I’m facing trouble with my zigbee network.
To be precise, some routers rather connect with the far away coordinator, than connecting, or at least tranceiving messages through the router-network.
Rollo_Büro (bottom left) connects to the coordinator with a lqi of 28(marked red).
There seems to be a route marked green that has a min lqi of 134. But in the Zigbee2Mqtt-FE 28 is displayed for Rollo_Büro.
And controlling also feels like it has ony 28lqi.
Things that i have tried:
Replacing the coordinator (From ZZH to the newer Sonoff P-Version)
Reconnectiong Rollo_Büro
Moving all Wifi-Networks to channel 11.
My devices are in Zigbee-Channel 11. So it should be ok, by looking at this graph from another post regarding this topic:
Is there something I missed? In DECONZ it was possible to force a specific route for an endpoint, but there isn’t in Zigbee2Mqtt, right?
some routers rather connect with the far away coordinator, than connecting, or at least tranceiving messages through the router-network
I just check my map, and majority of routers even ones that are fair away from a cordinator are connected directly to coordinator except one. The one that is not connected directly to a coordinator is aquara smart plug, others are tuya zigbee smart plugs. So it might have something to do with router it self.
I have looked into the routing some time ago, and I do not believe you can determine if information is routed. It might be directly, or through a longer chain. To my knowledge this can not bee seen from the map. The map shows all routes available, as it also siblings. Mine look like yours, all mains poweres zigbee routers have a lot of connections, and most of them also directly to the coordinator.
Well, it behaves like a device, that has a very bad connection. Sometimes it works with no problem sometimes it has a delay of up to 30seconds and sometimes it seems like the messages get lost.
Reading your messages it seems like, that the connection to the router isn’t an issue, and the route a package goes isn’t determined by the map.
However, in that case I’m still wondering why the LQI is that bad. If the network is truely selfimproving concerning the generation of the best routes, shouldn’t the LQI be at least 13X instead of 10-30?
I had a lot of issues with my zigbee network and it all started when I made the switch to Z2M. I spent months trying to figure it out. In my quest to find a solution I did lots of reading about wifi and zigbee channel over lap. This was the key for fixing my issues.
I’m sure there will be some disagreement with this but my personal opinion is zigbee channel 11 is the worst channel to use hence why when I made the switch to Z2M I had so many problems. Changing my Wifi router/AP to channel 11 did have improvement but I would still have issues with devices either responding slowly or falling off completely. I believe is because of the wifi networks surrounding me are mostly using channel 1 or 6.
I tried every zigbee channel and found that channel 25 to be the best. I haven’t any issues since moving to channel 25. A friend of mine also had issues his with zigbee and was ready to abandon zigbee all together but I was able to convinced him to move over to channel 25 and that solved his issues as well.
Channel 25 seems to have higher lqi numbers but even if a device has a low lqi it still responds instantly. Looks like you have lots a zigbee devices but you may want to consider moving off of channel 11. I’ve done it several times so I know its a PITA but it might resolve your issues
I had LQI starting at 0 increasing over time to 255, the fall back to 0 when I was using a Raspbee II. Now I have a Sonoff ZBDongle Plus and LQI is between 0 and 30.
I‘d say LQI has no meaning. If devices do not respond or even fall off the network, put some more devices powered by mains close to them. They implicitly act as routers.
And there are some battery powered devices who tend to stick to the router they initially connected to. If I remember correctly, that’s the case with some Aqara devices.
