Hi All
For the most part I am quite happy with the deCONZ hub and Zigbee in general, and yet I am really struggling to add and keep existing (mostly Aqara) closure and motion sensors connected. I have added a 7 Osram smart plugs to provide the mesh. Am I overloading the mesh?
I can reconnect most/some of them, but not all, and I lose the connection after a while. The “Door - Laundry” sensor is 3m at most from the “Plug - Laundry”, and I have replaced the original sensor because it kept on losing connection.
Any suggestions on how to troubleshoot? Is the network overloaded?
Definitely not overloaded, but I did experience issues with Osram components (see Zigbee2MQTT page on caveats and heard Aqara devices being very selective on routers they connect to.
I don’t know, if that makes sense or complicate things.
I have 63 devices / 255 entities on a 11.000m² area using a single coordinator. No connection issues whatsoever, but most of the devices are bulbs on mains, which act as routers as well. The battery-powered devices are Hue and IKEA motion sensors.
I had OSRAM / Lightify bulbs and sockets, but they kept failing so I replaced them with Tradfri and Innr components.
Each ‘hub’ or coordinator forms its own mesh/network. So doing that could make it worse (more coordinators, more networks in the same RF space, potential interference)
A powered device (like your plug) should act as a repeater. (youre seeing that)
Devices will connect to the device they feel gives them strongest access to the mesh at the time of JOIN
And they should attempt to reconnect after some time not seeing thier parent for a certain period (usually about 20 minutes but due to check in cycles can be double or more) In theory this means devices should self heal after going into panic (this is usually true in practice but…)
Taking all that into consideration, if you add a bunch of repeating plugs the end devices that existed before wont use them if they can reach the coordinator at all. (even weak choppy bad comms) sometimes you’ll need to kick them in by forcing them to. Many devices this is as simple as making the coordinator unavailable (unplug the stick) for 20-30 minutes. End devices panic, hunt, connect to the nearest repeater. Some devices don’t do this and may need to be re-joined in place, if you can’t re join in place you dont have good signal.
Ahhh… And then the Aquaras. Yeah these things are special. For some people they work fine… But for some… Well. There’s a bunch of threads out there on many forums. They’re cheap yeah, but they heavily use manufacturer specific clusters in thier communications. (they were designed and tested to only be used on Aquaras own gear - they dont test against external stuff). You will almost have to engineer your network around them with specially selected routers that the Aquaras are known to be able to use (Ikea plugs are one of the few for instance) to get them to use a routing mesh and actually keep them online.
Personally I don’t have the time nor energy to re-engineer my zigbee mesh around one nonstandard device so I simply will not use them
Agree with others, the mesh is not overloaded at all.
The end devices not connected, are you sure they are not connected. Try and use the ZHA Card (from HACS) and look at the RSSI/LQI. The visual diagram can have end devices with no connection due to different “timeout” before it removes the line.
I always have a few end-devices without any “line”. They show connected in ZHA card and they work/fire events as expected. No issues at all, except from the visualisation.
I have an Enocean momentary switch, that uses energy harvesting. Since it does not receive anything, but is a sender only. It‘s „offline“ all the time , but when clicked, it returns to move immediately.
