Help Terrible battery life, firmware issue?

edge or gprs, it wasn't important in my question. my point was that battery saving doesn't require messing with apns if your measurements are right, i hope they are.

Ahh, so there's not much of a battery difference between edge / gprs?

I didn't know that... I assumed that GPRS was less power hungry than EDGE/2G was less power hungry than 3G.

I'll check that one again tonight, but as far as I know they're right.

DrCereal (if you're still looking at the thread). I haven't seen sleeping issues with my Galaxy (admittedly I've not had it as long)... I've noticed a few apps try and prevent the phone sleeping but if you're careful with them it will sleep most of the time...

Interestingly, a battery graph app I saw stops the phone sleeping - so if you were using something like that it might actually cause the problem you were trying to look at.

ZeeToo

Call me mental, but I found one thing that made a huge difference with my Magic. Or so I thought.

So I used to listen to music on my way to and back from work (1hr15 commute or so). My battery used to get hammered. I'd have it set to 2G, no GPS, no WIFI, low brightness, quick time out. I'd usually get through the day OK, but often by the time I got home I was on the red.

Then my headphones broke one by one which left me with no music. So then I started using my old Sony Ericsson (which I charged to full on monday and is still above half! Though it has no working SIM card in it, though! It does have an orange one, and it does connect to a network (i get signal bar and stuff) but it's invalid) as a de facto MP3 player. The headphones are better anyway, and this means I can free up space on my Magic. Even when I get new headphones (for hands free) I might keep doing this, as it saves battery even if I have to carry both devices.

Anyway, the first day of my doing this, I expected big, big increases in battery. But nope, nothing. In fact, if anything it was worse. My phone actually lost all its battery on the first day. I have background data use turned on, and auto-sync, but that was always the case. Likewise, I always terminated apps. The only single thing I found that was different on that day, aside from not listening to music, was having an alarm on. Normally I set it every night before bed, but this time I set it to run from Monday --> Friday. The next day was much the same. The day after that, I turned the alarm off in the morning and bam, problem solved. My battery is now often above 50% when I get home.

It seems absolutely mental that an alarm could do this, and maybe it's just a coincidence and it's something else I'm missing - but maybe try that, if you have an alarm on?

Interesting...

I have the alarm on most of the time - it's not too reliable as the phone has shut down overnight before, but I didn't think that would have much of an effect. I didn't mention, but I had the alarm on for all of these tests - just goes to show what you'd not think was important..

Another variable to try out, thanks

I finished a hack'd mod to the net meter app, it's nasty but it keeps a graph record of a few valuable items over the up to 24 hrs the original will track for.

I noticed that the netmeter chucks away the difference between system and user processes, so I re-extracted and displayed the system process.



I'll hopefully post a snapshot of the battery (on my actual phone rather than the desktop sim) tomorrow morning once I've finished the overnight run.

ZeeToo

See now this is odd, i turned of my Background data last night after the phone was fully charged to see how long it would take before full discharge. It seems that it didnt make any difference, 14hrs of runtime and im down to 64% and it was the same with the Background data on.... Any toughts? Do i have a defective phone?

Ok, I tried again last night with "wifi off, sync off, background off, apn on, alarm off"

This time I got 76% - 57% over 8hrs exactly.

I was running my modified netmeter to track the battery and CPU usage but other than that it was clean..

However looking at the graph it might not be as bad as all that..

I did a bit of a stupid thing, I was hoping to charge the phone so I got a good comparison (ie 99% start on both rather than one at 99% and one at 54%), only after having it plugged in a for a while I noticed that it wasn't charging very quickly at all so I unplugged it and left it for half an hour to drop the initial spike... Only I'm suspecting that it didn't do that for a while, maybe something to do with a change in the battery temp?

Anyway, here is a bad shot of the graph (sorry about that but I can't take a direct snapshot as the android SDK I have doesn't like 64bit for debugging) and the flash washed the screen right out.



You can see the battery climbing at the start as it's charging.. Because this app doesn't stop the phone sleeping we zero pad the information once the phone wakes up again based on the elapsed time, the zero in the middle is the most of the time the phone is sleeping.

Interestingly it appears jittery because the graph view is smoothing the spikes that occur about every 5 mins (zooming in shows this, but I've no photos, sorry) - I don't know if the phone is auto waking up every 5 mins, or if the app was doing that.

I added the green line to show something interesting - as far as I can tell the battery dropped off back to around pre-charging levels quite quickly and then stayed fairly constant... I'll really need to run the test again with the battery fully charged rather than part charging it as batteries are fickle (and it was on about 54% before I started charging it..)

Bozzy, what else was set on or off? I've not had the phone long enough (and I've had to throw away data when the phone shut itself down) to get a very accurate picture of how the battery life is changed by what's running.

I'm actually hoping to be able to write an app that'll sit in the background and just turn off things that aren't needed when the phone's idle (and automatically turn them on again when I start using the phone again) so that I don't manually have to do these things to preserve the battery - it's the main reason I'm gathering this data.

