Warning: don't leave your Pandora plugged in for too long


I think that for people that need to (for some reason, like that their pandora wont run on AC only) keep their pandora mains powered for extended periods of time it would be good to have a daemon written that would keep the battery at ~40% SoC (configurable, of course) by constantly varying the charge current so that it approximately equals the usage of the console (as has been suggested by MW IIRC). The battery would still see small charge/discharge currents, but atleast it wouldnt be constantly running above 4V.


It might actually be useful to have this as a feature that is engaged after the pandora has been plugged in for eg. 24 (or 48? - better make that configurable also) hours, with a menu option to force it to charge to 100% (also just re-plugging the charger would work).
 
When it locks up like that, you don't need to pop the battery out - just hold the Pandora button and flip the power switch as you would when powering it on. It does a hard reset.

Of course, that's moot if even that wouldn't get it out of such a state, but I've not come across that myself. :p

Tried that of course, every time, never worked.


The daemon stuff is a good idea, I hope something like that is done. Otherwise, what should be the procedure? Unplug it when it's 100%, plug it when it's low, repeat every day? Sounds like a huge hassle.
 
I didn't even know that was possible. :blink: Well, I learned something new today, at least.
 
I think that for people that need to (for some reason, like that their pandora wont run on AC only) keep their pandora mains powered for extended periods of time it would be good to have a daemon written that would keep the battery at ~40% SoC (configurable, of course) by constantly varying the charge current so that it approximately equals the usage of the console (as has been suggested by MW IIRC). The battery would still see small charge/discharge currents, but atleast it wouldnt be constantly running above 4V.


It might actually be useful to have this as a feature that is engaged after the pandora has been plugged in for eg. 24 (or 48? - better make that configurable also) hours, with a menu option to force it to charge to 100% (also just re-plugging the charger would work).

This seems like this would be the best way to handle this. But based on mweston's explanation it doesn't sound like the pandora has the ability to stop the battery from charging. It sounds like as long as it's plugged in it's charging.


If that's true then the only I can think of to handle this problem would be a little logic in between the charger and the pandora. A micro-controller and a mosfet to cycle the charging power on and off. Sounds like a fun weekend project.
 
It has been explained before that the TI chip does not allow the battery to be isolated from the power input circuit, therefore trickle charging is continuous.

Uh oh.


We use a lithium-ion polymer (Li-Po) battery, right?

Wikipedia said:
It is important to note that trickle charging is not acceptable for lithium batteries; Li-ion chemistry cannot accept an overcharge without causing damage to the cell, possibly plating out lithium metal and becoming hazardous.
 
Pssh, Wikipedia. My former economics professor vandalizes their business articles. Seriously though, that sucks. Does USB power trickle charge too?
 
IIRC we could kill the charge cycle (as it needs to be enabled by the kernel, so I think its possible to entirely kill it), but that isnt needed.


Charge current (now its set to 1A) (as seen by the TPS chip) is really system current + actual charge current (that can be measured by the bq chip).


So setting charge current = system current once a second should approximately keep the battery current near zero. That + setting charge current some 100mA over or under system current when the SoC is reported under or over the target percentage should keep the battery current low and state of charge floating around the set percentage.
 
LONG time. My uptime was 73 days. zodttd says he had his plugged in for over 4 months.
It's REALLY too LONG :unsure: But, thank you for info!


by the way, if the battery become so hot when you charge it, maybe better to charge without back cover for better cooling? and put Pandora to more cool place (to the cold window or air cooler or to the massive cold metal table, for example)


And one more question - this problems with battery coming only when user leave the Pandora on charge for a long time, or when just using Pandora as usual?
 
Last edited by a moderator:
As Wizardstan points out, if you can get the AC input draw down to below 80mA, then the charging circuit will be shut down but since the battery is not isolated, that means the entire system plus the battery trickle charging has to be under 80mA which is pretty much impossible.
Then I can almost consistently do the impossible. It sometimes takes an hour or two once it reaches 100% (although sometimes as little as a couple minutes) but the charge current in low-power mode eventually drops low enough that it shuts off the charge.

or old ones for that matter, there should be some way to shut the charging circuitry off from software
There is a way, it's pretty easy, the driver just doesn't support it at the moment.

So is it possible to have a daemon re-configure the charging chip if it can detect charge has been at >98% for an hour, and allow the battery to discharge (for 30 min or till it reads <90%)? Not sure if this actually makes life better for the battery or not (is active +/-Q better than static 100%?).
I think that for people that need to (for some reason, like that their pandora wont run on AC only) keep their pandora mains powered for extended periods of time it would be good to have a daemon written that would keep the battery at ~40% SoC (configurable, of course) by constantly varying the charge current so that it approximately equals the usage of the console (as has been suggested by MW IIRC).
This seems like this would be the best way to handle this. But based on mweston's explanation it doesn't sound like the pandora has the ability to stop the battery from charging. It sounds like as long as it's plugged in it's charging.
It does have the ability to software control the charging so all kinds of things can be implemented. Which way is correct though?
 
Yeah. I took a picture after I put it in a trash bag. The battery is going in the garbage tomorrow. Kinda freaky. The Pandora works fine on AC still.


View attachment 86
Was hoping for better news from your first post to this forum. Some juicy psx4pandora news maybe. :p

The fact that charging stops at all for some people feels like some undocumented errata issue in the chip TI will never acknowledge as I have said before.
Damned if you do!

