I removed all unnecessary power consumers from the board such as LEDs and light resistor but there would have been more potential to improve longevity of the batteries. By setting a fixed IP to the module I could have saved a little bit of online time.
While I could measure the energy consumption during the deep sleep phase quite well it was rather hard to guess the consumption during the awake phase. My pessimistic estimation was that the batteries would last 30 days. And now the setup just passed the magic 1440 hours mark!
That means that with a little bit austerity engineering (sorry, not politically motivated) one could have this running for a long time. Depending on your application you could wake up the device only into a non-wifi mode, check the periphery and go back to sleep if nothing requires you to connect to the internet. Since Wifi is the real burner here that should save you a lot of your battery reserves.
I take bets for how much longer the setup will run.
Current voltage of the batteries
30 days ago when I posted the last update I was asked to measure the voltage of the batteries. I chickened out until now because I was afraid to influence the experiment. But curiosity won over fear once again and I measured the voltage of one cell: 1.323V after 60 days, compared to 1.623V of a battery of the same batch. This means that there are only about 4V fed into the voltage regulator and the difference to the 3.3V of the ESP8266 is only 0.7V….
|Voltage of one of the three AA cells after 60 days.
An unused battery of the same batch has 1.623V