[CM7] Help Track Down PBD (premature battery death)

[fattire]

WHAT IS PBD?

Some have reported issues with recent CM7 builds > 86 and premature battery death. This is being called PBD ("Premature Battery Death"). The main feature is a sudden loss of power at 30% or less without any warning.

Associated symptoms may include misreporting battery charge, where it suddenly jumps up (or down) 20% or more.

To get a hold on what might be the cause, let's collect info on whether people are having problems and with what configuration. Maybe we can figure this out together. For example, maybe everyone who's experiencing this problem is using the same combination of system/kernel, or maybe a certain configuration or procedure may make it go away.

PBD REPORT FORM

TEST CONFIGURATION:

BOOT DEVICE: eMMC
MLO: BN 1.2 built by fattire (MD5: a2c9a23efdbe05a40e0066500c5cfea1)
U-BOOT: BN 1.2 cyan build by fattire (MD5: c3db4062d1e67d50a382245471c1cef8)
KERNEL: uImage 2.6.32.9 from Nightly 86 (MD5: 9192c60a333c6ae32ab100c9eba16dca)
OVERCLOCKED?: to 925
GOVERNOR: Performance
SYSTEM: CM7 Nightly 86
PBD: NO
TEST COUNT: 2x- once with "untimely" u-boot. Which should be functionally the same. Both did not have BPD
TRIED REVERTING TO OLDER HC/CM7: NO
TRIED WIPING CACHE/DATA?: NO
DID IT WORK? N/A
PREVIOUSLY INSTALLED:

☑ Stock 1.0
☐ Stock 1.1
☐ Stock 1.2
☑ Nookie Froyo (any)
☑ Honeycomb (any)
☑ CM7 2.29 (any)
☑ CM7 2.32 beta 3
☑ CM7 nightlies > 84 (or whenever .32 appeared)

NOTES: I did have premature-battery death a week ago with the b3 system, the original 1.0 u-boot & MLO (built from BN source), and an overclocked kernel. Can't remember which one. Here is an image of 3% battery from Tues night (5/24/11).

THEORY

If you have a theory on what's going on, put it here.

 

[GabbleRatchet]

PBD REPORT FORM

TEST CONFIGURATION:

BOOT DEVICE: eMMC
MLO: BN 1.2 built by fattire (MD5: a2c9a23efdbe05a40e0066500c5cfea1)
U-BOOT: BN 1.2 cyan build by fattire (MD5: c3db4062d1e67d50a382245471c1cef8)
KERNEL: uImage 2.6.32.9 from Nightly 86 (MD5: 9192c60a333c6ae32ab100c9eba16dca)
OVERCLOCKED?: to 1200
GOVERNOR: Conservative
SYSTEM: CM7 Nightly 86
PBD: NO
TEST COUNT: Twice with current setup, each time nc shut off at 2%. Working on 3rd test now.

TRIED REVERTING TO OLDER HC/CM7: NO
TRIED WIPING CACHE/DATA?: NO
DID IT WORK? N/A
PREVIOUSLY INSTALLED: All of the below at least once

☑ Stock 1.0
☐ Stock 1.1
☐ Stock 1.2
☑ Nookie Froyo (any)
☑ Honeycomb (any)
☑ CM7 2.29 (any)
☑ CM7 2.32 beta 3
☑ CM7 nightlies > 84 (or whenever .32 appeared)

NOTES: I HAD PDB starting the night beta 3.1 was released. Shut down at 21% every time on every build from beta 3.1 through nightly 85. This plagued me through multiple data/system/cache wipes, different kernels, etc. When I went to do my full wipe for n86, I made a rookie mistake & formatted something I wasn't supposed to. The road to recovery led me to restore stock 1.0.1, upgrade to stock 1.2, ManualNooter, install clockwork recovery, full wipe, then install my current configuration. (I know, in a perfect world, that was way unnecessary, but I was at work & those were the only files I had access to. On the bright side, I was on the clock while doing this. Sorry boss.) I know others have had success flashing stock, discharging, charging fully, and back to cm7. I did not discharge & charge. I did all of this with about 60% battery, & on that charge I made it to 2% for the first time since .32 beta 1.

EDIT: I also wiped battery stats while flashing at full charge to try & fix. No dice there.

