[ltp] Re: [Thinkpad] "Resetting" 600 Li-Ion Battery Cells

Jim Harvey linux-thinkpad@linux-thinkpad.org
Fri, 03 Oct 2003 09:19:57 -0500


James McKenzie wrote:
> Jim:
> 
> If I use a V-O-M to read the resistance of the battery I should read a 
> complete open and no voltage.  I think I would read the same thing if 
> overheat detector were to open the cell.  If this is not the case, 
> please advise.
> 
> James McKenzie
> 
> Jim Harvey wrote:
>> James McKenzie wrote:
>>
>>> Frank and the rest of the list:
>>>
>>> I got the URL from an older message and attempted to reset the 'top' 
>>> two batteries with no noticable change in the performance of the 
>>> batteries.  I would like to know the 'normal' responses from a 
>>> totally dead Li-Ion cell.  I know the reponses from a partially dead 
>>> battery.
>>> It looks like I may need a new battery.
>>> James McKenzie
>>
>> I've taken a couple of IBM packs apart - Dead Lions go open.  If they 
>> didn't,
>> IBM wouldn't be able to put three of them in parallel in the 770 packs.
> 

True.  I don't know about the internals of the cell itself, I havn't the nerve
to saw one apart.  I suspect there's some nasty alkali inside.  The cells in both
packs are Panasonic Lithium CGR18650H. About $10 each from Digakey but they are
welded to the internal wiring membrane so mere mortals can't rebuild them.
There is an external temp sensor (or possibly it's a circuit breaker) nestled
between two of the cells, also a blue thermistor.

The 770 battery has three series connected groups of three paralled cells.
This explains why the charge meters go to 30 percent then directly to zero -
one of the cells in one of the groups has gone open and that group has only
two thirds the capacity.  I did notice one of mine would go to 65%, then
to zero for a few cycles, further confirming this theory.  Each of the three
cell groups has a lead to the monitoring circuitry so the internal electronics
can monitor the cell groups individually.
-- 
     Jim Harvey, Naperville, Ill.   Linux person - WB8NBS/9
            He who dies with the most software wins.