#6092 HIGH Never A: Battery Charging problems. (Collector bug)

[Zarro Boogs per Child]

#6092: Battery Charging problems.  (Collector bug)
----------------------------------+-----------------------------------------
  Reporter:  rsmith               |       Owner:  rsmith        
      Type:  defect               |      Status:  new           
  Priority:  high                 |   Milestone:  Never Assigned
 Component:  embedded controller  |     Version:                
Resolution:                       |    Keywords:                
  Verified:  0                    |    Blocking:                
 Blockedby:                       |  
----------------------------------+-----------------------------------------

Comment(by rsmith):

 Replying to [comment:4 frankprindle]:

 > Here you go, Dr. Smith... I anxiously await your diagnosis. It would be
 really great if I can get this puppy up before next Thursday!
 >

 First: Upgrade your firmware.  You are running Q2D05 which is a brick
 waiting to happen.  If your RTC gets an invalid date in it your machine
 will quit booting and the only way to recover it is to re-flash the SPI
 flash with special tools.  Upgrade to D07 (stable) or D11 (proposed
 Update.1 candidate).  Details on the wiki.

 > By the way, your procedure resulted in the laptop running on battery for
 only around 8 minutes before shutting down. Nearly as long as the first
 time I removed power, but clearly the procedure itself didn't fix the
 problem.

 It wasn't supposed to fix anything. Its just so I can see whats going on.
 All that did was clear the % capacity so that the EC thinks the battery is
 at to 0% and it to try to charge.  Otherwise the EC has it marked as full
 and won't try and charge.

 diagnosis:

 1202009368,,,,,,,
 1202009378,75,6469660,1227213,2016,864,Charging,Normal
 1202009388,75,6570920,1392057,2044,874,Charging,Normal

 On insertion of the battery this looks ok

 ...
 1202009682,78,7127850,1262109,2418,1131,Charging,Normal
 1202009692,78,7526180,260,2425,1136,Charging,Normal

 About 5 minutes later the voltage on the battery crosses the CC to CV
 threshold and your charging current drops from 1.2A to essentially zero
 (260uA which is measurement noise). This is the abnormality.  It should
 have taken much longer to get to 7.1V

 Right now this fits the mold of many of the other charging problem
 reports.  I don't yet know whats up but I suspect that the battery has a
 problem.  I don't yet have all the battery RMAs so I can't make any grand
 claims yet.

 batman.fth has the ability to manually charge your battery.  Lets force
 some charge down the batteries throat and see what happens.

 load up batman.fth and do:

 1w-init bat-charge <enter>

 Now the EC is in reset and you have forced the charge circuit on. The only
 way its going to be turned off is to remove all power from the XO or hit a
 key on an attached USB keybaord.  So please do not leave the unit
 unattended.  And please don't leave it on for an extended period. If you
 grossly overcharge the battery it will damage the battery.  If your
 battery is funky then I need it in that state so I can work with the mfg
 to figure out whats up.

 Values printed to the screen are the ACR, Voltage, Current, and delta ACR.
 There is no logging yet.  Thats my next feature.

 Watch the voltage and current readings and see what they do.  Watch the
 current to see if it follows what happend in the olpc-logbat data where it
 suddenly dropped.  Also watch the delta ACR.  The batteries are about
 3000mAh batteries so since you are very low you should be able to put
 2500mAh in them without any problem.  Note that delta ACR is reset every
 time you run so keep track of the total ACR delta if you stop and re-start
 the charge.

 I'm currently not tracking battery temperature either (also next on the
 list) so periodically hold the XO with our palm on the battery and see if
 it gets really hot.  While charging it will get hot but make sure its not
 really hot.

 So start small and see if the battery continues to take charge after your
 5 minute mark. If so then continue on up to .5 hour.

 If the battery actually takes a charge then put it back in the XO run
 olpc-logbat. Note that using batman.fth to charge will confuse the ECs
 understanding of the charge state but it will eventually sync back up.
 The % capacity number reported may be totally wrong.

 Then see what the EC does when you try to charge it normally (with olpc-
 logbat going).

 If it works then repeat the process with about 1h charge time, noting how
 much delta ACR was added. Then run it on the XO with olpc-logbat going and
 then try to recharge normally (with olpc-logbat going)

 Then lets look at all your data again.

 Sorry the above is somewhat vague but this is new territory.

-- 
Ticket URL: <http://dev.laptop.org/ticket/6092#comment:5>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system