Battery level property problem or faster discharge than original firmware

[vaspa]

Checking the battery level indicator I've noticed that it is sometimes inconsistent. E.g. it goes to 80% in one day, then jumps back to 90%, then goes down sporadically.
Could it be the lack of calibration or there is some running stuff?
As I understood the top battery-eaters are:

1. CPU calculations
2. BLE transfers
3. Screen updates

[althink]

That's part that I'm not much competent ;) Since two days my battery reaches 90% max after full night charging. I have two watches and max/min voltages are different on both.

When you enable accelerometer it's even worse. I've switched it to the lowest power mode when it should take only 6uA... and after 3 days from full charge battery was empty. So I'm trying to find a reason what's the problem.

[zigastegu]

I do some approx. measurements:

• minimum is about 15-30 uA
• on every 7.6 ms (120 Hz) there is a 1340 mV -> 1350 uA (1.34 mA) draw for about 50 us with 50 us capacitive tail

FW: github latest
method: uCurrent on 1 mV/ 1 uA rating
oscilloscope on 500 mV and 50 us scale

TODO:

• maybe do an interrupt analyse, for every feature with respect to interval:
  app_timer_start(timer_id, RTC_INTERVAL, NULL);