I came downstairs and once I got into the kitchen I smelled melting plastic. That's never a good sign either.
The charger had three flashing lights, which means something went wrong. I immediately unplugged it and broke apart the wiring harness to prevent any failing cell from taking the rest with it. 18 of the 20 connectors unplugged; the other two did more of an ungooey: the plastic was liquified. Yikes!!!
With the batteries disconnected from each other, I took apart the gooey stack and found this:
Quite a bit of melting. Definitely not good. Two of the batteries let their magic smoke out:
Two of the plastic sleeves are done:
As are the two gooey connectors:
So what the hell happened? My post-failure analysis revealed two major design flaws:
First off, let's look at the spacing between the cells:
There's 1/8" between each pack of cells. This distance is determined by the thickness of washer used between the packs. When everything's together, the two shown surfaces mate (above), but when separated, they look like this:
(FYI: I added a second set of washers to show the rubber washer coverage; i.e. there's only two washers, not four).
So what happened? Well, if we push the spring terminals together (which happens when the batteries are inserted), the springs push up against the plastic retainer, but can pop out when shaken. In the picture below, I'm pushing the spring that the battery would normally push up against. Notice that the wire is sticking out quite a bit.
If we do the same thing to the other side of the connector we get similar results:
It turns out these wires cross over each other and the rubber washer I assumed would nest these wires in didn't work properly. This explains why when I removed the batteries, the coil in the picture above was resting on the terminal of its adjacent battery pack:
Notice that the coil is cut right where the two wires would cross if shorted out.
So now we have the failure mechanism, but we still don't have the failure mode... What caused the leads to short? I've ridden the bike more than 60 miles at this point and none of the battery packs have leads hanging out.
So let's look at the battery voltages on the two backs adjacent to the failure:
The four batteries with voltages written on them are the where the failure occurred. Notice that cell 2:5.16 (53-56.6 volts) is at 0 volts. Cells 2:4.16, 2:3.16, 2:2.16, and 2:1.16 should all be at the same potential as 2:5.16, but instead they're all at 1.6 volts. One might initially think we pulled all of our current through the cell at 0 volts (2:5.16), but this couldn't have happened because cell 2:5.15 is overcharged to 4.1 volts. Post-analysis, I ran this cell down to 3.6 volts to prevent cell damage.
So I believe that the cell at 0 volts got that way due to a chemical failure. Remember that when I first got these batteries two of the cells out of 160 were at 0 volts. I believed at the time that they would charge just fine and work properly, but the above shows that that wasn't the case.
So the failing cell was at 0 volts throughout my ride. Once I plugged in the charger (placing all cells in parallel), the 0 volt battery pulled power from the other 9 cells, which warmed up the small voltage sensing wires due to the high current draw. Once the wires had sufficiently warmed the plastic, they suddenly made contact, thus shorting out the 25.3 volt and 28.8 volt lines. At this point, we probably drove a few hundred amps through the small wire, thus severing the coil and flash melting the plastic.
So how do we prevent this?
1. I will definitely be leaving the seven charging harness wires plugged into the batteries at all times so that no one cell can decharge more than it's other 9 partners.
2. I will be removing the small rubber washers and in their place installing a 1.5"*3.0"*0.125" rubber mat to prevent wires from shorting out.
Overall, this isn't a huge offset. I've already ordered 3 new sets of batteries. Better to lose two sets versus twenty. Damn design flaws :).
