My satisfaction with the Pandora after owning it for 2 years now

[TrashyMG]

@All: I have a totally new idea concerning my WiFi and charging issues. What has NOT changed throughout all the 3 Pandoras (2 originals, 1 rebirth) I have had is ……… the battery!!! Maybe this is causing both my issues with charging and potentially WiFi, as the WiFi chip is directly connected to the battery, according to notaz. It is likely that at the OpenPandora workshop the testing was done with their spare batteries, and therefore the issues were unnoticed? I will ask ED & Askarus directly, and if I get infos, follow up here.

Could be... I have three Pandora(s), but each have their own battery and I've never mixed and matched them yet. maybe why I've never seen any odd behavior.

 

[Askarus]

@All: I have a totally new idea concerning my WiFi and charging issues. What has NOT changed throughout all the 3 Pandoras (2 originals, 1 rebirth) I have had is ……… the battery!!! Maybe this is causing both my issues with charging and potentially WiFi, as the WiFi chip is directly connected to the battery, according to notaz. It is likely that at the OpenPandora workshop the testing was done with their spare batteries, and therefore the issues were unnoticed? I will ask ED & Askarus directly, and if I get infos, follow up here.

I don't know if we were using your battery or our own.

Did you send your one with you?

 

[porg]

I don't know if we were using your battery or our own.

Did you send your one with you?

@Askarus: Thanks for joining the public discussion! AFAIR I sent my battery with it, and if so, surely separated from the unit (to avoid unexpected electrical situations) and well packed. Maybe my battery was so well wrapped, that you left it aside and took you own spare battery?

Wipe the contacts with isopropyl alcohol?

@Binky: Thanks for your hint.

@Askarus: As an official OP serviceman, do you recommend that as well?

 

[Askarus]

I don't know if we were using your battery or our own.

Did you send your one with you?

@Askarus: Thanks for joining the public discussion! AFAIR I sent my battery with it, and if so, surely separated from the unit (to avoid unexpected electrical situations) and well packed. Maybe my battery was so well wrapped, that you left it aside and took you own spare battery?

I slightly remember you had a pretty old battery, right?

I really can't remember if I used it.

Wipe the contacts with isopropyl alcohol?

@Binky: Thanks for your hint.

@Askarus: As an official OP serviceman, do you recommend that as well?

Yes.

Clean the contacts on the battery (Don't think you can brake much here)

Then clean the contacts on the Pandora. It's secure if you first put the alcohol onto some rag and then clean the contacts with it.

 

[porg]

1) Booting the Pandora without a battery just connected to the original OpenPandora charger (5V @ 2000mA) had no positive influence on WiFi (my report here).

Before going into cleaning the battery contacts with isopropyl alcohol, I did these further test:

2) Charging via USB charger (Asus ADP-188W A Rev.01, 5V @ 2000mA):

My Pandora also stopped charging far (2500mAh ~68%) before reaching 100%.



No advantage compared to the mains charger.

3) Maybe cleaning the battery contacts will help. If not, I have to conclude that:

a ) my battery was broken quite from the early beginning (could a battery be "so faulty" to cause the charging issues?), as I recall charging issues ever since having the power-savvy OS, or

b ) that my Pandora is so power-savvy (all radios & services off, screen with middle brightness) as few other Pandoras, and that therefore the treshhold value (safety feature of the TWL, as described by WizardStan) has to cause failure! (Some other users reported similar issues, i.e. A3000)

4) If the contacts cleaning test fails, my next attempt would be to change the treshhold value. Instructions already now please, so that I could start right away! Thanks!

