Some observations about the battery and charger


MrConfusion

Very Active Member
Joined
Apr 11, 2013
Messages
331
So, this is weird, I noticed it yesturday, is this a known issue or something unique?

The Pandora thinks it's powered by the poweradapter (i.e. charging) all the time.

As in, I boot up, the remaining battery percentage gets updated. This is why I didn't notice this earlier; the percentage goes down on bootup and I've booted a lot. It has looked good when just glancing casually.

I got some SD cards yesturday and tried some emulators and started wondering why the percentage did not go down at all. Well, it did on next bootup!

So, essentially, it's as if the powerlevel only gets updated at startup.

What then caught my attention was that the "estimated time remaining" in XFCE keeps at 00:00 and when I hover the mouse over it, it also says "charging". If I plug in the charger, it still says the same. If I unplug the charger at that point, it remains the same. Always "charging".

I also tried "dropping" the battery out of it's slot while the Pandora is running charger attached. The Pandora reboots, but naturally then it's supposed to be "charging".

(

I hadn't realized the Pandora can run completely without a battery, but that is excellent, that's been high on my wishlist for almost every smartphone/PDA-like device I've used!!! Definitely a good feature in a gaming oriented device, because you can save the battery for when you really need it.

)

I also checked this page late last night: http://pandorawiki.org/Power_modes

I didn't have time to check things out in detail, but one thing was weird: Status in /sys/class/power_supply/bq27500-0 was constantly discharging.

So, part of the Pandora thinks it's constantly charging, part of it thinks it's discharging :).

I'll try to charge it fully when I get home tonight, though I originally intended to run it almost dry first.

But if any of you have seen something like this before, do let me know.

It's a CC unit, by the way.
 
It could just be a bug. Reflash to latest firmware and see if it will work correctly then.


Running the pandora without battery inserted won't work for every Pandora. You got lucky.


But most importantly: DO NOT REMOVE OR INSERT THE BATTERY WHILE HAVING THE CHARGER CONNECTED.


Strange things can happen and discalibrate/destroy your chargingchip.
 
It could just be a bug. Reflash to latest firmware and see if it will work correctly then.


Running the pandora without battery inserted won't work for every Pandora. You got lucky.


But most importantly: DO NOT REMOVE OR INSERT THE BATTERY WHILE HAVING THE CHARGER CONNECTED.


Strange things can happen and discalibrate/destroy your chargingchip.
I think it is already the latest firmware, I only got it this week. But is it still worthwhile to reflash to latest?

Thanks for the headsup on the battery, like I said in the post I already tried removing it once and trying to plug it in while running was on my "things to try next"-list :-(.

Addition: Charged it for a while. I don't even know whether it's full now, but minimenu shows 100%, also the SystemInfo app shows

"Charge: 3903.000 mAh, 92.05189 % of design (4240 mAh)"

Last night the last percentage I saw was around 25, so obviously it has charged during the last three hours.

(Edit: there was stuff here of the XFCE battery meter which was not true after login/logout.)

I did notice one thing though: When I attach the charger, SystemInfo notices this right away and starts showing some extra stuff, but the charging LED does not turn on. Something fishy there.

One more addition:

Now I've catted lots stuff from /sys/class/power_supply and everything I see looks OK to my eyes. Like, I'm on battery, status is Discharging, etc.

time_to_empty_now, time_to_full_now all say "No data available", which might explain some of the things I'm seeing (except for the charge LED). Are they even supposed to display anything?

Anyway, since the Pandora is clearly "getting juice" my interest in this issue is going... going... gone for the moment :p
 
Last edited by a moderator:
If it bothers you, a clean reflash might solve this.
 
If you haven't, pull the battery and leave it off for 30 minutes, give the capacitors a chance to discharge and force the charging circuit to reset.
 
FWIW, on my system /sys/class/power_supply/bq27500-0/time_to_empty_now and time_to_empty_avg show a decimal number when discharging, and time_to_full shows a decimal number only when charging. At other times I get 'No data available' if I try to cat them. As far as I know those are only used for reporting in the system tray icon (and _wb_'s System Info app).


