[Thinkpad] TP600 Battery Pack problem

[Steve Carter]

I'm really tempted to agree with JG, but can't quite yet. IBM 
discusses somewhere about 80% being equivalent to a full discharge, 
but never actually admits to taking action based on the discharge 
cycle count.  Some battery experimenters have reported success 
simply by replacing paairs of failed cells. I find this _VERY_ _VERY_ 
hopeful -- but NOT definitive. 

It seems most reasonable (to me) to act only on cell imbalance 
and let the battery last as long as the cells will allow. But 
this is more of an algorithmic issue, as opposed to one of design.  
Perhaps IBM simply took the easy way out.  After all, they 
supply (most of) the replacements.  
         
We've used individual cell-monitoring circuits (eight of them!) 
to measure the voltage of each cell in a series-string battery, 
and activate a specially designed individual cell charge by-pass 
mechanism to prevent overcharge. 

This would be nearly trivial in a 3-cell battery, as opposed to 
exotic dual redundant 8-cell batteries [fabricated from unobtanium!] 
where every bit of weight counts. 
        
Without a schematic, or access to the IBM designers, it may be 
difficult to accurately determine battery pack behavior at end-of-life. 

It might be an interesting experiment to monitor the voltage 
of each of the three pairs of cells during charge and discharge 
to see how a "bad" battery behaves. It might be possible to deduce 
one behavior or another from such careful measurements. Temporarily 
paralleling a well-chosen resistor across the cells that finish 
charging first might then allow the remaining cells to fully charge. 

This is also an opportunity for "venting with flame".  Do you have 
a Class D fire extinguisher at hand?  Have you priced one lately? 
Seems to me like battery prices have dropped a lot recently.  
Maybe it would be more cost-effective to simply buy a new 
battery assembly than to replace the internal cells. 
                                                
+++++++++++++++++++++
On 12/5/03, Jonathan Graham wrote, in part:
>Seems unlikely that they count charge cycles as there are folk who do repack
>600 series batteries for a living.  I've been waiting for some downtime to
>try repacking the ones I have here.    >J.

(much snipped)