Hi,
i have three networks of Aqara motion sensors on a raspberry pi 4B wih conbee II and deCONZ. Each network has about 190 Sensors. After some work on the network i had to reconnect the sensors to deCONZ. The sensors of two networks reconnected nearly complete and immediate. From the third network only 40 of 190 sensors reconnected. I actually red and did quite a lot of the recommandations from the phoscon and git-hub hot to establish reconnetion and tried even deleting and repairing sensors. Nothing worked. After a while I was thinking about changing the batteries - because i am using the sensors 2.5 years. Checking the batteries it shows that many of them are up to 25-50%, so actually i think it is a good idea anyway to change them
So i changed by now 30 batteries (and will change the rest next week) but after the battery-change the sensors still not connect even if i press the 10 sec the button on the sensor with the phoscon “new sensor” button.
So my question is: Is it possible that there are some sensors in the network functionating as a repeater have still low batteries and that because of them even the sensors with the fresh batteries do not connect?
Hello, sensors don’t work as router (99 % of them).
A Battery under 70% is dead, it’s not proportionnal, you have 100% during 1 year, then 90% to 70% in one year for exemple.
but after the battery-change the sensors still not connect even if i press the 10 sec the button on the sensor with the phoscon “new sensor” button
This is not good
It mean you can add new devices ? Because pressing 10s is for a new inclusion. What is your sensor ? you have the correct blink ?
Perhaps the used router is full, but the network is able to find a new one near the sensor. You have the issue on all of them ?
Do you have the GUI ? can be usefull to check logs. Perhaps a connexion issue ? (Try with USB extension and not on usb 3.0).
A router power cycle ?
Hello Smanar,
thanks for the answer.
I am using the aqara motion sensor RTCGQ11LM.
“the router/configuration node is full”
Yes, in the critical configuration there are 198 sensors on one deconz II stick. But I thought the old sensor swould proberbly reconnect with the old connection after changing the battery. That worked with other sensors in one of the other configurations.
The other issue is that in the other two configurations which are working i have two configuration nodes in the deconz node list. In the configuration which is not working proper there is only one.
hello, @smanar: you asked yesterday about the blinking of the sensors.
I opened up the “connecting new sensors” in phoscon and pressed one sensor for three seconds. It blinked three times short, one time long.
I opened up the “connecting new sensors” in phoscon and pressed one sensor for twelve seconds. It blinked three times middle long, one time long.
Finally I opened up the “connecting new sensors” in phoscon and pressed one sensor for twelve seconds and than twenty times quick.
But no connection.
The same after router power cycle and deleting ten sensors (so there should be room for new connections) and cleaning the cache.
I set up an alternative system of raspberry/deCONZ/phoscon II and the sensor paired. So it seems to be a problem that the deCONZ.
So deCONZ connects to ca. 40 of 187 (sensors). But it is not possible to reconnect further sensors or add new ones.
Ok so for the moment the only answer I have is “This network can’t exist” ^^.
You can confirm me, your network have more than 100 sensors and no router, and ofc it works ? (at least for 2/3)
Yes, one network has 197 Aqara motion sensors sending messages to one raspberry 4B, with deCONZ and a Conbee Stick 2. The other one 155. Starting deCONZ it says “limited to 200 network nodes”…
There it is said that the number of devices can be more than 32 but it needs routers. So i wonder where my routers are…
What devices can be routers? Can I identify routers in deCONZ (as i can identify configuation tools)?
All the end devices are installed in an area of about 30m². Is this prpberly has an impact by the necessarity of having routers?
Thanks
I just see one blue one, rest is grey.
I checked in Phoscon 55 sensors if they were reachable. They are all updated in the last hour, some in a bit more than the last hour.
Is there a way in deCONZ to check the way of the binding of an end device?
as the Aqara sensors will not use the routers automatically, reconnect the sensor by pressing the sensor-button 10 sec
Smanar told me: every powered device can be routers. As I do not need any device with a functionality (light, switch,etc.) in the network: Are there special zigbee routers (did not find any, only a “repeater”) which just have the routing-functionality and which work best with deCONZ?
Thank you for the support
Mimiix writes:
“Sensors do not have a limit.
You might have too many end nodes opposed to routers.”
