Has anybody seen this ZigBee effect? In our customer project, we have detected a state in which two end devices get the same short address from the coordinator. We do have a Snifferlog that shows the effect. In text form:
1. Network with 1 Coordinator and 2 Enddevices.
2. Enddevices are turned on and send "Beacon Requests"
3. Coordinator is turned on and forms a network
4. End Devices try to join nearly at the same time
5. As soon as the first End Device has joined (Addr. 0x796F) our application on the End Device starts sendig a packet (AF_DataRequest incl. Ack Request)
6. The paket is not acked by the coordinator! It seems to be busy with the join procedure of the second device.
7. In the mean time, the second device has joined the PAN successfully (0x7970).
8. The second device starts sending application packets. They are acked correctly by the coord.
9. The first device (0x796F) resends the unacked packet (Stack functionality). No ack is received either.
10. We assume a failure, leave the network and restart (new join).
11. It works but the coordinator gives us the address 0x7970.
We end up in a network in which two devices with different IEEE addresses have the same short address.
Its also possible that we simply missunderstood something concerning the usage of the stack. I would be very grateful for an answer what we could do to circumvent this behaviour.
Many thanks for any hint on this topic!