Lets back up here a bit on the diagnostics. The first thing that comes to mind based on some of your posts and even your zigbee table is that you have few good repeaters in your mesh. This should be expanded. Also the batteries being chewed up seem to be a sign of weak mesh.
Next I would definitely take your zigbee channel off of 13 and put it on either 15 or 20. This will ensure separating it from the default channels a lot of those routers use. (they can be a source of interference and knock stuff off)
As for the aqara sensors. They can be problematic because they don't quite follow the zigbee standard. Combined with a channel issue, these things can drive you insane. If you like them in general, the ikea repeaters apparently work well at stabilizing them. Personally I prefer the sonoff sensors.
Now at this point I'm only speculating because I don't know what you have mains powered wise in your mesh.
The most common preset wifi channel is 6. So Zigbee 13 is actually well outside this (and indeed just outside of OP's local channel 4 (*if not set to 40Mhz width). As OP hasn't done a scan of the neighbourhood it's hard to predict which zigbee channel will work best for them.
I actually use 13 (my wifi is set much higher up so no interference there. Interestingly my hue and Z2M networks are on 20/25 and these are rock solid too. I have made sure there 40Mhz is disabled on my APs too.
This is interesting. @aaiyar might be able to shed more light, but my experience that once a xiaomi routes in a certain path (when initially set up) they sure hate things being messed up like a router going out. (That's why all my Tuya repeaters are on battery back up - they will never fall off except in an extended outage).
@IllFatedIPA other than the mesh configuration others have mentioned and changing zigbee channels the only other thing I can suggest is to completely remove the Aqara devices (ie delete them from HE) and re-install them where they are to be used (not next to the hub). This will force a new route especially if you have previously disconnected other mains devices from the mesh).
The usual advice for forcing the mesh to reoptimise is to shutdown and turn off the hub (leaving the devices online) for 20 minutes and then turn back on. The mesh should stabilise in about a day. (TBH I'm not entirely convinced this works for Xiaomi stuff hence my first recommendation)
This is 100% correct. Xiaomi devices stick to a path (at least the older non-zigbee 3.0 ones do). I had the same concerns as you when I saw that @IllFatedIPA had unplugged a couple repeating devices.
The gateway uses the wifi gateway driver and the child device for the gateway and the WH31 both have Ecowitt RF Sensor as their drivers, which are grayed out so I can't change it. Are you saying there's another driver that I need? Because what I have seems to match what the directions say to use.
@kkossev Is there a page that isn't highly technical that I could read in order to actually understand what these terms mean? As far as the mesh goes, can zigbee router devices talk to the hub via other router devices? For example, the floor fan is in an intermediate position between the bulk of my living room and my bedroom where the hub is, so would all the other router devices be routing their own traffic through the floor fan? Or are they just trying to skip over that and send info directly to the hub?
@rlithgow1 Are the Gen 1 Iris plugs good repeaters? I have more of those that I can just plug into outlets for the sake of strengthening the mesh, but I assumed that since those are ancient technology that they would be finicky. Otherwise I can buy more S31 lites to leave plugged into random outlets. But as for the batteries being chewed up, I don't think I've indicated that they are - the batteries that I've tested out of the devices that kept falling off seemed to be fully charged. And the reason I have my zigbee mesh on 13 was an attempt to find a channel that the Aqaras liked, as one thread I found for troubleshooting these suggested that some people have better luck with certain channels that the Aqara hub uses anyway. I believe my channel was 20 when I first got it? But I can't say for sure. I'm not into buying more battery powered Sonoff stuff since the motion sensor gave me bunches of false readings and the button keeps falling off the mesh as well even though its in my bedroom no more than six feet from my hub with fresh batteries and the only nearby router device is the floor fan S31 lite that seems to be a good one.
