Battery consumption analysis
 

Yesterday a new nice application for battery monitoring came to Extras-Testing repository: Battery-Eye.

I've installed it and found that It continuously monitors the battery consumption, even when the phone is off.

I've taken a screenshot and I'm trying to draw some conclusions.





- First of all, it takes about 2 hours for charging completely (in the graph, from 22:50 pm to 1 am)
(good, if compared to many other phones)

- My battery can't recharge completely (rechargeing stops at 90-95%)
(too bad, damn!)

- After the charging stops, the phone starts consuming battery even if it the charger is plugged in (from 1 am to 8:45 am)
(very bad, what if I let the phone one day in charghing? I will find the battery half-consumed?)

- A reboot consumes a lot of battery (from 10:50 to 11 am)
(normal)

- Playing Angry Birds consumes battery heavily :). (from 21:45 pm to 22:50 pm of the 8th Feb.) Playing Angry Birds while connected to a wifi network will completely kill a full battery in 4h.
(understandable...)

I'll keep this thread up-to-date with more screenshots and analysis of my battery consumptions.

-

Re: Battery consumption analysis
 

Interesting - I installed this a day or two, and have some somewhat odd results.

Yesterday, the phone charged up overnight and remained pretty-much fully charged from 3AM to about noon (I'm a late starter at the weekend), whereas last night it charged up at about the same time, but immediately started dropping off. You can also see the change in discharge rate when full-time MfE checks kick in.

Re: Battery consumption analysis
 

Very very interesting. Why our phones discharge themselves? Can you remember what has been changed in your phone between the 2 screenshots?

Re: Battery consumption analysis
 

I think there were one or two updates to apps yesterday, but nothing major's changed. My understanding is that the phone should charge up to near full, then discharge to about 90%, then charge again (rather than trickle charging at near full, which is not as good for the battery). I'll see how it does tonight anyway.

Re: Battery consumption analysis
 

In your second screenshot it seems that your phone charged up to 95% and then discharged to 80%, not 90%. Were you using your phone?
BTW I've just changed my main post's image with a more detailed one.

Re: Battery consumption analysis
 

I noticed the drop after the phone charged whilst still being plugged in as well this morning. I thought that was a little odd.

Re: Battery consumption analysis
 

That's right - it dropped off a long way. What I meant was that the intended behaviour should be to discharge only to about 90% before charging. I'm not entirely sure how battery-eye is measuring this though, as most battery meters fail to pick up any changing information while the device is charging.

Re: Battery consumption analysis
 

I have this issue when using wifi...

im turning on wifi...use for 10 minute and then turn off...after that my battery drain fast..even on standby...with 2 hours from 30% drain to 10%...

only way that i can find for solve this problem is reboot...every time i do reboot and then my battery remain very good...

--------------------
So i think if u r using wifi might be related ;)

Re: Battery consumption analysis
 

Here it is a new screenshot.


The last line has been drawed in the graph while my phone was connected to a wifi network with automatic refresh of emails every 15mins.



I've tried to extend the hours grid and...


As you can see, my phone would be able to stay connected to wifi and refresh mails every 15 mins for about 23/24h.

Re: Battery consumption analysis
 

and how much battery das battery-eye use?
noticeable?

Re: Battery consumption analysis
 

Can't notice it, so I'd say it use very little battery, but I can not be sure.

Re: Battery consumption analysis
 

I'm in with analising

Re: Battery consumption analysis
 

Just run some tests of running a converted DVD 'Lord of the Rings' constantly and seeing how long the battery would last with wifi and most things disabled.

It lasted about 4 hours and 10 mins.

Re: Battery consumption analysis
 

could you share what your application exactly measures? (where does those readings come?)

Re: Battery consumption analysis
 

Install Battery-eye and you can see these reports of your battery use.

Re: Battery consumption analysis
 

no, where does the reports appear to battery-eye?

is there a random number generator? does it read something from dbus? does it read something from /dev/? etc

Re: Battery consumption analysis
 

I'm just guessing here but if I'd make that program, I'd use
every 10 minutes or so, and store the values.

Re: Battery consumption analysis
 

Looks like it reports 3 values from "lshal -u /org/freedesktop/Hal/devices/bme" :

battery.voltage.current
battery.reporting.current
battery.charge_level.percentage

Re: Battery consumption analysis
 

Large Kudos to the author. This application has lots of potential (estimating usage at current level etc.). Thanks!

