If I understand this correctly, this is not true CSMA, as the transmissions are inhibited by noise and not signal. It would seem like a better idea if transmission was only inhibited by a signal matching the current LoRa modem configuration. In shared unlicensed spectrum, all "users" must accept interference. There is probably no real motivation to avoid transmitting over "foreign" signals (LoRa or other modulations) that produce some level of energy in the receiver passband of the LoRa transmitter. Furthermore, the blocking signal might act as a "hidden transmitter" and only be detectable at the LoRa transmitter's location and have no affect on the SNR at the LoRa receiver's location. I'm not sure if such an alternative mode of operation is plausible. However, I would like the option to choose whether or not the listen before transmit mode is used, either in a fashion similar to the ability to set a socket into a blocking or non-blocking mode, or as an optional parameter to the socket send() call, for raw LoRa sockets. Come to think of it, perhaps I should have suggested this as a feature on github, but I've already typed it here, so I would be interested in any thoughts on this. Update: feature request on github is here.

@eron93br The SX1272 and SX1276 have no problems communicating, as long as you set the same modulation it's OK. But your request for node code is difficult, the SX1276 is just a radio transceiver it doesn't run any code, to use it you need to connect it to a host processor (can be a microcontroller, a SoC like the Raspberry Pi or even a computer via USB). All of these will have different code. For example if you host is an Arduino and since you mentioned LoRaWAN something like Arduino-LMIC would be suitable for a node. You can even use the ttn ABP example just remove the LMIC_setupChannel() lines to match your LoPy nano gateway setup (so have only ONE LMIC_setupChannel line with the same frequency you set on the LoPy)

I have to add some information: I can't use the lastest firmware because the modules are spread and so out of my reach :( I flashed all LoPy modules in April so that's the latest firmware I'm currently using. While searching the web for my problem I found this: https://github.com/Lora-net/LoRaMac-node/issues/104 It says that there is a bug in the SX1276 driver and so maybe in the SX1272, too. It misses a division by 4of the SNR value... And thats the factor I need. So: Does anyone know if this bug was a thing in the LoPy's too and had been fixed in the last weeks after my last update? I can't find anything about it in the last changelogs...