deCONZ - Official thread

I made the horrible descission to move all my lights to deconz. The idea was to have all my lights, which is mixed ikea, hue and other brands on one platform. What a bad experience. Ever since i did that nothing has been working properly. Switches only work half the time. Automations only turning on/off half the lights it should.

Just look at this example. My usual automation that turns off the lights outside at night only turned off some of the lights, yet it seems that Home Assistant thinks the light IS actually off, while deconz rightly thinks its on, which it is.

What is going on?

1 Like

Have you put your conbee on an extension cord ?

Yes and tried several lengths

Hey all
I have like 15 multiple purpose sensors from SmartThings, connected to deconz

All works good aside from one of them which only reports 2 out of 4 states
Any idea why or how to fix?

The values are found when I look in deconz directly, just not passed to HA

Or is this a case or remove and add again?

Try to put your problem lights in groups (in deCONZ) and only control the groups. For some reason that made my deCONZ setup much more reliable.

Yes that is an important advice

In Philips Hue and Ikea Hubs you normally define rooms (groups) and put the lights in these rooms. And then you associate the switches with the groups. In Zigbee world this means that when you turn a room on, one group message is sent.
Maybe you have som scenes with individual settings and then more messages are sent. But overall the number of zigbee messages queued up at the same time is reduced. And remember in a mesh network these messages are relayed by routers so that doubles or triples the traffic.

Home assistant groups are just logical groups and you can mix zigbee, zwave, RF etc devices into same group. But Home Assistant does not know which zigbee groups devices belong to so all it can do is send an individual message to each device.

By defining groups in deconz, these are presented as a single entity in Home Assistant. You can put these deconz groups in a Home Assistant group so you can turn on e.g. 10 zigbee lights in a zigbee group and 2 Tasmota switches in a single action. The result is that one message is sent to Deconz which sends a single turn on message on the network.

There are two things that makes this more robust. I already noted the huge reduction in virtually simultaneous traffic on the Zigbee network. The other is that even if there is a temporary poor mesh route to a single device, the probability that the individual device hears the group messages is high. I have occationally seen a light bulb that did not respond to on off but it worked when I turn the entire room on and off.

And in normal life 90 % of what we do is turning rooms or groups on and off

And finally for dimming, use this method… Never step increase or decrease a large number of zigbee devices.

Yes, I can confirm, controlling devices via deconz groups helps a lot. Especially the scenes are much more reliable than via HA.
But anyway. I’m every day facing the problem, that one or another bulb or switch is unavailable. Sometimes it’s more annoying, i.e. when the stairs bulb doesn’t switch on during the night or the ventilator switch doesn’t react, but we might have to live with it. Last but not least we’re talking about 2.4Ghz. WLAN, mobiles, microwaves and maybe the neighbours wlan and devices disturbing the mesh and we have no chance to handle that. The interesting thing is, that the hue hub was more reliable. But maybe we just haven’t noticed the interruptions.

1 Like

One weak link I eliminated were smart plugs from Osram

I had 7 of them. On Philips hue they always appeared offline. And on Deconz I noticed that devices routed via them would not always work. I replaced them all 7 with Philips smartplugs and that improved the stability of the mesh.

I still suspect some older Ikea 1000 lumen bulbs are not as good at routing but Ikea finally released an update for them.

The four button Philips remote were often unstable in the beginning. They were even worse on Philips’ own hub than on Deconz. I gave Philips hell about it. After a year they finally released a firmware that made them stable.

So check out for weak links. Not all devices have good firmware.

1 Like

Innr plugs are rock solid as well.
I replaced two Osrans in the living room and never had problems again. Thanks for the tip withe the philipps ones I will try them out.

