Release Hotfix 6 Final released!


WizardStan

Mega GP Mania
Joined
May 24, 2008
Messages
16,733
I don't think this is the charge limiter. It should be switched off by this point, if not by setting enabled to 0 then definitely by setting the limit to 0.


The only thing I can think of is that the percentage is wrong. Here's the situation: the TWL chip is SUPPOSED to stop charging when it reaches a certain threshold automatically. My original one did this, as well other users have reported that theirs did this (albeit usually as a bug report ;) ). Anywho, that is entirely expected behavior. The problem, first, was that not all Pandora's would stop charging as they were supposed to: some slight variance of the boards kept the detected charge rate from dropping below the expected threshold long enough to stop the charging, which resulted in more energy being dumped into the battery than it could safely store, slowly damaging it over time. The second problem was that, even those that did stop charging would not restart again until the battery was quite low, resulting in people waking up in the morning with 70-80% charge remaining; many a bug report was filed. The charge limiter fixes both problems.


Now, on the other hand, there's the bq chip. While the twl charges the battery, the bq keeps track of that charge and reports an estimated maximum and remaining amp hours. If it has miscalculated the maximum (and it's a learning chip, so this value can go up and down over time, depending on usage and the actual abilities of the battery) to be 5% higher than the actual maximum, then it could be that the battery is actually full, the twl is doing the thing it was designed to do and shutting off the charge, just that the bq has an improper measurement of the battery capacity now for some reason. I've been experimenting with how to get the bq chip to quickly recalibrate. I've cycled my battery more times in the last week than I have in the month before, but I think I'm finally onto something :)


When I've got the details worked out, I'll post a how-to on recalibrating the bq chip. I'm not certain why it would have become uncalibrated in the first place though. It's kinda weird, but since none of it happened before HF6 there's a good chance it has to do with the safety limits.
 

EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,668
Age
43
Location
Ingolstadt
Well, we FLASH the bqchip before shipping out the Pandoras with values the MWeston measured.


If we don't flash it, the battery reading it totally uncalibrated.


However, I thought this was a fixed value we're flashing there... does that change somehow?
 

DaMummy

Soldier Paste
Joined
Nov 5, 2009
Messages
4,417
Age
34
Location
Ohio
its official, my pandora cant hold a charge worth a damn in low power mode. at some point in time, it dies completely. sometime within 24 hours. seems like it just loses voltage and wont power on, as i can get out of low power mode after 5 hours, and still show only less than 5% drop in battery. somewhere down the line, it just dies from wherever it is, even from 90% straight to 0
 

WizardStan

Mega GP Mania
Joined
May 24, 2008
Messages
16,733
However, I thought this was a fixed value we're flashing there... does that change somehow?
The default values were all there, and all looked good. There's no doubt that these were the values that MWeston found to be good: they work fine in nearly every Pandora it would seem, but there's just something still not quite right. Even natural drift doesn't fully explain this. I've forced a recalibration and the bq is now reporting that my Pandora can only store 3500mah. Both of my batteris. That's a far cry from the 4200 that these were designed with. Yet, even with the apparent 15% loss in storage, I'm still getting the same play time as before; I should know, I've had to wait while cycling this thing several times now!


So weird...


edit: DaMummy's issue is something completely different. :/
 
Last edited by a moderator:

EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,668
Age
43
Location
Ingolstadt
You mentioned you're in contact with someone to solve this issue... are you on some track here or is there something I can try to do to help?


This is an issue that needs to be solved as soon as possible.


We're working on HotFix 7 already and have some stuff for an Alpha 1 release already (just need some time to actually compile everything together), would be neat if we could get some patches for this issue as well.
 

DaMummy

Soldier Paste
Joined
Nov 5, 2009
Messages
4,417
Age
34
Location
Ohio
sad part is, my pandora can stay powered on with battery draining at regular pace if i just leave it on with lid closed, of course thats until it gets to that magical 17% for me when it likes to initiate the critical battery power down
 

WizardStan

Mega GP Mania
Joined
May 24, 2008
Messages
16,733
I'm pretty sure it's an issue with the sense resistor (a 10mohm resistor between the battery contacts to measure current use) being incorrect. According to the tech at TI, if the resistance is a little bit higher for some reason (a bad resistor or, I dunno, corrosion on the solder pads for example) then it will dectect a lower current flow than actual which will result in exactly the situation I saw. He was a little surprised that I had to enable the IT algorithm manually though. He said it should be enabled on all devices going out to production so they calibrate to exactly this situation: it may be detecting an incorrect current flow, but it doesn't matter because it balances that out by detecting a smaller capacity battery, even though there's nothing wrong with the battery.


