Oh wow, i didnt realise that zigbee had groups and bindings, thats new to me.
I really wanted this to work, as it would solve soooo many of my problems (even more than what i asked for here). But unfortunatly, it seems like im just unlucky. As you suspected, Aqara... Just so happen to be that the one smart switch i had access to here (which i got more than dozen installed), just so happen to not have bindings working... sigh.
Yup, mine too 
Could you elaborate on this one? This is new to me - what are these "custom firmware", and where do you get/find them?
Ive even gone as far to change my entire network/gear around couple of times, and even went as far as to wire up most of the house with ethernet
Ive also gone as far to swap zigbee coordinators about 4 times now over the few years, to the point where im now using the SMLIGHT coordinator via poe (no way near the usb ports or the HA instance). Even my HA is running via 10th gen i5 via proxmox (so waaay too much power).
Ive also move house above 3 times, and the issues are the same. So, its not exactly related to neighbor/house interference either.
...so yeah, not sure what to tell you, you may have won the zigbee lottery. Dont get me wrong, not everything is falling apart or anything. With my current collection of devices, things are fairly solid - but i do definitely get hiccups if a light is accidentally turned off from the wall (which used to happen a lot with guests). One of the major issues i have is, none of the battery powered zigbee buttons ever work reliably enough to be usable haha (i have dozens from few different brands including ikea, aqara, etc).
...maybe my standards are high? haha
It's likely that people that don't have issues probably go the extra mile:
None of these points are fatal, but the more issues you don't fix (or worse introduce) the more likely the network will have problems.
If you have a large network with limited router coverage frankly you can't afford to be power cycling routers - they need to be permanently powered.
We are a bit off topic, but feel like i should answer this anyway.
You are not wrong, absolutely. But my comment still stands haha. You have some very valid points (hope i saw some of those much much earlier in my smart home journey). That said, you sort of also just explained the happy path though.
- Make sure the RF band they are using is clear.
DONE! Many Times! 100%
- Flash the firmware to known good versions.
Also done couple of times
- Make sure they have plenty of routers.
I personally have about 2x routers than end units (currently about 60:30 ish), and routers are distributes well (lights in every room for example)
- Leave all routers on permanently.
...well, i mean, this is the point i was pointing out right? Self healing means you CAN turn it off. My whole point is that if you (or a guest) accidentally turn off a router unit, the network starts misbehaving. If you can ensure that ALL routers are on ALWAYS (100%), then if you have an unreliable network (can happen - has happened to me), you have other problems to look into (in my case, some cheap smart plugs were messing with the network - maybe dropping messages etc).
- Don't spam their Zigbee network with excessive traffic.
Well... Not really upto you right? If you have a misbehaving node, then theres no easy way to figure it out.
- Control the size of their Zigbee network - the larger the net the more devices that will be talking.
True, but whats the alternative? I could also not have a smart home, and walk to the light switch to turn it on/off like a normie - likely 100% reliable and 100% self healing
...I think the most important thing i learnt in my journey was that: buy only reliable zigbee devices (from good known brands). One bad actor can ruine your entire network.
Yes there is. Open Z2M> Settings> Health. Sort by the "Messages per sec" column. Anything above 1 is a spammy device
OMG thats so useful! TYTY
I should put back some of the misbehaving devices and test it out haha. Currently i have highest at 0.087. That said, i cam imagine anything like a smart plug that does energy monitoring etc (with devices that are always on plugged in) would be fairly spammy.
Would be nice be able to setup some alerting based on this so people dont have to actively go have a look at this list all the time.
So far I have not read anything that makes any guarantees about IMMEDIATE self healing meaning that: if there is a route Zigbee will heal the route so that devices continue to operate, but there isn't a specific time frame for that - read you might get some interruption after turning off a router while Zigbee heals.
Break your large network into smaller networks - ideally based on geographic node locations.
The general advice has always been to place the coordinator in the center of the network, so if you are breaking your network into two, ideally each half should be separated as much as possible from the other and each coordinator should be in the center of their half.
Note: I don't have this problem so this solution is theoretical to me - my network is small (50 devices) and dense enough that turning off a node doesn't cause a problem - however I leave all routers on anyway.
Its driven by the clients.