1 Like

Conbee II stick stopped working, If I plug it into a windows 10 pc it just keeps connecting and disconnecting, This started after updating the firmware in phoscon… Anyone else had this issue or know a fix?
Thanks

Yes i had this issue. I had to power down the machine and plug it in another usb port …

This sounds like my recent experience (and others), you might be experiencing similar to https://github.com/home-assistant/core/issues/37372

Hass state is updated based on what deconz reports so the communication seems to be working which would suggest that it’s a different issue

I would rather say that the integration in HA is broken. HA shows me several entities and devices as unavailable, but I can switch them via Phoscon. I can also access the entities via OpenHab.

1 Like

I didn’t say it wasn’t an issue with the integration. I said that it was a different issue from the onw mentioned above. Please create a new github issue and post debug logs

Don’t agree with that. Had one innr SP120 and the two Xiaomi motion sensors connected always dropped out after a day or two.
Unplugged the innr switch, reconnected the sensors (now connected to a Feibit switch), the motion sensors are working reliable since weeks (again).
My experience regarding Xiaomi sensors and battery powered rocker switches: it is crucial to find the “correct” node. This might be a node even two floors above or below the end point device. If your Xiaomi end point devices drop out on a regular basis, look up to which end point device they are connected and show this behaviour. Turn off this device and reconnect your Xiaomi device. Iterate until you find the best end point device.

It’s so weird. I have the Conbee II stick with DeconZ in Home Assistant for a year now without 1 single issue, dropout or whatsoever.
I have 20+ Philips Hue Lights, 6 Ikea lights, 7 door sensors, 4 pirs and a couple of buttons from Xiaomi and also a couple of Philips Hue dimmer switches and a couple of Pirs.
I think that interference is a really big thing also. Zigbee works on the same frequency spectrum as your WiFi. Always move your Wifi router away from the Conbee stick, use a USB extension cord for the stick, even a powered USB hub would be a good idea.

i never had issues with wifi interference, no mather what channel i choose… the only thing i struggled for a long time were my xiaomi sensors, seems xiaomo, once they are paired to a repeater/router, they dont change anymore … so if you reallocate some smart switches, make sure, you unpair/pair again the xiaomi stuff , so they choose another router then

Hi, anyone knows if a rain sensor exists? I have a xiaomi water sensor, but offcourse not the same effect for rain

for some reason my add-on was suddenly stopped? i have no idea why? this was from log :
what could be cause?

2020-08-14 12:08:54 ERROR (MainThread) [pydeconz.websocket] Client connection error

12:07:48:277 don't close database yet, keep open for 900 seconds
12:07:51:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:07:52:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:07:53:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:07:54:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:07:55:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:07:56:232 don't close database yet, keep open for 900 seconds
12:07:59:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:07:59:286 GW firmware version: 0x26330500
12:07:59:286 GW firmware version shall be updated to: 0x26350500
12:08:00:275 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:08:01:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:08:01:464 don't close database yet, keep open for 900 seconds
12:08:02:274 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:08:03:275 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
12:08:03:341 Current channel 11
12:08:03:373 Device TTL 2400 s flags: 0x7
12:08:04:275 skip binding for attribute reporting of ep: 0x00 cluster 0x1000 (end-device might sleep)
[12:08:04] WARNING: Your direct VNC access is not protected!
[12:08:04] INFO: Starting VNC server (local/no)...
2020/08/14 10:08:04 [notice] 323#323: signal 15 (SIGTERM) received from 313, exiting
In exit
Closing socket listening at 127.0.0.1:5901
2020/08/14 10:08:04 [notice] 1851#1851: exiting
2020/08/14 10:08:04 [notice] 1851#1851: exit
2020/08/14 10:08:04 [notice] 323#323: signal 17 (SIGCHLD) received from 1851
2020/08/14 10:08:04 [notice] 323#323: worker process 1851 exited with code 0
2020/08/14 10:08:04 [notice] 323#323: exit
[cont-finish.d] executing container finish scripts...
[cont-finish.d] done.
[s6-finish] waiting for services.
[s6-finish] sending all processes the TERM signal.
[s6-finish] sending all processes the KILL signal and exiting.