[olpc-help] Low Battery Shutdown although full battery (problem)

[Richard A. Smith]

Emanuel VonAnkh wrote:

> 
> a) Direct hardware intervention (ie. open up the battery)

Perhaps.  The plastic is welded so when you crack it open you will have
to glue it back together.

> b) Testing various charge/discharge scenarios...

Possible if its some sort of recoverable cell defect vs a manufacture
defect.  You would need to study up on LiFePO4 battery chemistry and see
what the possible cell failure modes are.

> a) The idea was put forward on some posts that the charge voltage is too 
> high
> (what about discharge?), making the circuitry think the battery is 
> full before it is. One should then be able to insert a resistive diode 
> bridge limiting the charge current. (Ie a parallel connection of 2 
> diodes with a resistor in series.)

Its not that simple.  Its not "just" the charging voltage.  The problem
is the absolute voltage rise.  Charging starts off normally but the
voltage on the battery rises beyond "normal" levels (a function of the
battery not the charger)  Eventually this voltage rises above the
internal overvoltage cutoff.

I have duplicated the same behavior using a power supply with different
( much lower) charging currents.  All that changes is the time until the
voltage crosses the shutoff level, the net result is still the same you 
don't get a full charge.

> But I would need the internal
> schematics for the battery for this. (Where can i find this?)

They are only available from the manufacturer under NDA.

> I also 
> read there is an EEPROM inside the battery. How does that work? And what 
> is it doing?

The eeprom is part of the gas guage chip which is a DS2756 (google and
you will find the datasheet).  You can talk to the device via the 1-wire 
protocol if you have a 1-wire bridge. The additional data inside the 
eeprom is mfg specific and only available under NDA.

However 95% of the data in the eeprom is only for the NiMH chemistry.
Nothing in the eeprom influences the charging of the LiFePO4 batteries.
  the LiFePO4 algorithm is very simple whereas the NiMH algorithm is
very complex.

> b) One old post mentioned the formation of crystals within (LiMH and 
> NiCd) batteries. If this is also the case for the OLPC LiFePO4 
> batteries, these "crystals" would preferably be "melted". Melting is 
> usually done with heat and thus we should apply heat!

Nothing I've read suggests that LiFeP04 has the same issues but I've
also not read anything that says they don't.  Heating the pack would be
tricky if you get it too hot the cells will vent.  The max
operational charging temp of the cell is 50C.

> :) A second and 
> related fact is that charge/discharge characteristics strongly depend on 
> temperature. I thus suggest the following tests:

One of the key reasons OLPC is using LiFePO4 is because its low variance 
wrt temperature.  Even at -20C the cell voltage only drops about 1V per 
cell. At high temps there is very little difference.

> 1) complete discharge at >30 C and recharge >30 C
> 2) complete discharge at < 1 C and recharge >30 C
> 3) complete discharge at >30 C and recharge < 1 C
> 4) complete discharge at < 1 C and recharge < 1 C
> 
> My intuition tells me perhaps (1) followed by (3) is the most likely to 
> succed in getting your battery operating,

In one case I had a user play with putting a problem pack into
the freezer and then letting the XO charge it while cold.  The results
were similar to the room temperature case.

It was not an exhaustive test and didn't involve any higher temps.

> if it is just a matter of 
> fixing the internal battery chemistry. However from what Richard says, 
> perhaps this is not the case.
>

I'm doubtful but I'm not a battery chemistry expert.  I suppose you
can't make it much worse.

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