ZeeToo

turning off background data does not help a lot - it seems that the phone still connects to 3g for some reason. The best solution is to either disable 3g (so it only uses 2g-EDGE) or use APNdroid to enable/disable 3g when you need it. After that I have no problems with battery life. I don't think alarm or anything else is important - 3g seems to be the main thing.

Ive also found out that something is causing my phone not to sleep. Total time spent without sleeping: 4:45 hrs.... Time spent with screen on 1:59 hrs....

Well I ran overnight again with GPRS on, wifi off, background off, sync off, and it appears to back up my original numbers.

99% - 98% after 8hrs 19mins..

On my phone at least (and I don't know about the impact of EDGE or 3G yet) it appears that background data is painful rather than the need for APNdroid.

Bozzy, do you have a lot of apps running? Or at least installed, something like "tasKiller" is good to work out what's randomly started for some reason. Also try netmeter (also free on the marketplace) as it'll help you know when your phone is actually sleeping.

ZeeToo

No, i only run few when im actually using the phone and I have a task killer and i kill all the apps before i put the phone in my pocket

Battery problem seemed to be fixed with the h7 build but actually it isn't. There is something that keeps the CPU awake and this is the one that's draining the battery. You can see in battery history (you can access this with "spare parts" app or type *#*#INFO#*#* in the dialer) the time when the cpu was running and the time when the screen was on. Yesterday when I first look at the battery history the device was running 99,6% of the time from the last boot while the screen was on for around 10% of the time. Right now I don't have that problem anymore I've used several task killer apps and killed everything suspicious.

Ahh,

I've not had processes keeping my phone awake... I use an app to monitor what the phone is doing and keep track of anything that might stop it sleeping.

Battery life is ok atm, but I could wish for better... I'm probably getting about 2 days out of each charge, but it's new and I'm using it a fair bit...

Cool phone though

ZeeToo

Ok,

After playing around with the phone and android itself for a couple of days ive realized that its not the battery, the phone or the apps sucking it up. The problem lies within android itself and how it calculates the usage of the battery.

Let me explain. Yes having apps open 3g active, wifi, etc... open sucks up battery, but not as much as it looks, see the problem is, with all those applications open or active the system starts calculating battery life based on those applications, therefore brings down the battery life significantly. So now your phone is sitting at 1% and then it shuts down because thats the shut off switch and system thinks it cant sustain power anymore, as to where in reality you got bout 10-20% battery left.

Thats why you go from 80 to 75 with bout a min of wifi surfing and as soon as you turn it off and kill all apps the battery jumps up again.

All in all google needs to change the way the battery life is calculated in next firmware upgrade (Donut), and only then will we see significant change in the battery.

Ahh, that would explain some oddities I'd seen.

I didn't quite put it together like that Bozzy, but it makes sense...

Only I'm not so sure it's android thats an issue, rather than the battery tech itself (I've looked into some of the lipo batteries, it's similar to the lithium ion batteries I think are in these phones).

As you increase the draw on the battery the voltage will drop, and I think it's the voltage that is used to determine how much is left, so heavy usage will depress the voltage and make it seem like there's less left. Whereas the actual battery is measured in amps, not volts I just don't think there's any way to determine the number of amps the battery has left, just the voltage it is supplying at.

Once the voltage drops below a specific point then the phone can't run anymore, I presume the voltage regulators will turn the 4.something volts fully charged and the 3.something volts when almost discharged into the same voltage for the phone to use.

Even so, I agree, I hope that google write a better algorithm to read the battery state and convert it to a more realistic figure.

ZeeToo.

I think its Android because I've observed other phones with different OS and none of them do that, as to where all Android users (Hero, Galaxy, G1, MyTouch) are experiencing the same issue. But I could be wrong and its like you say.

(On side note my Macbook, Windows and Linux machine do same exact thing, but only with the clock of the battery, not the %, so as time before its empty varies on usage the % is constant

Hmm, could be.. (I agree with you it's something android needs to improve, but I don't believe we can ever be accurate with current battery tech as there's no actual reading of the energy left in the battery).

I've done some electric racing boat fun a few years back and had to do lots of things to condition the batteries, (NiMH but from what I read the new lithium batteries are worse for this) so I've a little experience and what I've read on the newer lithium batteries tends to show they're nastier, can't discharge too quickly else they'll explode, can't discharge too deeply (depressing the voltage below some value per cell is bad for the battery) can't charge too quickly, much more temperature sensitive and rather than 1.5v (or 1.2 really for the rechargables) they run from 3v to 4v+ per cell.

My gf has an iPhone 3GS and she's seen some similar problems (not quite as bad as I've had with the i7500) so I guessed it was endemic..

But as you say, maybe not

I ran last night to check out the difference with the alarm on or off:

Wifi off, sync off, APN off, background off, alarm on
48% - 48% after 7hrs 44mins - pretty much the same result as the alarm off, so at least on my phone it doesn't appear to make a difference.

As I was pulling up the details, and on my drive (not whilst I was driving ) to work today I noticed that the battery was bouncing all over the place, from 48% to 32% to 44% etc, I suspect this is as Bozzy says, android not compensating for the voltage shifts in the battery.

Also, just making a copy in this thread of Hawklord's results so I don't lose them (thanks for the info btw ).

