Page 1 of 4 1234 LastLast
Results 1 to 20 of 70
  1.    #1  
    I'm a mechanical/electrical/computer/nuclear engineer. I have about 11 years experience in Statistical tools and some advanced techniques.

    Here you will find a Taguchi DOE (Design of Experiment), using five different factors at 2 levels each. The five factors are: Bluetooth (off/on), Beam Receive (off/on), GPRS Network Connection (not connected/connected), Camera (after cradle reset: not used/used) and Sound (off or vibrate/on).

    The test involved setting the phone to the proper settings, charging to full charge (Green light on and reading 100%), then an in crade reset, continued charging until full charge shown, switching phone on, then disconnecting. Periodic checks and records of the time, % charge and the voltage level were made.

    A Taguchi L8 array was used for the analysis. While the L8 can record and work with up to 8 factors, only 5 of the columns were used. At least 3 consistant readings were used for the data in the matrix. Under some configurations, longer times were required to achieve a reasonable reduction in battery charge.

    01/16/2007

    A few notes on the Experiment:

    All software removed by hard reset prior to expeiment with only battery graph added
    Screen backlighting was set to one step above the lowest setting
    No Phone Calls were made or received during the test
    Auto Off was set to 15 sec, dim set to 30 sec
    Date and Time Synch set to manual (I should have made this a factor)
    Data Was taken using Battery Graph and recorded in an Excel Spreadsheet
    Prior to running the tests, all manner of resets, including an extended charge with the SIM out, were tried
    Prior to each test, the battery was charged until the light turned green and read 100%. At this point, the battery was removed to reset the Treo 680 while still connected to power, then replaced and the unit was allowed to charge to green/100% before each experiment was started.
    All settings, except camera, were changed to experiment position before disconnecting the battery
    For the Camera setting, I snapped 3 photos after disconnecting the power.
    Last edited by red-beard; 01/17/2007 at 07:08 AM. Reason: Added Notes
  2.    #4  
    The graph shows a log relationship. Higher is better. Network, Bluetooth and Camera use show that they have little effect on the battery usage, compared to Sound and Beam Receive.

    Beam Receive has the most effect on battery life. Turning the Beam Receive off adds much to the life.

    Sound switch. I was surprised that this had an effect (a last minute add to the experiment), but I was equally surprised that turning it to the "ON", or sound active position, gave more life. I cannot explain it.

    I am a bit surprised that the Network and Bluetooth had little effect. The Camera shouldn't have an effect, and the data shows it doesn't.

    So, keep the sound switch on, the Beam receive off and don't worry too much about the rest. Experiment #2 is pretty close to this and the data shows about 30 minutes per percent of battery life. This would be 50 hours of standby. While it ain't 300 hours, it is a very usable time.

    This was my phone, a ROW, unlocked GSM version 1.04. I had the screen turned almost all the way down, 15 seconds to off and screen off after 30 seconds during a call. No calls taken during the experiment.

    It would be good for others to set their phones to similar settings and see if they get improved results.

    James
    Last edited by red-beard; 01/15/2007 at 05:22 AM.
  3. #5  
    I mightily applaud your efforts, red-beard.

    I still don't know how much I trust the results, though. I am not critisizing your technique, which is first-class, but the fact that your experiment is at the mercy of the measuring device -- namely, the built-in battery meter. That thing I don't trust at all. Unless that is reporting correct values, your experiment's results are moot.
  4. #6  
    My standby time is about 50 hours too.
  5. #7  
    redbeard,

    Interesting results, although the last line graph is confusing to me.. Do you mind explaining it a bit? The slope is negative where the others are positive, which confuses me a bit.

    However, I think the 680 has some oddities that happen outside of the norm.. Such as something happens and next thing you know it's chewing battery, then later it's fine again.

    I think beam receieve is very interesting.... I knew it'd use *some* but yikes. I'll have to turn that off. I have it on with my Work profile as I swap files with it *very occationally* with other folks, but I can turn it on temporarily then.
  6. #8  
    Awesome work! I applaud your patience!


    The graph shows a log relationship. Higher is better.


    Can you explain that line? I cant really understand the graph. Like PatrickS said earlier, did you happen to measure the battery manually or through the builtin battery meter.

    Can you repeat one experiment by testing the battery with a multimeter. One that has an average battery use?

    I think there is some bug that is keeping the phone on.
    -----------------
    http://kunjan.net
  7. #9  
    With sound off, vibrate is always working instead of sound notifications. That could be the reason why battery consumption is higher.
  8. mslade's Avatar
    Posts
    16 Posts
    Global Posts
    18 Global Posts
    #10  
    The data confirms my simple BatteryGraph test.

    Now if we could only get Palm to look at it and then fix the basic standby problem.
  9. #11  
    Quote Originally Posted by mslade View Post
    The data confirms my simple BatteryGraph test.

    Now if we could only get Palm to look at it and then fix the basic standby problem.
    I agree. I really don't fully understand these results. All I know is the battery is not performing correctly and it is Palm that has to explain this - and not by saying its a smaller battery.

    I'm speaking to a level 3 tech support person tomorrow at Palm Corporate at 2pm. It'll be interesting to see what he has to say. If nothing else, I can point him to all these threads. But from my perspective, the only thing that will prevent me from returning my device is for Palm just to say they have a problem, they are working on it, hopefully they know what it is, this is the workaround and there will be a fix. As far as I'm concerned, if its a simple thing such as not using the camera, then no problem. I don't use cell phones to take pictures anyway. In terms of all the other variables pointed out from the tests above, I have everything turned off and the ringer turned on. This isn't what's causing the battery problem. An in cradle reset seems to fix it but then something happens to change its performance. We just need to know what that something is and either avoid it or if we can't wait for a fix.
  10.    #12  
    Battery voltage was measured using Battery Graph.