@rocketwiz I can try that but ultimately I'm trying to use the Ecowitt devices since the sensors have the screens on them so I don't have to pull out my phone, and either way if they're so touchy that they can't deal with having to find a new path if I move a plug, then I don't want to use them anyway since that's just an extra hurdle I don't need. But since things seem to be continuuing to try to route through the Washing machine plug and 615 wakeup plug that aren't connected anymore, I will remove them fully from the hub and shut the hub down for 20 minutes to try to rule that one out. Ultimately doesn't seem that it will help why Ecowitt is only reporting certain things with any regularity, but I don't see that it will hurt anything else at all and will probably only fix issues with my mesh.
You've lost me with your reply. I think you are confusing the ecowitt device issues with the non connecting zigbee sensors, they are completely unrelated.
Ecowitt
First did you get the correct error feedback (in the hubitat system logs) when setting up the hubitat for the first time to receive data with no devices installed in HE for the weather station
If you get this periodic error (every 80s per your screenshot) in the logs (with no devices set up) then HE is correctly receiving data and the rest of the install should be pretty straightfoward.
Delete the HE devices and examine the system log to see if you get the above error message.
Zigbee Devices
First I assume that you have deleted the washing machine/615 plug devices from HE (and either reset the plugs or make sure they're unpowered - resetting will ensure they're no longer paired to HE.
Next delete the aqara/tuya sensors and reset them and re-pair them to HE (it is critically important that you first remove them completely as devices in HE). No need to shutdown the hub.
Give it some time and report back,
As I've mentioned measuring the no load voltage on a lithium cell by just hooking up a multimeter to it is insufficient to tell if it is depleted unless it's really dead. Hence the suggestion to swap out the battery anyway as a test.
Neighbor table's showing which repeaters the hub is tracking, but being in the table doesn't imply they have a good enough signal to make them viable. If they have poor connections to the hub they'll be designated as 'high cost' paths to allow them to be avoided when routing messages (repeaters exchange status info at 15 second intervals with the hub and other neighbor repeaters). The worst rank is 7, indicating links with weak signals/high error rates.
Devices showing outCost=0 failed to send valid status info for several check-in intervals, so the hub won't use them to send routed messages. This could be due to signal blockage or interference; if you can, experiment with relocating a repeater to see if you can get a few more neighbors showing low (but nonzero) inCost/OutCost numbers.
The fact that there are repeaters in the neighbor table that have been unplugged shouldn't have any affect on routing since the hub knows (from the zero outCost) that they aren't communicating. They would normally be 'evicted' from the table if there were more than 16 in-range repeaters and you'd see another repeater taking their slot.
If your battery devices have chosen a parent repeater that is a'poor' neighbor, it will need to route through another repeater with a better connection to the hub (as it stands you only have two viable neighbors showing up in your table: Floor Fan and Owl Lamp-- the only two devices showing up as 'via's in the route table enries).
Zigbee range in real world environments (attenuated by walls and furniture) is much less than the 'free air' number you might see quoted.
When the hub initializes its Zigbee network for the first time, its supposed to do an energy scan and pick the clearest channel. Before overriding its choice with one from a wifi scan tool, try adding or relocating repeaters first. One thing you won't know from a wifi scan is that there may be other unknown Zigbee networks in your vicinity that the hub has chosen its channel expressly to avoid. If you've got an Amazon Echo Show (even if they haven't been configured to use Zigbee), smart electric meters, or even cable box remotes they all can have actively broadcasting Zigbee networks. I found that I have 3 Zigbee networks in my vicinity besides my hub; one is on my Echo Show 10 (I don't use it and AFAIK it can't be turned off); other two are still mysteries.
True. I have a half dozen original Xiaomi round buttons; they require careful pairing through a compatible repeater or they'll drop off within an hour. However, with Markus's drivers (and the proper choice of parent when joined) they're trouble free on my network. Not saying that I recommend their use for someone not familiar with their mesh or these devices' special quirks....
