Yet Another Battery Drain thread - Don't kill your phone - give it life!!

Welcome to YABD – Yet Another Battery Drain thread
or Everything you wanted to know about Battery Drain but was too afraid to ask about




Corrupt media database
Explanation: thumbnails and some database's could get corrupted making the media scanner running berserk, resulting in battery drain.

Fix: Using SGS Tools, Press [Apply script] select [clean_thumb_db.txt] and execute the script, this will delete the files that might get corrupted and reboot the phone so the database files and thumbnails get recreated.


WIFI Bug
Explanation: This is a mix of bugs and wrong settings, hardcore found some leftover debug code in the driver, tc80tc80 tested different routers seeing this effected the WIFI too, other users reported that the distance to the router effects battery drain as well. On top of this there are some settings that makes the WIFI running all the time that are not easily found.

Fix #1: Check your settings,[Wireless and network] select [Wi-Fi settings] uncheck [Network notification] press the menu key and select [Advanced] in [Wi-Fi sleep policy] select [When screen turns off]

Fix #2: See if you could try your phone with another router and compare to your home router, also check your signal that you have good coverage on the WIFI.

Fix #3: Use an updated Kernel, either hardcore's or trasig's and apply the wifi-fix (wifi-driver from JPA): View attachment update-wifi-dxjpa.zip


SGS Email 2.2.1
Explanation: The new email client has a bug making it drain the battery a lot if you set it to update every x minutes or manual update instead of push (exchange accounts)

Fix #1: Setup your account to push both on peak and off-peak, if you use imap or pop you can't do anything.

Fix #2: Use another email application, K9 or Improved Email have been reported to work much better. If you remove Samsung email you might need a fix if you have applications using your camera available here:View attachment CameraappsNoEmail.zip


Battery not "calibrated" - we don't have real facts on this one yet, so consider this our best guess
Explanation: It is said that the battery looses track of it's capacity (or rather the phone does) so it needs calibration.

Fix #1: Always flash new ROM:s with full battery.

Fix #2: Follow the guide TechnicR have on recal:
· Turn phone off and charge till full
· Start recovery and wipe battery stats
· Reboot and start the phone
· Use it heavily until it turns of by itself
· Connect charger and turn on the phone and wait until fully charged


Hardcore 256Hz vs. 500 Hz
Explanation: this is the system tick in the Linux system, number of interrupts per second for timesharing, gives a smoother appearance

Fix: None needed, the difference is not noticeable in battery drain.


Battery level drop to 95-98% when disconnecting charger
Explanation: We are used to that the battery is 100% chraged when we disconnect the charger, but with 2.2.1 we gat anything from 95-98%.

Fix: Not needed, this is a change in the charging cycle to prolong the life of the battery, but please take this in consideration of reporting battery drain, start counting from the actual value and not from 100%
For more info, check under Expert witnesses in the next post.


Could someone help me to explain the following phenomens as they are mentioned in so many various ways:
3D Gallery bug
Sensor driver hangs
Auto rotate bug

Link to my current phone config and battery life:
http://xdaforums.com/showpost.php?p=10521024&postcount=613


If you wan't to support me more than by contributing to this thread, consider if you would like a easy and automatic backup for your PC, 2Gb free plus a 256Mb bonus space for both you and me, thats 2.25Gb of free backup space for your mail and documents (and anything else you want) and for each referal you make in the future you get an extra 256Mb space added to your account.
I have been using Mozy for 3 years now and I'm very happy with it.
Sign up from here: https://mozy.com/home/free/ at the end of the registration please enter my referal code: BQBF53

How is my phone now days?

