We are testing 6 new prototypes based on Micorchip RN2483A modem. They are all placed at fixed position realyl close to a lorixone gateway conencted to TTN. The distance between the gateway and the sensors is less than 10meters and all 6 sensors are at less than one meter fro meach others. Nothing have been moved since few weeks.
On of the the 6 node have seen his RSSI gradully reduced during the last 2-3 days. It was between -35 and -50 like the 5 others for the last 2 weeks. The is started to gradually reduce down to -80 and finally stepped to -110 to -120.
I immediatelly suspected an antenna issue on this node but after checking it everything was normal. I resetted the sensor and RSSI was back to -40 !
The 5 other sensor have stayed stable between -35 and -50 during the same time, excluding a gateway issue.
Have you checked if Adaptive Datarate is enabled? This is a feature where the end-node reduces its transmit signal strength to as low as possible, while still ensuring transmissions make it back to the gateway.
This looks like ADR is reducing the TX-power. Sometimes I have the feeling, that ADR is active although it was switched off. I have a node with ADR=off and SF7. But suddenly this node was transmitting with SF12. I am still searching for the reason.
An other idea: You are neither in free space nor in a fully anechoic chamber. The direct and a reflected signal of a node can sum up at the antenna of the gateway in the way that the signal is nearly extinguished. Moving the node only a little bit can change the RSSI-value significantly.
And you are right: The SF does not influence the RSSI.
The Lora Alliance explicitly forbids network operators (like TTN) to allow devices on their network with SF fixed to SF12. So ADR is mandatory for any device set to SF12.
You can’t actually completely turn off ADR, because certain other operations like configuring the node’s channel map involve commands that also have a field telling the node what datarate and transmit power to use.
It would appear though that there’s something else going on in this case.
For starters, the reported initial RSSI’s indicate that the nodes are really too close to the gateway to begin with, and possibly overloading it. It’s possible in this situation that some of the weak signals being seen are also false ones caused by the distortion of that overload making it appear that there was a weak transmission on a different channel than the actual transmission was on. It would probably be best to move the nodes a bit further away, and then really look at detailed logs over time - see what spreading factor is used for example. I’d think it should be possible for firmware in the custom part of the node to query and serial log the settings that the canned LoRaWAN stack in the module is using. Various setups for measuring node power consumption could also see if the transmit power is changing.