The four phases

While a charge is running, Home names the phase it is in now, how long it has lasted, and which phases usually follow. The watch names the phase a charge is in now, too. When the session ends, the phone's detail screen segments the whole curve — the same four phases, in the same order, painted in the same colors everywhere in the app, and the watch's own detail screen draws them too. A charge is not one event; it is these four, and which ones dominate is most of what a curve has to say.

Starting up

The first seconds after connecting, while phone and charger settle on how much power to use: a low, held current step before full power arrives. Usually too short to see on the curve. ChargeLog records it and claims nothing about why the charger waits — contact checks, thermal checks and source detection all fit, and the battery gauge cannot tell them apart.

Charging

Constant current: the charger holds its setpoint while the battery's voltage climbs. The fastest stretch of a charge, and the one that shows what the charger can give. Found by the flatness of the power, never by how high it is — a charge that starts above roughly 70–85% has none of it, and the app says so instead of inventing one.

Topping off

Constant voltage: the voltage is held at its limit and the current falls away as the battery nears full. The long finish — this is where the last 20% spends its time.

Full

The charger has stopped; the battery is full and no longer drawing power. Time on the charger after that is holding, not charging: the session's row says how long the full cell was held, and the hold is kept out of the efficiency numbers.

What the analyzer won't do

Phase detection runs on smoothed power, looking for where the current loop gives way to the voltage loop and where the charger stops. It names only what a charger's own datasheet names — no ramps, no peaks, no step-downs, no trickle: the hardware does none of those, and an earlier version of this app said it did. Samples that fit no phase's definition stay unclassified — the analyzer never stretches a label to cover data it doesn't explain, and a blip too short to be a real phase is not promoted into one.

Why segment at all

Because "this charge was slow" is not an observation — "twelve minutes at full current, then fifty minutes of topping off while the phone ran warm" is. Phase makeup is comparable across sessions, chargers, and conditions in a way raw curves are not, and it is what turns one charge into evidence about the next one.