I would say I'm quite satisfied with the battery life now, lets start with my current config:

I use my gmail account to sync with my personal outlook for calender and contacts, on top of this I also have one exchange account plus my personal imap mailbox.
Sync settings:
General, Background data and Auto-sync is ON
Gmail all sync ON
Exchange sync every hour, 3 day mail, Contacts and Calendar sync ON
IMAP sync every hour
Exchange and IMAP use AOSP email client
WIFI is switched ON
Network notification OFF
Advanced, WIFI sleep - When screen turns off (no hacks on this so 15min delay)
Mobile network
Use packet data ON
Network mode WCDMA only
Bluetooth OFF
Location
Use wireless networks ON
Use GPS OFF when not used
Use sensor aiding ON (is turned on together with GPS)

Display set to auto brightness
using Screen Filter set at 62%

Widget Beautiful Weather update every 4 hours

System Panel is always running Monitoring settings everything is ON (incl. High priority)

Active applications:
AdFree
BLN control
Dropbox
ClockSync set to sync every 12 hours when awake
TrueCaller
SMS Backup+
CalendarPad Pro
SwiftKey
Zeam launcher
Doc Theme V3 AOSP lock screen

The ROM is from Doc's and Stefunel's romkitchen.org using hardcore's K12K kernel, JPP modem and after the generation of the ROM I replace the email.apk from system with AOSP:s email.apk, replace libcamera.so so BarcodeScanner and other camera apps work without the original email.apk

I don't think I forgotten anything, the result for the latest 20 hours



First 7 hours is sleeping for the both of us. First high usage is 40 minutes of XDA browsing and some Angry birds, next big peak of 40 minutes is Andoku (soduko) and some XDA reading, then a 10 minute Angry bird and the last 50 minutes is gathering all the data for generating the diagram, I used Dropbox on the phone to sync the files to my PC so it's a few Mb data included and looking up all settings so it was active all the time.
Only 2 calls about 10 minutes total, 25 emails, no SMS!

If you like to try this ROM, go ahead, it's a no wipe version, so if you want to wipe, do a factory reset before installing the ROM, I take no responisibility of any of your data or anything else.
Bokbackens ROM

If you don't know how to flash a ROM from CWM, then go to another thread and learn, I would recommend Doc's or Darkyy's in the Development topic.

Less then 3 weeks ago (when I started this thread) my phone didn't last 8 hours in sleep mode on a full charge!


Yes I know my setup is not optimal for battery life, it's not supposed to, I bought the phone to use it.
We pay more than 500€ for a Smartphone, then we should be able to use it fully!
If I wanted an Idiot-Phone, I wouldn't be here were the smart people hang out, would I?

I didn't start this thread to tell you how to turn off everything in your phone to get decent battery-life, I don't want to carry a zombie in my pocket, do you?

Gathering the suspects

Let’s get rid of the “I think” “If you only do” and “I’ve heard that” and start arresting the suspects, line them up, and execute them (oops, did I say that out loud?), and put them in quarantine until further notice.

To keep this thread cleaner, we’ll only focus on 2.2.1 mainly XXJPU/X/Y and not bother with to old 2.2 ROMS

Let's end the guesswork and start a scientific approach on this, it’s ok to suspect something, but you must have some evidence.
We need repeatability on the problems, so if you did something that started to drain your battery, describe this in as much detail as possible so someone else can verify your findings.



I have gathered some suspects but now I need your help in Collecting Scientific Information (CSI) to round up more suspects or eliminating them so we might find the truth about the drainage problem.


Forensic tools:
Free/Pay, Application name + author
Both, SystemPanel by NextApp
Free, OS Monitor by eolwral
Free, Android System Info by ElectricSheep
Free, Shark for Root by Elviss Kustans (not native version)
Free, Shark Reader by Elviss Kustans

Battery Monitoring / Saving tools:
Free/Pay, Application name + author
Free, JuicePlotter by Latedroid
Free, JuiceDefender by Latedroid
Free, Battery snap by Xhelas
Free, Brightness Level by CurveFish - let you lower the brightness another 8% and give you some extra milage
Free, Screen Filter by haxor industry, reduce the brightness a lot! thanks for the tip lpy

List of suspects:
#1 SGS Email 2.2.1 sync set on every x minutes - will be charged as an environmental villain, is put in custody pending trial
#2 Battery not “calibrated”
#3 Corrupt database files for media scanner
#4 Modem version X
#5 Voodoo lagfix
#6 WIFI bug - could be organized crime here, many villains involved. hardcore has infiltrated them and so far he has put 2 of them behind bars!
#7 Non KIES updates - could Samsung original updates be better

