Low Temperature Limit for Puck and MDBT42Q

Good idea, and here are the results:

Puck 2018-07-25T16:35:15.258Z { temp: 25, bat: 2.998 }
...
Puck 2018-07-25T17:25:02.873Z { temp: -19.75, bat: 2.668 }
Puck 2018-07-25T17:26:02.885Z { temp: -19.75, bat: 2.674 }
Puck 2018-07-25T17:27:02.946Z { temp: -20, bat: 2.663 }
Puck 2018-07-25T17:28:02.996Z { temp: -20, bat: 2.665 }
Puck 2018-07-25T17:29:03.024Z { temp: -20, bat: 2.664 }

First conclusions, that are actually observations:

• Puck.js works down to -20 °C
  
• a fresh CR2032 (noname, cheap purchase) also works down to this temperature
  
  

Indeed, my problem from last winter seems to be related to low cell voltage instead of low temperature!
First of all, I changed the code from sending GATT service data to sending the data as ManufacurerData during advertising. Hoping that this change makes the battery last much longer.

Another question (mainly for myself but input is welcome) is whether there are cells out with a slighly higher voltage (like 3.3 V - and well above 2.6 V at -20°C) that can be used through Puck's external connector, getting me close to 1+ year endurance without maintenance.