My Zigbee Experiment-Ongoing

Not happy with all the zigbee radio reboots with attendant ST Arrival sensors mis-operations and 3 of my Centralite 4200C repeaters going off line, I thought I'd do an experiment. The experiment has just begun.

I unplugged all 9 of my mains-powered Centralite 4200C plugs, which are the only repeaters I have. These are the only mains-powered zigbee devices I have.

Maybe 4 had a lamp plugged into them, and the rest for Christmas lights, but nothing that ZEN04 plugs couldn't replace.

I have 16 battery-powered zigbee devices:

9 Indoor Hue Motions
1 Outdoor Hue Motion
3 Xfinity Contact Sensors
1 Third Reality Temp/Humidity Sensor
2 SmartThings Arrival Sensors, V4

Z-wave includes plenty of mains powered switches, some motions, garage/gate relays, and some plugs.

I walked around the place and, so far, everything seems to work!

I thought it was cool at how quickly it recovered. Much more resilient than z-wave would have been, I believe, but alas, not so much lately.

The Child/Route table show all 16 of the battery powered zigbee devices accounted for.
The Neighbor table shows outCosts of 0 for the 9 mains-powered repeaters, which makes sense since they are all unplugged.

Zigbee radio power is set at 20.

We'll see how the zigbee radio reboot thing goes now.

32 is the max number of devices without a repeater, correct?


32 child devices per hub is the number that we've seen quoted before.


Did you just unplug them, or removed them from your devices, then unplugged? If you simply unplugged without actually removing them, it won't be long before your devices will stop working.

That is correct, but on paper. In reality, the rule of thumb is to have no more than 6 battery powered devices/repeater in order to achieve a strong mesh that is resilient to potential interference from other devices on the 2.4GHz band.

Why should that be? The devices should join their new parent without any issues.


Yes. I initially just unplugged them.

Then, I went back and plugged them back in and deleted them.
I still have a couple in the route table though.

With the configuration you have now (all devices C-8 children) you can get a good idea of their RSSI whenever they transmit via the Zigbee logging page ( the last hop RSSI is guaranteed to be from the device itself, rather than a repeater). How do the RSSI's look?

I just cranked it up. Get back to you a little later.

You'll want to keep an eye on the devices since you can't directly affect their transmit power (they'll just do more retries that would impact battery life). Anything more negative than -80's will be on the fringe, signal strength wise and bears keeping an eye on... that said, you might be pleasantly surprised. I've got an Iris V2 contact sensor in my mailbox about 80' from the parent repeater it picked in the middle of my living room), going through aluminum siding as well. No issues with battery life; never misses a beat winter or summer in the NE.

I guess they'll all bear watching, lol.

Those all could be OK except for dev 325....

Not a complete list.
May try restarting hub to get rid of neighbor nulls.
Plus have to do some brake work on the car. I'm procrastinating here, lol.

Oh, the higher negative number the worse. I get it now. Hadn't looked at this log for a while.

That device is a contact sensor in the package bin at the end of the driveway. Reception is by definition at the fringe, but it's been pretty reliable even so.

1 Like

Doubt you'll see any functional impact from those.

It's not a good look, lol.

Well this will be an interesting experiment. Keep us informed.


I did a shutdown/power cycle/zigbee rebuild. Cleaned up the neighbor table.

It took a little while for an outlying sensor or two to report in, but they all have, save for my wife's car's arrival sensor, which is not here.

I moved the 3R temp/humidity sensor in the basement a bit, and that seemed to get it going again. But, it continues to be its chatty self now in its original position, so who knows.

Thankfully, I did not have to go through any extraordinary measures, like setting the hub on wifi (untried by me so far), powering with a battery, and dragging it out to the end of the driveway, so I wouldn't have to dismount a sensor or two.


Oh, and there's this:


Neat. The only rationale I can think of for beefing up the hub's Zigbee radio (as was done with the C-8) is to permit exactly that... allow removal of unnecessary repeaters. If you don't need them for their end device capacity (or range extension, still a TBD in your experiment) they just add link status traffic not to mention route discovery broadcasts throughout the mesh...

1 Like

So far, so good.

1 Like

Hey, check it out. I'm one of the cool guys now.

edit: I updated to


Download the Hubitat app