Re: Battery consumption analysis
 

It's interesting looking at the different graphs - the mAh graph doesn't show any drop off after charging, but I'd guess this is "backdated" after unplugging, as this bme value doesn't update while plugged in. Similarly, the % graph is presumably interpreted from the mV graph during this period (the mV figure seems to be the only one updated while plugged in).

Re: Battery consumption analysis
 

It could be improved a bit to run as a daemon (start, stop) and store data in a file. Options to send them via e-mail? Or even you one can connect to the tablet's battery-eye daemon and read the values remotely and showing them in real time?
May be store a top-like log every 5 mins (so you know what's running?)

Lot's of ideas for improvement...

Re: Battery consumption analysis
 

If this uses data from bme in hal, which comments seem to confirm, then these things probably apply:

Meter is reset and started from a new, guessed state on reboot.
Meter doesn't update during charge, except jumping to full when full.
Meter corrects itself when near full, near empty, or right after charger has been disconnected, and the values aren't vey accurate.


In general I suspect the N900 meter uses a voltage profiling methodology, which has the advantage that the actual battery meter uses less power, but is less accurate. In particular I'd think it's not entirely linear, and this will probably get worse as the battery ages...

Re: Battery consumption analysis
 

this is what I was after with my questions: accuracy of values monitored.

Re: Battery consumption analysis
 

Very interesting! Thanks.

Re: Battery consumption analysis
 

I think the reporting.current mAh and percentage values in hal are just guesses based on battery voltage...

Re: Battery consumption analysis
 

nice thread;) app needs a lots of features

Re: Battery consumption analysis
 

probably not the most accurate battery monitoring, but the graph it extrapolates on a timeline does allow you to better determine when and where your battery life is going

Re: Battery consumption analysis
 

I'm in too! will report in the morning how it went ;-)

Re: Battery consumption analysis
 

Hah,

Here is my system. Basically first 3-4h light usage on Wifi and after that in sleep. Something is pretty fck up :) I have installed stuff from different repositories, but when checking stuff with top i noticed pulseaudio is hittin constantly 1-2% cpu. Going to restart and charge battery full. Lets see how it goes after that.

So looks that right now i´m probably hitting this bug:http://talk.maemo.org/showthread.php?t=34863
http://talk.maemo.org/showthread.php?t=34863

Seems that there is no bugzilla entry and after reading couple of timeless comments to poor reporters i think that i will pass this one also :| I cannot reproduce this or have enough time or adequate skills to report.

Re: Battery consumption analysis
 

My device almost seem to discharge as fast as it charges... I've only been listening to music for like 45 minutes... Browsing for 45 minutes... And sent like 5 sms'es... Of course it could be the wifi bug... but I've been running with it off... using a program from extras-devel which should unload it from the kernel... I just activated wifi again :) But when I look at your graphs this doesnt seem normal at all?

Re: Battery consumption analysis
 

Wow, there's already a thread for my app. =)

I'll spend my reply on answering some of the questions on this thread.

(1) What is being measured, and how?

As it has already been pointed out, the data is collected with lshal. The daemon listens for the OS to report that the values have changed (there's a DBUS signal when HAL updates the values), and records the new values and current time. So the raw data it collects is a group of points. The graph in the GUI is created by simply connecting the points with straight lines.

Now, everyone who knows something about scientific data visualization also knows that simply connecting the dots is just plain wrong, as it makes the visualization to suggest that we would know something about what has been happening in between the observations. I know that too, but it just makes for a nicer graph. =) And for the most of the time, if the data collection is continuously running, the actual values are not that much off, and the connected line gives an ok estimation.

However, when the data collection is not active (e.g. phone off), connecting the dots from that time might suggest that we can also record accurate data from a power down period. Obviously, there is no data being recorded when the device is not powered, it just looks like that in the graph. In a future version of the app (likely the next one) this effect will be fixed by not drawing the graph for periods when the daemon has not been running.

As you might expect, a 3rd party application like this shouldn't be relied to give 100% accurate measurements. So don't depend your (battery) life solely on this app.

If someone cares about the raw data, it is being stored in an sqlite db in /home/user/.config/battery-eye/ . If you know your SQL, the schema should be simple enough to work with.

(2) Does it shorten my battery life to I use the app?

In short, yes. Everything you run on a battery powered device will consume power from the battery.

