IssacZachary wrote:You guys are exaggerating. Diode losses wouldn't be frightful
No? I was imagining delivering into a single cell at about 3.8V though a full-wave bridge with two forward drops. I expect Schottky diodes would be unacceptbly leaky (it's considered a design challenge for the BMS to just *measure* the cell voltages without draining the pack unduly), so that would mean delivering only 3.8/(3.8 + 1.6), or about 70% of the input transformer power. I call that frightful. I may have picked a bad topology to imagine, though; if the transformer is operated in a flyback configuration, only one diode would be needed, so that would cut the 30% loss in half. But that's still pretty bad.
IssacZachary wrote:and the BMS in our Leafs is a whole lot more complicated than just a mosfet and a resister.
Yes, but I'd expect it to be heavily integrated. From occasional curious glances at trade-magazine articles/adverts for such ICs, I gather that they are capable of handling a small number (like eight or so) of directly-connected cells, with appropriate level shifters for measuring the cell voltages. I imagine they can also generate the gate drives for directly-connected balance-resistor
FETs as well, so from the component count standpoint, the LEAF's per-cell charge-balancing arrangements DO consist simply of a transistor and resistor (or just the mosfet as lorenfb suggested), plus 1/8th of an IC and its attendant powersupply and communication infrastructure.