 

[ace7196]

PBD REPORT FORM

TEST CONFIGURATION:

BOOT DEVICE: eMMC
MLO: BN 1.2 built by fattire
U-BOOT: BN 1.2 cyan build by fattire
KERNEL: uImage 2.6.32.9 from 5/23 OC kernel
OVERCLOCKED?: 1.2 GHz
GOVERNOR: conservative
SYSTEM: CM7 beta 3.1
PBD: YES*
TEST COUNT: 3x*
TRIED REVERTING TO OLDER HC/CM7: YES*
TRIED WIPING CACHE/DATA?: YES
DID IT WORK? N/A
PREVIOUSLY INSTALLED:

☑ Stock 1.1
☑ Stock 1.2
☑ CM7 2.29 (any)
☑ CM7 2.32 beta 3

NOTES: Installed beta 3.1, updated bootloader 1.2 grabbed from stock image (after wiping EVERYTHING [cache, boot, system, data, dalvik]). Upon flashing 1.2 and letting battery run down to 2x%, had PBD. Repeated twice more.

I FIXED IT by flashing full 1.2 stock, running battery down until it shutdown (~4%), then flashed fresh copy of beta, 5/13 kernel. Fixed it. Since then I get down to 3%, and this is after installing the cyan edition of fattire's bootloader.

Something in the stock 1.2 is not being carried over.

 

[mutant13]

I know this said cm7(im on stock withMN) but... it happened to me just last night @ 20% very strange

BOOT DEVICE: eMMC
MLO: BN 1.2stock(MN 27)
U-BOOT: BN 1.2 stock
KERNEL: emmc 2.6.32.9 from 5/13 OC kernel
OVERCLOCKED?: 1.2 GHz
GOVERNOR: ondemand
SYSTEM:stock froyo bn1.2
PBD: YES*
TEST COUNT: 1x*
TRIED REVERTING TO OLDER nope*
TRIED WIPING CACHE/DATA?: nope
DID IT WORK? N/A
PREVIOUSLY INSTALLED:1.0.1 AN

 

[dim_adr]

PBD report

PBD REPORT FORM

TEST CONFIGURATION:

BOOT DEVICE: eMMC
MLO: BN 1.2 built by fattire
U-BOOT: BN 1.2 cyan build by fattire
KERNEL: uImage 2.6.32.9 from 5/13 OC kernel
OVERCLOCKED?: 1.2 GHz
GOVERNOR: smartass
SYSTEM: CM7 beta 3.1
PBD: NO
TEST COUNT: 2x
TRIED REVERTING TO OLDER HC/CM7: NO
TRIED WIPING BATTERY STATS?: YES
DID IT WORK? YES
PREVIOUSLY INSTALLED:

☑ Stock 1.2
☑ CM7 2.29 (any)
☑ CM7 2.32 beta 3

NOTES:

Contrary to public belief that PDB only occured in latest CM7 nightly/beta, I first experienced PDB in phiremod6.2 (based on CM 7.0.2) using dalingrin's OC kernel 04/24 when NC suddenly died at batt level 30%. I noticed that when connected to USB, the battery level suddenly jumped from >70% level to full 100% (it didnt jump when batter level is below 70%. It ). I strongly believe this causes PDB, as the battery calibration is incorrect.

I solved the incorrect battery indicator by wiping battery stats in CWM recovery boot from SD card (Wiping battery stats from CWM recovery in eMMC did not work for me). It worked and NC showed correct battery indicators. I regularly wipe battery stats whenever I encountered jump in battery indicator and I did not experience PBD.

This week I flashed CM7 2.32 beta 3.1 with OC kernel 2.6.32.9 from 5/13. I still consistenly encountered battery indicator jump when connected to USB (always) and power adapter (seldom). I always wipe my battery stats after that, and until now have not experienced PBD. I managed to use my NC until 6% yesterday when I decided to charge it. To be honest, it's quite painful to wipe battery stats everytime battery level jumped to 100%.

I think this PBD is caused by incorrect battery indicator level when connected to USB/power adapter, my opinion.

 

[paleh0rse]

I know this is an oddity, but I just experienced a true PBD at 45%... wtf?!

 

[greenmky]

I went from stock 1.2 -> CM7.0.2 stable -> CM nightly 87

BOOT DEVICE: eMMC
MLO: ?
U-BOOT: stock B&N 1.2 (I think? unless nightly 87 messed with it?)
KERNEL: .32
OVERCLOCKED?: to 1.1
GOVERNOR: Conservative
SYSTEM: CM7 Nightly 87 / OC Kernel 5/23
TEST COUNT: 1x
PBD: Died at under 4%. I don't think that counts, frankly - so I would so no problem here. I didn't see any funky percentage jumps either.

 

[skwalas]

Fattire, in the 1.2 update zip, there's a separate sub-zip labeled "charger". The updater script for that particular zip has a series of flags (0, 1, and 10) that appear to direct how the update will work based on the battery level when starting the update. Now, I've seen those flags elsewhere in the larger update zip and use in ways that don't appear to relate to installing the update directly. So, I am wondering if there is some call to a procedure that has been missed when isolating the uboot and mlo? Something that requires those flags to properly initiate certain activities when the battery hits milestones like 15%? But don't work properly if the flags or calls or whatever are absent?