Same here. I used to use ikea repeaters but have switched to Tuya and have never looked back. Also it helps that I moved all my plugs to zigbee2mqtt and lighting to hue or Z2M (tip: smartthings/aotec plugs don't work well with xiaomi stuff)
I use a couple of @iharyadi 's environment sensors for my Xiaomi's parent devices. Xiaomi's also work fine if you can get them paired directly as child devices of the hub (but its radio is so weak you literally need to shut off every other repeater nearby when you join them or they'll be attracted to it like a magnet). Initially it was a frustrating nightmare getting them stable but as long as they keep working it's more trouble to change my automations to swap in new ones.
@aaiyar Do you have recommendations? If they're simply okay but not the best, then I'd like to use them since I already have them for free from when my parent's had Iris. But if they'll actively make the mesh worse then I'll ditch them.
@rocketwiz I'm aware that these are two unrelated issues. I combined them in the topic I guess somewhat unwisely while I was typing this and having the blood pounding in my ears after having yet another "solution" to unreliable temp sensors fail me and waste my money.
I don't believe I actually checked for that error in the logs and I'm not seeing that if I scroll back. I may have installed the driver before waiting for that error message. So should I delete the Ecowitt devices from my Hubitat and start from the beginning?
I have completely removed the Washing machine and 615 wakeup plugs from HE. They've been unplugged for weeks now anyway, but now they're removed from the system altogether.
For the battery, I wasn't sure if the suggestion that the batteries were possibly dead was being attributed to me as opposed to a suggestion from someone else (which it was). If I can find another newer CR2023 kicking around then I will install it, but in the meantime, do you happen to know if there's a way to know if these "maxell asia product" cr2032 batteries are actually lithium vs something else? It doesn't say lithium on it, so I want to be sure before I spend money on more batteries when I literally have a sack of potentially good ones in the closet.
@Tony Does reliability of a router device indicate anything in regards to how it actually functions at passing along messages from other devices? For instance, the couch lamp that seems to be absolute garbage at passing along messages is pretty reliable when manually or automatically turned on and off. But I'm assuming that means that it's just doing its own thing while the HE hub knows it's a bad router? The corner where I have the Couch Lamp and Litter Night Light is in the farthest spot in my apartment from the hub, so that would make sense, and I can replace the S31 Litter Light with an Iris Gen 1 plug and move the S31 to a better position, but I'd like to hear @aaiyar 's take on the Iris situation before I plug something back in and make things worse.
Since you have them already, I would use them. I know they are Xiaomi-compatible. Just follow @Tony's recommendations on building your mesh carefully, ensuring that Xiaomi/Aqara devices are paired either directly to the hub, or through repeaters that are Xiaomi-compatible. And don't unplug repeating devices ......
Personally, I've gone the @rocketwiz route - all my repeaters are Xiaomi/Aqara-compatible USB Devices that are on backup power.
Since I'd like to use the Ecowitts more than the Aqaras at this point, I will try to redo the mesh later on. Would a valid plan of attack be
rearrange routers in a better manner to minimize distance while unplugged
plug in the compatible ones
shut the hub down again for 20 minutes to reset the zigbee mesh and give it a day to build itself
do a fresh pairing of the Aqaras while only compatible ones are plugged in
plug in the rest
Also not that I think I'm financially capable of going this route, but how are you supplying backup power to those repeaters? Just a bunch of UPSs scattered around the house?
If you're seeing your 'garbage' devices (based on their link stats) working reliably when individually controlled, I'd reckon it's because of retries (the lower levels of the protocol don't get logged ) that mitigate their poor connectivity. There would be extra latency you might measure but it must be pretty low if you're seeing good response. Still, Zigbee's smart enough to know that they shouldn't be given preference for routing multi-hop paths if there are alternatives available.
Not everything in the neighbor table needs to have great stats (since they won't get blindly used as routers), you just need enough neighbor routing devices to make the mesh perform acceptably well.
Don't quote me but this advice is probably relevant to huge networks (the 100K+ Zigbee device Aria hotel complex in Las Vegas comes to mind). You should see dramatic improvements in minutes for a relatively small mesh..
