Connection problems with ZHA

Configuration Zigbee
29

Having a few dedicated Zigbee Router devices will make a huge difference when just getting started with Zigbee and you have not yet have added many non-dedicated Zigbee Router devices, but those dedicated Zigbee Router devices might be ok to remove if and when you added many other non-dedicated Zigbee Router devices that make up the difference. Regardless, you can never have enough Zigbee Router devices. Again that is also covered in my guide → Zigbee networks: how to guide for avoiding interference + optimize using Zigbee Router devices (repeaters/extenders) to get a stable mesh network with best possible range and coverage

1 Like
30

I got my USB2 hub (this one here: Amazon.de)

I turned off my HomeAssistant. Connected the hub where the Skyconnect was and plugged the skyconnect - with its extension cable - inside the hub.
The newly added USB2 hub is on the left side of my homeassistant, the skyconnect is on the right.
So I didnt change the distance from HomeAssistant to zigbee coordinator. (its about 2m).

But it didnt change anything.
Now I followed all the tipps in the guide and I really have no idea what to try next.
Any tipps?

31

I really wonder if its time to try z2m?

32

Have you tried changing to a different Zigbee channel after checking WiFi channels neighbor uses?

And did you try downgrading the firmware to latest 7.1.x.x or 7.3.x.x (as at least the 7.2.0.0 was known to be unstable, and the 7.4.x.x versions are still cutting-edge so not proven yet if stable enough for production).

FYI, the ZHA documentation has not been updated but it is now possible to change Zigbee channel (radio frequency range) from the ZHA UI/GUI interface and usually you will only need to re-pair sleepy devices (meaning normally just battery-powered devices and mains-powered products that are not Zigbee Router devices such as those that do not have neutral connected).

Important though is to first read and understand this article on (that was linked in the guide) as it covers what you should consider before changing the Zigbee channel:

PS: There is some more info on that ZHA UI interface option (not documented in ZHA docs) here :

33

Argh
I got your first post 18d ago wrong. And downgraded my stick to version 7.2.0.0.
This was probably the worste firmware to chose :o)

But Im a bit confused now.
If I open the github respository (silabs-firmware/EmberZNet at main · NabuCasa/silabs-firmware · GitHub) there is version 7.1.0.0.
As you said Í should downgrade to the latest 7.1.x.x. version I looked inside the archive and found 7.1.3.0 in the archive.
Putting something inside the archive looks like sayin “its there, but better dont use”.
Which version you recommend?
Edit: as the flashing website tells me, that there is nothing wrong with the firmware the stick was shipped with - Im going for 7.1.0.0.
This should be better than what I have, and flashing to another version (if you recommend anything else) isnt that hard.

edit: I know my zigbee is running on channel 11. Thats all I know.
If the firmware downgrade doesnt change anything, ill take a look at the channels (and try to understand whats happening there)

34

That is how intepret it too, if it is in archive then it is not recommended, (so my guess is that they have first had that under the beta directory but then it was proven not to be as stable as older version so they moved it to archive instead).

I also suggest that you go with the 7.1.0.0 version for now, (though you can keep an eye on the beta directory too as expect that will probably see that a newer 7.3.x.x will become new stable sooner rather than later)

But if you do want to experiment with later version today then check out thee unofficial (NCP) community builds by darkxst that are very popular among testers and early adopters → silabs-firmware-builder/firmware_builds/skyconnect at main · darkxst/silabs-firmware-builder · GitHub

So you are saying that you did not read the guide and are not followng the best practice tips there :stuck_out_tongue:

At the very least read this → ZigBee and Wi-Fi Coexistence | MetaGeek

Can you at least check your own WiFi-router or WiFi access-point(s) what WiFi channel(s) are using? While logged into it you should make sure it is set to a static channel and not to auto so it won’t change.

Preferably also install a free WiFi-survey/scanner app (like WiFiman by Ubiquiti) on an Android-based phone or tablet and have it quickly analyze what WiFi channel(s) your neighbors are using and what strength they are so you with that knowledge can make a choice to avoid those too if their signals are strong.

Zigbee and Wi-Fi channel numbers may seem similar but they are not. So if you are for example are using Zigbee channel 11 as you said then you and your neighbors should preferably not be using the overlapping WiFi channel 1 (which is the most commonly used WiFi channel and hence it is generally a bad idea to use Zigbee channel 11 to 14.

