Rockbox Technical Forums
Rockbox General => Rockbox General Discussion => Topic started by: TexasRockbox on July 19, 2007, 04:54:56 PM
-
Hello,
I'm a Rockbox newb (but have designed a WPS) and an ipod less-than-newb. In another topic it was mentioned that we don't have all the ipod information therefore we don't know what is exactly being turned on or off by Apple and that the battery indication is unreliable in Rockbox.
OK?
For fun, I booted to the Apple ipod mode and went through their menus and turned "OFF" anything that was turned "ON". Everything I could find, maybe I missed something. Even turned down the brightness control all the way to lowest setting.
Still with me?
I boot up to Rockbox (using my own WPS as I was before) and what was once 4h 49m of battery time left is now 14h 0m. Is it possible that some or all of the "OFF" settings in "ipod-mode" are still in effect in "Rockbox-mode"?
Am I being fooled? I guess we'll find out 14 hours or so later.
Thoughts?
-
The estimated time remaining in Rockbox is not at all accurate. And when they said "turned off things" they meant specific pieces of hardware, not software functions as most of those are. For example, properly shutting down the USB hardware when it's not in use.
-
OK.
For experimentation purposes anyway I've used it slightly over three hours and the battery remaining shows 73%. All .OGG files, a 7/16/07 build. Minimal navigation or lighting up the display.
From previous usage, to me, the battery estimation is done "on the fly" and is adaptive. That is, the more one navigates or lights up the screen, the battery estimation drops quickly, but if that kind of activity is stopped, the estimated battery usage decreases and the time time remaining actually increases.
What has me intrigued is that previous to the "ipod-mode" changes the display the battery life was a consistent 4h 49m (100% fully charged) until those "ipod-mode" changes. Then the sudden leap to 14h 0m.
-
Look, the best way to do this is full battery benchmarks using the plugin so that we can see the actual voltages over time. Both percentage readings at any given time and the estimated time remaining are known to have inaccuracies, so please simply create full, proper battery benchmarks.
-
I'll do that when this charge runs out. Thanks.
-
Stopped, recharged, Battery Plugin underway. I guess we'll find out in the morning or that I ran the plugin incorrectly.
-
You'll need at least two benches won't you? One with the settings on, one with them off, all other conditions exactly the same?
-
True. Thiis test is with the "ipod-mode" settings as they were altered (things turned off -- FM, Photo, whatever seemed like it could conceiveably utilize power). I can run another one with the ipod settings reset to default.
The item that bugs me is the "remaining time". It was a consistent 4h 49m with full charge until the ipod changes, then it jumpted to 14h 0m after the changes. Has anyone else reported a "remaining time" of 14h 0m with the ipod 5g Video 80gb?
-
I told you that number is virtually meaningless right now. Change the Battery Capacity setting (if it's changeable on the 5G iPod right now) and see how the time remaining changes, for example.
Objective tests are the only way to go, depending on that at all is pointless.
-
Just out of curiosity I just did this and did notice my remaining time get a boost I've never seen it get before. I understand that this number is not to be relied on, but it does have me curious. I'm going to do a bench with factory settings and one with everything turned off myself in the odd event there is something to this.
-
Hmmm...Looks like I may have to do a third test. Currently, I'm running a test to see if having speech enabled or disabled impacts the battery cycle...I may also have to try one or the other or both with all possible factory settings turned off and see what happens...Hopefully, this will prove useful.
-
Having speech enabled will currently result in the CPU being boosted at all times (see bug report at http://www.rockbox.org/tracker/7447), and hence lower battery life.
pondlife
-
It appears from test #1 (ipod settings set to off) and using Sony EX90LP as a load, and a slight bass and treble boost (normal listening conditions for me) and repeating The Beach Boys "Pet Sounds" over and over, the battery lasted 12h 1m 19s.
From what I've been reading on the forum, this is pretty good time (stock battery, low mileage).
-
Booted into "ipod-mode", reset to defaults. As a test I changed one of the settings, ran reset to default again, and the setting changed back to default. Ok there. I rebooted into "Rockbox-mode"and let it charge up a couple of hours (not quite full yet) and took a quick peek at the time remaining while playing a song and it showed 12h (cant recall)m remaining on the battery. Hmmmm.... Will run battery test #2 as soon as full.
-
. I rebooted into "Rockbox-mode"and let it charge up a couple of hours (not quite full yet) and took a quick peek at the time remaining while playing a song and it showed 12h (cant recall)m remaining on the battery.
This reported time is inaccurate, means nothing, and is beyond worthless.
Seriously, ignore what the WPS says. It has no bearing on the conversation and only adds to the jumble of numbers floating around.
I hope your prior report:
It appears from test #1 (ipod settings set to off) and using Sony EX90LP as a load, and a slight bass and treble boost (normal listening conditions for me) and repeating The Beach Boys "Pet Sounds" over and over, the battery lasted 12h 1m 19s.
came from a battery bench file, not from the WPS.