A well behaved child âpanicsâwhen itâs parent is unavailable for 20 min. and is supposed to start hunting a new parent.This is why a common reset step is shut your repeaters off for a half hourâŚ
So imagine the network clustered around the four or five devices sharing the parent. At 22:00.001âŚ
STORMâŚ
If your network is already saturatedn that could be enough to cause badness.
Some custom firmware files are available on GitHub. The one's I'm using are for Tuya-branded presence sensors and thermometers. I've got 60 devices. Half of my devices are Third Reality (their plugs make excellent repeaters) and the other half is a mish-mash of Aqara and Tuya brand. No Hue devices. I know they're quality products, but I see plenty of complaints about pairing and stability.
0.087 isn't too bad. That's like 1 message every 11.5s, so I wouldn't call it spammy at all. Having said that, my most chatty device is, indeed a smart plug. It's marked as 0.0296, so about a message every 33s. For me, that's the sweet spot between too much & too little reporting, so make of that what you will.
Why though? A device is either spammy or it's not. Just keep an eye out on that table for a couple of days after pairing a new device / updating firmware / changing reporting settings and you're done. I glance briefly at that page every few months & call it a day - don't need alerts set up because it doesn't really change.
I run a Zigbee switch and smart bulbs in my living/dining room â 6 KAJPLATS in 2 clusters of 3 off a single Zigbee switch. I did this mainly for dimming and to use the 2 clusters as separate lights. NO real issues, just a couple of clever automations to minimise any rapid turning on and turning off the lights. Also a delay, so any dimmer commands are held up for a minute when the switch is turned on. I get the occasional hiccup when one of the lights "sticks" on the wrong brightness, but it's very rare. Just switch off wait 10 seconds and turn back on
I'm in the UK, so all my Zigbee switches have no neutral, so they are not repeaters. However, the KAJPLATS are. This is the main issue with mixing Zigbee switches and Zigbee lights: the Zigbee mesh. Without a strong mesh, turning the power off on the bulbs can cause havoc. I have Zigbee main outlets throughout the house, which act as repeaters; this gives me a really solid mesh. The other trick is to turn off availability in Zigbee2MQTT This way the bulbs don't drop off the mesh when they lose power, and Home Assistant always thinks they are available.
Dang, that is bad luck. Hopefully you find some way to sort this out, replacing a bunch of switches is no fun for the wallet, but that might be one of the few ways around it.
It's times like this I wish we had more ability to make custom firmware for these Zigbee devices. Like how ESPHome gives us the ability to make custom firmware for arbitrary ESP32 devices. I know there's a handful of devices I'd give much different behavior if I could just figure out how to make custom firmware (including a bunch of Aqara stuff).
If it's any consolation, my current house hates anything Z-Wave, but Zigbee works really well. My Z-Wave setup sounds similar to your Zigbee setup. It works, but it's not "good." Sometimes things just don't work out with radio frequencies, despite what all the literature and theory says. I know you've moved, so I'm not suggesting you need a new house. Just saying that RF is weird sometimes.
What is "decoupled mode"? I have a lot of smart switches on all three main protocols, and should my Home Assistant server be offline every one of them is "Grandma friendly".
No, no and no. That completely defeats the purpose of having a Zigbee device. When you take a router out of the network, your whole Zigbee network is degraded while existing devices hunt for new routes.
In an open feild with no other electronic devices nearby, you might see 10 meters. Zigbee is very low power, much lower than WiFi on the same frequency band. When you have a mix of Zigbee and WiFi in the same building, the reliable Zigbee range is much lower. I have over 100 WiFi devices and almost 50 Zigbee devices working together in my house. What works for me is that no Zigbee device is more than three meters from a router.
Zigbee is self-healing, but the recovery time could be in hours. But it will repair itself.
This is how all of my Zigbee, Zwave and WiFi switches work. If you have a switch the requires connection to a hub or to the web just to function, throw it in the bin. I can turn off the host PC, unplug the router from the web and Grandma wouldn't know the difference. Everything just works as her 90 year old brain expects.
I buy cheap stuff. I have only had one out of 47 devices installed that didn't work out of the box.
Again- unless you absolutely can't live without color control - Smart switch + dumb bulbs will do everything that a smart bulb would do.