Vincent, I think you should get in touch with the RFLink guys
The decoding of the RF data is usually done by reverse engineering (or “guessing”). So the received data is analyzed and looked for the binary representation of the expected values (“node-id”, “battery value”, “measurement value” etc.).
If you notice the id is changing for negative values, then very likely the id bits are wrongly guessed (the sign bit to indicate negative values is wrongly considered to belong to the device id).
Therefore you see the id changing…
Though you already found a way to circumvent this flaw, it’s better to get this fixed by the RFLink firmware itself
The RFLink mainainers have been quite responsive in the past. If you point them to your findings here and provide useful data, I’d expect them to fix it quick