Prosecutor’s witnesses:
For suspect #1, SGS Email 2.2.1:
http://xdaforums.com/showpost.php?p=8554755&postcount=546
http://code.google.com/p/android/issues/detail?id=9307 - very strong witness

For suspect #2, Battery not calibrated:
http://xdaforums.com/showthread.php?t=722862
http://xdaforums.com/showpost.php?p=9075485&postcount=5653
http://xdaforums.com/showpost.php?p=9149156&postcount=143
http://xdaforums.com/showpost.php?p=10092321&postcount=11053

For suspect #3, Corrupt database files for media scanner
http://xdaforums.com/showpost.php?p=9075485&postcount=5653

For suspect #6, WIFI bug
http://xdaforums.com/showpost.php?p=10038645&postcount=29
http://xdaforums.com/showpost.php?p=10042758&postcount=6627 - hardcore found some leftover code in WIFI driver
http://xdaforums.com/showthread.php?t=873779 - tc80tc80 tested different wifi routers with scary result!

For suspect #7, Non KIES updates
http://xdaforums.com/showpost.php?p=10057892&postcount=69
http://xdaforums.com/showpost.php?p=10061398&postcount=73

Defender’s witnesses:

For suspect #6, WIFI bug
http://xdaforums.com/showpost.php?p=10043778&postcount=49


Expert witnesses / interesting battery info:
http://www.androidpolice.com/2010/12/14/your-battery-gauge-is-lying-to-you-everything-you-need-to-know-about-bump-charging-and-inconsistent-battery-drain/
http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries


If you want to support me me more than by contributing to this thread, consider if you would like an easy and automatic backup for your PC, 2Gb free plus a 256Mb bonus space for both you and me, that’s 2.25Gb of free backup space for your mail and documents (and anything else you want) and for each referral you make in the future you get an extra 256Mb space added to your account.
I have been using Mozy for 3 years now and I'm very happy with it.
Sign up from here: https://mozy.com/home/free/ at the end of the registration please enter my referral code: BQBF53

lpy's measurement - a summary

This is a summary of lpy's measurements for our referens, our thanks go to hime for this great work.

Comparison of ext4 and rfs

Ok not sure if I'd call this inconclusive or if people are happy enough with these results to label the EXT4 excess battery usage a "myth" just yet (it really is hard to tell, especially if the results would vary from tests vs real life usage), as I'd probably need some sort of test app to do these properly. Plus it's 5am here.. Anyway, these are my results (running JPY 2.2.1):

[EXT4] & [RFS]

Quadrant Advanced -
1/4 File system writes = N/A (too quick)
2/4 File system reads = N/A (too quick)
3/4 Database writes = 220-230mA
4/4 Database reads = 320-330mA

Copying 48MB file from -
Internal SD card to /data = 190-230mA (Write test?)
/data to internal SD card= 190-230mA (Read test?)
/data to /data = 190-230mA (Read/Write test?)

The Benchmark app which tests creating 1000 files, deleting 1000 files, etc... showed

[EXT4]

using up to 330mA,

[RFS]

up to 260mA.

However, the EXT4 test finished a lot quicker.

Idle (no file transfer) -
~100mA

Different current draws see the JPY wifi it does draw a lot more:

So how much current exactly does stuff use?
2.2 JP5 AU & 2.2.1 JPY Firmware

Phone off: 0mA
Screen off, Idle: 1-3mA
Screen off, Idle, 3G on (full signal): 2-4mA
Screen off, Idle, Wifi on: 3-7mA
Screen off, Idle, Wifi on: 25-250mA (JPY firmware with weak Wifi signal)

2.2 JP5 AU Firmware

TWLauncher, default background, no icons, minimum brightness: 93mA
TWLauncher, Aurora animated bg, no icons, minimum brightness: 158mA
TWLauncher, default background, no icons, maximum brightness: 205mA
TWLauncher, Aurora animated bg, no icons, maximum brightness: 228mA

