FORUMS

[REF] Leo power consumption breakdown

1,734 posts
Thanks Meter: 13
 
By kilrah, Senior Member on 13th February 2010, 08:30 PM
Post Reply Email Thread
With the HD2's pretty short battery life compared to my previous HTC phones, and also as I was a bit bored , I decided to have a look at how much the different systems drain in the HD2, and did some tests.

So for anyone interested, here are the results:

Base consumptions: (pick the one that suits your situation)
Standby, phone on, WIFI+BT off, no connections active: 5mA
Processor running idle, screen off, connections as above: 55mA
100% CPU usage (Coreplayer benchmark), connections as above: 315mA
USB connected, connections as above: 125mA

Additions: (add this to the base one depending on what you have on)

Connections:
BT on, idle: +1mA
Wifi on, idle: +5mA BUT!! If you enable wifi during standby (with BsB Tweaks or the WifiNoStandby CAB), the processor does NOT go to standby
anymore, so the base needs to be the 55mA above!


Connected to EDGE, idle: not measureable
Connected to 3G, idle: +4mA
Beware those 2 will occasionally and unpredictably send data every now and then, so this doesn't necessarily mean much.

BT transfer (file copy through ActiveSync, 120kB/s, CPU usage ~2%): +80mA
Wifi transfer (file copy with Wifi Remote access, 1MB/s, CPU usage ~3%): +200mA
EDGE/3G transfer: Pretty impossible to measure due to so many variables, but can be extreme (total current consumptions of 850-1150mA, so very approximative 500-800mA draw from the data connection during page loads are common!), and leads to most of the energy draw when using the net a lot. EDGE uses more power than 3G, consumption is higher when network coverage is lower, bad network throughput or congestion also mean lots of retries/overhead and less effective data transfer, using power for a longer time until the data is finally there.

Memory access (large file transfer):
Internal read/write through ActiveSync (speeds 2.5 / 2.1 MB/s respectively): +20mA / +130mA
Card read/write through ActiveSync (speeds 3.2 / 2.7MB/s respectively): +25mA / +40mA
Card read/write in USB disk mode (speeds 7.5 / 5MB/s respectively): +40mA / +45mA

Backlight:
10%: +65mA
20%: +78mA
30%: +90mA
40%: +105mA
50%: +121mA
60%: +136mA
70%: +154mA
80%: +170mA
90%: +190mA
100%:+210mA

GPS:
+95mA

Flashlight (with HTC Flashlight):
Level 1: +32mA
Level 2 (same as camera flash on): +107mA
Level 3 (same as camera "bright flash during shot"): +530mA

And remember the battery capacity: 1230mAh

So with this you can calculate your battery life for various activities. For example:
- Playing a video with full backlight: Let's say the video is well encoded (30% CPU use, for example 720x400 1Mbps DivX), that's about 150mA base, + 210mA backlight = 360mA, resulting in about 3h20 battery life.
- Same at night with only 40% backlight: Life goes up to about 5h!
- Wifi during standby for an 8h night, or a program that somehow prevents the phone from entering standby: 60mA or 55mA respectively, draining about 40% of the battery during that time.
- GPS program, with 70% backlight: Let's assume 150mA for the processor as it has some work to do, 90mA for the GPS, 154mA for the backlight = 394mA, or 3 hours.
- Music through wired headphones, screen off: the 55mA base plus a little 5mA as MP3 decoding is nothing for the CPU = 60mA or 20h.
- Same with a BT headset, with about half the BT bandwidth: 60+40 = 100mA or 12h
etc.

Charging
[EDIT 14.04.10]
This is now a dedicated section as I did some more thorough charging tests.

So, as some of you might know, the HD2 has 2 charging "modes".
- One is USB, that is used if the phone is connected to a PC, or an unknown device. In this mode, the current the HD2 will draw from the port/charger is limited to approx. 470mA, to stay within the maximum of 500mA a USB port can supply.
- The other is "dedicated charger", which is recognised on the original HTC charger (and some others, it's becoming a standard for a "dedicated charging USB port") by shorting of the 2 data pins of the USB connector in the charger.

USB charge
Important to know, during USB charge, the phone will NOT go to sleep, as it's supposed to be connected to a PC, and be running either ActiveSync, Disk drive mode, or modem, and in all 3 cases would be expected not to shutdown. So not only the current supplied to the HD2 is low, but the phone draws some of it for itself, leaving very little for actual charge - so expect loooong charge times.

USB charge with screen on (backlight dimmed, 10%): 285mA
USB charge with screen off (standby): 345mA (which shows the processor still runs and draws the "base USB" current)

Dedicated charger
When used with a charger that has the 2 USB data pins shorted, such as the original charger, the HD2 will draw a current that is proportional to the voltage on the USB power lines. To measure this I have used the original supplied USB cable, a variable regulated power supply, USB socket (with data pins shorted), and 2 meters for voltage/current. Voltages are measured at the "PC" end of the USB cable, so not taking account of losses in the USB cable. Will talk more about this later on.

