posted 2002-10-25 to: https://www-1.ibm.com/support/fbmast.html

I hope this is an appropriate forum for feedback. If not, please let
me know what is.

The design of the battery for the ThinkPad 600 series has a serious
defect. After a few months of usage, the battery's on-board "smart"
computer seriously underreports the battery capacity, rendering the
battery nearly or totally useless even though the physical cells still
have plenty of capacity. It has taken quite some time to come
definitively to this conclusion, but now the evidence is compelling.

The support forums for the ThinkPad 600 show a much higher than
normal flow of problems with the battery. In addition, reports from
users abound on ThinkPad mailing lists and other fora. IBM's response
on the support forum is to repeat the generic battery advice. This
advice is unhelpful, to say the least, and misleading if indeed IBM
has knowledge of the problem.

In particular, the advice of three full charge/discharge cycles does
not work. Lithium Ion cells, unlike many other types, do not show a
memory effect. A full charge/discharge cycle actually takes more life
off the battery than a partial cycle. Even so, the "smart" computer
actually prevents a full cycle, and the attempt does not improve the
underreporting of capacity.

There _is_ a workaround, however, discovered by painstaking
experimentation on the part of the user community. By physically
isolating the pin on the battery which reports the capacity in digital
form, the battery can be discharged for many minutes (in my
experience, about an hour and a half) after the "smart" computer
reports 0% charge. Upon recharge (this time with the pin re-enabled,
to prevent overcharging), much of the capacity is regained (in my
experience, over an hour).

Others have found similar results by fiddling with their BIOS settings
so a reading of 0% battery does not force shutdown of the computer. I
have not verified these reports. This approach would seem to have the
advantage of not requiring physical intervention, but the disadvantage
of being somewhat sensitive to the exact BIOS version. Flashing a new
BIOS can be a somewhat risky operation.

IBM's response has not been adequate. In particular, the articles
MIGR-4PJN9H and PFAN-3QNQJN do not address the real problem. The
former document begins, "It is normal for all batteries to lose some
capacity over time." This is a true statement, but in this case the
capacity loss reported by the battery's onboard computer is
significantly greater than normal, and indeed than the actual capacity
lost by the cells. The latter document recommends the three cycle
technique, but as discussed above, this procedure is not effective.

At a minimum, IBM should admit that the battery design contains this
defect. It's also possible I'm wrong, despite the thorough research
put into this issue by myself and others. In that case, IBM should
provide a detailed explanation of the battery issues, consistent with
the observed data.

Given that there is a workaround to improve the capacity reporting
accuracy and thus the useful battery life, IBM should also post a
technical support article outlining how to do it. Of course, I expect
IBM engineers to verify and perhaps refine the technique, drawing upon
their access to the internals of the battery design.

Finally, I find this marketing statement galling: "Thanks to IBM's
advanced power management systems, the ThinkPad's battery life has
earned a powerhouse of a reputation." Perhaps this is true in the
sense that Firestone tires on the Ford Pinto have earned a reputation
with respect to safety, but I believe it is misleading and should be
withdrawn.

For more technical details, reports from users, and links to other
resources, please see the Web page I have created dedicated to this
issue: http://www.levien.com/tp600-battery.html

I eagerly await your response.