I have a simple temperature / light monitor that worked
quite well on one arduino, so i added a breadboard design and an RF transmitter and receiver so one ATmega328 was talking to another wirelessly.
I got this working reasonably well for a little while until I decided to try some batteries instead of a 5V supply.
I used 4 AA batteries (6V average) but used a 3.3V regulator to tie the voltage to a nice stable one. This worked fine for a couple of days until I notice that my Cosm feed had nothing in it one morning (this morning). So I metered the battery pack and it was around 2.5V
A fresh set of batteries got the ardunio chip up and running again, but it started giving odd readings from the TMP36.
At first didnt think anything of it (on testing frozen to -25 the warm up curve is quite slow), but after it was running for an hour or so it was still reading between -11 and -7.
All code has remained the same so I know it is not that, I have tried a further ATMEGA and put this sketch on it and am now confused.
The reading of the TMP36 from a meter is 0.206mV and the reading of the ground line is 0.000V so is a good clean line. To me it looks like the voltage supplied was below the minimum voltage for it and has caused an issue with it. Unfortunately I do not have another to play with, so unable to check, but is this common?
If so what is the best way to monitor a battery voltage to turn off the supply to the TMP36 for the future?
I have thought of using a potential dividor to set the *fresh* battery voltage to just above 4.6V, tieing ARef to a 5V (not sure how to make as would need to be very low drop) supply and then monitoring this, but not sure if monitoring a battery voltage greater than the Vcc would be a good idea. Very confused here, but this is a 2nd point.
Any ideas are welcomed.
Edit, cosm feed is as below (will last around 6 days before new data pushes it out...