I'm not sure if this will become a widespread problem (not likely) but if it does the fix is pretty simple: enable the IT algorithm (I'll write a program to do just that) and then deep cycle the battery.
 

Darkknight512

Active Member
Joined
Sep 7, 2010
Messages
637
Age
27
Location
Canada
It also depends on the charge cycles the battery has gone into, from my experiences with LiPo batteries for RC use, they kinda start at 80-90% full rated capacity and after about 10 partial charge cycles they can take a full almost 100% charge before slowly declining.


This might be of use, I wrote a presentation on rechargeable batteries for a chemistry class a while ago. This was one of the best resources I have used. http://batteryuniversity.com/learn/


What is the precision of the sense resistor (from the manufacture)? Keeping in mind to check the resistance you need to take the resistor out of the circuit.


What is our cuttoff voltage for the battery? 3.5 Ah might be correct if the Pandora cuts off at around 3.6 volts because 4.2 Ah is rated at something like 3.2 volts cutoff. Using voltage to measure battery capacity is close to linear, but it has curves to the beginning and end of the capacity vs voltage graph.


Also I have noticed that our battery gets quite hot when drawing more then 350 mA, we could be losing quite a bit of capacity due to that, that is something the sense resistor will not see.


I'll charge my Pandora's battery to full, and then take it out and let it sit for about an hour (after a charge the battery voltage will be higher then expected) and check the battery voltage with a DMM, we might be charging the battery up to 90% in the first place instead of the expected 100%.
 
Last edited by a moderator:

notaz

Certified Guru
Joined
Aug 23, 2005
Messages
4,913
Location
Lithuania
Website
notaz.gp2x.de
This is an issue that needs to be solved as soon as possible.
Well if you want to revert to pre-HF6 behavior just set [battery_charge] enable to 0 in /etc/pandora/conf/eventmap, that's it. It's pulled into firmware from pandora-libraries.git.

I'm not sure if this will become a widespread problem (not likely) but if it does the fix is pretty simple: enable the IT algorithm (I'll write a program to do just that) and then deep cycle the battery.
What makes you think IT is not already enabled here? Also check Your PM.
 

WizardStan

Mega GP Mania
Joined
May 24, 2008
Messages
16,733
What makes you think IT is not already enabled here? Also check Your PM.
IT Enable register was 0x00 :p


Issuing an ENABLE_IT command flipped it to 1 and everything started working again.


Unless something happened and IT became un-enabled somehow.
 

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,724
Ok, got a 100% with HF 5.


Rebooted with HF 6 with the adapter, saw 98 %, then 10 seconds after, 100% and the charge led went off.
 

slaeshjag

¯\_(ツ)_/¯
Joined
Apr 8, 2010
Messages
2,687
Location
~Stockholm, Sweden
Hmm... My pandora can't see the battery at all (and it doen't turn on when a charged battery is inserted,) could this be bacause of buggy settings in the bq-chip? I have been able to observe a tiny difference in the displayed charge level with it inserted compared to when it's not (about 1 or 2 percent points lower with battery inserted.) I have yet to see the charge LED light up (I've had it for more than a month,) and if I unplug the charger with the battery in, it dies pretty much instantly. Could it be a software issue, or will i need a new board?


EDIT: I got it with a beta of Hotfix 6
 
Last edited by a moderator:

WizardStan

Mega GP Mania
Joined
May 24, 2008
Messages
16,733
That sounds like a completely different problem unrelated to anything in HF6. Did you create a thread about it in the support forum?
 

slaeshjag

¯\_(ツ)_/¯
Joined
Apr 8, 2010
Messages
2,687
Location
~Stockholm, Sweden
Neh, ED said it was probably because of board damage caused while shipping it, so I didn't bother. But due to the lack of replacement boards and some people having problems with premature charging cutoff made me kinda hopeful that I could have a usable pandora without waiting another two months :p
 

notaz

Certified Guru
Joined
Aug 23, 2005
Messages
4,913
Location
Lithuania
Website
notaz.gp2x.de
Top