On start-up the sensor sends a Get_Salt-message to the base station. The
base station responds and sends the salt, which the sensor uses from
there on.

The nrf24l01 used on the boards is probably a counterfeit chip. It
cannot acknowledge received packets (probably a bit flip of the NO_ACK
bit described here:
https://ncrmnt.org/2015/03/13/how-do-i-cost-optimize-nrf24l01/) if the
dynamic packet length is disabled.

Without it the bits are sometimes read incorrectly.

The address was set for the wrong address and thus the module could not
receive acks from the base station

The packet length is now fixed to 32 bytes.

The algorith used is for the CRC-16-CCITT, also known as CRC-KERMIT.

The ID is used as the last byte of the RX address. It is also sent as a
part of the data packet to allow assigning the sending station to a
packet.

Several issues were found after receiving the PCBs that should be
rectified for the next version.

This is done to avoid transmission erros as described in
https://forum.mysensors.org/topic/10452/nrf24l01-communication-failure-root-cause-and-solution

The pin definitions are updated and the uart debug interface is removed
as it is not present on the actual board.

Currently the data values are written to the debug output.

The firmware is developed using an aduino micro pro with an atmega32u4
as the hardware for the weather station is not ready yet.