It seems there may be some discrepancies between theoretical extremes and real-world experience.  Here's a link to one of the forums on which people are reporting on their results (highlights mine):

Re: building my own lifepo4 batt question.

by tostino » Fri Sep 17, 2010 6:05 am

By doing discharges on the bench with the cells exposed, and being able to tell that each string holds almost exactly the same voltage all the way through the discharge, I am pretty sure I can tell they are not horribly out of balance. I did IR matching of the cells, but no capacity matching when I built the pack, so there is a little room for deviation i'm sure, but it is not much I assure you, or I would have a really out of balance pack at this point. I did a few 100% dod cycles on the bench to test and see how out of balance they got... It went right down to 3.2v/cell (resting) (2.9v under load) without them going out of balance at all, and they then charged right back up to full and stayed balanced.

You seem so certain that i'll kill my cells if I didn't have a cell level LVC (with out the buffer space I provide it). My LVC is set to 46v with a 14s pack. That means that it is 3.28v/cell. The resting voltage for each cell when I stop the discharge is about 3.5v. I would have killed them by now if it were going to happen, and they would have gone out of balance now if that were going to happen too.

I found this thread to be informative and a good place to start on studying people's actual experiences building and using these batteries:  http://endless-sphere.com/forums/viewtopic.php?f=2&t=2633&start=705&hilit=thyristor

Michael

On Fri, Jan 21, 2011 at 8:35 AM, Hamid Wasti <hwasti@lm50.com> wrote:

Michael McMahon wrote:

I do not want to start an argument,  but it is frustrating for people to say they don't like ideas based on out-of-date or incomplete data.

It is even more frustrating when people proposed ideas based on
incomplete data. Your "EV guru" friends are correct, you do not HAVE to
cell balance a pack, but only as long as you are willing to live with
the limitations imposed by that choice. Do you know what those
limitations are?

When you have a number of cells of any chemistry in series in a battery
pack, they all receive exactly the same current when charging. Some
cells are a little more efficient than others so they get fully charged
before their colleagues in the string. If at that point they continue to
receive charging current, they will over-charge. Some chemistries are
able to handle this over-charging. Other chemistries like LiPo are very
intolerant of this over charging and quite literally blow up. Cell
balancing attempts to make sure that the charge on each of the cells in
the string is identical so they all get fully charged at the same time,
maximizing the charge that the pack can hold. That maximum number is the
one everyone throws around and that is the number you will expect to get
from your battery pack.

As I mentioned, you do not HAVE to cell balance. For a LiPo or LiFe
pack, you can just monitor the voltage on each individual cell and stop
charging when one of the cells gets fully charged. Over time, the
discrepancy in th charge state between the most efficient and the least
efficient cell in the string will keep increasing, with the usable
capacity of the entire pack being controlled by the charge in the least
efficient cell. Taking this to the theoretical extreme, at some point
the pack will not be able to deliver any energy because one cell will be
fully charged and another will be fully discharged. In real life, you
will declare the pack useless and stop using it before you get to that
point. If you are willing to live with this diminishing capacity, then
cell balancing is indeed not required. Just remember that your pack is
no longer going to have the same capacity as the pack that has cell
balancing and you must design the rest of your system to account for that.

Quoting the late Paul Harvey: Now you know the rest of the story.

Regards,

Hamid



--
For archives and unsub http://mail.lancaironline.net:81/lists/lml/List.html