Current vs Voltage diagrams are attached. Charge current is proportional to voltage, linearly until it reaches the max charge current, approx 830mA. This was measured so that the only draw is charge. If the phone is turned on while at max charge current, it will draw extra, until it reaches about 980mA, and will then stop to respect the 1A rating of the stock charger.

Now, to the influence of USB cables. I initially had some trouble with inconsistant numbers, phone only drawing 670mA from the stock charger, i.e matching neither the ~350mA from USB, or ~830mA from stock charger in "normal" condition. Turns out that to make it more convenient on my workplace I was using an USB extension between charger and HD2 cable. It was a $2 extension I bought on Dealextreme. Removing it solved the problem... and after making those measurements I poked with it again. Turns out that at 1A current, the voltage drop in the extension (which by the way isn't longer than the HD2's USB cable) was 1.8V! Yep, nearly 2 Ohms for a 1.5m extension! Couldn't believe it.
I have a cheap Chinese microUSB cable that wasn't as bad,but still significantly more resistive than the stock one, hence me noting I used the stock cable for my tests. So, quality of the cables, extensions, adapters IS important! Note the phone correctly reaches full charge current a little bit under the 5V USB spec, so everything is well tuned.

Now, important to know, Most 3rd party chargers will not have the 2 USB data pins shorted, and will thus result in the same behavior as mentioned under USB charge, the processor will also be running continuously drawing the "base USB" current.
It is often possible to modify 3rd party chargers by opening them and shorting the pins, speeding up charge. The voltage/current curve behavior is actually helping there, because thanks to it if the charger is overloaded its voltage will most likely fall a bit, and the HD2 will thus draw less and find a nice balance point. This DOES NOT mean there's no possiblilty of damaging the charger, but all 3 I modified did well. One that was really weak resulted in not much more current being drawn after the mod than before (i.e voltage fell very low, approx. 4.4V), however the gain from not having the processor running like in USB mode still sped up charge a little.
Attached Thumbnails
Click image for larger version

Name:	I_U_full.jpg
Views:	307
Size:	14.3 KB
ID:	309245   Click image for larger version

Name:	I_U_linear.jpg
Views:	253
Size:	10.2 KB
ID:	309246  
The Following User Says Thank You to kilrah For This Useful Post: [ View ] Gift kilrah Ad-Free
13th February 2010, 09:02 PM |#2  
ephestione's Avatar
Senior Member
Flag Rome
Thanks Meter: 61
 
More
Thank you, that is one useful chunk of comprehensive information
Could you please explain the exact measuring method for these tests?

40% backlight at night is a tough example tho, Lumos on my device is set to 10% for 0 sensor value, and only because I can't set it to lower than 10%... nice to know battery drain goes up twice from 10% to 60%!
Also, 100% is usually really necessary under direct sunlight, in a normal lit room probably 40% backlight is more than enough to watch a video... all in all, with your superinteresting info, the battery doesn't look like lasting "too short" now, but more or less "the right amount considering the battery capacity".
My iPaq 210 has a 2200mAh battery, just to make a comparison... that's why I could go for 8 hrs during some bus trips while watching tv series, and I just needed to swap battery and used a little of the second one.
13th February 2010, 09:28 PM |#3  
Shasarak's Avatar
Senior Member
Flag London
Thanks Meter: 127
 
More
What's your estimate of the drain caused by activating push email? I've recently been doing some rather crude experiments myself, and one provisional conclusion is that push email on a hotmail account uses a lot more battery than push email on an Exchange server.
14th February 2010, 08:00 AM |#4  
pa49's Avatar
Senior Member
Thanks Meter: 15
 
More
Excellent stuff, you are to be congratulated.
I have a long held theory and I wonder if you are in a position to to test it?
I believe that battery consumption is greatly increased when an app is run from mem card and would be intrigued to see a comparison between an install to this as opposed to phone mem.
Any chance?
14th February 2010, 10:03 AM |#5  
Junior Member
Thanks Meter: 1
 
More
Battery levels
I found that power consumption of the battery got even worse after I went to Rom 1.66.707.1. However, after a few days, I let it run all the way out, switched back on, it ran out after a few minutes. Then after an overnight charge I found the battery (on standby) only went down by about 10 -12 % in 24hrs. I'm hoping this performance will continue.
14th February 2010, 10:43 AM |#6  
Senior Member
Thanks Meter: 3
 
More
What did you use for the testing?
14th February 2010, 10:52 AM |#7  
ephestione's Avatar
Senior Member
Flag Rome
Thanks Meter: 61
 
More
Quote:
Originally Posted by pa49

Excellent stuff, you are to be congratulated.
I have a long held theory and I wonder if you are in a position to to test it?
I believe that battery consumption is greatly increased when an app is run from mem card and would be intrigued to see a comparison between an install to this as opposed to phone mem.
Any chance?

That would be reasonable as reading from an external media will need energy to access its contents... still, once the app files have been loaded into the device's RAM, it shouldn't matter much
You could also test by comparing:
1) copy say 30mb from one location to another of the internal mem
2) copy from microsd to microsd
3) copy from mem to microsd
4) copy from microsd to mem
14th February 2010, 10:56 AM |#8  
kilrah's Avatar
OP Senior Member
Thanks Meter: 13
 
More
Quote:
Originally Posted by ephestione

Could you please explain the exact measuring method for these tests?

