• Introducing XDA Computing: Discussion zones for Hardware, Software, and more!    Check it out!
Search This thread

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
Hello all,

I've been suspicious of the power draw of the screen on the GNex, and so decided to figure out how much a difference brightness made.

This is to test the difference extreme brightness settings make to battery run times. Including text, browsing or having radios on will dilute the results. I'm only interested in the efficiency of the screen and its supporting infrastructure people! :)

Here's my testing method.

1. Charge phone to 100% using a PCs USB charging (should give a more full charge than the wall adapters quick charge, I think, otherwise no harm done). Leave attached to USB cable
2. Switch the phone into Airplane mode.
3. Switch off Auto Brightness and turn brightness to Max
4. Restart Phone. Leave it to settle for a few mins
5. Fire up "Just Pictures" and the image "TotalWhite.jpg" (attached). No other photos in the folder where it sits.
6. Disconnect USB charging
7. Start Slide show. Record the start time.
8. Every now and then (about 45 mins to an hour and bit) I quickly pop into settings and take a sneak look at charge level - don't want to get caught out!
9. Resume Slide Show.
10. Wait until the phone switches off, which is less than 2% charge

Now, it's not a quick process as I'm going from 100% Batt until auto shut off. So I'll be updating this post in installments. Today is the Max Daddy Full Brightness White image!

Full Brightness, White Image:
100% 00h00m
85% 00h39m
60% 01h53m
35% 03h08m
0% 05h00m​

Pretty good I think!

Battery status stated that Screen was 91% and Android System was 9%. Screen on time was the same as the run time. It never went off.

There's a heavy set of disclaimers to go with this though: The White Image isn't quite the right aspect for it to display across the full amount. Check out the image "White Full The Phone.jpg" - it's pretty close though. Just Pictures isn't so full screen that it takes over the Android buttons - Anyone know a 100% full screen picture view BTW? Between the point where it comes up with "Please connect your charger" (15%) and 6% it said that message over the White image and partly dimmed the screen - I'm at work and was in the middle of something when it happened.

Minimum Brightness, Black Image:
100% 00h00m
85% 01h26m
82% 02h05m
18% 08h54m
0% 10h47m​

Battery status stated that Screen was 85% and Android System was 16%. Yes, that's 101%, but there you go... Screen on time was within a minute of the run time.

Again, some notes on the testing: From the "connect your charger" point (i think 15%) to 8 % it had the message on the screen. Just before the end I accidently touched the screen and ended the slide show for up to 15 minutes and the screen went off. Switched it back on and it was at 2%. It then lasted something like 30 mins before auto shutdown. Black isn't Black! check out my the photo "Black Dim The Phone.jpg" - ignore the strange black specs, it was 1 seconds exposure at F1.4, with the camera facing down - think there's some crap in my camera body/lens.

Minimum Brightness, White Image:
100% 00h00m
0% 10h20m​

Again, Battery status stated that Screen was 85% and Android System was 16%. That's the same as when I ran it on Min Brightness, Black Image...

Maximum Brightness, Black Image:
100% 00h00m
0% 10h52m​

Battery status stated that Screen was 90% and Android System was 11% (again 101% also probably due to some rounding going on). Screen was on 100% of the time, Android OS was awake for 100% of the time, CPU time 23 Seconds

Overview:
Maximum Brightness White Image: 5h00m
Maximum Brightness Black Image: 10h52m
Minimum Brightness White Image: 10h20m
Minimum Brightness Black Image: 10h47m

Well, well, what's happened here? Maximum brightness with a black image scored the longest run time - unbelievable, right? Yeah pretty much: On the minimum brightness black run I did check the battery stats a few times. The black image run had a greater Android OS drain (11% vs 9%) compared to the white image. As I switch the phone off in preparation for the test it's possible I inadvertantly bump charged the phone. In addition to this, I was trusting the phone a lot more in this, the final test and didn't check the battery status near the end, so perhaps drawing a lower current happily sitting there doing nothing other than displaying a blank image the phone managed to syphon off the last dregs of power, as opposed to spending CPU time with me rummaging through the settings screens for battery stats. Let's put it all down to being in the margin of error and consider the Black images to have the same run times, despite the Android buttons blazing away on Bright (but they aren't many pixels).

Conclusion:
Back to my original concern, that the screen was abusing the battery more than necessary. I'm mostly happy with the outcome - Black bright and dim screens draw substantially less than a very bright white screen on this phone, as it should. It's surprising to see that on dimmest setting the white screen got so close to the run time of the black screens. It's producing a lot more light than either black screen. It's drawing c5% more power than the black screens, but I would expect a lot more. I'm guessing that the circuitry to run the Screen and the small amount of power used to make the "Black" OLEDs glow dark dark grey draws only a little less than the white screen.

