How To Properly Benchmark Your Android Phone's Battery

It's common that whenever a new technology finds its way into mobile phones, a company feeds us a line about how much better it will be for battery life. They know how to get us like that. The Achilles Heel of many smart phones is the battery life.  If a phone can't make it through a day under your unique usage pattern, it is of little use. When AMOLED displays started making their debut on phones, the battery life was supposed to be much improved.  

A new analysis from Laptop Magazine shows that, in fact, many AMOLED Android phones perform worse than their LCD-using cousins. The Droid X with its large LCD display bested the smaller AMOLED handsets. Informative as this is, it was an artificial benchmark. If you really want to know how your battery performs, the best way is to just use the phone, and track your battery usage. Read on as we show you how to do just that. 
 

Tracking battery use to benchmark

Since Android can run processes in the background, it's a simple matter to dump your battery level data into an app for later review. One such app is Battery Graph. This is the simplest method of tracking you battery as it discharges. And that's just what you should do. This tool uses virtually no battery power we can tell, so it's safe to leave it running. You will be able to pinpoint the time you remove it from the charger, as well as your usage pattern over time. If you check out multiple days, you'll likely notice a pattern emerging. This is what you're interested in. 

To really benchmark your battery, it's best to let it drain all the way down, then plug it in the check the graph. This gives you a good idea of your real battery life. Why bother with the graphing app, you might ask? It's best to let battery graph run for a few days before running the drain test. That way you can compare the patterns on the day you did the test, to those previous days. It's a way to make sure your usage that day was not abnormal. You want numbers that are indicative of the phone's true capabilities. If there is an unusual drop in battery one day, you might need to look for a problem before declaring the result valid. The graphs also provide a solid record of how long the phone lasted each day. Battery Graph even supports export to a .CSV file. 

To have the most reliable data set, you might want to do this a few times. Don't overdo it though; you don't want to completely discharge a battery too often as that can reduce its life. At this point, you should have a pretty reliable data set to find an average personal battery life for your phone. Arrange your data and look for outliers, i.e. data points that are significantly different from the rest. If you have numbers around 15 hours, but then a value of 8 hours pops up, it's best not to include that in your final calculations. Rather, you might want to investigate what caused the poor battery life that day. 

If you're comfortable letting software do the benchmarking for you, we would recommend an app called Battery Left Widget. This app actually calibrates itself based on your personal usage patterns. When using battery Left, it is not uncommon to need to wait through a few days of usage before you have an accurate number. As you use your phone with this app installed, it begins to track what voltage readings correspond to which battery level readings the system is reporting.  

When it has properly calibrated itself, the widget will display the battery level as well as the time your battery is expected to die. In the past when we've used this app, it has been very accurate. After you've finished calibrating the app, it's best to turn down the update frequency. The default is 5 minutes, and that can actually drain the battery if you aren't using the phone much. Battery Left Widget is free, but a more full-featured version is available for $1.99. An app called JuicePlotter also offers similar functionality, but the widget is not as informative.  

We'd recommend trying both the manual operation with graphing apps, and the fully automated app-based method to get the best idea of what kind of battery life you can expect. 

Troubleshooting issues

As you endeavor to properly benchmark your real-life battery use, you might start to notice some irregularities. Perhaps a large drop happening at regular intervals, or overall higher drain starting at some point in the past. It's not unheard of for an app update to significantly alter the app's battery use profile. If that's the case, put on your Sherlock Holmes hat, because you're in for some detective work. 

The first place to look is in the standard Android battery usage app. On most phones, this is accessible from Menu > Settings > About phone > Battery use. If you aren't seeing anything terribly out of the ordinary here, best to move on. In the Android testing menu, there is a second hidden battery use app that can help you track down battery drain and get a really accurate benchmark. Just type *#*#4636#*#* in the dialer to bring it up. Then tap battery history to dive in. From this interface, you can view battery use by component. You can pick from CPU, Network, sensors, etc.  

What really makes the Battery history app useful is you can specify the data window as total for all time, or since last boot. The default, and only option in the regular battery use app, is just since last unplugged. If you're doggedly tracking down battery drain issues, you can use Any Cut from the Market to make a direct shortcut to Battery history or Battery use. 

These methods should manage to get you pretty close to a real world benchmark of your phone's real battery life. Android phones can do a lot of great things, but you don't want to take you eye off that battery meter for too long. If you know with a degree accuracy how much longer you phone is likely to last, you can at least have some piece of mind. Let us know what sort of battery life you get out of your phone in the comments. If you are feeling ambitious, try a few of the methods detailed here, and let us know the results tomorrow. If enough people respond with real numbers, we might have a truly interesting data set.