Tuesday, February 10, 2015

Battery Life: the Unexpected Extinction (OR how we lost it)

Since the very beginning of Android everyone has been talking about Battery Life... and how bad it is; but why that much hate?

Let's investigate:

For the following, i had only based the research on the devices i owned/used:

Impacting points

But... What are the culprits??

Here are some, but i'm sure there are many more things that impact battery life.

  • Device global power and Screen size/technology vs battery size
  • Antenna quality and network coverage
  • Firmware: Kernel, ROM... (implementation, and optimization regarding hardware)
  • background services (viber, facebook, google services [location seeking, google fit, google wear, ...], ...)
  • User habits
  • ...
There are lots of variables in the equation so we cannot "benchmark" them as we do for performances; the following won't give answers, but more questions that anybody should ask themselves when talking about battery life.
A good study would need known and fixed variables to investigate on one specific variable, with negative and positive controls, triplicates, and statistics to show significant differences. But we have none.


Power, Screen Size/tech, Battery Size...

We have 4 (5) devices from 3 generations:
  • The "old phone"
  • The early smartphone (The e10i was shipped with android 1.6 donut!) and
  • The current gen entry/middle class smartphones (XT1022-XT1032-XT1039)
The oldest features a 128*160px screen for a 900mAh battery and could handle more than a week and a half without the need of a charge (up to 2 weeks with moderate usage)
The X10mini features a 240*320px screen for a 950mAh battery and could only stay on for about 24hrs-48hrs
The MotoG: 720*1280px for 2070mAh (24-48hrs)
The MotoE: 540*960px for 1980mAh (48-72hrs, and up to a week with fair usage)

Here is chart about battery capacity (mAh) and screen surface (px) coefficients between my first phone (Z530i) and the newer devices:

device length width surface (px) battery (mAh)

Z530i 160 128 20480 900 surface increase battery increase
E10i 320 240 76800 950 3,75 1,06
XT1022 960 540 518400 1980 25,31 2,20
XT1032 1280 720 921600 2070 45,00 2,30

Battery and screen surface increase coefficient with Z530i as base
Z530i --> XT1032 : 2.3x battery capacity increase, but a 45x wider screen!!
Luckily, battery life had not been decrease by 20! (only by 7~14)

Screen technology also plays a role in power consumption: 
A standard TFT/IPS screen uses a whole backlight that is either ON or OFF (+ brightness levels) meaning similar power used regardless the displayed content.
On the opposite, AMOLED based screens don't have any: these screens features LEDs for each pixel that are completely OFF when a screen part is black meaning NO power used at all!
The previously named devices are all using TFT/IPS based screen though.

CPU and GPU are not much relevant (regarding clock since MHz/W has been increased a lot making newer devices a lot more efficient); this point would need a whole in depth article, so we won't study this much now.

Antenna quality has a lot to do with battery life, the better the antenna, the better the life: the device would need less power to get same signal, saving some more juice.

Network Coverage isn't directly linked to hardware (since it is not relateed to the device itself, --in fact yes if we consider the bands handled by the device and the ones provided by the MNO--)


Hardware isn't the only one
Implementation and optimization:
A well optimized software for a specific hardware can move mountains, as easily as a bad one will sink in the deep see...
That's easy to show for benchmarks (see my comparison between stock and CM11 on MotoG and impact on MFLOPS), but hard for battery life (too many other un-fixable variables).

All we can say here is that a process that needs more CPU cycles to perform an action will keep it awake longer, thus resulting in more power consumption and a shorter battery life.

Background services:
Google IS Evil, yes, their services are running in the background at any time and everywhere. Most of them won't be used by the end user, but f***k it, they are alive!
"- i'm looking at you location seeking, google fit, google wear, google search,..."

Facebook (app and messenger), Viber, ... All these apps use some services to keep refreshing for potential notification, calls, pokes...

What can we do?
Seek for wakelocks using Wakelock Detector,
Hibernate apps you don't use with Greenify,
Ask in the Forums for some advice

User habits

99% of bugs are between screen and chair

Who said the device was all culpable?

Look at yourself, what are you doing with your device right now? maybe reading this article from you phone/phablet? Playing all day long? Answering emails? Checking twitter/facebook/whatever? Browsing 9gag?

Didn't you changed your usage of the so called 'smart'-phone? are you doing the same as you were on your old 'not-that-dumb'-phone?

No comments:

Post a Comment