserves margin and shuts down while the cell still holds charge. Leave the phone off and forgotten, though, and self-discharge keeps going. Battery University’s discharge guidance notes that this grace period can last several months until self-discharge lowers the cell to about 2.50 V, at which point the protection circuit opens and most packs become unserviceable with a regular charger. That is not degradation. That is a dead device that a normal charger will not revive.

The practical version: a phone left at 0% for a weekend is fine. A phone left at 0% in a drawer for six months may not come back. Apple’s long-term storage guidance is to charge to roughly 50% first, and the reasoning is the same on both sides, low enough to limit voltage-driven ageing, high enough to stay clear of the protection cutoff.

Named limitation: the one legitimate use for a full discharge is fuel gauge calibration, which is a software correction, not a chemistry benefit, and it is worth doing at most a few times a year on a device that asks for it.

Why your battery drains so fast, and why that is a different problem

A battery that empties quickly is usually a power consumption problem, not a capacity problem, and the two have completely different fixes. Capacity loss is gradual and one-directional; a phone that lasted 14 hours last month and lasts 5 hours this week did not lose 60% of its cell capacity in four weeks, because that is not a rate at which lithium-ion cells fail.

Capacity loss versus power consumption

Capacity fade is the permanent reduction in the charge a cell can store, measured against its original rated capacity. It arrives at a pace of single-digit percentages per year under normal use. Power consumption is how fast something is spending that charge, and it can double overnight because of a single misbehaving app, a lost cellular signal, or a display setting.

Rising internal resistance sits between the two and confuses the picture. An aged cell with high internal resistance may hold most of its capacity yet still shut the phone down early under heavy load, because the voltage collapses when the processor draws a burst. That reads to the user as sudden battery death when the capacity number still looks reasonable.

Symptom	Cell wear or energy drain	What to check
Runtime dropped sharply within days or weeks	Energy drain, almost certainly	Per-app battery usage in settings, recent app installs, recent OS update
Runtime declined gradually over 18-24 months	Cell wear	Battery health or maximum capacity figure, cycle count
Phone shuts down suddenly at 20-30%	Cell wear, high internal resistance	Battery health figure; whether shutdowns cluster in cold weather or under load
Phone gets hot while idle in your pocket	Energy drain	Background activity, poor cellular signal forcing high transmit power, a stuck app
Battery drains overnight with the screen off	Energy drain	Background app refresh, sync settings, a wearable or accessory holding a connection
Charges to 100% in a suspiciously short time	Cell wear	Battery health figure; a smaller usable capacity fills faster
Percentage jumps down in large steps	Gauge estimation error, not necessarily either	Whether the phone has had a full charge recently; gauge behaviour after a full cycle

What to check before blaming the battery

Start with per-app battery usage, which every major mobile operating system exposes. A single app holding 40% of the day’s consumption is the answer more often than the cell is.

Signal strength is the underrated one. A phone in a weak-coverage area raises transmit power to stay connected, and it does that continuously, in your pocket, with the screen off. This is why a battery can seem fine at home and terrible at a specific office.

Check whether the drop coincided with an OS update. Freshly updated systems reindex files and re-optimise apps for a day or two, and the elevated drain during that window is temporary. Check the weather too: cold ambient temperature reduces available capacity temporarily, and the charge comes back when the phone warms up.

Named limitation: none of this rules out a genuinely worn cell, and the checks above cannot distinguish a marginal battery from a well-behaved one on their own. If the battery health figure is low and the shutdowns happen under load, the cell is the answer.

What actually helps, in order of effect

Ranked by how much difference each one makes, the list is short: keep the cell cool, keep it away from the extremes of the charge range, avoid deep discharges, and stop worrying about charger wattage. Everything else is noise.

Factor	Direction of effect	How strong	What a user can actually do
Cell temperature during charging and use	Higher temperature means faster capacity loss, roughly exponential	Strongest single factor; Battery University’s storage data shows 25 °C at full charge losing 20% in a year against 40 °C losing 35%	Take the case off during heavy charging, keep the phone out of sun and off soft bedding, do not game while fast charging
Time spent at high state of charge	More time near 100% means faster loss	Strong; Battery University’s storage figures show 100% charge losing 20% at 25 °C over a year against 4% at 40% charge	Use the built-in charge limit or optimised charging; charge to 80-90% on ordinary days
Depth of discharge per cycle	Deeper swings mean fewer cycles	Strong; roughly 300 cycles at 100% depth against roughly 2,000 at 20% depth for NMC, per Battery University	Top up in small amounts rather than running flat and refilling
Charging at low cell temperature	Charging a cold cell plates lithium	Severe when it happens; 20% loss in 50 cycles at 3.5 C and 0 °C, per Yang et al. 2018	Let a cold phone warm up before fast charging; the phone usually enforces this itself
Calendar time	Capacity falls with time regardless of use	Moderate and unavoidable; roughly square-root-of-time behaviour over the first months, per Krupp et al. 2021	Nothing, beyond storing spare devices cool and near 50%
Charger power rating	Little direct effect at moderate temperature	Weak; the phone’s own controller sets the current, and it tapers above roughly 60-80%	Use a compliant charger and stop thinking about it
Number of full discharges to 0%	Each one is expensive	Moderate and entirely avoidable	Charge before the phone shuts down; do not leave it flat for months

