This morning I wrote:
So the battery didn’t survive more than 300 hours because the current demand is in reality 2 milliamperes. I suspect that the voltage monitor circuit is causing this, so maybe it is time to insert a jumper to isolate that part.
But the suspicion was wrong. I debugged almost all evening to find out where the problem might be, it can be fun, hunting for the last milliamp like they did in Apollo 13 returning from the moon. The eventual solution was: to put all the sensors on a digital line like I do for the voltage divider, and to remove that back piece of ESD foam on the FTDI pins, because the foam forms a 5k resistance between vin and ground, and this is what you exactly don’t want. So, let’s see where we get from this point on.
20-9-2014 10:30 Voltage seems to be holding at 5500 mV at 10 hours, so this was probably the solution.