ANd as deCONZ supports 200 nodes…
But I just paired them and it worked. Did not know i need routers. I will check tomorrow or on tuesday the network with the 155 sensors if they are all reporting. But i already did that as i implemented them and than regulary every 6 month. And yesterday i did check 55 sensors of 155 and all of them where in the network. So actually i am shure they do work. But i’ll check again.
thanks for the repeater links. I will buy some repeater for the network which does not work and set this network up with routers.
I have a long tuto from ebaauw, I think I can share it
Zigbee is a packet switching network that doesn't maintain connections. Think of each message delivered as a separate letter. Each message (packet) is routed separately. A request message and the corresponding response message (or simply the ACK) are routed separately.
There's a difference between FFD (routers, yellow nodes) and RFD (end devices, grey nodes). Routers forward Zigbee messgaes, end devices do not.
There's a difference between devices with Receiver On When Idle and without. The first have their radio on all the time, the second have their radio off most of the time (typically to preserve the battery). In DDF words: they are sleepers. Routers need to have Receiver On When Idle (as they need to receive and forward messages), end devices might or might not be sleepers.
There's a difference between battery-powered vs mains-powered devices. Routers tend to be mains-powered, but there's no law for that. We've seen plenty of example of mains-powered end devices, although they typically have Receiver On When Idle.
Again, there's no hard relationship between router vs end device, main-powered vs battery-powered, Receiver On When Idle vs sleepers, even though the API distinction between /lights and /sensors assumes this. See the Node Descriptor in the Node Info: this lists the Device Type, Power Source and Receiver On When Idle separately.
So how do I send a message to a sleeper? How do I send a message to some-one without a mailbox? You use a P.O. Box instead. This P.O. Box is located at the sleeper's parent router. The sleeper polls the parent router periodically, checking for messages.
For this to work, the parent router must know what sleepers to keep messages for, and the sleeper must know what router to poll. Unfortunately, this part Zigbee seems less standard, and sleepers and routers from different manufacturers (or rather Zigbee stack implementations) tend to misunderstand each other. When the parent router is no longer available (because some morons still think it's cool to power down Zigbee lights with 20th century wall switches), the sleeper needs to find a new parent. When the old parent becomes available again, it should no longer answer to route requests for their former child.
Parent routers only keep messages for slightly over 7 seconds. After that, they are discarded. They should send a "return to sender" message, but many don't, discarding the message silently. Light sleepers poll their parent every 5 seconds, so messages won't get lost. Effectively, they react like Receiver On When Idle, albeit with a delay. Deep sleepers poll their parent less frequently, and messages will be lost. This is what the Sleeper attribute in the DDF indicates: deCONZ should only send messages to this device, when it knows it's awake (i.e. polling its parent router).
Because each P.O. Box takes memory on the parent router, the router firmware only supports a limited number of children. The RaspBee or ConBee, in this aspect, is just another router. deCONZ doesn't refuse anything, even the RaspBee or ConBee doesn't refuse anything; they simply don't accept another child. It's up to the sleeper end device how to deal with not being able to find a parent router. Some might leave the network, others might still remain in the network.
Note that "being in the network" simply means that a device has the network parameters (panid, transport key, channel, extended panid, ...) stored in non-volatile memory. There are no connections in Zigbee.
Orphaned end devices are still able to send messages, so they might appear to work. However, they are not able to receive messages, incl. the ACK for the messages they send. Some Zigbee stacks don't like that and will leave the network, others don't seem to mind.
I don’t know what devices are in the above picture. Most likely, they’re end devices with Receiver On When Idle. Alternatively, they could be orphaned sleeper end devices, that are happy to remain in the network.
Wow, ok in the moment i am quite relieved that i did not construct something supernatural which should not work
So in the next week i will change batteries in the non working network and set it up completly new. it is a good possibility to do that. I will report if it worked.
Thank you for all the support. It was very eye-opening…
Hello,
i implemented now the recommended solution and it works perfectly:
set up a fresh raspberry os and fresh deCONZ
buying 14 zigbee repeater (Tuya Signal Repeater Zigbee 3.0) together with the phoscon stick i managed to connect to 180 Sensors (repairing them). I know it is not the ratio (end-sensor/repeater) you recommended was 5:1 but as it costs 10 EUR per repeater i tried to skip costs a bit and as long as it works…
I fought my fight with the new raspbian os (bookworm) and chrome browser - by changing to firefox. which released me from bad performance of the phoscon app in chrome.
So I would close this post - thank you very very very much for the support. It helped me a lot !!!
