at what voltage would zero percent be considered as?
3 Volts. Technically you could probably run down to about 2.7 volts or so before it became too low to run anything. Don't. Anything below 3 volts and you risk damaging the battery.
Isn't there some way to manually calibrate the battery monitor to the battery's highest voltage and the Pandoras lowest working voltage?
Batteries don't work like that. There isn't a direct correlation between voltage and battery level. If you attach a load to the battery, there will be a sudden drop in voltage; remove that load and the voltage will rise again: the bigger the load, the lower the drop. If you just based battery level on voltage, you might see, for example, a 10% drop (from 100% to 90%, or 50% to 45%, etc..) every time you started Mupen64. Conversely, if you plug it in to charge, the voltage will spike upwards: you might unplug it at 50% for some quick gaming only to have it read 45% a minute later.
The bq chip works by starting from an empty battery (3 volts) and measuring the current going into the battery as it charges until it detects that the charge has stopped. Then it adds up all the measured current and says "this must be the total capacity" and marks that as 100%. From then on it measures the discharge rate and subtracts that from full in order to arrive at the current capacity.
MWeston predid all the calculations and testing and made a golden file with which all the bq chips are flashed so they already know ahead of time how big the battery is. In the course of my experimenting, I forced it to recalibrate from 3 volts again, and that worked great for weeks. Then just this past week, I don't know what happened, but it recalibrated in such a way that it thinks my battery only has about 1000mah. From 100%, it discharges to 0% in about 2 or 3 hours, and then sits at 0% for the next 7 or 8 hours before it finally reaches 3 volts. Charging it up goes the opposite: it charges from 0 to 100% in about half an hour and then keeps charging for another 2 hours before it's actually full. I've tried re-recalibrating, but so far it hasn't worked. If all else fails, I've got the golden file, I can just reflash the chip if I have to, but I'd rather get it to behave properly.
Now, if only the nubs were as hacker friendly as everything else on the board.
edit: to clarify, a battery that is not under heavy load (positive or negative) does have a correlation between capacity and voltage, and that's how most home battery testers work, and how a lot of things work to give quick estimates on battery capacity. It's not a linear correlation, but you can graph it out. The problem comes when you start using power off the battery, or charging it up. Makes it hard to measure accurately.
