Until battery technology catches up with today’s electronic gadgetry, we have to constantly be vigilant of battery banditos that snatch our juice before we know what happened. While I am not a fan of battery conservation measures that may affect my user experience, I do aggressively search out/stamp out battery sappers, those apps and functions that steal millivolts without purpose. Knowing something was not right, I set out a month ago to figure out what was stealing my juice. Now, 31 overnight tests later, I have more questions than answers, but that’s something, right.
My unscientific testing was pretty simple. Turn off different radios and background tasks on my phone and let it sit untouched overnight. Then record the results in the morning. All my tests ran for 7.5 hours, about half a typical user’s day, and about as long as I could go without touching my Surround. The first dozen days revealed some unusual results, till it was suggested that I should probably be starting each test with a full battery, rather than a starting/ending point. Turns out that batteries drain differently at different points, so you need to eliminate as many variables as you can to garner anything at all from the test results. For the next 19 days, I fully charged my phone, making sure my indicator light was green, and I was trickle charging for at least 15 minutes. Pulled it off the charger and ran the battery down to 99%, at which point I would turn off the screen and start my test. About as accurate as I could get without real test equipment.
The first step for any testing is to create a baseline, so you have something to measure against. For the purposees of my test, I disabled/removed all of my background tasks and turned off WiFi and Data, leaving only my GSM radio active for voice/text messaging. Several repeat tests yielded the same result; a 10% drop in battery (with one at 9% and one at 11%). Don’t know exactly what causes this substantial drop (putting my phone in Airplane mode with ALL radios off yielded the same result) but I believe it’s simply the way my battery/device reacts to the battery meter, and I am not really losing 10% of my battery with nothing running. I have observed that my battery draws down the first 10-15% very quickly and then levels off, moving much more slowly after that, speeding up again when I get down around 40%. Note the graph below that shows how my battery dropped quickly at the beginning and then leveled off, all while it was sleeping. I am sure that each device will yield a different result, and possibly even with different batteries, which is why you need to establish this baseline on your own device before doing any other testing. The baseline, being a constant, is deducted from any of the results noted below.
I ran various tests on background tasks, my WIFI & 3G radios, and a combination of both. Each time I thought I found a battery bandito, a second validation test would produce a different result. Frustrating would be an understatement. Despite the variances I did draw some conclusions as outlined in the attached chart. For example, my WiFi radio consumed about 2% per test, or about 4% per day, so don’t worry much about keeping it turned on 24/7. My 3G radio on the other hand burned up 5-6% per test or 10-12% per day, and that’s without moving the phone around. Note that was with it sitting in the best reception location in my house, typically showing 3, or sometimes 4 bars. I am sure the actual idle 3G drain per day, moving in and out of signal areas, is much higher. I did test the Homebrew Battery Status app, running at both 10 and 30 minute updates. Total drain after 7.5 hours; 1%, or as I can’t see decimals, probably less.
The reason I started testing, and most likely a bandito, turned out to be Calendar Calculator. The app had a background task, but I could not toggle it on/off. And there was nothing in the app settings requiring a background agent. In my early tests, which were slightly skewed because of the starting point. with WiFi & Data turned off, I dropped 25% overnight (15%, plus the 10% baseline). After deleting this app the overnight drain dropped to 10%, basically representing my baseline number. Unfortunatley, or fortunatley, when I reinstalled the app to validate my initial findings a few weeks later I noticed that it had recently been updated and despite several tests, it does not appear to cause any drain at this point. The app now has a toggle in settings to turn on Live Tile support, which I currently have disabled. The moral here is to be wary of any Background Tasks in the "Advanced" section, that you cannot toggle on/off, and are not associated with something like a music app that may simply want to run in the background.
In my last few tests, I turned all seven of my background tasks on; Battery Status, CleverToDo, Fox News, Mehdoh, Network Dashboard, USA Today & Weather Channel, along with WiFi. Total drain after 7.5 hours; about 4%, or 8% a day. Not much to keep those fruity tiles fed. Turning Data on chewed up a bigger chunk of the battery (22%), which is no surprise. The bottom line though, is that "most" background tasks consume very little energy, all less than 1% per day. So turn them on and enjoy all that Windows Phone offers. Although I wasn’t looking for it, I did discover that Fox News will not update unless Data is turned on (the background task turns itself off when on WiFi only), so for that app you will be using Data regardless of your WiFi status.
A few of my tests varied more than expected and those results were tossed out. For example, one night I set my phone down with only Data turn on. I was surprised to find only 7% battery remaining the next morrning. Somehow, I completely lost my cell connection and my AT&T tile indicated, "Emergency Calls Only". A soft reset remedied the connection issue. Might explain some of the mystery power offs that some users have experienced. I do suspect that the GSM/Data connections are the biggest cause of variability in battery performance. The silent banditos that snatch up your power when you least expect it. Unfortunately, there is not much we can do about that.