So I'm in the same boat as a bunch of others here with Zigbee devices falling off the network since moving to the C8 from a C7.
There's a ton of threads on this topic, but I wanted to start a fresh one for some things I've noticed.
The most obvious is looking at the zigbee child and route info page, all of my LQI's are drastically lower than they were on the C7. I used to have two plug/repeaters that were always 255. Now they are both around 150-170.
I do see more devices connected directly to the hub than I used to though.
Also despite the much lower LQI's, in zigbee logging my overall RSSI is actually better. Maybe too good? (-37 on a bunch of devices).
I'm on the same channel I think I've been on (unless the migration tool changes it) and the default power of 16.
So my questions come down to this. Is it possible the increased radio output is too high and not letting the mesh do it's natural thing? I've seen some folks talk about lowering the power but there seems to be mixed results.
At this point I'm hesitant to hunt and poke at trying random things in fear of making it worse.
To your point there have been lots of discussions on this topic. I have been running mine at 8 for maybe 6 weeks, which is the power setting on the C7. The external antenna makes a difference, too. But I have noticed no difference between 8 or 16. I have also tried shutting down nearby APs in case the external antenna made the C8 more susceptible to interference - no change there either.
Yeah a good point. I think I'm at a place where I have tried just about everything anyone has suggested and I'm going to have to wait for Hubitat to find the problem and issue a fix. If they aren't able to, I'll migrate back to a C7 - though that would be painful.
As it is I have taken some critical sensors (both z-wave and zigbee) and moved them off the C8 back onto the C7. But I can only do so much before I have to start migrating more and more devices to make sure the mesh remains strong. At some tipping point running both the C7 and the C8 makes no sense.
Gotcha. I've double checked my wifi channels for overlap and all as well.
I must have missed those threads. My bad.
Have you seen the same with lower LQI's on the C8? I've already repurposed my C7 to the garage, but devices that are about the same distance to it have much higher LQI's than similar devices similar distances to the C8.
I wonder if it's a signal/noise level thing or just how those values are being calculated. Maybe it's part of the issue, maybe not but I wanted to bring it up.
Yes - my LQI numbers are significantly lower on the C8 and I think most folks are seeing the same thing. Whether it's a reporting issue on the C8, or over-inflated numbers on the C7, not quite sure. But it is a known issue. Incost and outcost, though, seem on par or better.
From what I can tell, the lqi is not accurate on previous hubs (C-5, C-7). I’m seeing much stronger rssi across all devices on the C-8 as well as incost/outcost numbers. I’m also seeing routes on the Zigbee route page that make sense to me, where it often didn’t on my C-5. I’m assuming this is the reason that my Sonoff dongles are now showing up routing for most of my end devices, which seems to have made a noticeable improvement in latency.
Here's a summary of the suspected reasons for the different/lower LQI's on the C8 from @Tony :
On the C-3, watching Zigbee logging for about 5 minutes shows virtually every LQI showing 255 (only a few outliers in the range 217-249). Yet the RSSI's asssociated with events showing LQI 255 ranged anywhere from -73 to -43, making the "LQI = 255" bucket virtually useless to differentiate stronger signals from weaker ones.
With the C-8, the LQI-RSSI correspondence seems to have been recalibrated; there's a much wider spread of LQI values showing up. It would make sense to do this; it would allow better path selection (RSSI's and resulting LQI from both ends of the link ultimately get mapped to 3-bit cost numbers that should give preference to less error prone links).
Lqi On the C8 is actually more accurate on the signal going to the router than the C5. I can see a lot more of what's going on by using Xbee's and things really don't look good signal wise for router signals coming back to the C8. Remember there's a route going to the repeaters and a route coming back. I just don't get why the C5/C7 can handle these weaker signals better from not so great repeaters.
This is how the C8 see's the signal to the Sonoff The Sonoff starts out under 200 these devices are in the same room about 10 feet max away from the C8 Xbee 3 devices nice signal both ways
I would get the P version of the dongles. The other version (chip that is in the C-8) is still listed as experimental and not recommended (no I can’t remember where I read that, even though it was yesterday, but it wasn’t on the Onion)
I have 5 of them, 3 on my C-8, and they are working fantastic on it, but just seem okay on the C-5. I’m wondering if it has to do with the way the C-5 reports lqi vs the C-8.
Also, I’m using the “Zigbee Repeater with Presence” driver on HPM set to bind (xbee setting). I had issues with them occasionally dropping off of my C-5s and set up a rule to notify me if it changed to “not present”. The driver will also attempt to reconnect the device if it doesn’t respond to the ping.
Aeotec Zigbee repeaters seem good for the people that don't want to mess around with Flashing the Sonoff dongle. I'm sure the Sonoff dongles are good so no point in testing just going to Flash them and start using.
Maybe too close the C8, so devices are just connecting to the hub instead as it's no more work? I have my SonOff dongles farther away from the hubs and one wall (minimum) in-between them. Biased closer to the devices...