Calendar ageing deserves its own numbers, because it is the part almost nobody accounts for. Battery University’s storage table, measured over one year, is the clearest published summary:

Storage temperature	Capacity retained at 40% charge	Capacity retained at 100% charge
0 °C	98%	94%
25 °C	96%	80%
40 °C	85%	65%
60 °C	75%	60% (after 3 months)

Read the bottom row twice. A cell stored hot and full loses more in three months than a cool half-charged cell loses in a year. Battery University does not publish the cell type or measurement protocol behind this table, so treat it as a shape rather than a specification, and see the contradictions section below for a peer-reviewed study whose absolute numbers are far smaller.

The 20-80% habit follows from all of this rather than being a rule in its own right. Staying inside that band shortens the time at high voltage, keeps depth of discharge shallow, and keeps you clear of the deep-discharge cliff. The cost is real: you are choosing to carry a phone with 60% of its rated runtime available. Anyone whose day genuinely needs the full charge should charge fully and stop feeling bad about it, because a phone that dies at 4 p.m. is a worse outcome than a battery at 88% health in two years instead of 91%.

How to read your battery health numbers without fooling yourself

The battery health percentage on your phone is an estimate produced by an algorithm, not a measurement taken from the cell. It is useful as a trend over months and close to meaningless as a reading on any given day.

Here is how the estimate is actually built. Texas Instruments’ Impedance Track algorithm, which is representative of how modern fuel gauges work, derives total chemical capacity as Qmax = PassedCharge divided by the difference between two depth-of-discharge measurements taken at two rest points. The gauge cannot update that figure whenever it likes. The published application note lists the conditions: the cell must be well relaxed, defined as a voltage drift below 4 microvolts per second or a five-hour wait; the charge passed between the two reference points must exceed 37% of the design capacity; the temperature must be between 10 °C and 40 °C; and neither reference measurement may fall in the 3737-3800 mV window, where the voltage curve is too flat to read a state of charge from.

The accuracy figures in that same note explain the jumping. A fresh gauge starts with a maximum error of 100%. After the first learning phase the stated maximum error drops to 5%, and after both parameters are learned it drops to 1%, then grows again by about 1% every 20 cycles until the next update.

Three practical consequences follow.

A percentage that sits unchanged for two months and then drops three points in a day has almost certainly just completed a learning update. The capacity did not fall off a cliff that morning; the estimate caught up with reality that had accumulated over weeks.

A phone that never gets a long rest at a stable temperature, or never passes enough charge in one stretch, may go a long time between updates and show a stale figure. This is precisely why Google asks Pixel owners for a full charge every tenth cycle and why Apple lets a limited iPhone top up to 100% occasionally.

Comparing your number against a friend’s is close to pointless. Different gauges, different learning histories, different rounding.

Named limitation: none of this tells you whether a specific phone’s battery health figure is trustworthy, because manufacturers do not publish the exact algorithm or update conditions their devices use. The Texas Instruments documentation describes a widely used approach, not necessarily the one in your handset.

The bottom line

Charging habits matter far less than the environment the battery lives in. A phone charged with the fastest compliant adapter available, kept out of the sun, kept out of a hot pocket, and topped up before it runs flat will outlast a phone charged slowly and carefully but left on a car dashboard in summer.

If you want one change, make it thermal: take the case off during long fast charges and keep the phone off soft surfaces overnight. If you want a second, turn on the charge limit your phone already has and accept the runtime you lose for it. If neither fits your day, charge to 100% and use the phone, because a two-year-old battery at 85% health is a normal battery, not a damaged one.

The number nobody wants to hear is that some of the loss was never yours to prevent. Calendar ageing runs whether you charge well or badly, and a phone that spent two years in a warm room lost capacity while it sat there.

Frequently asked questions