Repository navigation
Sleep apnea recording draws ~55–80 mA on Galaxy Watch8 (SM-L320) with 0.10.12; watch runs flat overnight #27
Description
Battery report from a Galaxy Watch8, following up on the sleep apnea battery work in #13.
Setup
- Watch: Galaxy Watch8 40 mm (SM-L320), One UI 8 Watch, Health Platform 1.7.00.05 (dev mode on), battery 318 mAh rated.
- Phone: Galaxy A56 (SM-A566B), One UI 8.
- LibreWrist 0.10.12 (build 61), Health Monitor only: no Bridge, no Alarm fork, no Health+. Watch SHM 1.5.1.022, phone SHM 1.4.6.001.
- Sleep apnea screening on. Irregular heart rhythm (IHRN) also on.
Summary
While the watch detects sleep, it draws about 55–80 mA on average, against 5–8 mA idle with the screen off. On the first night it went from about 59 % to empty in roughly 3 hours, at 1 % every 3–4 minutes with the screen off and the device idle. It shut down before morning.
The start on sleep and the stop on wake both work on this watch: the 0.10.12 stop on GET_DATA kicks in within about a minute of waking. The cost comes from the recording itself.
Measurement during a 44-minute nap
Source: BatteryService current_now from the watch logcat (about one reading per minute). Sleep onset and wake come from SleepDetectionController in systemui.
| Period | current_now |
|---|---|
| 5 min before sleep onset | −5 to −7 mA |
| Sleep onset (T+0) | −66 mA within seconds |
| T+1 to T+44 min | 10-minute averages −54, −55, −62, −81 mA; per-minute readings −42 to −134 mA, never above −42 |
| After wake | back to −5 to −8 mA within about 1 minute |
- Battery went from 43 % to 29 % during the nap.
com.samsung.android.service.health/.tracking.requesthandler.WearTrackerServicestarted as a foreground service 6 s after sleep onset.
batterystats daily history
| Period | Projected life per full charge | Screen-off discharge steps ≤ 5 min per 1 % |
|---|---|---|
| Previous 10 days | about 35–40 h (27–50 h range) | 0 |
| First day with screening on | 11 h | 65 (54 with device idle) |
Comparison with your measurement
docs/spo2-adapter.md gives about 38 mAh/h while recording against 12 mAh/h idle on a Watch5 Pro, about 26 mA above idle. Here the recording adds about 45–75 mA above idle. A Watch8 40 mm cannot record a full night on one charge at that rate.
Unknown
I couldn't tell whether IHRN's PPG_GREEN/ACCELEROMETER sessions also run during sleep and add to this. The HTS client doesn't show up in dumpsys sensorservice. They might explain the spikes above the ~42 mA floor.
I can export a LibreWrist device log or try a nap with IHRN off if that would help.
Activity