 

[Gin1212]

I think this PBD is caused by incorrect battery indicator level when connected to USB/power adapter, my opinion.

I have never received PBD (don't get excited) because I've never let my battery hit under 30% because of all of the reports.

However, last night for the first time ever, I couldn't be bother to get out the adapter for the usb cable, so I just let the computer charge it all night.

Well, I'm a Sub. Teacher and I find on certain days I have a lot of time to read books, so I turn off the wifi and lower the display brightness and get excellent battery. There are days that I spend just reading (8hrs), and I get only as low as 60-80% depending on how heavy I really get to read or if I'm reading something like manga.

Today however, I subbed for a Kindergarten class that did not let me read pretty much at all and by the end of the day I only had like 55% battery with 3 of the 8hrs being heavy reading of a pdf. (The other 5hrs the Nook wasn't even turned on for a second.)

What's my point? I barely touched my nook, and yet had 55% battery by the end of the day with very low display brightness and no wifi. No games were played, and I really only used Adobe Reader and Ezpdf reader.

Hell, I had a Snesoid/Gbaoid day the other day (almost all day because the kids had a field day and I was left with a small group of kids doing independent work) and still only hit 55-60% by the end of that day.

So just my thoughts on "Usb Charging" over the Wall socket charging.

Edit: My settings are N86 OC (5-23) (300/1200 Conservative)
Edit 2: I am now seeing reports of heavy battery usage issues in the CM7 Dev thread, so it may be even possible that it's just using up a ridiculous amount of battery as well... However, that seemed like quite a bit for only 3hrs of real use.

 

[skwalas]

Well I'll be damned. It is for sure a calibration issue. Somebody Googled these batteries and read that they could not go below 3050 mv safely. I got pbd repeatedly just a few minutes ago at 20%... and 3000 mv.

Put the charger in and juiced it up a couple percent (showed 3600mv with charger in), unplugged, and let it run down again. Shutoff at 20% and 3000mv. I was able to replicate this multiple times. I already know that my 100% of scale is always right around 4200mv so it seems as if the 0% of scale is out of alignment.

How to recal the 0%?

 

[keyodi]

PBD REPORT FORM

TEST CONFIGURATION:

BOOT DEVICE: eMMC
MLO: BN 1.2 built by fattire (MD5: a2c9a23efdbe05a40e0066500c5cfea1)
U-BOOT: BN 1.2 cyan build by fattire (MD5: c3db4062d1e67d50a382245471c1cef8)
KERNEL: uImage 2.6.32.9 from OC 5/23
OVERCLOCKED?: to 1200
GOVERNOR: Conservative
SYSTEM: CM7 Nightly 86
PBD: NO
TEST COUNT: 2x- once with "untimely" u-boot and once with "cyan". Each time nc shut off at ~4%
TRIED REVERTING TO OLDER HC/CM7: NO
TRIED WIPING CACHE/DATA?: NO
DID IT WORK? N/A
PREVIOUSLY INSTALLED:

☐ Stock 1.0
☑ Stock 1.1
☑ Stock 1.2
☐ Nookie Froyo (any)
☐ Honeycomb (any)
☑ CM7 2.29 (any)
☐ CM7 2.32 beta 3
☑ CM7 nightlies > 84 (or whenever .32 appeared)

NOTES: I have dualboot setup on my nook with stock 1.2 installed on parts 9/10. Installation steps: Formatted /system(ext2), /data(ext2) and /cache, Installed stock 1.0.1 then updated to 1.2, setup dualboot partitions, moved 1.2 to dual boot parts, formatted /system(ext4), /data(ext4) and /cache, then restored CM from backup. I often switch between CM7 and 1.2.

 

[I Am Marino]

BOOT DEVICE: eMMC
MLO: ?
U-BOOT: Whatever came with N87
KERNEL: 2.6.32.9 OC 5/23
OVERCLOCKED?: 1200
GOVERNOR: Conservative
SYSTEM: CM7 N87
PBD: No
TEST COUNT: 1x
TRIED REVERTING TO OLDER HC/CM7: No
TRIED WIPING CACHE/DATA?: No
DID IT WORK? N/A
PREVIOUSLY INSTALLED:

CM 7.0.3 Stable.
Stock 1.1

EDIT + NOTES: Battery finally died at 4%. I went near or below 3,000mv when it happened.
I had 13% with 3606mv. Voltage started free falling once I hit under 5% battery, I watched it go from around 3400 to 3060 all within the 4% range, right before it died, was trying to screencap when it died on me. I was a little over 12 hours off charger, I took it off at 96%-ish.
It seems 3000 to 3050mv is that cutoff point.

