scianiac said:
Interesting, was unaware of the voltage spike issue from disconnection an alternator but makes sense, also probably depends on the exact alternator in question.
Anyway just thinking out loud some ways to fix this problems with some simple off the shelf parts, because obviously you can fix anything with a custom programmed BMS and software but then you have the reliability problem that may cause. So a supercap bank may be a decent option to both protect the alternator and provide additional starting current (funny enough I messed around with some for this exact purpose, it seems a supercap bank can extend the useful life of an old lead acid battery a bit and they'll pretty much last forever in this use case). Then either get a large enough BMS for starting (or one that can be programmed with a long enough peak time), use a smaller BMS and to control a contactor/mosfet (disabling it's current limiting functions) or connect the starter not through the BMS. Then use a diode to control charging through a suitable resistor or possibly polyfuse along with a thermal switch that limits the charging current to a suitably low current when the temp is below freezing. Throw in some TVS and such for good measure (possibly as a backup to protect the alternator if the supercap doesn't absorb enough) an maybe that would work, I'm sure I've overlooked something.
Supercapacitors / ultracapacitors are also quite sensitive to overvoltage. I'm not sure how much battery, capacitor, or a mix is required to absorb the flyback spike without too much bounce / without worrying. Should be able to catch it on an oscilloscope to verify. A lead acid battery seems to do this just fine.
For clarity there are 2 voltage spike issues to consider. One is disconnecting all alternator, it overvolts and pops the diodes. Second is flyback from inductive loads, starter is the worst I can think of.
As far as I'm aware all alternators can die if they are disconnected while energized. The diodes can sometimes last a few times.
A normal BMS can shunt maybe 1 amp on a cell group. If there's 60 amps coming in that 1 amp very likely won't be enough to prevent overcharging a cell. Not sure I worded that well last post.
Limiting charge current might not be as straight forward as it seems. Voltage regulators might freak out. Newer cars have "smart" voltage regulators that are probably even more sensitive.
Separating the starter wire ads extra challenge into a normal install. Separating the alternator wire would be a similar challenge.
Adding a dc-dc charger could work but would be pricey and adds an extra failure area.
Porsche and a bunch of others seem to have this figured out without all the extra stuff were thinking through. It would be interesting to see some of their BMS' to see what they did and what isn't actually necessary.
There are lots of failed Lifepo4 batteries available on the second hand market. These would also be interesting to understand what they missed and what went wrong.