Last weekend, I was preparing my electric tractor for a new season of mowing, when I inadvertently touched the charging leads to the wrong battery tabs. The batteries for my tractor are from a totalled Chevy Volt, and they can source a lot of current. It’s what makes them so great for running an electric lawn mower!

The power supply I was using to charge has protection against electrostatic discharge using TVS zener diodes, but that was not designed to survive if I even momentarily connect the leads backwards to a source that is happy to supply hundreds of amps… One of those diodes was blown partway off the circuit board, and the other dissipated into a cloud of “magic smoke” and a pile of stinking char.

The great thing about having an open source power supply is that the technical information is all online. And the designer also kindly responded to my request for debugging guidance, too! I replaced the dead diodes (and the physically adjacent relay I broke in the process of replacing the dead diodes, oops), and now I have a working power supply again.

The magic smoke has been re-installed, the genie is back in the bottle, and I’m back to charging my tractor.

In order to protect against doing this again, I also bought some high-current schottky diodes and made an interposer board for my charging cable. The diodes are rated to 30V, and I might encounter 48V in my battery array, so I stacked two of the diodes in series. I am using remote current sensing to the other side of the interposer, rather than just adding the voltage drop to the selected voltage, because Vf depends on temperature. I could get a slightly lower (10-15%?) Vf with a series-parallel arrangement and another heat sink, but that would also double Ir, which is what I’m trying to protect against, so it would defeat the purpose to some extent. It isn’t necessary and would add complexity to no purpose.

In theory I should use load sharing resistors to make sure that the diodes see the same Vrmm if I supply inverted voltage again, but the diodes are thermally coupled (attached to the same heat sink) and I’m hoping this doesn’t matter in practice. Ir is high enough already without making it several times worse with load-sharing resistors.