Unreliability issues

I installed a Home Assistant based automation system about a year ago but I am very disappointed because the system has not proved reliable. It all works when I am there, but too often devices become unavailable. When that affects something like the heating it becomes rather important.

It is in a house that is not my own so I am not there often to fix issues. I have remote access but there is a limit to what I can fix remotely and I cannot monitor it 24/7 anyway.

The most common problem is disconnection from the WiFi or Zigbee network. Often, turning the device off and on again will reconnect it; other times restarting HA does the trick. Sometimes a reset and reconnect is necessary.

I have a mix of devices. Those on WiFi are mostly connected with Smart Life and integrated with Tuya Cloud or Local Tuya. Zigbee devices are all connected to a SONOFF hub and integrated with ZHA.

The WiFi devices are connected with a quite dense TP-Link Deco M5 mesh WiFi – 6 nodes for 12 rooms. I have a similar number of Zigbee repeaters. There are something like 50 devices of each type. It is an ordinary detached house so there are not many other WiFi networks nearby and as far as I know the house has no more than the average amount of 2.4 GHz band interference.

It is not clear that this is really an HA issue as such, but where else can I ask? If I cannot solve the problem then I am going to have to give up the project!

I would therefore very much appreciate any tips on how to make the whole system more robust – that is to say devices should rarely disconnect at all, and if they do (due to power cut or similar) then they should reconnect without manual intervention.

It would also be helpful to know how to get HA to send me alerts when a device disconnects.

My first inclination would be the AP setup is interfering with both itself and the ZigBee mesh. I would start there.

I would:

  1. Make sure the ZigBee radio is on channel 15 or 20 first.
  2. Check the WiFi channel assignments on the APs. Try to stick to channels 1, 6, and 11. The other piece is to try to keep adjacent APs off the same channel.
  3. Lower the power of each AP.

This process takes some tweaking. There’s a litany of WiFi analyzer type apps out there. I’d grab one of them and walk the house.

EDIT: After a quick Google search, it doesn’t appear that you can manually set the channel for the WiFi with the Deco, so you might end up needing to replace those to really solve this.

Another thing that may also help would be to make sure you have a separate 2.4 GHz only wireless network for the IoT devices. Most that I’ve ran into really don’t like band steering (the process where the AP tries to connect the device to a particular band when you have 2.4 and 5 GHz enabled for the same SSID).

Thanks. I did most of that…

  1. Zigbee is using channel 15.
  2. WiFi is currently on channel 8. Would 11 be better in some way? Deco selects the channel automatically; there is no manual override, so I cannot change it overall or for each AP.
  3. Deco cannot change the AP power as a direct parameter but there is something called beaming that allows the beam shape to be adjusted. Would that help? I guess it is only useful if the Decos are fixed to the wall?
  4. I tried several WiFi analyser iPhone apps but did not yet find one that easily visualises local signal strength or interference. Most do nothing the Deco app does not already have. Can you recommend one?
    (4a). I recently bought a HackRF One that I plan to set up as a signal analyser that should also detect interference, but there is a long learning curve on setting that up! Has anyone used a HackRF One for this purpose?
  5. Deco does not allow channel or AP power to be manually set. If these are necessary, I would have to buy another mesh WiFi system that does allow them to be set manually – any ideas?
  6. The Deco WiFi network is set to 2.4 GHz only. There is a separate TalkTalk WiFi from the router that has both bands, but none of the IoT devices use that SSID. Should I stop laptops and phones using the IoT network?

Generally, as I say, when I look at it closely everything looks fine, though the internet is sometimes inexplicably slow on my laptop. Nevertheless, IoT devices keep disconnecting. So frustrating!

A TalkTalk engineer called recently and said everything is good with the router and broadband. He also thought the Mesh network would be better if it were fixed on channel 11 with channel hopping turned off.

BTW: When I set up the Deco network I found I could not get it to work with the TalkTalk server unless I allowed the TalkTalk server to do the DHCP (not the Deco, which is the default). It seems to work OK (and I can fix IP addresses if necessary using the TalkTalk UI), but could this possibly be a contributory factor?

I would very much like to hear from any reader on

  • A. Any mesh WiFi system that you have found works particularly robustly with IoT / Home Assistant.
  • B. Any tips you have for diagnosing in HA logs etc. why a WiFi device disconnected.
  • C. Any tips you have for diagnosing in HA logs etc. why a Zigbee devices disconnected.

That depends on what channel width you are using. At 40, I imagine there will be some slight overlap, but not by much.

It might. Beamforming is supposed to reshape the RF lobes to be more directed towards connected devices to cut down on EMI for other devices. Can’t say it works that well though.

WifiMan from Ubiquiti is what I often use.

I don’t know that I’d replace the system on that premise alone.

I’ll always vote for Ubiquiti gear here. There’s definitely “better” options, but for an out-the-box, general consumer option, this is it.

No…HA will just complain that it can’t reach the device. Accounting for the why is between the device, AP, and router.

Pretty much the same as WiFi except less information. Closest you can get (that I can think of) would be looking at the mesh map and LQI. Low LQI would be more subject to disconnects. Looking at the map, and knowing what router a device is going through, can explain things. Like, if a bunch of end devices suddenly drop off and are all using the same router, probably should look at the router first.

I downloaded the WifiMan on an iPhone and it looks nice, but for signal analysis it needs the Ubiquiti Unifi WiFiMan Wizard – a small white box. While this is just $99 in the USA, the cheapest I found in the UK is £189.56 ($238.96!) including shipping from the USA. That’s too steep for me as I am already working on a spectrum analyser solution using a HackRF One (which cost me £128.85)

for Zigbee devices …

The mesh map looks fine – the devices are connected to the nearest repeater (sometimes I had to reconnect them to force them to do that).
There are plenty of repeaters, so I do not have low LQI – that is why it is so mystifying.
:thinking:

My mistake. I didn’t realize that WiFiMan can’t scan from an Apple device. Appears it’s a limitation with the iPhones where the app doesn’t have the same access to the WiFi hardware as on the Android side.

Ah. So if I resurrect an old Android phone, I might use it for this purpose without the $238 add-on?

Strong possibility.

I’ll dig one out and try it. Might even be cheaper than $238 to buy one!

For zigbee devices I use uptime kuma for monitoring devices avability using mqtt. If devices is offline uptime kuma send me a notification on my telegram account for that device. You can use the same for wifi devices using ping. But as you use tuya devices connected to the cloud…
Well I dumped most of it because of very poor reliability with out possibility for improvement.
Expect flashing esphome firmware on it. And that is not easy task sometimes.

Depressing news! I do not use MQTT. I do have some ESP Home devices and they are mostly OK but also not 100% reliable in their connection.

Why this should be depressing news? I also have esphome devices and some of them are using mqtt. It’s easy to setup esphome device to use mqtt.
But mqtt is not your problem. You probably have some kind of interference or poor router or one or more blind spots or maybe this is combination a little bit of everything.
As from esphome devices goes this things never disconnects from wifi network. I never experience that esphome devices disconnects it self from wifi.