Matthijs Kooijman

I think the most reliable way would be to use watchdog, which will cause a real reset (but that requires a bootloader that clears `MCUSR` as well). In any case,...

Add `__attribute__((__weak__))` in its declaration IIRC (but Google should help with that).

What is the mess exactly? I think the IDE might incorrectly sort and default to the -0 version, but -3 should really be the latest version and it should work....

Looks like you configured the wrong baudrate?

Ah, completely forgot about this one. The printf implementation is AVR-specific right now. Could you try the https://github.com/matthijskooijman/arduino-lmic/tree/non-avr-printf branch (or better, cherry-pick the top commit of that branch onto master...

I've seen the power usage from the RFM95 well below 10uA (perhaps only 0.2uA, haven't measured that precise), so I suspect there is something in your setup that leaks current....

I haven't looked closely, but the bottom line is that TxData only queues the data for sending (and calling it twice overrides the queued data). You'll have to wait for...

IIRC, the frequency plan defaults to just the mandatory frequencies from the spec (3 for joining, 6 for data): https://github.com/matthijskooijman/arduino-lmic/blob/ba1265d5d2f775177cdc7c82186724e4f0bdc3a8/src/lmic/lorabase.h#L73-L82 In ABP, you are expected to configure the frequency plan...

In one project I updated the `millis()` counter rather than the LMIC ticks, which is more general (but also needs to access internal Arduino variables). See https://github.com/meetjestad/mjs_firmware/blob/master/mjs_firmware.ino#L232-L242

> @matthijskooijman Thanks, is the hack on millis() working on all platforms? No, just AVR (others *might* work, but I expect their implementation to be different).