The Challenges of Achieving Accurate Battery-life Results for Mobile Devices

I’m in the process of evaluating EVGA’s Tegra Note 7 tablet, and of course, that involves a lot of testing. The last time I took a look at a mobile device was with NVIDIA’s SHIELD, and for that article, I established some simple methods of testing the device’s battery-life. Most notably, I ran a 720p movie for 1 hour, and estimated the total number of hours that such content could be viewed before the device would take a forced nap.

After that one-hour test, the battery was left at 95%. It’s not hard to do the math here: The SHIELD should last 20 hours. That’s an impressive feat to say the least. But is 20 hours truly accurate? It could be, but it’d be for a “best case” scenario. One thing I’m certain of now is that it’s impossible to deliver a completely accurate result.

After running similar tests on the SHIELD again, along with the Nexus 7 and the Tegra Note 7, I’ve noticed some oddities. In one particular test on the SHIELD, the battery showed 98% full after the one-hour test. That’s a bit hard to believe; there’s no way the SHIELD could last a staggering 50 hours without some sort of backup battery. So, I charged back up, and retested. The second time, the battery sat at 95%. After two hours, it was at 92%.

As you can see, the scaling isn’t linear – at all. Even though I’m playing the same film over and over, and charge the device to a definitive 100% each time, it’s as if the device needs to work out kinks as it begins to lose battery power. Ultimately, this has led me to settle on a 5-hour test instead. Even that’s not perfect, but again, nothing can be. Not even running the film until the device is dead, because not everyone runs the same film and settings, and even then, I could wind up with a slight variation if I redid the entire test.

Nonetheless, after a 5 hour test, the SHIELD sat at 72%. This more or less suggests that the device should get about 17.9 hours of battery-life. Does it?

Not quite, though it comes close. In a special test, the SHIELD lasted 16 hours and 20 minutes playing back that 720p film. That’s off from the 20 hours I speculated in the review, but I do believe a couple of more hours could be eked from the device if the brightness was turned down further, and maybe the audio. Our methodology involves us setting the brightness to 50%, as well as the audio. On this particular device, both of those values make for a pleasant experience in a normal household. You could afford to decrease both; but countering that, you’d want 100% on both if you were on an airplane. See how complicated this is?

Something this testing does highlight is that the SHIELD has great battery-life. In the same 5-hour tests, the first-gen Nexus 7 dropped to 56%; the second-gen Nexus 7 dropped to 49% (thanks to Jamie for this test), and the Tegra Note 7 dropped to 54%. Of course, the SHIELD is a beefier device than all of the others, so better battery-life could be expected. Even so, to be able to watch 720p content on a flight from Toronto to Hong Kong without the device dying is impressive.

That all aside, given what we’ve seen, all movie streaming will be conducted for 5-hours in our tests, which involves both having the film stored on the device and a home server to be streamed via Wi-Fi. We’ll also continue to do our game testing, which will involve a Super Mario World loop via SuperGNES, as well as Beach Buggy Blitz, an Android game; both left to sit for two hours.

Over the weekend, Jamie and I went back and forth constantly on this battery-testing nonsense, and I’m left with the feeling that we simply can’t be perfect. And that bugs me, since that’s what we strive for in our testing. Nonetheless, 5-hour tests for movies and 2-hour tests for games will certainly be more effective than simple 1-hour tests.