Battery Drain

[WizardStan]

Ok, it's consistent now. I don't need to towel wrap my Pandora or anything, still using the same Sony charger that's been charging my PSP for years, when it gets to 100% it just stops charging. Three nights in a row now it has shut off its charging and been down a couple of percent when I wake up. It's sitting on my desk beside me now so I can monitor it. If I unplug it and plug it back in, the red LED comes on for about 5 minutes or so before shutting off again. Something fishy is going on and I no longer believe it is an overheating thing.

 

[Prometheus]

That is odd, Stan. Incidentally, I've switched back to using my Flypower charger, and have encountered no more oddball behaviour.

In the really off state, you'd need to leave a pandora alone for at least a week to get a good idea of the battery drain (and then I'm not sure if the battery gauge would still be accurate).

I've just completed leaving my Pandora alone for one week and one day in order to give a rough idea of this. It was horrible, but someone had to do it. I will note that I did use it for a couple of minutes, twice during this time.


The total battery drain over the course of one week and one day, with a few minutes of usage on two occasions, was 6%.


EDIT: While I think of it, the few minutes of usage only seemed to total about 1% of usage.

 

[WizardStan]

So I was poking around at the internal registers of the charge chip, reading the spec sheet on it, and come to conclude that there are some circumstances where, even trickle charging, the chip will just say "ok, battery is full, I'm done" and stop charging. However, even though it isn't charging anymore, it is still monitoring the charge. In the past, I've assumed that it stopped charging meant there was a problem. Now I know better. Armed with knowledge I left it plugged in and waited. And waited. And waited. At around 85%, the charge light came back on.


So what I'm asking is for people who have had this "problem" to perform a similar test: when the LED shuts off and the charging has stopped, don't unplug your Pandora: leave it plugged in until it has discharged to about 85% (maybe leave it to 80% to be sure) to see if the charging turns back on. You may want to turn on the wifi and use it to try and force it to discharge faster. If a reasonable sampling consistently get the same result, I'll talk to Notaz and MWeston about getting a fix in for it.

 

[MarioPandio]

Okay I saw your related posts on both forums Wizard and as I had noticed the charge light go off before, I have tested as you wished and you are correct.


After 100% and still on the charger for over 1hr, I flipped it to standby and almost immediately the light blinked twice, then after five minutes it blinked five times and then switched off. At this point I left it awhile before resuming it and saw it had dropped by 1-2%.


I then OC'd to 1ghz and ran Ultimate Doom via PrBoom at full volume, with wifi on and chromium and a few other basic apps running. This was at 1pm and by 3.30pm my kids were back from school, so I escaped Doom to the Pickle Launcher and was flashed the XFCE desktop at 87%, at which point the charge light resumed. Its now at 92% at stock speed after just 10 minutes or so.


Hope that helps somehow.

 

[WizardStan]

It does help, it means I'm not alone in my observation

I escaped Doom to the Pickle Launcher and was flashed the XFCE desktop at 87%, at which point the charge light resumed.

So the charge light didn't come back on until after you quit Doom? There is an aspect of CPU required to turn the charge light back on, but it's trigger driven so shouldn't be affected by CPU usage.


I also noticed when I woke up this morning that mine had dropped to 81% in standby, but the charge led clicked on seconds after resuming speed and noticing it had dropped to 81%.


This is some weird interaction that I'm going to have to investigate further. The good news is I'm 99% certain there's an easy software fix to keep it trickle charging forever, I'm still just curious as to what the actual causes of it dropping out are, and the conditions that trigger it to start charging again.

 

[tsh]

The good news is I'm 99% certain there's an easy software fix to keep it trickle charging forever

I think that is technically a bad thing for the battery (based on random Wikipedia stuff, not a real datasheet). If the battery chemistry is not able to burn excess charge as heat in the same way as a NiCd or NiMH does, the charge current should stop at 100%. Probably explains why the HTC desire ramps between 95 and 100% on charge, and my ac100 rate of charge drops significantly from 90% (but it does stay level at 100%, and AFAIK used a TI PMIC).


Maybe there is some tweaking which will mean the PMIC can be pushed back to a charging state sooner than 85% - that would be good. Probably this is some almost working feature in the PMIC which is intending to be nice to the battery, but is a little buggy. Maybe worth checking in the android GIT to see if there is anything similar mentioned there - the Desire for example is better now than it was 6 months ago.

 

[WizardStan]

Just finished another test: Wifi on, OPP5, clocked to 1000, running N64 emulator, draining the battery like some kind of battery draining vampire, the charge circuit kicked back on at 92%.


According to the spec sheet, the charge is supposed to come on when it drops below a specific voltage, so I'm confused as to why there's such wild fluctuation. The bq chip does not report sudden changes in voltage but the charge chip is reporting different voltages depending on current draw. High current drags the reported voltage down, low current boosts it up.


MWeston? Little help here?

 

[tsh]

The bq chip does not report sudden changes in voltage but the charge chip is reporting different voltages depending on current draw. High current drags the reported voltage down, low current boosts it up.

Check where the current sense resistor goes. I think MWeston commented on this before, one of the sensors is not as accurate as the other.

 

[MarioPandio]

Was probably coincidental timing the first time, as I just repeated things and was away from the pandora whilst it ran Prboom demo's and the charge light had kicked back in at 86%.

