Looks terrible to me I guarantee your C5 isn't reporting Lqi like that. Do you know how far away your repeaters are from the C8? When people start telling me Lqi doesn't matter I start to worry, especially when the folks that make the hub tell me my signals are trash. how would the Hub ever know which repeaters to route through if they're all terrible. If I was to report your zigbee logs and said I was running a C5 and said I'm having things drop off. People would be like wow man your mesh sucks Better add 500 more repeaters.
2 Likes
Thanks. Confirmed what I thought. I have several repeaters (at a guess at least 10 to 15 or so) consisting of Ikea Zigbee plugs and dedicated repeaters. Within the room with the C8 there are 2 repeaters (probably about 20 feet away from the C8), with the balance spread out throughout the house. For a test, I'll put a repeater about a foot away from the hub to see if that improves anything.
1 Like
Yeah, unless your hubs in a bank vault I'm pretty sure you should be seeing better signal then that.
Zigbee doesn't directly use LQI to select routing paths, it uses 'cost' numbers (derived from LQI) for each link segment (they're additive for multi-hop routes); also links get evaluated by reception and transmission quality, not just reception; LQI figure indicates reception only so you need more...
So the hub and routers rank the neighbor links they track. For each, RSSI gets derived from radio receiver, algorithm generates LQI from RSSI, algo maps that into inCost (3 bits representing a number from 1-7) from LQI.
Four times a minute, it includes those 3 bits in the link status message it sends to the router at the other end of the link. At that end, Zigbee router's doing the same thing; the hub gets those 3 bits with every link status and uses them as the outCost figure for the link. So now both hub and router have metrics for reception and transmission (how well they are hearing, and being heard). Routing algorithm chooses best path (if more than one exists) according to costs for path segments.
I made a sketch (forgive me):
So the RSSI gets mapped to a 8-bit number (LQI) that gets mapped to a 3-bit number (cost); if the algorithm is working right, it will generate a cost of '1' for when it gets a signal with low probability of being received with a bad frame check sequence (indicator of corruption). Why bother mapping 8 bits into 3 bits? Probably because they needed to shoehorn it into the link status message and didn't have bits to spare (and you cut out transmit time and power by shaving bits; Zigbee green power takes that to the extreme).
If all these low LQI's were indicating terrible reception, I'd expect to see high inCost and outCost numbers (reception at both ends of the link matters). I just have a little C-8 test setup to play with; I do see LQI numbers that are quite a bit lower than expected, but the inCost/outCost numbers look normal, like I would expect to see on my C-3.
If I had to guess, I'd say it's likely they (they = Silicon Labs) made the LQI calculation more granular. On C-3 it was pretty much all or nothing; LQI's anywhere from 240's to 255 mapped to 1 or 2, you needed to get to 230's before you saw a 3. Anything below 200 was toast (4 or worse). Using less of the range means less discrimination (half the class gets an 'A', quarter gets a 'B', quarter an 'F').
Can't say for sure since I can't tell how the devices now on my C-3 would run with the C-8 yet. But bad cost numbers would correlate with bad signals....
9 Likes
The way the C-5 reported lqi never matched (or was even close to) what my routers reported. It was pie in the sky numbers and I certainly wouldn't compare the numbers between the 2 hubs as they are obviously not calculated the same. The C-8 numbers seem much more realistic.
2 Likes
I'm not just talking Signal numbers though. I'm talking RL performance. Quite honestly I don't care what the numbers are. When I have devices that won't connect at all from a few rooms away but then I turn the C5 on and I can connect them with no other repeaters there's definitely A problem. I'm talking anywhere in my house and outside I can connect these same devices to the C5. Like I said when I get the replacement hub I'll share my Xbee's data.
my single hue outdoor ver 2 that i was having trouble with has now stayed on for 24 hours on .113 fingers crossed..
Just lost an ST water leak sensor that had been stable. Does not look like .113 has helped.
Most of the routers are through multiple walls. Sonoff repeater 3 is around 30 feet away on the other side of a 4 ft wide double refrigerator and a wall mounted in the garage. Sonoff repeater 1 is 3 feet from the hub. Sonoff repeater 2 is through one wall and 25 feet away. Toilet is an in-wall GE dimmer one level up. Kenny’s Bedroom Samsung Outlet is ~40-50 feet from the hub, upstairs and above the garage.
Also, all 4 Hue outdoor motion sensors and 3 Hue indoor motion sensors have been solid since I moved them back from my Hue bridges. I’m using device activity check to monitor them every 2 hours. My other 86 Zigbee devices on the C-8 have been solid as well. Except for having to rejoin one Iris V2 after migrating, I haven’t had a single issue.
1 Like
Are you having any issues? Those Signals look good to me nothing like I'm getting on my particular C8. Like I said I didn't even want a replacement until I talked with the hubitat guys on the phone. I'm having performance issues and the signal numbers to back the problems. Want to help take a Sonoff outlet place it in the same room tell me what Signal numbers you get. I'm not confident I'll see an improvement when I get the replacement.
1 Like
Zero.
I don’t have any on this hub and am not a fan of the 2 I have on my other hub, only the Sonoff Zigbee 3.0 dongles flashed with router firmware. The Samsung outlet kitchen is around 15 feet away behind a pantry door.
ok
I'm not either, i just seen you had some sonoff stuff, Your Zigbee logs actually prove my point. One room over I can't even get a good signal on any device except Xbee's and those are way overkill to compare to any normal repeating devices.
1 Like
The Sonoff Zigbee 3.0 dongles were so worth every penny. Not so much on the C-5, but they really shine on the C-8
1 Like
They are your friend. 
1 Like
Can you clarify?
You had the motion sensors on the C8 then moved them to the Hue hub?
You have recently moved them back from the Hue hub to the C8 and all is good?
I had them on the C-5 and moved them to the Hue hub because a few of them would occasionally disconnect. The C-8 has been so good that I decided to move them after .110 was released. I have had no problems.
1 Like
I moved most Hue motion to the Hue hub and are working well for me through HA. This approach though is a step backwards. Pre-Hubitat I had devices, apps, etc. everywhere which was super hard to maintain. Since Hubitat I have moved almost everything off of other platforms. So.... moving motion sensors to the Hue hub AND HA is not desirable. I'll keep watching this thread and maybe move them back to Hubitat when suitable.
1 Like
I always found Hue sensors and button controllers to be more difficult to keep connected than any of my other devices.
I do have 3 Sonoff Zigbee 3.0 dongles (flashed with router firmware) connected to the C-8 that are turning out to be phenomenal repeaters. Every Hue device that shows up on the routing table has been repeating through one of them. I have one 3 feet from the hub and the other 2 on opposite ends of the house.