The general best practice is to use Zigbee channel 15, 20, or 25 (if you only have newer Zigbee 3.0 devices then Zigbee channel 26 is an option but should be avoided if have any older Zigbee devices).

If you are using Zigbee channel 15, 20, or 25 then you want to lock down your WiFi to it is set to static channel (not auto) and only using WiFi channels 1, 6, or 11 as you get the least overlap with Zigbee.

If possible talk to your neibors to lock down their WiFi to static as well (also using WiFi channels 1, 6, or 11 but not the same WiFi channel as you are using).

Personally I will not be help you anymore here until you actually done your best to follow my guide:

I mean, why should we spoonfeed you here a trickle at a time when you can not be bothered to read and try your best to take action in following those best practice tips to help yourself. Help us help you!

image

35

Thanks again cor your help!

But I hardly need to contradict this:

I read ALL the links, some of them more than once to understand whats going there.
And I did ALL mentioned there (except the channel thing, and I switched to false firmware, because I got it wrong).
And furthermore I posted what Ive done, in the order I did it, because I didnt want to change everything at once to maybe see where the problem was.

So Im really really lucky to have some1 like you - and all the others here ofcourse helping me here.
And if you got the impression im not reading, or not following your suggestions youre defenitely wrong!

But again, changing all in one, wont make me find the issue.

36

So. I took an intense look at the channels thing.
I read the guides more times, but please bear with me - Im having some problems understanding it.
At the end of this gude ZigBee and Wi-Fi Coexistence | MetaGeek there is a “Zigbee Wifi Chart”.
If I get it correct, its just essential to use WIFI channels and zigbee channels that dont overlap.
So using Zigbee 11 and Wifi 1 is a bad idea, because they are in the same blue area.
Maybe Wifi 1 and Zigbee 21 is a good idea, because they dont overlapp?
And ofcourse dont miss your neighbours Wigi channel.
Is this the conclusion of this guide?

So the Wifiman apps tells me its using Wifi channels 100,60 and 52. But in my FritsBox its on Autochannel, so this value isnt worth anything.
I find one neighbours WLAN (Fritz Box 7490) on channel 6.
In my Fritz box (with Channel auto settings) it tells me Im using channel 6 on 2,4ghz and channels 100, 52 and 36 on 5ghz.
I think 5ghz WLAN isnt of interest (as it shouldnt interfere with 2,4ghz zigbee).

So if you say Zigbee 15,20 or 25 is best practise Ill choose Channel 25.
Because channel 15 and 20 do a bit overlapp with my neighbours WIFI channel 6.
(and there are only companies around here - chances to make them change their channel are very small).
In addition to this I would change my Wifi channel from Auto to channel 6.
But a question here.

There is no overlapping with zigbee channel 25 and Wifi channel 6. (i think)
Why should I use Wifi 1,5 or 11 only?

37

That was a typo-o on my part, I meant that you should aim to only use WiFi channels 1, 6, and/or 11.

Using Zigbee channel 25 with WiFi channel 6 is a good idea (as long as lock down that WiFi channel).

Another good alternative is to use Zigbee channel 15 with WiFi channel 11.

For why those WiFi channels see this video → https://youtu.be/49JBYSv3Nig?si=dPafqszjQAZP7rgN

As explained in that video it is best to also move as many WiFi devices to 5GHz bands as possible.

Again also check what WiFi channels your closest neighbors are using and how stong signal they have.

The goal of following best practice tips is to first try eliminate all common and probable root causes of problems before you begin to look for unusual and improbable root causes. The point I was trying to make is that it is not worth the effort to try to dig deeper and try to troubleshoot a problem that might simply be caused by you not following all the best practice tips.

Please understand that it can be hard to impossible to troubleshoot further to try to find possible and less probable root causes if your current environment has radio noise and is unstable because you are not already following all existing best practice tips recommendations. Nothing good can come of that and at best that will only make troubleshooting harder, and it will for sure cause unnecessary frustration along the way regardless of how we continue here.

Keep in mind that there are an estimated 200,000+ users of ZHA today (and many more Zigbee2MQTT users than so), do you think would be a sustainable community if each and every one of those users would only willing to try a single best practice tip at a time and not willing to make the effort to do “all in one” recommended actions.

Now I wrote that guide to avoid that situation but most of those best practice tips are also available under troubleshooting sections in the ZHA (and Zigbee2MQTT) documentation, and if you create a new issue before following those tips then respect that you are wasting the time (and patience) of developers.

