The problems appears to be linked to JeeLib and the RFM radio. It doesn’t appear to be related to undervoltage, temperature or any other obvious cause.
During my recent tests, two failure modes occurred.
- The one reported where transmissions ceased, and a press of the reset button was needed.
- One not reported, where the message count and energy totals were reset to zero but transmissions continued.
The second happened very frequently, though neither had been seen in the original testing in the summer of 2018.
I established that the first fault happened whilst waking the RFM, transmitting the data and sleeping the RFM again.
I replaced the “full” JeeLib with a cut-down ‘transmit-only’ version, and that appears to have provided a significant reduction, if not a total removal, of the problem. The full details are here: EmonTx stops sending data - no led activity until reboot - #18 by Robert.Wall
Note the warning about the higher risk of r.f. collisions.