Below is a description of my 400Ahr LiFePO4 bank using 4 * 400Ahr CALB cells for the house bank of our 46ft cruising yacht, plus a link to an interesting video on LiFePO4 charge/discharge characteristics.
McG you may find this link interesting of a video by a Carnegie Mellon Robotics Institute PhD talking to a group of electric
car enthusiasts about LiFePO4 charging and conditioning.
Robotics Institute: Videos
He recommends an initial charge/discharge cycle with your cells all in parallel, and stopping at a well defined curve point (e.g. very near full charge or full discharge), to initially ensure all your cells are at a very similar SOC. This is a step further than the good advice offered to you by Electric1.
Also, some of the earlier posts recommend matching your cells for capacity and internal impedance very closely at the factory, and I also think that is critical.
I purchased 4 * 400AHr cells off CALB direct earlier this year, and asked them to match as closely as possible. When they arrived, with the equipment
I have I couldn't measure a difference in capacity or impedance, and parallel charge/discharge showed no imbalance (by monitoring individual cell current).
I was also very impressed with CALB's customer service. The sales person still contacts me every few months to see how the cells are, and answers tech questions very quickly.
I run the cells as a 12V (nominal), 400 Ahr bank. They replaced a 600Ahr Trojan SLA pack, and provide more useful energy per discharge. Another side benefit is that I don't have to worry about self-discharge ruining the cells when I leave the boat for an extended period.
I'm currently charging using a 120A alternator
through a Sterling Power regulator
, and a Victron 50A galvanically isolated mains charger. Both can have their charge parameters changed to fit the ideal charge curve for whatever brand of LiFePO you're using (I note that different brands have some quite significantly different charging parameters).
I use a Vetus Battery
Watch unit which controls low voltage cutoff at the pack level to the consumers. Combined with the high voltage cutoff provided by the charge sources, this produces the sensible separation between charge and discharge as suggested by OceanPlanet (i.e. after a low voltage cutoff, the cells can still be charged, whilst a high voltage cutoff doesn't remove the ability for the consumers to draw power from the battery).
I was originally going to add a cell-level BMS, but agreed with GoBoatingNow's comments back in Mar/Apr 2010 that cell level probably isn't necessary for a relatively low current situation (like a typical cruising house bank), especially when there are only a few closely matched cells (4 in my case). My usage thus far supports that view, so we'll see how it goes in future.
I figured that as worst case I might have to do a parallel charge/discharge cycle manually on an annual basis to realign the cells, but I can't detect any drift thus far, so that may not even be necessary.
Anyway, I hope that this info helps some people. Also, thanks to those of you (Gael & OceanPlanet spring to mind) for starting to put real user data on this post - I find it helpful, and gives me areas to think about. Actual user data and experience trumps manufacturers' propaganda and people's general conjecture every time.