But you can go ahead and waste your own time trying to directly find and fix that maybe non-existing and improbable root cause before you have done your best to eliminate all common issues like radio noise.

image
image750×600 79 KB

38

Bro, thanks for the clear words!

I really got it.
Those best practice tipps are nothing I should follow in some parts to eleminate my errors.
Its something I MUST FOLLOW COMPLETELY, before its even worth spending time on my problems (if they are still existing then).
But again. Even if I understood it wrong at the beginning, I now followed ALL! best practise rules except the channel thing.
As I really ask you for answering my questions in my last post.
Edit: I found your edit regarding the Wifi channel 6.

39

Sorry for all the double posts. But I think a new post here is easier.

What I did until now:

  • change coordinator firmware to 7.1.0.0 (from 7.4.?.?)
  • added a lot of new Thirdreallity plugs as routers
  • changed my 2 aqara plugs with thirdreallity ones (so all my mains powered devices are from thirdreallity, except 2 Aqara LED stripes)
  • in addition to all the plugs I added 3 Sonoff Zigbee Dongles with repeater firmware at very good places (I think)
  • I’m using a mains powered USB2 hub on my IntelNuc and on this I have an 3m extension cable with my Skyconnect. So the coordinator is about 2 - 2.5m away from the NUC
  • I removed zigbee plugs very close (<30cm) to shelly plugs (in case they receive a signal and cant forward it because of the shellys)
  • I changed my WIFI 2,4ghz channel from Auto to channel 6
  • I changed my zigbee channel from 11 to 25
  • and one thing - not directly mentionend in the tutorials: after sending this I will delete all the unused AP of WLAN devices (I have some shelly 2pm for example connected with WIFI but still creating thei own WLAN access point (which isnt needed)

Ofcourse I have no idea if my problems still exist (as I changed the channels 10 min ago).

But I really think I followed every single best practise tipp (after some attempts, what I really really can only appologize for)

edit:
when changing my zigbee channel from 11 to 25 via UI it told me that all my devices will rejoin the network and it takes about an hour.
But after confirming this (about 30 min ago now) just nothing happend.
All my 97 devices stayed inside zha (even the aqara battery ones), but in the ZHA settings channel 25 is displayed.
In zha i clicked on config, choosed channel 25 and clicked change channel.
After this I clicked on actualize configuration (at the bottom of the page).

40

If your problems remain after you have done all that then you should now enable debug logging and try to capture and collect debug logs from the point in time where the problem occur so that it is likely to be seen in the debug logs. See:

and

In summery that mean getting debug logs + collecting diagnostics and opening a new issue here in the Home Assistant core repository on GitHub (where you can also search for already reported ZHA and other issues) → Issues · home-assistant/core · GitHub

There is also additional information about reporting issues you should read before open a new issue:

But yes, to make it worth the effort for ZHA developers to help with log analyzing you need(ed) to first have already done your best to take all those actions in order to at least follow the most important best practices tips before moving forward with collecting debug logs, (or else there is a chance they will only show that you have a lot of radio noise in your environment, i.e. EMI/RMI/EMF jamming your Zigbee devices and preventing them from communication properly). If already following best practices then hopefully ZHA developers can help with log analyzing and deeper troubleshooting to find any remaining root causes.

Sorry if I snapped at you but I read your posts like you felt following all best practices tips would be too complicated or time-consuming for you and therefore we should use our time instead to try to find your root cause directly without you needing to take all the recommended best practice actions first, which if it was the case is an outlook and attitude that I find unacceptable as practically all those best practice tips are relatively easy to follow and should not take much time or effort from yoursekf to take actions on. Because in essense the hardest parts boiled down to you making sure to use the Zigbee Coordinator via a long USB extension cable connected to USB 2.0 port to get it away from known sources of EMI/RMI/EMI (like USB 3.0 ports/peripheral/cables/), as well as adding a few additional “known good” Zigbee Router devices in strategic places around your home, and then locking down your WiFi channel and changing to a non-overlapping Zigbee channel (to something that does not conflict), and finally maybe being forced to re-pair most sleepy/battery-powered devices (if they where not awake when the Zigbee channel change commands were sent out to your Zigbee network).

PS: I still think that you should consider also reading this guide too before posting next time though:

41

Well.
Ill wait until tomorrow (as I cant imagine all zigbee devices stay in the network after changing the channel) and repair whats not working.
Then I know if I still am having issues or not.

Ill give message after.