And that's a shame. I wanted to run an app such as "Off-Clock" to have a clock on the screen like the Nokia N8 used to do, but i'm thinking this will eat the battery in 10 hours 20 minutes. Only one way to know for sure....

My initial annoyance that Black never is actually Black whilst the screen has faded. Many of you good folks have pointed out (and after more Googling) that all OLED screens have this, although the reason for it isn't clear. It's either to ease the transition from black to lit by having the LED at the threshold already, power seeping into the circuit somewhere or something else. Who knows, maybe we'd get better run times if the screen was on full brightness but used PWM or a fast strobe to show a dimmer image!

That's all folks, discuss, and remember - continue testing...
 

Attachments

  • TotalWhite.JPG
    TotalWhite.JPG
    5.7 KB · Views: 1,539
Last edited:

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
Some tidbits: The whole phone was warm, front and back. Not scary hot, but like a nice hand warmer. The screen shots were actually done after I connected the charger and restarted the phone. You'll see the graph shows it starting to charge, but all the stats are for when it was last on the battery - in the test.
 

gogol

Senior Member
May 20, 2005
3,735
216
Thanks for testing this.

How do you define min brightness?

The black color, do you mean #000000 hex color code?

That will last days!

Sent from my Nexus S using XDA App
 

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
Thanks for testing this.

How do you define min brightness?

The black color, do you mean #000000 hex color code?

Not too scientifically - I made it black in MS Paint. It's a JPG. I tried it out in portrait and landscape on full brightness and I can't see the edges of anything on the screen, so it's as black as the rest of the black in Gallery. If you have a better way, then I'd gladly take a 16x10 black image if you think it is more black.

Min brightness will be using the brightness slider in settings. I haven't rooted the phone so don't see any way to get it lower. I tried the app "Dimmer" but it's the same brightness as far as I can tell as min setting. Also I figured the min setting was relevant to more users than something they can't easily get to (or probably see on screen!).

Do you have any suggestions? I want to root, but my home PC can't see my GNex. It's being a pig about it and I haven't got Broadband ATM... Hope to root as soon as I can.
 
Last edited:

Nebucatnetzer

Senior Member
Feb 4, 2011
5,820
6,598
Just a quick input. If you just show a picture for 5h how does the cpu perfom in this time? Isn't it really bored?

In addition if you switch it to air plane mode it doesn't use any power for transmitting a signal. I know this point is really tricky due to the fact that not everybody receives the same signal strength.

Isn't there a more realistic way to test it? Like a combination of stability test and the white picture?
Or a slide show with specific pictures, for example a picture with lots of green, one with a lot of red, one with blue, one with white and one with black?


I really like the idea but if we could create something like a standard which tries to simulate real usage that would be great for future testings.

So that we could say something like:

Galaxy Nexus with Vanilla Android -> 5h with XDA battery benchmark

Galaxy Nexus with CyanogenMod 9 -> 5.3h with XDA battery benchmark

We could compare different settings and see how good they perform based on a "standardized" procedure.


Hmm is this probably a bit too much?
 

pewpewbangbang

Senior Member
Feb 13, 2010
2,046
257
Ann Arbor & NYC
I like the idea of a standardized battery procedure for XDA.

Only thing is, i feel like just web browsing would give a better idea, mostly white website sites and it uses data and CPU/GPU still to give real world usage results.
 

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
Isn't there a more realistic way to test it? Like a combination of stability test and the white picture?
Or a slide show with specific pictures, for example a picture with lots of green, one with a lot of red, one with blue, one with white and one with black?


I really like the idea but if we could create something like a standard which tries to simulate real usage that would be great for future testings.

Hi Nebucatnetzer, I get what you're going for and think it would be a great test. My original goal was just to see how much of a drain the screen was. I used it on Airplane mode as I do not want to have any other drains diluting the result. As you say, signal strength changes, and so this becomes an uncontrollable variable. I'm only showing a white image as it's the most uniform full power test of the screen I can think of. As the display is RGBG, maybe a greenish white would draw more, but I can live without that - web pages are predominantly Plain White and Text in ICS is white.

I actually think that 9%/10% drain for Android OS may be a bit high for 5 hours. I think Just Pictures may be the cause for it as it does every minute change the image in a slide show... to itself...
 

Nebucatnetzer

Senior Member
Feb 4, 2011
5,820
6,598
I like the idea of a standardized battery procedure for XDA.

Only thing is, i feel like just web browsing would give a better idea, mostly white website sites and it uses data and CPU/GPU still to give real world usage results.

It was one of the things which annoyed me the most here. Everyone says he has the ultimate tweak to save juice but you never really could measure it.

We probably have to skip the data part due to asimilar signal strengths.

And it would be really cool if we could find something which works for every phone.
 