 

[Narfx]

Well I'll be damned. It is for sure a calibration issue. Somebody Googled these batteries and read that they could not go below 3050 mv safely. I got pbd repeatedly just a few minutes ago at 20%... and 3000 mv.

Put the charger in and juiced it up a couple percent (showed 3600mv with charger in), unplugged, and let it run down again. Shutoff at 20% and 3000mv. I was able to replicate this multiple times. I already know that my 100% of scale is always right around 4200mv so it seems as if the 0% of scale is out of alignment.

How to recal the 0%?

Yep, my pbd happened at ~3-3.2V. Seeing as some people are reporting only ~10%-20% at ~3.6V, seems like our batteries were SEVERELY depleted. I hope they were not damaged because of that :/

I'm letting it run down now. Will post my u-boot, MLO et al. when/if it reaches 3.3V.

 

[I Am Marino]

I think it would help if people starting finding our their voltages when or near their battery dies. Seems to be a pattern developing.

 

[Animec]

This is odd, my 100% is now right around 4000 mV and I know for sure it's been closer to 4200 before :/

 

[sinanju]

This is odd, my 100% is now right around 4000 mV and I know for sure it's been closer to 4200 before :/

I have noticed that the meter will read 100% a while before the "n" light on the cable will go green, further pointing to a calibration issue. Perhaps that's your case?

 

[Ronin3178]

I think it would help if people starting finding our their voltages when or near their battery dies. Seems to be a pattern developing.

Here ya go, I got down to 4%, went to plug it in, and it shut down right before I got there. I got a screenshot of the 3041mV I believe. When my Nook comes back on (showing the charge screen right now) I will upload the shots.

Here is the datapoints I got 18-4%


18% 3568
17% 3573, 3566
16% 3563
15% 3538
14% 3562
13% 3557
12% 3565
11% 3505
10% 3508
9% 3501
8% 3499,3490
7% 3482,3475
6% 3390,3382,3356,3336,3321,3302
5% 3279,3256,3232,3198,3158,3123,3101
4% 3101,3041

A majority of this was sitting at battery booster screen watching at full brightness and with screen timeout set to never, so the cpu was pretty much idle.

 

[derekr]

It's real scary (and it's not Halloween!) I just finished a battery discharge test where I monitored the voltage and percentage as the battery was used up. The plot as a pdf is below:

http://dl.dropbox.com/u/22355237/NookBattery.pdf

Look at the rapid drop-off in voltage below 40% to 2860 mV at an indicated 34%. No wonder we are seeing PBD if others have the same problem!

I was watching the voltage drop before my eyes. I actually shut it down at the 2860 mV, and I can't turn it on again (I'm at work) - no charger.

Something is very wrong with the calibration... At an indicated 34% the battery is dead, deceased, gonzo, perished, pushing up daisies......

Edit: n87 - OC523 - data taken with Circle Battery widget.
If you can't see the plot - here's the data:
86 3917
84 3886
79 3839
76 3795
70 3758
67 3751
58 3696
46 3639
40 3604
37 3556
35 3503
34 3271
34 3172
34 2861

 

[Ronin3178]

It's real scary (and it's not Halloween!) I just finished a battery discharge test where I monitored the voltage and percentage as the battery was used up. The plot as a pdf is below:

http://dl.dropbox.com/u/22355237/NookBattery.pdf

Look at the rapid drop-off in voltage below 40% to 2860 mV at an indicated 34%. No wonder we are seeing PBD if others have the same problem!

I was watching the voltage drop before my eyes. I actually shut it down at the 2860 mV, and I can't turn it on again (I'm at work) - no charger.

Something is very wrong with the calibration... At an indicated 34% the battery is dead, deceased, gonzo, perished, pushing up daisies......

Edit: n87 - OC523 - data taken with Circle Battery widget.
If you can't see the plot - here's the data:
86 3917
84 3886
79 3839
76 3795
70 3758
67 3751
58 3696
46 3639
40 3604
37 3556
35 3503
34 3271
34 3172
34 2861

If you look at my post above yours, yeah, def a calibration issue, mine dropped out @ 4% 3041mV

ps, I would switch axis on the graph 0-100% bottom to top, 4000mV to 2400mV left to right

 

[derekr]

If you look at my post above yours, yeah, def a calibration issue, mine dropped out @ 4% 3041mV

ps, I would switch axis on the graph 0-100% bottom to top, 4000mV to 2400mV left to right

As you wish :
http://dl.dropbox.com/u/22355237/NookBattery2.pdf

however, let's discuss this over a beer sometime - which is the independent variable here? I think the first plot shows the precipitous voltage drop-off in a more graphic fashion... But, as always, your wish is my command

BTW - MATLAB won't let me (simply) reverse the x-axis order...