Really no problem man! I think helping me isnt that easy, as I did understand many things here wrong. Making me a person you dont want to help.
But it maybe took a bit longer than needed - but finally we got it :o)

Thanks for continuing giving me advise! Really!

2 Likes
42

Sounds like something maybe went wrong with your Zigbee channel change as normally all your Zigbee Router devices should come back online within 10-15 minutes (because those are usually online so recieve the change command directly) and then after that you would normally need to manually re-pair all battery-powered and other Zigbee End-Devices.

You can try power-cycle each of your mains-powered to see if they automatically connect after that, if not then you are only left with the option to re-pair them (or restore the whole Zigbee network from backup).

Regardless, note that you should not ”remove” any devices, instead only re-pair them, meaning that you follow each devices manufacture’s procesure for pairing. You do not want to remove the device from ZHA because then it will get a new entity name and that means any automations using the old entities will need to be manually reconfigured to use the new entities.

1 Like
43

Im gonna bet this ends up being the culprit. Just marking it down for posterity… Don’t go off path and Continue troubleshooting. You guys are on a good clip now. :wink:

44

Nope.
The problems still exist.
I found out there is an issue already opened on gihub ZHA page.
As I never had such issues for 2 years and it came out of the sudden and there are lot of people having exactly the same issue than me, im thinking it really might be a ZHA issue.
Even more after excluding all the inteference stuff.
But as the uploaded logs in this issue look exactly the same as my logs do (amd I dont have any “working” logs), i cant provide any helpfullinformation there.

But I wanted to test z2m anyway - but never did because zha was working.
After changing the zigbee channel I would need to repair all my battery powered devices anyway. So Ill install z2m tomorrow.
Some of my devices are supported better there (with more entities, I would use if I had them). But as mentioned I didnt want to change the running zha.

45

Thats what im doing right now.

46

Some days are gone now.
In the meantime I changed the zigbee (and the wifi channel) again. Maybe something went wrong the first time or changing didnt work.
This time my network was broken, just nothing worked anymore.
So I needed to setup all my zigbee from scratch.
I found out there is an issue for zha regarding exactly my problem, but as my network crashed I dont have any logs to present anyway, and I wanted to try z2m for long time now - but just didnt, because zha worked well for me (even with some device function restrictions), and I didnt want to set up everything new. (about 100 devices now)
As I had to anyway Im now trying z2m.
Dont get me wrong - Im not saying zha works not z2m does.
But what I can say after rebuilding my whole zigbee network everything works perfectly fine now.

Thanks to all here helping me building up an almost perfect mesh network!
And obviously especially thanks to @Hedda for sharing his knowledge here and wrtiting such an awsome zigbee guide!

47

Thanks for the tip @Hedda - WiFiman led me to the source of my troubles.
I had DELIVERY-FAILED-3074 errors for about a week and was pulling my hair.

  • First I moved ZHA from CH20 to CH25.
  • Then I moved ZHA from a Sonoff P to a Sonoff E stick. This really improved the overall LQI-values in my mesh, but still some devices were either very slow in responding or didn’t react at all.
  • Then I moved my remaining Hue devices to ZHA and got rid of the Hue Hub (CH15).
  • Then I figured that maybe >100 devices were too much for ZHA and I installed a parallel Z2M network (CH11). I moved a lot of devices there, mainly battery operated end devices like temperature sensors and motion sensors (but also some strategically positioned routers).
  • I also searched for WiFi interference, but used WiFiAnalyzer (which did show me a nearby AP that I didn’t know, but was no further help).

It was WiFiman that finally provided the answer. It showed more information. The unknown AP also used Bluetooth and the vendor was Espressif. With the strength of the Bluetooth signal I was able to locate the source in my son’s room. He had installed a smart device himself that created its own WiFi network, which was not only strong, but also very broad, overlapping a range of frequencies.
I pulled the plug on this monster and my ZHA network is now very stable again.
Well, due to the problems I now have ZHA and Z2M running parallel on a Home Assistant Green. Let’s see how this will work.

48

I still think this is a HA issue and not a general ZHA/Zigbee issue. I’ve tried every fix in the book and can even go stable for 2+ days. But then, out of the blue, DELIVERY-FAILED-3074 error. Absolutely nothing has changed. My neighbors are doing what they’re doing, my USB distance for my Sonoff is the same as it always was, nothing added to the network. Yet somehow, just out of the blue, HA can’t deliver to the Zigbee device. Incredibly frustrsting.