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
i feel like just web browsing would give a better idea, mostly white website sites and it uses data and CPU/GPU still to give real world usage results.

Hi! I'm not trying to do a real world test as such. I'm trying to find out how much of an impact a white screen has compared to a black screen. I have a suspicion that the Black screen will draw more than it should do. I've been using my phone on Auto Brightness and at night, min brightness and have found the screen is caning the battery. There's more to the screen than the AMOLED - there's also a MIPI Framebuffer controller, plus whatever interfaces that has to the rest of the phone.

I'm aware that ICS and the other internals of this phone seem quite efficient, so any savings I can make on the screen (which really does suck the juice) should translate into big run time gains... Surely...Right?

If it turns out that there's very little different in battery life between a Black screen and a white screen, then I'll crank up the brightness to revel in its retina destroying beauty. If there is a difference, then i'll stick with my black homescreen background....
 

pewpewbangbang

Senior Member
Feb 13, 2010
2,046
257
Ann Arbor & NYC
The black screen shouldn't draw any power virtually as long as its truly black. SAMOLED functions where black doesn't turn the pixel on so no energy is used. I think the black test has been done by someone, you can probably google it.
 

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
We probably have to skip the data part due to asimilar signal strengths.

And it would be really cool if we could find something which works for every phone.

would be good if someone could make a benchmarking program which pumps out signals in a set pattern on all devices at set frequencies and at a few different power levels. It would't need to wait for a response from any cells or WiFi routers, just talk to itself and only itself. Maybe this would need too low level access to work, below root.

Then you'd have your predictable Radio part of the test. If it's even possible.
 
Last edited:

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
The black screen shouldn't draw any power virtually as long as its truly black. SAMOLED functions where black doesn't turn the pixel on so no energy is used. I think the black test has been done by someone, you can probably google it.

Well, I hope a black screen will draw considerably less. We shall see. Or it will be a revelation... Or everyone will get to witness me finding out that my phone is defective. Hahaha... oh...god I hope not...

I may run the black test overnight. It should last the night. If it doesn't I can always hook it up and check the last "On Battery" status as I did earlier. Then I can run two tests tomorrow!
 

Nebucatnetzer

Senior Member
Feb 4, 2011
5,820
6,598
would be good if someone could make a benchmarking program which pumps out signals in a set pattern on all devices at set frequencies and at a few different power levels. It would't need to wait for a response from any cells or WiFi routers, just talk to itself and only itself. Maybe this would need too low level access to work, below route.

Then you'd have your predictable Radio part of the test. If it's even possible.

This sounds more than difficult. Not a clue how one could do this.

As soon as I have my device I'll probably start a thread where we can think about a standardized (what a stupid word) battery test.

Curios how your second test goes.
 

Davidsmonkeyroost

Senior Member
Jun 30, 2007
53
76
Hertford
Second test - black on lowest brightness - is underway.

Some bad news, which you may already know about: Black isn't black. It's very dark, but the phone isnt switching all the pixels off. I took a photo with my SLR in the dark but can't upload at the moment, will upload tomorrow.

Just took a sneaky look at the stats.
85% 01h26

The draw is 85% screen and 16% Android OS. Yeah, I know they don't add up, but that's what it says.

So far, the draw is higher than I expected for lowest settings with a black image. Odd. The phone isn't warm though, like when the screen was showing a white, max brightness image. It's cold.

Sure hope it doesn't start moaning about running flat whilst i'm asleep...
 
Last edited:

Top Liked Posts

  • There are no posts matching your filters.
  • 71
    Hello all,

    I've been suspicious of the power draw of the screen on the GNex, and so decided to figure out how much a difference brightness made.

    This is to test the difference extreme brightness settings make to battery run times. Including text, browsing or having radios on will dilute the results. I'm only interested in the efficiency of the screen and its supporting infrastructure people! :)

    Here's my testing method.

    1. Charge phone to 100% using a PCs USB charging (should give a more full charge than the wall adapters quick charge, I think, otherwise no harm done). Leave attached to USB cable
    2. Switch the phone into Airplane mode.
    3. Switch off Auto Brightness and turn brightness to Max
    4. Restart Phone. Leave it to settle for a few mins
    5. Fire up "Just Pictures" and the image "TotalWhite.jpg" (attached). No other photos in the folder where it sits.
    6. Disconnect USB charging
    7. Start Slide show. Record the start time.
    8. Every now and then (about 45 mins to an hour and bit) I quickly pop into settings and take a sneak look at charge level - don't want to get caught out!
    9. Resume Slide Show.
    10. Wait until the phone switches off, which is less than 2% charge

    Now, it's not a quick process as I'm going from 100% Batt until auto shut off. So I'll be updating this post in installments. Today is the Max Daddy Full Brightness White image!