Wifi off, 3G off, sync off, bacground off, all apps killed (phone not used for that time).
99% - 99% after 9hrs 40 mins

ZeeToo.

And todays test wifi off, background data/sync off, 3G off but leaving all the programs on [factory test (?), wssyncmlnps, servicemodeapp, calendar, chompSMS, google mail, email, settings, taskiller].

99-97 in four hours (including several pings (didnt check screen whether it came on for a while)) when emails arrived (email program checking every 15mins).

Anyway it seems to me that the battery is actually fine on a standby mode that works for calls, texts, emails providing the wretched sync & 3G is off.

Of course I am not advocating leaving the phone like that all day, rather leaving it like that when one is not using it IF one wishes to prolong the battery life

H

ps havent tested effect of background data being on or off qualitatively (and ZT I didnt have APNdroind on yesterday for full-retro standby test)

pps a cautionary tale about forgetting to sync sometime - i thought i had sent an email (from "where am i") but hadnt ... would be nice to have a startup macro type thing to do this

ppps i know wrong thread but driving me crazy - no way to delete all the emails in my inbox in the email app?? When I am at pc sorting them and deleting them I then dont want to go thru 20 one by one on the phone deleting them all over again...

Hawklord,

I don't quite follow, did you have a data connection active? By APN off I mean that I've used APNdroid to alter the APN settings so the phone can't get a data connection - it's what I assume you did for a retro dumb phone?

If you can clarify I'll edit the post so it's correct

Thanks.

ZeeToo

So for retro phone I turned 3G off, wifi off, syncy&background data off ... I didnt use apndroid (wasnt sure whether doing so might block recpeipt of MMS). Anyway pretty similar impact.

Anyw progress on finding what bacground data sync is up to when sync is disabled and no programs running under taskiller?

Thanks for the clarification Hawklord

I've not tracked into background data yet - I need to find out how to trace data going into and out of the phone, mostly I'm still black box testing things with recording results and how I use the phone.

Once I work out how (and it should be possible as it's a form of linux afterall) I can write an app to track it and any processes that fling data around..

I think something is communicating via the background data network, but what that is I don't know yet. It might be worth taking a look at taskiller's ignore list as that's supposed to be a list of system procs we shouldn't kill there - mebbe there's the culprit.

I've noticed things like market launching themselves, potentially it's something to do with checking for updates etc.

Still, I'll keep you posted

I've finally got 3G working on the phone so I can get some figures with that enabled.

ZeeToo.

I have done some testing with Background data on and off, with the rest of the settings being same, the difference is 2% at maximum throughout the whole battery life.

And now im playing around with APNdroid and going on 33 hrs with 52% left and the ussage of bout 2-3 hours of wifi on and surfing the market and browser, few phone calls, few text messages, 15 min Ebook (Aldiko) and bout half an hour of farm frenzy (its soo addicting)

Cool, thanks very much for the info Bozzy

A couple of questions though - the background data on and off over the lifetime of the battery, what did you mean by 2% difference?

Did you mean the phone lasted 2% longer (or not as long) with background data off?

Do you have the figure in hours or something - and what were the other settings, APN, wifi etc..?

I'm impressed with a full battery life figure - the only info I have is the overnight tests I do and other similar tests that people have added.

Thanks again, every bit of information is useful

ZeeToo.

P.S. My apologies for asking lots of questions about the testing environment, it's a bit of a habit from doing a lot of testing at work and I'm trying to make something of a model of how the phone responds in different circumstances.

Oh my bad,

I meant i am left with extra 1% with background data off in the same amount of hours it took the battery to 50% and 2% by the time the battery was "supposed to" be empty.

All other settings were off as i was targeting only background data. So 3G off, WiFi off, Sync off (i sync manually)....

How do you do overnight tests? What do you test it with?

And not a problem on the questions I do the same thing. So if I'm not clear on something ask away...

Ahh I see Do you happen to know how long it took to get to 50% and to the 5% empty?

My overnight tests aren't exactly a high tech solution, at least not yet.

I check all the settings, update a text file with the time and the battery percentage then shut down any programs I'm not expecting (I'm just using Note Everything).

One other thing I do is I run an app I've written (modified from netmeter) to monitor the battery and CPU overnight - it lets me know how the phone's been sleeping and if the battery decayed evenly over the time of the test. It's a very low impact app and doesn't change the sleep behaviour of the phone but it helps me catch things like mistakes I make with the charging profile of the test - there's a screenshot of an early version of the app in an earlier post. I'm still updating it as I think of new things to track and things that interest me about how it works.

All my tests atm are around understand the no-load / sleep profile of the device, unfortunately I only have one phone so can't run too much all the time - and this is why information from people like you is so useful, thanks very much

I am wondering how to do things like load tests, but I'm not sure how to get reproducability unless I just write an app to simulate a workload - which I will probably do as I get more experience with android apps.

ZeeToo.

P.S. I know this much effort into understanding the phone seems insane but I hope to write an automatic app to do all the stuff that works to automatically compromise between functionality and battery life.

you can use 'top > out.log', then parse it on pc....