Graphical Analysis on Battery (3G vs WiFi, Black vs White, etc)

I was able to add some real time graphing to the battery calibrator app where the current draw thru the battery is sampled every 2 seconds. This gives a very fine tuned accurate picture of what the battery is doing. note the battery driver had to be edited to allow for current readings taken at a faster sample rate, so all that work was done initially.

This first test is the phone switching from connected to 3G to switching over to wifi. notice the extreme difference between the spikes and then how smoooth it levels off around the 14 second mark. shows how much battery is saved using wifi (i'll come to this more later).

Switch from 3G to WiFi data by RogerPodacter

The second image is just showing the phone left completely idle 3g data, but with the screen on, min brightness, and timeout set to never. curious to see those regular periodic blips of current draw. notice -170mA is about spot on for the black amoled screen showing all black color like that. and this shows how efficient 3g data is during idle when no data transfer is taking place. This graph view is what was displayed during this time, mostly black.

Idle 3G Connection, Screen on Min Brightness by RogerPodacter, on Flickr

the next screen shot shows switching from screen brightness at minimum to maximum brightness, but showing the same black graph image. notice the slight increase, but it stays dead flat still. it switched at the 36 second mark.

Switch from Min to Max Brightness, Showing Black by RogerPodacter, on Flickr

This screen shot shows what happens to amoled when displaying a fully white image at maximum brightness. the power draw goes thru the roof! it started at the 16 second mark. i used the app "dead pixel detect" to display solid full screen colors, this one showing white.

Displaying a Solid White Screen by RogerPodacter, on Flickr

Same as above, but now switching back to the black graph view, the current drops down to normal. Swtich happened at about the 15 second mark.

White Color Max Brightness by RogerPodacter, on Flickr

The next screen shot shows why wifi gives so much better battery life. this shot shows receiving a new gmail email message over 3g data connection. right at the 20 second mark the email comes in, and the current draw spikes over -500mA. when the same email message is received while connected to wifi, there is not even a spike at all, it stays flat.
Receive Email over 3G by RogerPodacter, on Flickr

This is due to the much faster latency that comes with wifi. on 3g, the 200ms ping time, each direction to and from the server, all while the radio is fired up, is what causes the spike. now think about that happening all day long with all your google services constantly talking to their servers.

This next one shows loading a web page with a light color, and scrolling around. I loaded a thread here at XDA with the stock browser, without invert color turned on. look at the enormous spikes.

Load and SCroll Light Color web page by RogerPodacter, on Flickr

And finally, this is loading and scrolling the same web page, same thread, but with invert color option turned on, so the color was basically black and dark. (For these last 2 web page tests, i forgot to turn the screen brightness back from max to minimum brightness. oh well.).

Load and Scroll Dark color web page by RogerPodacter, on Flickr

If anyone wants more test comparisons done, just post your request. For the above tests, i didnt think things out too much, so i'd like to really do things over and show more/better scenarios of how power draw is effecting our phones' battery.

I had one more, from today. Watch what happens when i go and read a topic on the engadget app (entire white colors) while i was outside at max brightness. power draw sky rockets up, then back down when i return to the black graph view.

Browse the White color Engadget App by RogerPodacter, on Flickr

fyid said:

great data - can you check some apps, like words with friends & weather channel (background widget). words with friends tops out my cpu for no good reason.

Click to expand...

Click to collapse

how exactly would you like it tested? i've never used words with friends, and i never installed the weather channel widget either. the second one might be harder to test. is there a way to tell it to initiate its update or data transfer or whatever it is that it does?

zeus_chingon said:

The current graph is a lil bit confussing. Is the rightmost the oldest plot? and the leftmost is the most recent?

Click to expand...

Click to collapse

The right direction is older and the left is more recent. The full current graph only shows in seconds, around a minute or so. So this is like a true profile of all power, in real time, going thru the battery's resistor, and hence device power draw. So its great to see exactly what different tasks do to power behind the scenes.

I've been doing many more tests just for fun. My overall discovery is that the screen is bad, real bad, and there's a disproportionate difference between min and max screen brightness. Min is really good, but max is so far off the charts bad that its horribly power hungry. So I'm practicing keeping locked at min brightness. Then comparing things like CPU usage loading a web page in the background, then over WiFi then 3g, etc.

I'm enjoying seeing the bad offenders.

Jack_R1 said:

Excellent analysis to visually confirm what has been known and constantly overlooked by people complaining about their battery life: the 2 major power-hungry things in the phone are the display and the 3G data - when it fires up.

You could also show GPS power draw, to have a good comparison between values. I'd guess that running Google Maps on full brightness during the day over cellular data will show near the maximal power draw possible from the device.

Click to expand...

Click to collapse

Good ideas, I'll do a gps test next. I've already done it but with no screen shots, and I can say it really puts I high load on the battery by far.

I also just now did a 3g phone call test, and it was higher than I thought it would be. I'll post the screen later.

I'm now a huge proponent of min screen brightness. If you can use min brightness then do it. If that's too dim during the day then 40% gives the best combo of still usable with decent power. But auto brightness, or max brightness which will happen in sun light, is just totally off the charts bad. And when loading a web page outside at max brightness, the power is thru the roof.

So I'm now locking my screen at 40%.

Edit: I also did color comparisons. Full screen of white, red, blue and green. Nothing too crazy but still cool.

Wow I'm finally trying out stock gingerbread and the idle power readings are significantly lower than froyo. On froyo I'd see around 20mA and I'm seeing about 7mA now consistently. My app doesn't work on stock so I don't have fine control but it definitely makes sense from reading what people are saying.

tangcla said:

Nice little review.

Out of curiosity, do you think there's much difference between AMOLED, SAMOLED and SAMOLED+ in terms of power consumption?
I'm eyeing the Galaxy S II as my next phone and online reviews all rave about how power efficient it is. But we all know how that isn't entirely true!

Click to expand...

Click to collapse

Samsung claims a 30% increase in power savings, bit I don't believe it for a second and here's why.

First, both the nexus one (amoled) and nexus s (super amoled) both get IDENTICAL browsing test results, both make it 3.7 hours at anandtech. Nexus s has slightly larger screen but also has larger battery.

Second all you have to do is look thru galaxy, nexus s, and galaxy s2 forum. Not a single one has anyone getting more than 4 hours of screen on time in the battery threads. Not a single bit of improvement, at all, between the amoled technologies.

Then to rub it in, iPhone 4 crushes it getting a full 10 hours of browsing time on the same test.

LCD is becoming my preferred choice more and more. Testing an HTC inspire with slcd, it pulls 250 mA Max brightness. But my test results in this thread alone shoot up to 900 mA just loading white pages.