Do you still get decent reports for charge_full and energy_now? I think that by dividing those you can work out the charged percentage, though if those are working the only reason I can think that the system trap icon wouldn't show that percentage is because it thinks it's charging. Is /sys/class/power_supply/bq27500-0/current_now positive or negative?


If it's basically working for you, even if the indicators are wrong then that's fair enough I guess. I'm just a little concerned that when it gets to 100% it may end up overcharging your battery if it's got no way of finding out what's going on. Hopefully there's some circuitry in the battery, or lower level in kernel modules or hard coded into the board to stop that happening, but presumably it would break the work done by the Pandora team to extend the life of your battery.


FWIW, the charger LED is controlled via /sys/class/leds/pandora::charger. On my system, I can't write into the brightness file in there even via sudo, presumably because the charging subsystems got it open for writing anyway, but I can at least read off it to confirm that what the system think's it's doing and what the LED is doing are the same.
 
Thank you for all this information (to everyone).

I will try flashing the firmware. Or actually I'm tempted to try the "Run the OS on SD"-trick that has been explained in the Wiki, I gained the impression it would essentially amount to same... so if it's the same, then "whatever", I can do tests on the SD card in the future too, if I feel like hacking and if it fixes things, I can flash the NAND too.

I tried to write to the brightness file and I could do that, but there was no effect. I then tried to do that to the bluetooth led and for that one it worked and the LED turned on and off as I wanted it to. I do get reports from charge_full and energy_now, whether they are accurate I'll tell later...

When I left the Pandora charging earlier today, the battery became warm. Since there was no reliable indicator, I charged until it felt cool to touch, from which I assumed charging was over. As can be seen from the figures from my earlier post, it didn't charge to full capacity but left some 300 mAh of full state. Whether that is because I cut power or because something else cut power I don't know until I try to charge again. But I will definitely watch it while testing it now initially... won't leave it alone untill I know exactly what it's doing.

When I try to fix the right nub over the weekend, I could of course also tip the charge LED just in case... though the probability for that coming loose is... well, within the error margin, really, next thing it would be bouncing in the case (oh dear... I hope I still see a LED where it should be ;-).

Additions:

Pandora was without battery for over an hour this morning, no change in battery status reporting.

In the end I just flashed the firmware "the normal way" (i.e. didn't start the SD excercise): no change in reporting. In the beginning it said of some updateable components being "already there, not updating. That was kernel and/or bootrom, if I understood correctly before the text whirred by onscreen.

Still havent tried fixing that nub and soldering the LED, will try in a moment.

The system info tool says: "Battery discharging", then on the next line: "Time to full at current rate: full now?". This is when I'm on battery, no charger. So SystemInfo is also "confused", though far more usefull in estimating the remaining juice.

Some signals are not "going places" on the MB? In addition to the LED that is. More stuff to solder?-)
 
Last edited by a moderator:
When I try to fix the right nub over the weekend, I could of course also tip the charge LED just in case... though the probability for that coming loose is... well, within the error margin, really, next thing it would be bouncing in the case (oh dear... I hope I still see a LED where it should be ;-).
LMAO :lol:

That's exactly why the LED didn't shine :lol:

There is no LED on the board!!! It's been there, since the traces have solder on them, but the LED is MIA.

Sometimes its the simple things...

Well. I've now got the Pandora in the alligator grips, so I'm still going to try fixing the nub issue...

But still... damn :p ...
 
I've not had time to think through the rest of your posts, but yes, installing the OS to SD and booting from that will have the same effect as reflashing your NAND on the issue at hand.
 
I'll continue this privately with ED, as it is almost certain by now that to have a 100% "as new" unit I would need access to spare components (like the LED)  I don't have access to :-(.

Any newcomer considering a Pandora purchase and getting frightened by this thread should know and understand that the unit we are discussing is a salvaged (i.e. "partially hand made") originally busted CC unit, this is not something that has come from the GC factory line!
 
Back
Top