Steve,
Lithium Iron Phosphate, to give them their correct name. LiFePO4 in short-hand.
There are many different "Lithium" technologies. All have different voltages.
Current best practice is to charge to 3.6 volts, no more, to increase cycle life.
Once the battery is full, the terminal voltage rises rapidly.
Once you remove the surface charge, they drop to 3.2 volts, and remain there for most of the discharge cycle.
Low voltage cutoff (LVC) is usually set at 2.7 volts.
Charging is normally managed by a shunt across each cell, to limit the voltage. Basically a precision zener. Coupled by control of the charging current, to drop the current to a reasonable level which can be handled by the shunts. Mostly it's about heat dissipation.
In my setups, the shunt fires an opto-coupler. Then there's some diode logic which generates 2 control signals; Any shunt low and All shunts low.
In the negative lead of the battery, I have 2 paralelled current control paths. One for bulk charge, one for finishing charge. Any shunt low turns off the bulk charge. In the vehicle application, All shunts low goes through an RC time delay of about 10 minutes, then turns off the finishing charge. In the solar application, All shunts low is dis-connected, and I just leave it on trickle charge forever basically. My scooter charger uses a Meanwell power supply with constant current limiting. The constant current has been modifed downward. By default, the current limit is set by the factory to more than 100% of the rated power output. The output voltage adjustment range has been tweaked so it puts out 60 volts. The mains input is switched by a relay. Charging is initiated by a momentary switch in parallel with the contacts. The relay is held in by a current sense circuit. So when charging is complete, All shunts low has turned off all the current, the relay drops out and the charger switches off. Very clever I thought.
In the solar setup, the bulk charge control is a 100 amp solid state relay. In the vehicle applications, it's just a mosfet switch. And the finishing charge is controlled by a constant current source with a negative temperature co-efficient. Same circuit I use for powering Led's.
LVC is implemented with a TC54 across each cell, again opto-coupled onto a bus. In the scooter, this signal switches a power transistor, which turns on the brake lights. The motor controller monitors the brake lights, and cuts power when the brakes are applied. Being a slacker, I haven't actually connected this.
In the solar application, the LVC signal fires an RS flip-flop. The output is connected to another solid state relay, which dis-connects the load. A push-button switch manually resets the flip-flop. Flip-flop state is indicated by a bi-colour Led.
In all applications, the logic functions are implemented with a 4093 quad nand schmitt, and some diodes. Kinda my trade-mark really. Too many diodes and a 4093.
If you're not seeing where the 3.2 volts comes into it, and the cells run at about 3.7 volts whilst dis-charging, then I'd say you've got Lithium Polymer (LiPo).
The original write-up on the scooter BMS is here.
http://endless-sphere.com/forums/viewtopic.php?f=14&t=6419.
Since then, the battery arrangement has been changed, the BMS was re-built in a new case, the bulk charge CC source was re-calibrated to be greater than the charge current (fail-safe), and the charger replaced by the single Meanwell 48 volt 350 watt. But the schematic is mostly correct.
I probably should gather up all the schematics, and take some photos, of my solar setup. One of these days.
Amanda
Zap has replied whilst I was typing this.
Yes, cycle life. Typically, LiFePO4 is quoted as;
2000 cycles at 80% DOD.
3000 cycles at 70% DOD.
How many cycles at 25% DOD is what I want to know. This is where my scooter is mostly operated at.
And pretty much no Peukert effect.