 

[centus]

Just finished another test: Wifi on, OPP5, clocked to 1000, running N64 emulator, draining the battery like some kind of battery draining vampire, the charge circuit kicked back on at 92%.


According to the spec sheet, the charge is supposed to come on when it drops below a specific voltage, so I'm confused as to why there's such wild fluctuation. The bq chip does not report sudden changes in voltage but the charge chip is reporting different voltages depending on current draw. High current drags the reported voltage down, low current boosts it up.


MWeston? Little help here?

Hey there, just a little piece of information from dealing with LiPo batteries on RC planes.


The voltage will always drop as the current draw goes up, usually, this is how the component in charge of stopping the current draw to prevent damage to the battery (at least in planes) decides when to drop it, when the voltage per cell goes below a certain value (I think around 3.6V).


The interesting part is that if you stop the current draw, the voltage will go back up so it is a poor indicator of the battery's capacity, but is the best possible one given this behavior.


That said, it makes some sort of twisted sense that the lower your current draw, the more you can get out of the battery before you hit the red line.


hope it helps...

 

[WizardStan]

Not really. It's an odd thing: the voltage reading from the twl charge chip differs depending on the load. A simple calculation indicates it can swing by as much as 5% or so from high load with wifi on to completely idle OPP1 with the LCD off. At the same time the bq fuel gauge chip does not do this same percentage swing: the capacity it reports is stable no matter the current.


I'm not sure how the fuel gauge chip is wired that it would report different from the charge chip. I mean, your explanation makes sense, except there's also this other chip that is not reporting differences in voltage. In any event, I'm testing different levels at which the charge is supposed to restart. If MWeston ever returns to discuss this, I'd like to get his take on the situation and make sure I'm not making things worse.

 

[urjaman]

It could be that the bq chip is sensing voltage at the battery and the twl chip is sensing voltage after the current-sensing resistor.

 

[centus]

What I meant in my last statement is that the higher your current draw, the sooner you will see a difference in the battery voltage.


That said, voltage of the battery under load is the best indicator of the battery's capacity, when you take away the load, it might go back up to a higher voltage value but it doesn't mean anything because as soon as you draw more out of it, the battery voltage will drop, until eventually it'll be below the red line for the voltage regulator.


Urjaman's explanation makes sense, if those two chips are looking at the voltage before and after the regulator, it might explain the differences.


But as you said, all knowledge of this exists, we just need to get MWeston's attention...

 

[Tor]

Note that the battery has internal resistance, it's like if it had a virtual resistor between a virtual internal 'battery' and the external connectors. In practice this means that the more current you draw the more the voltage drops at the connectors. Lower-capacity batteries act as if the internal resistance is higher than for a better-capacity battery (that's why old or too small batteries react so badly to load, where you may see the battery meter suddenly drop to warning level or worse).


This all happen before any circuitry is even coming into consideration (and that part can complicate things even further, but the voltage drop happens for purely electrical reasons in any case).

 

[WizardStan]

This all happen before any circuitry is even coming into consideration (and that part can complicate things even further, but the voltage drop happens for purely electrical reasons in any case).

So why does one chip report this voltage drop but not the other one?

 

[Tor]

So why does one chip report this voltage drop but not the other one?

I scrolled back for a number of posts, but I couldn't find where you mentioned two different voltage readings by different chip. So I'm unsure. Of course, the internal resistance which forces a voltage drop as load increases is always there, but the magnitude of it will differ and may or may not be significant compared to whatever happens in outer circuitry. A fresh, new, high-capacity battery will have low internal resistance compared to a warn-out, or low-capacity battery: The latter will have a much greater voltage drop when the load increases.

 

[WizardStan]

I've mentioned it in both posts, one of which you even quoted

The bq chip does not report sudden changes in voltage but the charge chip is reporting different voltages depending on current draw.

and

Not really. It's an odd thing: the voltage reading from the twl charge chip differs depending on the load. A simple calculation indicates it can swing by as much as 5% or so from high load with wifi on to completely idle OPP1 with the LCD off. At the same time the bq fuel gauge chip does not do this same percentage swing: the capacity it reports is stable no matter the current.

 

[Tor]

The fuel gauge I can understand, it's presumably trying to show how much juice is left in the battery and that'll be mostly guesswork, taking into account voltage as just one of the factors. And presumably the algorithm could even be changed by software (you may have mentioned something about that already, as you noticed I'm not always good at noticing all the bits when re-scanning old posts )

 

[sparkymark79]

Hello,


this is a question for Mweston or somebody else electrically clever


For months now I have resorted to shutting down the Pandora, opening the battery cover and using a slim piece of post it note to break the circuit to the battery (then replacing the battery and cover). When I want to play again, I pull on the post it note and hey presto the machine switches back on. I have to do this due to battery drain. I - like a few others experience 20-25% battery drain when the unit is switched off over the course of a night. With the post it note trick I get no obvious drain at all and can leave it alone swiotched off happy that my battery isn't discharging.


It is a pain in the ass though, having to open up the battery cover everytime I shut down so If I could get this fixed it would be brilliant. I was wondering what could be causing this (please don't say it's battery alignment). What about the copper shielding on the inside, could any of this be causing a short. Anything else you can think of - as I wouldn;t have thought it a universal board issue as only a few people like myself are experiencing it


Thanks


Mark

 

[Yannick]

There's a real volume wheel.


Why no real on/off switch?


one that breaks the circuit.