I see some clear patterns on battery drain/charge affecting several sensor readings. Am I the only one getting this?
I’m guessing this maybe happens because the battery gets warm when charging… I’ve now separated the battery from the board as much as I can, I’ll post new results.
See here data: http://www.flickr.com/photos/armadillu/8713648364/in/photostream
hello Oriol!
you are right! we found this hardware bug some weeks ago!
The problem is not the battery. The problem is the charger of the battery.
http://ww1.microchip.com/downloads/en/DeviceDoc/21984a.pdf
The MCP73831 continuously monitors the voltage at the VBAT pin in the Charge Complete mode. If the voltage drops below the recharge threshold, another charge cycle begins and current is once again supplied to the battery or load. This current heat the chip and because the sensor is near, the temperature change a little bit…
Remove this completely is impossible with this design (one more thing to resolve), but greatly improved if the sensor is at 90° and not touching the PCB.
Also could be corrected by software, because we know the state of the battery.
Maybe you can try to do this and give us feedback? 
thanks for your appreciation!
Best!
Hey Alex, I’m giving the “software fix” a shot now. Doing some early tests, bit it takes so long to see the results. It will be pretty hard to do this accurately, but I think an easy hack can offer decent results. Maybe the sensor should send both the “real” and “corrected” data… I’ll let you know. I’ll be posting here https://github.com/armadillu/Smart-Citizen-Kit
great! tell me!
You can see progress here!
Esto está ya bastante estable! ahora debería probarlo con otros SCK para ver la variación de placa a placa!