After installing AccuBatteryPro (btw is there any open source app with similar functionality out there?) I managed to “measure” the capacity based on integrating over the charging current and got to 3014 mAh (pretty close to the factory 3060)
The driver still claims it’s a 2000 mAh pack, which might cause initial calibration to be bad, so it possibly needs a few more charge/discharge cycles to come to a good calibration than normal.
After a few more loading/unloading cycles, and monitoring the battery utilization with AccuBattery, it looks like deep sleep is indeed working correctly, with the discharge curve in standby almost flat. The only cases where it wasn’t was when I had misbehaving apps drawing a lot of battery. Discord was one of them. Even with Wifi off, it somehow managed to keep the standby current at approximately 2% discharge per hour.
Unfortunately the stock android display to report app battery usage doesn’t do a good estimation on which apps are drawing current while the device is on standby, so in doubt one has to check that with trial and error.
It looks like both Wifi and SIM Cards only draw a surprisingly small amount of power in standby, which is a testament to the quality of components used in the Fairphone. What draws a lot of power in standby though is the CPU if any app keeps it awake when it technically should have gone into deep sleep doing nothing.
It’s surprising that a standard chat app like Discord, which relies on google push services for notification and wakeup and could go to sleep properly draws so much power in background, while at the same time, another chat app, Telegram, which relies on the “notification hack” instead of the “google way” to remain active in background and keep a network connection open, ends up using much less.
PS: I double checked, turning Wifi off alone does not trigger high bat usage. Battery usage is lower with wifi off than on, as long as no apps are running. As it would be expected.