    The graph shows that with most of the factors, it is better to have them "On" than off. With Bluetooth, Network and Camera factors, the graph is almost flat, meaning they really have little effect. I doubt that the factors really increase battery life.

    The graph shows that sound in the "ON" position conserves more battery life than in the off position. Beam Receive in the "Off" posiiton conserves more battery life.

    The Taguchi Signal to noise ratio factor (Last in the matrix and the one that is graphed) for optimizing Larger is better is:

    http://forums.pelicanparts.com/uploa...1168829779.jpg

    The graphed values come from averaging the Signal to noise ratios for each factor level. For Beam Receive, experiments 1,2,5&6 were averaged for factor "Off" and experiments 3,4,7&8 were averaged for "On". A Taguchi DOE allows optimaization of factors with a minimum of experiments. A Straight DOE with 5 factors of 2 levels each would require 2^5 or 32 experiments to test all possible combinations. It took me most of Friday and the weekend to run 8 experiments. 32 experiments would take over a week!

    The measured effect is minutes per percent battery. More minutes is better. Checking the voltage with a multimeter would be problematic. It would cause a reset in the treo680 each time it was removed.

    As long as the readings are consistent, the data should still be valid. The experiments appear to be consistent. After about 6 experiments, I pretty much could predict the final results. I had an idea about the Beam Receive from some testing last week, but I wanted to test more factors and test if there were interactions and get the relative importance of the different factors.

    I will be discussing all of this with Palm tomorrow. I am working with one of thier techs, but this experiment is my own design.

    Again, I would request others try setting the Sound On and Beam Receive Off and then reporting what this does to your battery standby usage.

    James
  11.    #13  
    I'm runnning one extended test tonight through tomorrow morning (about 9-10 hours), with all factors set to optimize life. I'll report the test result in the morning.
  12. #14  
    Most of that went over my head, but i do understand the graph better.. Thanks a lot James for taking this up with palm. Keep us posted on what comes out of your talk with the tech tomorrow.


    Thanks again!
    Last edited by kunjan; 01/14/2007 at 09:24 PM.
    -----------------
    http://kunjan.net
  13.    #15  
    NOTE: If any Palm Person wants to contact me, I have the e-mail and Private message options turned on in this forum.
  14. #16  
    red-beard: Awesome work!

    However, have you tried the battery reset? This seems to be the only thing that solves the camera bug. Just taking out the battery for a short while does not help on the camera bug!
    And with the battery-bug I see the same: 48-50 hours of standby.
    But after a battery reset, and all succeeding cycles without using the camera, I see 150-170 hours of standby. (With network on 2/3 of the day)

    // Egholm
  15.    #17  
    Quote Originally Posted by red-beard View Post
    I'm runnning one extended test tonight through tomorrow morning (about 9-10 hours), with all factors set to optimize life. I'll report the test result in the morning.
    After a bit over 7 hours with Bluetooth on, E-mail being checked every 5 minutes, No camera use, Sound on and Beam Receive off, the phone is 83%. 25.6 minutes per percent of battery, or a standby life of 43 hours. I could live with this.

    Egholm, if you are refering to the reset as set below:

    http://forum.treonauts.com/treo-mode...tructions.html

    I have tried this. I still think there may be other problems, but there is no doubt that the sound switch and Beam Receive have a huge effect on Battery life.
    Last edited by red-beard; 01/15/2007 at 09:04 AM. Reason: Beam Receive was off, mis-typed...
  16. #18  
    Quote Originally Posted by Egholm View Post
    red-beard: Awesome work!

    However, have you tried the battery reset? ... Just taking out the battery for a short while does not help on the camera bug!
    ...
    // Egholm
    What does it mean that taking out the battery is not the same as a reset? As far as I know, taking out the battery always performs a reset.
  17. #19  
    Quote Originally Posted by mdavis View Post
    What does it mean that taking out the battery is not the same as a reset? As far as I know, taking out the battery always performs a reset.
    There is a difference (though I am not sure anymore on the correct terminology for Treo's).
    If removing the battery while the device is connected to the charger (or powered USB port) it will also reset the battery calibration circuitry
    Digital since 1980, Handheld since 2001
    M105, TE, T3.5, Treo 680, Treo 500v
    Happy Datebk6.1 user
  18.    #20  
    Quote Originally Posted by holvoetn View Post
    There is a difference (though I am not sure anymore on the correct terminology for Treo's).
    If removing the battery while the device is connected to the charger (or powered USB port) it will also reset the battery calibration circuitry
    Soft Reset - Removal of the Battery and Re-installation
    In cradle reset - Removal of the Battery and Re-installation while power is connected to the device
    Warm Reset - Removal of the Battery and Re-installation and holding the 5 way switch "UP" until the 2nd palm screen comes up.
    Hard Reset - Removal of the Battery and Re-installation while holding the red on/off switch until the 2nd palm screen comes up

    There has also been talk of a "Charge with the SIM card out", which is referenced from Treonauts.com, above.

    Every one of these resets has been completed on the phone. An "In Cradle" reset was done before each experiment, to ensure that the unit was reset and, if it really does calibrate the battery meter, that the battery meter was calibrated.

    James
Page 1 of 4 1234 LastLast

Posting Permissions