 

[WizardStan]

Ok, download this: http://rebirthofxeen.com/files/pandora/i2c-tools_3.0.2-r2.6_armv7a.ipk<br />Install it with "sudo opkg install i2c-tools_3.0.2-r2.6_armv7a.ipk"<br />(I'd tell you to get the latest from the Angstrom repo but the latest version requires a newer version of libc than we have on the Pandora, so slightly older version is required)<br />On the command prompt, copy the following

sudo i2cget -f -y 1 0x4a 0x91
(should output 0x69, this is the default. If not, stop)
sudo i2cset -f -y 1 0x4a 0x85 0xf4
sudo i2cset -f -y 1 0x4a 0x91 0x39
sudo i2cget -f -y 1 0x4a 0x91
(should now say 0x39 to match what we just pushed onto it)

This will set the end-of-charge threshold much lower than default and you should see it charging for longer. If it still stops before reaching full charge we can turn it down further, or even off entirely (use 0x09 instead of 0x39. That '9' needs to stay unchanged for now). It should be safe to disable it, there's a userspace daemon watching to shut it off when it reaches 4.2volts, but I'd still watch it carefully the first few times it approaches 100%.<br />Note that you will have to run the i2cset commands (the i2cget commands are just to make sure things are correct) every time you reboot. You can stick them in an rc5.d script or something.<br /><br />Before doing this though, there's one more thing that just came to mind that I'd ask you to try: plug it in and let it charge until it thinks it is finished charging (ie, the LED shuts off even though it isn't 100%). Make a note of the percentage being displayed. Turn it off, remove the battery, and wait 10 minutes. Put the battery back in and check the percentage again. I'm just making sure it's a problem with the twl, not the bq. If it's a problem with the BQ, then the percentage after reinserting the battery will actually be 100%, or pretty close, instead of the 85% or whatever you've been experiencing lately. I'm pretty sure it's the twl but I just want to make sure before we go doing things with the potential to harm your battery.

 

[porg]

Dear WizardStan!

Thanks for your instructions. By the nature of the issue — full dis/re-charging takes time — my tests will take some days, but then I'll report back.

Before doing this though, there's one more thing that just came to mind that I'd ask you to try: plug it in and let it charge until it thinks it is finished charging (ie, the LED shuts off even though it isn't 100%). Make a note of the percentage being displayed. Turn it off, remove the battery, and wait 10 minutes. Put the battery back in and check the percentage again. I'm just making sure it's a problem with the twl, not the bq. If it's a problem with the BQ, then the percentage after reinserting the battery will actually be 100%, or pretty close, instead of the 85% or whatever you've been experiencing lately. I'm pretty sure it's the twl but I just want to make sure before we go doing things with the potential to harm your battery.

Will try that.

It should be safe to disable it, there's a userspace daemon watching to shut it off when it reaches 4.2volts, but I'd still watch it carefully the first few times it approaches 100%.

"Watching it careful" contradicts the idea of testing positively for "unattended charging"!

• Is it really that dangerous? If so, then the best considerable safe idea I have is to start charging, take sysinfo's predicted "Time to full at current rate", add up a little more, because charging slows down the fuller the battery gets "Estimated time to full", and at that time see if the charging LED is still on.
• If the LED is OFF, open the lid and see how charged it is, and if considerably below 100%, check sysinfo's log, when charging stopped.
• If LED is ON then I know it is still charging. But I cannot know wether it is charging ok, or already overstressing the battery. I have to open the lid, see the sysinfo log and determine how the current load came into being. If it is at 100% constantly, I have to assume charging is overstressing the battery, if it is bouncing 100 -> 95 -> 100 -> 95 -> … then I assume it's correct, if it's considerably below (without a higher load in its charging history), then I know it simply did not yet reach the final load, but as I intervened, the test is useless, and I have to start yet another test from a low to full charge!

Note that you will have to run the i2cset commands (the i2cget commands are just to make sure things are correct) every time you reboot. You can stick them in an rc5.d script or something.

man page mentions:

WARNING: i2cset can be extremely dangerous if used improperly. It can confuse your I2C bus, cause data loss, or have more serious side effects. Writing to a serial EEPROM on a memory DIMM (chip addresses between 0x50 and 0x57) may DESTROY your memory, leaving your system unbootable! Be extremely careful using this program.