    Full Brightness, White Image:
    100% 00h00m
    85% 00h39m
    60% 01h53m
    35% 03h08m
    0% 05h00m​

    Pretty good I think!

    Battery status stated that Screen was 91% and Android System was 9%. Screen on time was the same as the run time. It never went off.

    There's a heavy set of disclaimers to go with this though: The White Image isn't quite the right aspect for it to display across the full amount. Check out the image "White Full The Phone.jpg" - it's pretty close though. Just Pictures isn't so full screen that it takes over the Android buttons - Anyone know a 100% full screen picture view BTW? Between the point where it comes up with "Please connect your charger" (15%) and 6% it said that message over the White image and partly dimmed the screen - I'm at work and was in the middle of something when it happened.

    Minimum Brightness, Black Image:
    100% 00h00m
    85% 01h26m
    82% 02h05m
    18% 08h54m
    0% 10h47m​

    Battery status stated that Screen was 85% and Android System was 16%. Yes, that's 101%, but there you go... Screen on time was within a minute of the run time.

    Again, some notes on the testing: From the "connect your charger" point (i think 15%) to 8 % it had the message on the screen. Just before the end I accidently touched the screen and ended the slide show for up to 15 minutes and the screen went off. Switched it back on and it was at 2%. It then lasted something like 30 mins before auto shutdown. Black isn't Black! check out my the photo "Black Dim The Phone.jpg" - ignore the strange black specs, it was 1 seconds exposure at F1.4, with the camera facing down - think there's some crap in my camera body/lens.

    Minimum Brightness, White Image:
    100% 00h00m
    0% 10h20m​

    Again, Battery status stated that Screen was 85% and Android System was 16%. That's the same as when I ran it on Min Brightness, Black Image...

    Maximum Brightness, Black Image:
    100% 00h00m
    0% 10h52m​

    Battery status stated that Screen was 90% and Android System was 11% (again 101% also probably due to some rounding going on). Screen was on 100% of the time, Android OS was awake for 100% of the time, CPU time 23 Seconds

    Overview:
    Maximum Brightness White Image: 5h00m
    Maximum Brightness Black Image: 10h52m
    Minimum Brightness White Image: 10h20m
    Minimum Brightness Black Image: 10h47m

    Well, well, what's happened here? Maximum brightness with a black image scored the longest run time - unbelievable, right? Yeah pretty much: On the minimum brightness black run I did check the battery stats a few times. The black image run had a greater Android OS drain (11% vs 9%) compared to the white image. As I switch the phone off in preparation for the test it's possible I inadvertantly bump charged the phone. In addition to this, I was trusting the phone a lot more in this, the final test and didn't check the battery status near the end, so perhaps drawing a lower current happily sitting there doing nothing other than displaying a blank image the phone managed to syphon off the last dregs of power, as opposed to spending CPU time with me rummaging through the settings screens for battery stats. Let's put it all down to being in the margin of error and consider the Black images to have the same run times, despite the Android buttons blazing away on Bright (but they aren't many pixels).

    Conclusion:
    Back to my original concern, that the screen was abusing the battery more than necessary. I'm mostly happy with the outcome - Black bright and dim screens draw substantially less than a very bright white screen on this phone, as it should. It's surprising to see that on dimmest setting the white screen got so close to the run time of the black screens. It's producing a lot more light than either black screen. It's drawing c5% more power than the black screens, but I would expect a lot more. I'm guessing that the circuitry to run the Screen and the small amount of power used to make the "Black" OLEDs glow dark dark grey draws only a little less than the white screen.

    And that's a shame. I wanted to run an app such as "Off-Clock" to have a clock on the screen like the Nokia N8 used to do, but i'm thinking this will eat the battery in 10 hours 20 minutes. Only one way to know for sure....

    My initial annoyance that Black never is actually Black whilst the screen has faded. Many of you good folks have pointed out (and after more Googling) that all OLED screens have this, although the reason for it isn't clear. It's either to ease the transition from black to lit by having the LED at the threshold already, power seeping into the circuit somewhere or something else. Who knows, maybe we'd get better run times if the screen was on full brightness but used PWM or a fast strobe to show a dimmer image!

    That's all folks, discuss, and remember - continue testing...
    2
    I love you for doing this. Please don't stop.
    2
    Sorry guys, I messed up the Black Bright test. It was still running when I had to go to work and the power button must have been pressed when It was in my bag. Will re-run tonight.
    2
    Guys, stop suggesting he use normal brightness settings or browse the web etc. That would introduce variables that would ruin the results. He's not testing to see how long the battery lasts. He's testing to see if white vs black display images affect battery life.

    Sent from my Galaxy Nexus using XDA App
    1
    We we aren't criticizing you specific we just hijacked the thread a bit sorry for that :D.