And to my question about Brightness by Curvefish this was the result with a great tip of another app to lower the brightness: Screen Filter by haxor industry


It helps a little, but a bigger help is the Screen Filter app - unless you're in the sun

Dark background -
100% - >112mA
50% - >97mA
0% - >80mA
-8% - >78mA (CurveFish Brightness)
-8% + Screen Filter @ 20% - >71mA

Bright background -
100% - >383mA
50% - >243mA
0% - >109mA
-8% - >100mA (CurveFish Brightness)
-8% + Screen Filter @ 20% - >71mA

Bonus test

Vibration test (*#0842#): mA vibration on - mA vibration off =

Vibrate full = 65mA
Vibrate just over half = 25mA
Vibrate just under half = 10mA
Vibrate low = 7mA

As you can see, setting vibrate to half (or just under) can save quite a bit of power too.

TLDR lpy's quick tips:

Screen Filter app for night time/indoors
Set vibrate to just under half or less

Earlier measurements:

My own test results:

K10I 256HZ (default color), Tweaks on, ALL EXT4 -

Phone off: 0mA
Screen off, Idle: 1mA-3mA (very nice)
Screen off, Idle, 3G On: 2mA-4mA
Screen off, Idle, Wifi On: 4mA-7mA
ADW Launcher, blank screen, dark wallpaper: 86mA
ADW Launcher, blank screen, animated Aurora wallpaper: 178mA
ADW Launcher, blank screen, Froyo default wallpaper (building pic): 123mA
Winamp app playing mp3 (screen off): 52mA-60mA
Winamp app playing mp3 (screen on, minimum): 120mA-130mA
GPS Test app (GPS off): 150mA-160mA
GPS Test app (GPS on): 200mA-285mA
White screen (minimum brightness): ~130mA
White screen (maximum brightness): ~600mA
SetCPU app 100% CPU stress test (screen on, minimum): ~295mA
Camcorder standby black: ~320mA
Camcorder recording black: ~410mA
TuneIn Radio app (stream), A Online Radio app (stream), Winamp app (mp3), screen off: 166mA-200mA
J Benchmark 2 app (Disc Benchmark): 400mA-500mA

K10I 500HZ (default color) -

Exactly the same mA as 256HZ.

I will put a link to in post #3
@lpy: If you like me to add any info here please send a PM

1st Test Cycle with JPY

Ok. End of my cycle. It's been 32 interesting hours battery-wise.
I think i might have isolated the causes of the battery draining I am experiencing, maybe even found a proper bug. Needs more testing, but at least it seems very much replicable, which is essential for a bug hunt.

PREVIOUS SITUATION - JPX, flashed with KEIS reghack, not particularly bothering about charge level, wiping and factory resetting. I have done no proper battery testing for JPX since JPY came out shortly afterwards. Though the feeling was definitely like there were the same worse battery draining.

FLASHING METHOD FOR JPY - Charged to 100%. Wiped cache partition and then Dalvik cache, both through recovery. Did a HARD factory reset (*2767*3855#, this will wipe everything, including internal SD card, not even asks for confirmation!). Flashed with reg-hacked KIES. After 1st boot, HARD factory reset again.

ACTIVE MODs OR CUSTOM ROM - Everything stock, except for rooted CF-root JPY kernel, which is stock JPY + injected "root stuff".

BACKGROUND - If you read my post #59 you'll learn more about my use-cases, my battery testing ad why i strongly suspected battery mis-calibration for being the primary cause of my battery draining. Therefore, i decided to do the charging bump. To learn the theory behind this i strongly urge you to read this expert witness, if you haven't yet. To learn about the practice, there is a long and full of contributions thread about the HTC Desire battery (re)calibration here on XDA (which is unfortunately a perfect example of a poor scientific approach to battery issues that leads to a waste of efforts). There is no widely accepted and proved working method to recalibrate the battery, you will find tens of procedures, tens of successful and unsuccessful reports even with the same procedure. To me this is a clear sign that battery de-calibration can definitely be an issue and it's no suggestion, that recalibrating is definitely possible but up until now we are very much shooting randomly when trying to solve it.

