Anelok: saving more power (1/2)
werner at almesberger.net
Thu Apr 30 19:43:18 UTC 2015
Rafael Ignacio Zurita wrote:
> Okey, but before to continue: Why was the reason to not consider the
> turn-off turn-on
> idea in those five minutes per day?
Hmm, I don't understand the question.
Those five minutes would be the total time of active use during
the whole day, with the display on. Typically, this would be in
small sessions of 30-60 seconds each.
And yes, if you turn Anelok on but then don't use it, or forget
to turn it off after use, it will quickly drop to lower-power
E.g., after 10-30 seconds (depending on whether it shows content
you may just look at for a while, e.g., a password, or whether it
shows something you're expected to react to quickly, e.g., a
selection list), the screen will go dark. (State DARK1.)
If you're not done yet, you just touch the slider and the screen
content comes back instantly.
If you don't react to Anelok going dark, power management assumes
you're not in a hurry to have it back on and powers down the display
panel. (DARK2) Recovery from DARK2 takes longer because there are
delays in initializing the display hardware.
In DARK2, Anelok waits just one second to make sure the system has
settled in a stable state, then switches off the boost converter
(READY). All this happens within 21 to 41 seconds after you last
touched the slider.
If you're at the account list, your current consumption between
last touching the slider and READY would be
15 mA * 10 s + 3.1 mA * 10 s + 0.7 mA * 1 s = 181.7 mAs
READY is already quite low-power (130 uA), so it stays there for a
minute in order to be able to respond quickly to touch screen
activation. After that minute, it reduces the sampling rate for the
touch screen (STANDBY).
READY is thus another
0.13 mA * 60 s = 7.8 mAs
(4.3% of the 181.7 mAs it took us to get to READY.)
Also, when "on" (ACTIVE), the MCU sleeps (LLS state) whenever it
can. But the dominant power consumer is then the OLED panel anyway.
More information about the discussion