I understand the battery problems. (finally)

[Richard A. Smith]

Here's the 411 on the battery problems.

The gas guage IC in the batteries has EEPROM in it where the EC code 
stores lots of parameters on the type and state of the battery.  Under 
certain low voltage conditions when writing the data the timing on the 
one-wire bus is marginal and the EC ends up writing lots of 0xFF's to 
the EEPROM.

Recent version of the EC code have a couple of sanity checks when the EC 
reads this data.  The 0xFF's trip the error handling and the charging 
state machine restarts, reads out the bogus data again, gets errors, 
lather, rinse, repeat.

There is currently no method of recovery in EC code.  So once the EEPROM 
gets corrupted the laptop will will never charge the battery again.

I spoke with quanta about this issue on the phone tonight and turns out 
they have also discovered this and added some write verification checks. 
  They released PQB81 a few hours ago.  I will be releasing a new 
firmware (B81) tomorrow with this updated EC code.

Although B81 may stop future corruption from happening it will not fix 
all the current broken batteries.  I spoke with quanta about a method of 
re-initializing the EPROM data to default values so we can recover all 
the batteries that are dead.  They said they were going to look into it.

In the meantime I'm going to work on a few ideas that I have so that we 
can recover the batteries in the field.  Stay tuned.

-- 
Richard Smith  <richard at laptop.org>
One Laptop Per Child