CHARGE BUMP - The method i choose is just the one i felt more logical and straightforward, more likely to work with SGS. Might be a wrong one, or might be just one of the right methods. Here is what i did:

• Left 2% of charge on the phone, power it off;
• Plug the charger, let it charge while powered off.
• Left charging 1 more hour after 100% sign and vibration.
• Plug out the charger, powered on, let boot and media scan complete, wait 1-2 more minutes, power it off.
• Plug the charger, let it charge while powered off. It charged the phone for almost 10 more minutes, than 100% sign and vibration happened.
• Plug out the charger, powered on, let boot and media scan complete, wait 1-2 more minutes, power it off.
• Plug the charger, let it charge while powered off. It charged the phone for less than 5 more minutes, than 100% sign and vibration happened.
• Left charger in, boot into CWM recovery, wiped battery stats, reboot system, plug out the charger while booting.

POWER MANAGEMENT SETTING - Packet Data ON; Wi-Fi OFF; Sync/Push ON on GMail, Contacts, Calendar, Twitter, Last.fm, Skype; FancyWidget refresh ON (every 2h); Brightness on AUTO; GPS OFF; Bluetooth OFF; 2G/3G AUTO; Haptic feedback on keyboard and keys ON; AutoRotate ON;

(Let me know if i missed some settings you're interested in)

FORESCIC TOOLS -

Battery Statistics (*#*#info#*#*)
OsMonitor (Process list was sorted by CPU Load)

BOOT STAGE - By the time i unlocked the screen, it showed 99%. I left media scanning complete, quickly checked mail and tweet notification, then gave a quick look at OS Monitor to see if something strange was happening during boot, and then i switch Flight Mode OFF/ON to get rid of the infamous "50% of time without signal" bug. By the time i did this (2-3 minutes), it fell back to 98%. I went to sleep with a felling of impending battery doom.

1st STAGE (98% to 66%) - But that feeling was unjustified. I woke up temporarily 5:30h later, to find out that there was 90% left. I went back to sleep, and when it was time to wake up for good 3 hours later, there were 86% of battery left. So we are taling about 12% drop in ~8h, 1.5% per hour, a proper sleep drain.
Then i started using the phone according to my test-case (post #59, again), checking OsMonitor and Battery Statistics every once in a while to see if something strange was happening. It was not. I will make just this two remarks, they will be useful soon:

1. In OsMonitor List of active processes, sensor_yamaha occasionaly popped up at the top, with 2-3% of CPU load, but more often than not it was not in active use (remember, list sorted by CPU usage)
2. In Battery Statistics, in the sensor section, the only item present was Maps, and the time using sensor was exactly the time i kept the app open: few seconds, 25-30. Then i closed Maps, and sensor statistics shows no more use of the sensor.

After 16h and 1 minute, i reached the 66% left mark. Perfectly on schedule for 2 days.
Everything was pointing to a successful recalibration of my battery.

2nd STAGE (66% to 40%) - Then I had to take a picture with my SGS. I took it and then opened the preview that is now easily accessible through Camera with 2.2.1 ROMs. That opened the Gallery to show the image, which i then adjust for viewing by rotating my phone.

This is when the heavy drain started. I lost 26% in 5:30h.

All of this was shown using forensisc tool:

• OsMonitor now showed almost always sensor_yamaha in the active processes, with an increased average of 4-5% cpu load. It reverted the previous situation: now sensor_yamaha was in sight more often than it was not.
• Battery Statistics was even more odd: in the sensor section an item "Gallery" appeared. From now on there was no way to stop increasing time counter. Since i first opened the gallery, until my battery died the next morning, there were nearly 16h in between. That is the total duration os sensor using for the "Gallery" item.