However, the background data collection is designed to use minimal amount of resources (it only wakes up on DBUS signals and an infrequent timer event (60min) and does nothing else than record the current battery levels). In practise, I haven't noticed any difference in my battery life.

Naturally, if you use the app's GUI a lot (I like scrolling the graph back and forth a lot ;-) ), that will naturally have an effect.

(3) My battery doesn't charge to 100%?

Don't blame me, the program just reads the values from HAL as they are. If HAL says 97 percent, then that's what gets recorded and eventually displayed.

(4) My graphs look strange?

Yes, sometimes they do, especially the voltage graph. It seems that the OS updates the battery values more frequently when the device is actively used, and not at all when the device is completely idle. The voltage graph illustrates this effect quite well (first a straight 6~8hr line when I'm sleeping, and then a tight zigzag when I start to read my e-mails and slasdot in the morning).

At some point, I might try to figure out if there's a way to force the OS to update the values in HAL, without consuming too much battery. That might improve graphing accuracy from long idle periods.

(5) It seems that there are features missing?

Yes, it's still under development. I myself don't consider it ready.

BTW, the usual pre-extras disclaimer apply for battery-eye:

It's still in extras-testing, it may screw up your device!

Re: Battery consumption analysis
 

It actually does this. ;-)

Re: Battery consumption analysis
 

Hi juise,

as I said before, big thanks go to you for making this application. I am very much interested in the further development and will be glad to test :)

Best regards,
Corwin

Re: Battery consumption analysis
 

I'll say the same as Corwin, i would be glad to help test it out, i just got my n900 a few days ago (sweden + 3 =loss:()

my disclamier would be the fact that i have no idea around the xterminal or linux system at this point in time (first time using a linux OS is on the phone tbh) but if i could do anything to help the advance of this great app i will:)

And also i'm hoping in on the reporting as of today.

Regards.

Re: Battery consumption analysis
 

More likely this bug if it's only 1-2%. Can easily fix with "killall tonegend" or "killall pulseaudio" but it's a drag to have to check for this each time you take a call and care about battery life.

Fixed in next firmware apparently.

Re: Battery consumption analysis
 

i have this app installed but no time yet for experimenting. i'll post an update too shortly, maybe with my regular and massive use.

Re: Battery consumption analysis
 

I have made some things since yestersday:
i switched from dual to gsm
i removed the weather tool from the desktop
i uninsalled some things ...
I disabled gps

So yes , like in the picture i will add, you can see , with musik it last now longer ...

But what is wonderng me , from 24:00 to 6:00 i turned OFF the WIFI with some extra tool and turned on the OFLINE mode. SO , why !!! he still looses so much energy in 6 hours ??? If i would draw that line , the battary will be empty in 3 days DOING nothing ... i just cant understand it. I turned everything off , the only thing was on is the alarmclock ... so why would a mobile phone, while doing nothing get empty in 3 Days !!!

so here is a screenshot i made in battery eye.

http://img64.imageshack.us/img64/4994/screenshot06.png

The time is what i realy did , you can see it nor 100 % on this picture
1) The battary was full, WIFI OFF and OFFLINE MODE
2)´Alarm, serving 10 m
3)Musik about 30 m nothing else
4) Tunrned OFFLINE MODE OFF and switched to WIFi in the school , used it in the lesson to goole things
5) turned wife off
6) Musik ...
7) Just turned on wifi , and internet , and musik ,and some windows ...

Re: Battery consumption analysis
 

razialo, to make things clear. You charged N900 and plugged it off from charger 24:00?

An it looks like it discharged itself 5% in 6 hours so it would make from 95% to 5% in about 4 and half days.

.edit
So it looks like now that
http://europe.nokia.com/find-product...specifications
Is BS for some people. We should have more data for this.

..edit
Maybe you should check mAh not percentage.

Re: Battery consumption analysis
 

yes . i plugged it off from the charger at 24:00 ...

yes 4 days ok ; but some people sad here on fotum , their battary lasts a week ... so i hoped to get same results -.-

278 max stand-by ... that are 11 days ... and i was in the offline mod -.- so 4 days real and 11 in the tech specs ... hmmm -.-

Re: Battery consumption analysis
 

http://img64.imageshack.us/img64/2563/screenshot08.png

http://img43.imageshack.us/img43/5082/screenshot09c.png