Just try to use USB cable to charge from PC We want to check if bme show proper behavior, that way it shouldn't matter if You're charging from wall charger (data pins crossed) or USB. Unless we meet some strange bug with only fast-rate charge (enabled by mentioned earlier cross between D+/D- pin) - but i really doubt about that.

Flandry - this way You're right, of course. But please, don't do measurements (and don't recommend others to) of battery drain by using battery-eye. Battery-eye, using standard bme guessing methods, just... Well, guess - when we talk about battery charge percentage. Only info that may be useful from applications like that is Voltage - mAh, percentage of charge etc are just estimated values.

Even worse, this like to turn into total mess when You switch heavy/low usuage many times during one cycle, then, on another - for example - use only low use or only high use. For me, Battery-eye show under-zero capacity 1-2 hours before real discharge almost 100% times. Not to mention how it goes crazy when You replace battery with new one, or - scary word - hotswap it

Only one exception from that may be dr_frost mAH graph, tied strictly with voltage - but even voltage can fluctuate highly on power-cycle (yea, after turning off/on you can have higher voltage for even quite long time), so measurement by that way can show You that powercycle charge Your battery On the other hand, programs like battery-eye use some build-in artificial handling of powercycles - when you do it, it mandatory shows huge capacity drop, so, as i mentioned in other thread, You can do 50 power cycles, then use Your N900 for another day with -70 capacity all the time