I hope you tell my the correct registers and values!!!  :wacko:

 

[WizardStan]

Those are correct, don't worry about the i2cset commands.

The TWL doesn't let you randomly muck around with its data: the first i2cset sets a special flag that unlocks the the one register the second i2cset command writes to. If you remove the first i2cset, or send it the wrong value, the second command won't work at all. It's practically impossible to accidentally do any damage: you'd have to make two mistakes that coincidentally allow you to unlock the wrong register and then write to that wrong register; very unlikely.

"Watching it careful" contradicts the idea of testing positively for "unattended charging"!

Just the first time it approaches 100%, keep an eye on it to make sure it stops charging reasonably. It's not super dangerous, just that if you are charging it and see it is at 100%, and then an hour later you check it again and it is still charging (ie, it's been at 100% for an hour and hasn't stopped charging) then you'll want to unplug it and check back.Overcharging a little bit has a very minor impact on battery life. The problem comes when you're overcharging for hours at a time, day after day, for a few weeks. A couple of hours doesn't really hurt, but a couple of hours every day (or even week) can add up to a severely damaged battery. If it does shut down like it is supposed to, and bounces between 100 and 95 like you say, then you'll be fine to leave it charging completely unattended

 

[porg]

1) TWL I2C command: Writing to one specific register in order for another register to become writeable! Clever protection system!

I love the Pandora community just for learnings like this one. One really gets an understanding of low level operations.

2) Precautions with low treshold: Ok, now it is clear to me how I can do my tests.

 

[porg]

1) Cleaned my battery contacts with isopropyl alcohol yesterday.

Before doing this though, there's one more thing that just came to mind that I'd ask you to try: plug it in and let it charge until it thinks it is finished charging (ie, the LED shuts off even though it isn't 100%). Make a note of the percentage being displayed. Turn it off, remove the battery, and wait 10 minutes. Put the battery back in and check the percentage again. I'm just making sure it's a problem with the twl, not the bq. If it's a problem with the BQ, then the percentage after reinserting the battery will actually be 100%, or pretty close, instead of the 85% or whatever you've been experiencing lately. I'm pretty sure it's the twl but I just want to make sure before we go doing things with the potential to harm your battery.

2) Tested this today, and I can says: Something is NOT ok with my charging chips!

# Getting my reading

10:16, sysinfo 1576mAh, xfce-status-bar 40%

# Shutting down. Taking out my battery.

13:29 # Rebooting, did not read battery gauges right after booting

# Working for 10 minutes, then looking at the battery gauges

13:39, sysinfo 1789mAh, xfce-status-bar 46%

The battery charge is reported HIGHER (!) after having had the battery out for ~3 hours and using the Pandora another 10 minutes!!!

3) What do you conclude from that, and with what test shall I proceed next?

I'd now attempt to fully charge (with the cleaned contacts & the yet unchanged treshhold value).

If it stops, I'll note the charge, shut down, take out the battery for 10mins, then reboot, get another reading of the charge.

 

[WizardStan]

The battery charge is reported HIGHER (!) after having had the battery out for ~3 hours and using the Pandora another 10 minutes!!!

After being allowed to "sit" for a few minutes, it's not unusual for a battery to appear to have a higher voltage. I've used this trick to get a couple extra minutes out of a dead battery in a flashlight: if the light is dim, take the batteries out and wait a minute or two before trying again. Freezing them also gives a further temporary boost of power. That 6% isn't really important: maybe that's how much power it really should have had and some minor current fluctuations were keeping it from reporting properly, or maybe that's an artificial 6% boost and you'll simply deplete that 46% in the same time it would have taken to deplete the 40%. Doesn't matter, you can experiment if you want but I'm just worried about what might happen if you put it in and it was significantly higher, 20% or more, especially if it hit 100%.

 

[porg]

1) Repeated that battery out/in test at medium charge, and again I had:

40% (1575 mAh) —> battery out, wait ~20 minutes, battery in —> 46% (1791 mAh).

2) Will do the treshhold-altering test on the weekend, when I get time for it.