It's an undocumented feature of TI's chip. So far, I've only seen a few Pandoras (two to five) out of a few hundred that DO NOT work without a battery.


Most do.
and maybe damned if you don't!


@Exophase: Unplug it and time how long it takes to drain. Work out a pattern with your power consumption and then buy a timer plug.


Community challenge: I'm not good with electronics but if I was... I would get the battery status from the server and talk to my Arduino which is controlling the a circuit which supplies my Pandora with power. 100% off 20% on... something like that. Gotta be safer than Titans suggestions. :p
 
It's REALLY too LONG :unsure: But, thank you for info!


by the way, if the battery become so hot when you charge it, maybe better to charge without back cover for better cooling? and put Pandora to more cool place (to the cold window or air cooler or to the massive cold metal table, for example)


And one more question - this problems with battery coming only when user leave the Pandora on charge for a long time, or when just using Pandora as usual?
The Pandora should not get hot with normal use, not sure if messing with voltage & overclocking could heat it up. This battery problem was caused by people leaving the Pandora plugged into a wall adapter with a battery inserted. If you need to keep the Pandora plugged in for long periods, just remove the battery altogether is the easy fix.
 
The Pandora should not get hot with normal use, not sure if messing with voltage & overclocking could heat it up. This battery problem was caused by people leaving the Pandora plugged into a wall adapter with a battery inserted. If you need to keep the Pandora plugged in for long periods, just remove the battery altogether is the easy fix.

MWeston told us to never try running it with the battery not inserted, that it could be putting the hardware at risk.


I'm really not sure what I should do, whenever I get another battery.
 
Then I can almost consistently do the impossible. It sometimes takes an hour or two once it reaches 100% (although sometimes as little as a couple minutes) but the charge current in low-power mode eventually drops low enough that it shuts off the charge.


There is a way, it's pretty easy, the driver just doesn't support it at the moment.


It does have the ability to software control the charging so all kinds of things can be implemented. Which way is correct though?

Same here. Mine does as yours and switches off the charging periodically allowing it to drop to about 85-90% before resuming to 100% for awhile.

The Pandora should not get hot with normal use, not sure if messing with voltage & overclocking could heat it up. This battery problem was caused by people leaving the Pandora plugged into a wall adapter with a battery inserted. If you need to keep the Pandora plugged in for long periods, just remove the battery altogether is the easy fix.

I've not noticed any additional heat at 1ghz for prolonged times. Your next tip is definitely the way if your not taking the Pandora anywhere.


I'd love a charging station/ dock in future where the batteries are removed to charge in holders, whilst the Panda is direct to the mains for use as a permanent computer, as I sometimes found despite the mains charger in, the battery wouldn't always keep up against the demands of Usb hubs, Hdd arrays etc and would drain down. Running direct on the mains solves this issue I occasionally encountered. Then just switch off, pop in a battery and off you go again.
 
So let me verify that I've got this straight:


1) It's bad to continue charging your Pandora once it's at 100%.


aka. It's bad to have your Pandora plugged in for too long with a battery.


2) It's bad to have your Pandora plugged in *without* a battery.


3) Putting in the battery while the Pandora is plugged in resets it.


The combination of 1 & 2 would seem somewhat cumbersome to achieve. But even if you do #2, there's no way to get the battery back in without resetting? What about all my (future) state? What about poor zodttd's uptime?


Is it also the case that:


4) Removing the battery while the Pandora is plugged in resets it??


Because that one seems similarly unfortunate.


I'm also hoping, though I suppose I've never seen confirmed (or denied) that:


5) Plugging the Pandora in while turned on *doesn't* reset it.


(and similarly)


6) Unplugging the Pandora while turned on (also) *doesn't* reset it.


The gp2x did those, and I found it quite frustrating, particularly so with the poor battery-indicator mine had.
 
MWeston told us to never try running it with the battery not inserted, that it could be putting the hardware at risk.
He said not to insert a battery if running on AC, to make sure you shut down completely and pull the AC before attempting to insert the battery as hot swap is definitely not supported (although having read the specs on the chip it may be possible to write an app that will allow it. Don't hold your breath though). There was another post he made that I recall that said running off AC was an "experimental" feature until he had received enough positive feedback to suggest it did actually work, but I can't find that one right now.


edit: Dude, the post right above the one I just posted was you thanking Evil Dragon for clarifying about not needing a battery to run the Pandora :D


edit edit: ok, here he reminds that it is an UNOFFICIAL feature that is not guaranteed to work. It's like overclocking: if it works, bonus; if it doesn't, no harm no foul.


edit edit edit: MWeston saying that, while experimental, it is still a fairly stable feature and "encourages" thousands of people to report success ;)
 
Last edited by a moderator:
I'm also hoping, though I suppose I've never seen confirmed (or denied) that:


5) Plugging the Pandora in while turned on *doesn't* reset it.


(and similarly)


6) Unplugging the Pandora while turned on (also) *doesn't* reset it.


The gp2x did those, and I found it quite frustrating, particularly so with the poor battery-indicator mine had.
Neither of those reset the Pandora.
 
Last edited by a moderator:
Okay, guess I'll try powering it without the battery then.


Looks like it booted up.


Thanks WizardStan.
 
Back
Top