Rockbox

Saratoga, I re-opened this task because I feel we cannot simply classify this report as 'invalid'. The rockbox main page clearly marks the ipod 3g as a stable port, so people can reasonably expect features to work. In my opinion, if runtime estimation is not implemented for this target, we should not just show a bogus value instead (I consider 49 hours at 39% a bogus value). I'd rather see no estimation at all in this case.

Also I'd like to give the original bug submitter a chance to follow up on his comment that this was working properly in builds a week or so ago.

Does anybody have a battery bench of a iPod 3G (tested with a rockbox version from March 2008 or later)? I could only find the last minutes of a bench using a new 850mAh battery with unknown usecase (played file format, volume, ...) which ended up with a runtime of 9:18h.

I have noticed this happening on the Sansa e260 V2 with any of the daily builds. It often reports that I have 152 hours left. :-/

Eric, afaic Sansa V2 is still an unstable build and not officially released. Maybe you should create a seperate flyspray entry and describe some further details:
- Which revision did you test?
- Do you see this nonsense runtime estimates all the time or just occasionally during the usage?
- What does the battery percentage and voltage say in such case?

Oh, apparently the runtime calculation was broken intentionally:

"r21565 | saratoga | 2009-06-29 20:32:43 +0200 (Mon, 29 Jun 2009) | 1 line

Move Archos battery runtime estimation defines into config files instead of powermgmt.h. Replace old default values (which were actually the Archos Player currents) with a new default (1mA). This should give nonsense runtime estimates (instead of semi-reasonable but completely wrong estimates) on all players that do not have a battery current defined. Hopefully this will remind people that they should define the battery current when they start a new port."

I still think it's a bug to show unreasonable values, my opinion is still that we should either show a reasonable value or clearly indicate that the value is not reasonable (e.g. show no estimate at all).

Bertrik, you are right. Nevertheless we need at least a battery bench with a known and reasonable test case for the 3G iPod to be able to calculate the current. And databat described that "it often reports" nonsense values, not "always".

Sonicandtails - are you able to run the battery bench on your iPod 3G, that we can use to update Rockbox?

I found a similar problem with release 3.4 of the iriver H10. (It's the reason I went back to 3.3.) The battery life suddenly started showing hundreds of hours remaining, instead of 5-6 hours for a full battery.

Danny - does release 3.4 always do that for you on iriver H10 or was it a one-off? If it consistently shows the wrong battery time, does it work in a newer current build (i.e. a current build made since after 3.4 release)? .. If not, then yes, something must have changed in H10 config that needs to be fixed. Would be great if you can help us track down whether it's still an issue and still needs fixing.

Andree, Dave, it's pretty clear to me the battery runtime was intentionally changed to indicate an incorrect value for some players, see the comment and changes for r21565 again and the comment in powermgmt.h line 79. The config for h10 for example simply does not define a value for CURRENT_NORMAL, so it defaults to 5 mA in powermgmt.h. In my opinion, this is so clear that there's no need to ask people to re-verify this.

Now we are mixing up several players and the intentional change to show weird battery runtimes when there was no current set for the player. Let us get back to the first comment here -> battery runtimes on iPod 3G: Is there any (actual) battery bench available from which we can calculate the current consumption?

Dave - I just installed the latest build (r23249) on my H10 and the battery life error is still there. It currently reads "89% 146h 46m".

Danny, then please perform a full battery bench (fully charge the H10, then reboot to rockbox, then enable the battery bench plugin, then play an album in repeat mode until the H10 dies). Do not enable any EQ or crossfeed. Please use the default WPS. In the end you post the output file here and we can extract the needed metrics (current consumption, battery levels, ...) out of it.