OTAA fails with increased sensor distance

Hello; I’m running the full Loraserver stack using docker-compose (gateway-bridge, loraserver, lora-app-server), all the relevant Loraserver components are at version 3.

I’m using pretty standard configfiles, a Laird Gateway RG186 and three Laird RS1xx temperature sensors. The newest firmware is installed on all the hardware, channel definitions are the standard ones for my region (EU).

When the devices are kept within some 10/20 meters from the gateway, everything works fine - OTAA gets through and I get consistent packet exchange in both directions.

If a sensor is moved some 300/400 meters away, in line-of-sight with just one concrete wall in between, packets are still seen on the gateway with very good SNR. But the sensor must be missing the ACKs from the gateway, and so after some minutes it will enter a new activation cycle. I see Join Reqs on the gateway, followed by Join Accept packets with the right information and the usual 5s second delay (down to the microsecond at the gateway); but the sensor seems to never catch them, so it stays in the endless activation loop (I see hundreds of Join Req/Accept pairs) and never sends data again. Moving the sensor closer to the gateway again lets the activation go through.

But, is this really because signals from the gateway don’t arrive at the sensor? Is doesn’t seem right, given that the small, battery-powered sensor without antenna can easily send packets in the opposite direction with very good quality - and the 200/300m range in almost open air should be covered by Lora effortlessly. What else could I look at? Any hint? Please help, I’ve tried just about every possible investigation path!

Thanks in advance, keep up the good work!

For those interested, it turned out to be a hardware issue: some of the sensor units delivered were of the wrong region type (RS191 instead of RS186), and we did not notice at once. Although with the 861Mhz firmware the units worked as expected within short range, at longer distance some difference in the radio chain (filters?) made the unit misbehave. I can confirm that the “proper” RS186 units work as expected on the 200/300m range described above.