What do you mean when you say your Zigbee mesh looks weird?
1 Like
I mean that switches that are on the other end of the house in the basement show up as direct neighbors in the Zigbee table while switches right next to the hub don’t. I’ve tried powering off and unplugging the hub for several hours but that didn’t change anything.
Do you have anything else repeating besides the GE that could be causing issues? Also, how many repeaters are within 10 feet of the hub and not separated by a wall? I’ve found that I need to have several repeaters that close for everything to work reliably. Maybe there is a distance factor at play here.
I have one switch in the room with the hub. (My office) Directly on the other side of that same open door wall is a bank of 3 more and across that hallway from the three are another three. So there should be plenty within 10-15 feet of the hub.
Any bulbs in the mesh?
Three sengled bulbs which Hubitat support say work ok and do not try to act as repeaters.
Yeah, they're good. Any repeaters?
Just the jasco/Ge switches. There are so many of them, I didn’t feel the need to buy and dedicated repeaters.
I would get a long ethernet cable and try the hub in other areas. Also, how does the lqi look for the groups of switches? I split mine between my two hubs partially because the lqi was low when they were on the same channel.
Good thoughts Ken. I have tried moving the hub to 6 different spots around my house, including the main entry area by the front door which is the dead center area of the house and has about 9 switches within 15 feet of it. LQI numbers are excellent. I have been pouring over the Hubitat wiki docs and haven't found anything that can explain this yet. Even the RSSI numbers look solid.
1 Like
That really doesn't mean anything. In fact it might be slightly desirable that things are actually meshing and not only talking direct to hub. Mine does exactly what you describe, and is very stable.
Is this all you have? In other words do you have other devices other than the GE and Sengled? Not necessarily powered or repeaters, just Zigbee devices in general.
Did you try any other channels, 15 often is a good one.
Well to be honest, this isn't the same setup. It might be the same devices, but you are in a totally different environment, with totally different layout and spacing, and a bunch of other differences.
It could even be bad wiring in the house, dirty power, a switch that "died" upon moving it, heat ducts reflecting a signal, wire mesh in plaster walls, foil faced insulation, metal siding/roofing, or a bunch of other variables that are nearly impossible to compare the new house to your old house.
I know that isn't helpful in some sense, but you cannot assume that everything is "exactly the same". It is far from it.
And one more question that nobody else brought up.
Can you describe how you moved these devices and the hub? You mentioned you factory reset at some point, but did you originally move the hub and everything contained on it, and just paired the switches again, or did you factory reset every switch and pair fresh, or did you just power the switches up and they popped back into their old spot on the hub?
Probably doesn't make a difference, but maybe it will reveal something.
1 Like
The hub moved from the old house to the new one and was factory reset to start fresh in the new location. I factory reset it a second time out of frustration when nothing was working. All the switches are brand new and paired fine originally. When I did the second factory reset of the hub, I also at the time did a factory reset of the switches. Also, every time they completely stop communicating, I would go into zigbee discovery mode and do a full factory reset of the switches each time.
So, you left the original switches at the old house? Were they the last gen switches, 45857 & 45856? I only ask because there was another thread on here a few months ago with someone having similar issues with a large install of the newer model GE/Jasco Zigbee dimmers and switches.
1 Like
I know that this suggestion means a lot of work, but, if you've tried everything else and can't get it working, then this may be an approach for you.
This approach will show you how each device is working.
- From your hub outwards, do this procedure, one device at a time.
- For a switch, open up the device details page.
- Open up another tab in your browser with Settings -> Zigbee details -> Push the Zigbee logging button.
- On your switch's device details page, turn on the device. Immediately, go to the other tab that you have opened with the Zigbee logging. You should see something like this:
Make special note of the lastHopRssi. If it isn't between -65 and -75, you may have an issue with that device. - Repeat step 4 for each device.
From the late, great, Patrick Stuart:
Last two numbers are the most useful.
The lqi is an indication of interference or quality of hop between device and nearest router.
The rssi is the signal strength. Closer to 0 the better as that means less power was used to transmit.
Usage is to track down a zigbee endpoint device that is dropping off due to interference or weak signal.
Or to find sources of interference or range or a sensor.
Keep in mind ZigBee does attempt to optimize itself automatically. So a single response isn't worth acting on, its the averages.
Moving a device might result in better signal, but it could optimize through a bad router and lqi could decrease in quality.
What this doesn't show is the hops between the nearest routers and the hub for each device. So its not the entire picture.
For most, its not even needed to look at or monitor. But it is handy for troubleshooting edge ZigBee devices that might be having issues on the mesh.
In almost every situation, adding repeaters / routers will solve the problem. It is a mesh after all.
1 Like
Also worth noting that LQI shown in the neighbor table only tells half the story: how well the hub hears the neighbors. Links have to be good bidirectionally, and you can't tell how well the hub is being 'heard by' the neighbor routers from this figure.
The link cost figures shown there are key, and ultimately the only numbers that are used by the Zigbee routing strategy (it adds up link costs of each hop, not LQI's, to determine preferred routes). Hence you'll often see LQI's of 255 even though links are completely broken and unusable. Those are the ones showing zero outCost and age 7; meaning the neighbor hasn't sent a link status update in more than 6 update intervals, because its either dead or incapable of hearing the hub.
I'm curious to know what your neighbor table looks like when all your Zigbee devices all stop communicating. If there truly is no communication at the link level, your neighbor table would be empty (all neigbors evicted because their outCosts stayed 0 with age 7) and no eligible routers took their place. If you refresh it (every 15 seconds or so) and see age counters change or other neighbors appear in the table, things are happening, though not necessarily stable.
One other thought; have you always manually set the Zigbee channel? If so, you might try letting the hub pick its own channel instead. The hub should do an 'energy scan' when it initially forms the network based on what it sees as the quietest channel (at least, at that point in time). Maybe it is seeing someting in the spectrum that might not be otherwise apparent.
1 Like
I’m having exactly the same issue. Did you get this resolved by any chance?
This thread has been dead 2 years. You should start a new one describing your symptoms, what brand/model of sensor they are, what you've tried and what platform you are on.
1 Like
I never got it working. Matter of fact. I ended up moving. I haven’t bothered to add any new switches in the new house and the Hubitat is sitting in a box in my office. I gave up. Until it’s Jess of a science project and more of a true commercial product I’m done for awhile.
I think they did have another thread they started, but it sorta died out. They also posted a couple other places.
@tmatei85 it probably would be better to try and keep things in one thread to reduce confusion for everyone. It is hard to tell what has been tried otherwise.