-
Also, it is better if you use enough song in the playlist to fill up the buffer, rather than just one song.
-
I'm utilizing the entire "Pet Sounds" album, not just the song. The album was compressed with .OGG Q7. The battery benchmark indicated to use an album, so I did so.
OK, whatever the WPS battery readings indicate are meaningless. The Battery Benchmark plugin shows that the player played for 12h 1m 19s -- can it be assumed that that number is correct? If so, is 12h 1m 19s playing exclusively .OGG files with a headphone load and the bass and treble controls slightly elevated (not using either the software or hardware EQ) a bad, good, or outstanding figure? From reading various posts 12+ hours with Rockbox with ipod 5G Video 80GB would seem to be a very good figure.
Not trying to cloud the discussion but to further understanding: If the WPS figures are meaningless how to explain the sudden jump from a consistent 4h 49m to a consistent 14h 0m -- it must mean something to some piece of software or firmware even if we don't understand what it is -- it doesn't appear to be a random number.
I think to some extent all consumer battery remaining indicators are only (at best) generally accurate. So 14hours indicated lasted 12hours in actuality -- I don't think that's too bad -- 85% accurate. I'm certainly not going to sweat the minutes. I think the number indicated in the WPS is adaptive and represents a formulated figure based upon usage (current draw) over some period of time. Therefore, I would expect the actual, raw numbers to go up and down a bit (giving the appearance of meaningless, inaccurate, etc). Rockbox has the ability to display the raw numbers while the ipod uses a segmented display. The accuracy of the reported numbers may be good enough for a 4 or 5 segmented display (accurate to 25% or 20%).
Has this type of battery WPS behavior (sudden 7h swing upward) been reported before?
I wish I had copied the first battery test results off the ipod HD (I did rename it though). So I won't have anything further to report until either very late tonight or tomorrow morning.
-
12 hours is good in a battery bench.
For the last time - ignore the WPS time estimates.
I'm assuming you first saw the 14h estimate on or after the 14th of July?
Right after http://www.rockbox.org/tracker/task/7216 got committed?
Have you seen the 100%=4h estimate since?
Regardless, it doesn't matter. Actual runtime, not estimated runtime, is the test. (Unless you want to hook your opened iPod to a bench power supply and measure amperage)
-
Yes it was noted after the 14th of July -- on the 19th when I opened the topic.
My Rockbox might be a July, 16th build or a July, 19th build. Not shure about the 7216 fix but since it appears to be available beginning with the July 14 build, I woud think that it is installed.
Have not seen the 100% = 4h 49m on the WPS since the 19th but did see it on the 18th, 17th, 16th.... back to since I had obtained the player.
The second Battery plugin test is still running. I'll verify it all when it completes sometime late tonight.
-
Second test completed (returning ipod settings to default):
Rockbox version r13922-070716, Bass +5, Treble +5, Sony Ex90LP 16ohm, Beach Boys "Pet Sounds" album repeated, encoded Q7 ogg Vorbis.
The second test ran for 11h 59m 55s by the benchmark figures. Looking at the "%" figures thorughout the test, those percentages would seem to be generally accurate enough for a 4 segment battery indicator -- not perfect by any means but close enough to give a rough estimation of remaining time -- my view, anyway.
-
So, evidence then shows that the Apple firmware's settings don't affect Rockbox, as you were told in the beginning?
-
Not certain. I still can't explain the jump from 4h49m to 14h 0m in the WPS. Something changed. I was also told that the WPS battery indication is beyond worthless -- evidence shows the battery plugin data would seem to indicate that the battery data would have some worth (if utilized properly) as a general indicator of battery time remaining.
-
You were told the time remaining is worthless. I don't understand why you seem unwilling to accept this. What SVN revision were you using when it said 4h49m?
The percent remaining isn't useless, but it's not strictly accurate. But it's not the only thing used to create the estimate for time remaining.
-
The exact time remaining isn't really such a big deal to me -- I was just reporting a sudden change in value that seemed to coincide with making changes to enabled features in "ipod mode". When it was 4h 49m it was probably a build from a day or two earlier than the 070716 build.
I'm not trying to hold anyone accountable for the accuracy of the time or percentage remaining. I've seen it reported time and again that the firmware was based upon some educated guesswork and perhaps trial and error. Whether or not it is 100% accurate or 0% accurate is of no real matter. Data was requested, data was gathered. If some of the data gathered is meaningless (although it does seem to be generally close), so be it.
What does matter is that I'm getting 12h of playtime with some bass and treble boost playing into a 16ohm load with OGG Q7 files. That makes me very, very happy.
-
I could imagine that the change in the remaining runtime estimation is related to this change:
http://svn.rockbox.org/viewvc.cgi?view=rev&revision=13885
-
Yes, that fix was indicated earlier. Could very well be an answer to the question.