I noticed i was in a new, heavy drain stage earlier, when suddenly i saw 61% left when in fact i was expecting 64 or something like that. The fact is, i thought maybe some gallery process hanged, but couldn't find any, nor with stock Samsung Task Manager, nor with the awesome Task Manager from the market, nor with OsManager. So i tried, and launched the gallery again, let it load all of the thumbnails, then push the back button, go to the OsManager process list: there was a still active Gallery process. I killed it. I did this twice during this second stage, but it came to no help. Draining wasn't stopping. So i decided to take a different route.

3rd STAGE (40% to 28%) - I disabled AutoRotate through its toggle in the drop-down notification bar. Guess what?

Draining stopped altogether!

It just went back to prior 2.5% per hour, the normal "Using The Phone" draining, the same there were until 66% mark! The next 12% took 5:30h to discharge. It reached 28% left, and then i went to sleep.

Please notice: the fact that i turned sensor off stopped the draining, stopped the process sensor_yahama, but didn't stop the increasing statistical counter. That makes me think about a few possibilities:

• A bug in the Battery Statistics utility (very unlikely, Maps and other items were correct).
• The counter do not show the time spent using the sensor, as someone would suppose; instead, it shows the time spent by processes sending request to the sensor daemon/server. This would explain why a faulty/bugged gallery would keep it growing even though draining stopped (sensor disabled = process terminated by android = requests simply ignored = no more energy used).

So before going to sleep i made quick maths; with the correct draining rate restored, the phone would have last for the next 12h, more or less.

4th STAGE (28% to 0%) - And probably would have. I will never know, because a different issue showed up. While i was sleeping i received 4-5 SMS. Having HandCent SMS installed, both its popup and stock Samsung notification went on. The problem is, HandCent SMS for some reason never left the phone go into sleep mode again. You can perfectly see that in the screencaps about "In Execution" section. From the time i received the 1st SMS (9:09) to when i read it (12:20), Handcent was in execution (3:11h). The "Time spent without sleeping" counter was exactly where i left it at night... plus Handcent massive contribution!

I was very lucky i woke up in time to notice this and have a look at the stats, take the screencaps and make my speculation about this

MISCELLANEA - The signal strenght in my area wasn't very strong, but also wasn't a complete mess at all. Frequent changes from 2G (Edge) to 3G and HSDPA. Never seemed to affect the phone much, though.

WHAT'S NEXT? - Now that it seems my battery has been recalibrated, sensor is my main concern at the moment. For the next cycle i want to change the least possible variables, and so i would probably made another little bump (just once this time, and no battery stats wiping of course) and keep auto-rotating OFF. If all will go well, i will re-test with AutoRotating ON but avoiding starting the Gallery. I will then evaluate what to do, i heard there is a fixed version of the Cooliris Gallery (Samsung stock gallery) somewhere, maybe in the SGS Tools.

In the meantime, i hope someone could make use of this extremely long report and confirm/controvert this two issue. I think, apart from System draining, a list of faulty apps badly using power has to be one of our targets.

VOLTAGE - My 2 cents about voltages. I also think a combined monitoring of percentages and voltage values has to be made. It is a kind of analysis we lack of, never seen that. It will help to understand more about calibration, among other things. Once in a while i had a look on them (with 1s refresh), and i can definitely say that the higher voltage i have ever seen on SGS is 4.240 mV. It last as long as the charger is plug in.
The lower voltage i saw was 3.460 mV.
I have reason to think voltage is the actual value used when the system automatically shutdown itself on low power: several reports of recalibrations (it actually happened to me once) shows that 1st cycle after recalibration could end in a shutdown while battery shows 20% left or some; another thing that could happen is a battery suddenly dying while doing a power-intense task like, for instance, switching/disabling Flight Mode. The sudden drop in voltage value could mislead the system to think battery is empty. It's not rare that when this happens, after a little while you could power on again and enjoy even 10 more minutes of juice.
If i'm right and this is the case, it's very important to get near to the ideal situation where max voltage is recognised as 100% and minimum voltage match the 0 or 1% mark.