Simply using the built-in current sensor, getting the reads from AEBPlus battery information screen, and methodically turning things on/off once the others are evaluated and can be subtracted from the total reading.

Quote:
Originally Posted by ephestione

40% backlight at night is a tough example tho, Lumos on my device is set to 10% for 0 sensor value, and only because I can't set it to lower than 10%...

Really? You should force the backlight off then
I love bright images myself, so even in my bed in total darkness if I watch a video or photos I'll force 100% backlight I have Lumos set to force 100% for Coreplayer, Resco Photo manager and HTC album
Of course not for browsing or just messing around, in that case it's 20% for me

Quote:
Originally Posted by Shasarak

What's your estimate of the drain caused by activating push email? I've recently been doing some rather crude experiments myself, and one provisional conclusion is that push email on a hotmail account uses a lot more battery than push email on an Exchange server.

That's one thing I'd have no idea about... I've never used push email at all. And that's "standby usage", so hard to evaluate, as you never know when it kicks in.
Measuring that would need to be done on a long time. I'd say to leave your phone one night with push email off, one night with Exchange only, and one night with hotmail only, and then check the difference, preferably with a battery at about 80% charge at the start (mine seems to fall from 100% to 90% in a few minutes before becoming more regular, so I'd say the top of the scale isn't that reliable).
And I should really try push email once, that would be nice, but I *think* I have no provider that can do it for me... well I have a gmail account I never use, I should try to see if I can have it check my usual 3 mail accounts, aggregate and push... never really looked into that stuff.

Quote:
Originally Posted by pa49

Excellent stuff, you are to be congratulated.
I have a long held theory and I wonder if you are in a position to to test it?
I believe that battery consumption is greatly increased when an app is run from mem card and would be intrigued to see a comparison between an install to this as opposed to phone mem.
Any chance?

Thanks
Well I'd tend to refute that theory, because when I tested the BT and Wifi in use I tried read from internal memory, write to internal, read from card and write to card, and all 4 were identical. I should have mentioned it indeed, but I was mostly interested to seeing if reads (wifi/BT "sending") and writes (wifi/BT "receiving") would have an influence on consumption, which wasn't the case, as well as whether the throughput was different, which wasn't the case either.
But I've done a few more tests, see the updated first post

One intersting thing is firstly that when the HD2 is connected to USB, the current draw grows significantly, so I've made a new "base consumption".
Next, the card is actually faster than the internal memory both in reads and in writes, tested both through activesync for consistency. Writing to the internal memory eats a LOT more than writing to the card. Reading from the card eats a little more than reading from internal memory, probably evens out as the reads are shorter due to faster transfer rate.
I've added some charging tests as well. Apparently, even if the phone "disconnects" from USB when turned off, the processor still runs and uses about the "USB connected" base current.
14th February 2010, 01:03 PM |#9  
ephestione's Avatar
Senior Member
Flag Rome
Thanks Meter: 61
 
More
Quote:
Originally Posted by kilrah

Simply using the built-in current sensor, getting the reads from AEBPlus battery information screen, and methodically turning things on/off once the others are evaluated and can be subtracted from the total reading.

d'oh aebplus has a battery information screen, checking it right away In that case you'd have to take into consideration aebplus' current absorption anyway... which is not measurable as you cannot check the current intake of aebplus without aebplus being running
I have the cab on the sd already, but didn't install if after noticing that didn't work for button assignments with later versions of the rom... does it work for you on that side? I used the program all the time on my previous ipaq because it was oh so useful but never got around to notice it had a battery info subsection.

Quote:

Really? You should force the backlight off then

Wouldn't have much sense doing it in the dark would it (admitting it's possible altogether on the HD2!)
But I actually used my oooold casio cassiopeia, about 7 years ago, with backlight turned off, while reading ebooks with speed reader plus during train trips, as the neon lights created a reflection good enough on the display so that I didn't need backlight...
in the end, the backlight died altogether and until I bought a new device, I managed to use it with light turned off
14th February 2010, 05:02 PM |#10  
djet's Avatar
Senior Member
Flag Moscow
Thanks Meter: 2
 
More
A flashing LED (incoming SMS warning, e.g.) seems to add consumption of 1-3mA.

Running FlexMail in background with a push service (IMAP IDLE) adds up to 30mA.
15th February 2010, 08:14 AM |#11  
kilrah's Avatar
OP Senior Member
Thanks Meter: 13
 
More
Quote:
Originally Posted by ephestione

does it work for you on that side?

AEBPlus works fine for me yes... but I don't have a "latest version" ROM AFAIK. I don't like WM6.5.x new softkey arrangement, so I'm staying with 6.5.

Anyway more about Push, I configured that through gmail yesterday, and it works just fine. I left it on during the night, and this morning I had lost 8% battery. So it's pretty much negligible. I received 2 e-mails during the night and was on 3G network.
Post Reply Subscribe to Thread

Tags
battery, consumption, current, power

Guest Quick Reply (no urls or BBcode)
Message:
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes