End Node downlink problem

[Saar Eliad]

Posted on June 11, 2018 at 17:41

Update:

We solved our problem-

Our gateway (improved single channel gateway based on WAZIUP project) sent with invertIQ=off for downlink and that caused the problem.

Note: The answers given here were not correct, as LoRaWAN time requiremets + delays can be easly changed.

Hi,

We are working on a LoRa project using the NUCLEO-L073RZ +

SX1272MB2DAS 

and the end-node application (from STM32CubeExpansion_LRWAN_V1.1.5) On region EU868.

We use the end node with ABP and static network_key which is configured also in the server application.

Our gateway is composed of raspberry-pi and an SX1272MB2DAS LoRa RF expansion board. This is a single channel gateway, it receives/sends packets only on frequency 868.5 with datarate 0 (bandwitdh=125, spreading_factor=12), hence, we have configured the rx_window2 with the same frequency and datarate.

The gateway receives a LoRa packet from the end node and sends it to the server.

Then the gateway sends a downlink packet which came from the server.

The end node never gets this packet.

The time between the gateway's uplink reception and downlink transmission is around 10.5 seconds.

We tried to use CLASS_A and CLASS_C with different rx delays and receive windows but the end node didn't get the downlink packet (rx timeout).

When placing another NUCLEO board which runs the PingPong application, it does receive the downlink message from the gateway successfully.

We guess that the end node doesn't receive the downlink messages because of a timing issue, but can't be sure. (as we also tried using class C…)

Can it be a different cause?

How can we overcome this?

Regards,

Saar and Guy.

[Gidi]

Posted on June 12, 2018 at 17:15

For class A, there are 2 reception windows opened by node 1s and 2s just after TX.

downlink after 10.5s after uplink reception is definitely to long period.

[Saar Eliad]

Posted on June 12, 2018 at 17:20

Hi Gidi,

We increased the windows length (symbol length variable), also increased rx1 and rx2 delays to mach this, and it still did not work.

We also tried class C with giant window, so we susspect its somehting else.

[Saar Eliad]

Posted on June 15, 2018 at 14:33

It still does not work as we want.

We did manage to hit one time in class C.

the rule is that we must hit the window right when it opens – otherwise we get rx timeout.

but what if we can't be 100% accurate about this?

the sx1272 did open a RX window, but a small one:

            it is on only open for a small number of symbols called MinRxSymbols.

Increasing this MinRxSymbols   seems to ncreases the window, but we still don’t  recognize the packet.

Note:

(Our Ping Pong recognized the downlink packets)

We still don’t find explanation why Ping Pong recognizes and end node doesn’t.

We would appreciate your help on this

[Gidi]

Posted on June 18, 2018 at 08:15

PingPong example, it is not full LoRaWAN node implementation. IT implements PHY and lower MAC only. You needs to follow time requirements of LoRaWAN.