[Solved] Zigbee Instability is back!


#141

.117 and stable
Was and is Wierd behaviour to say the least.

Let's see how it is tomorrow morning when I'm back home.


#142

This hypothesis is invalid.

All the device protocols we're talking about here, ZWave, Zigbee and WiFi have a Collision Avoidance mechanism. They listen first, if the airwaves are in-use, they back off A RANDOM AMOUNT of time. Therefore, devices very quickly spread out across the time spectrum. The entire purpose is of course to avoid collisions. If two devices are both listening, both independently determine that the airwaves are clear, they collide and both will do the random delay.

The backoff doesn't count against a 'check-in'. It's an integral part of sending. It hasn't finished sending til it's done sending. Collision Avoidance is the beginning part of sending.


#143

So if you're stable on .117, then it doesn't make sense to point fingers at the platform. I would take a step back and look at the situation when you receive the Trådfri plugs. @gavincampbell commented on how impressed he was vs the specific (now discontinued) Sylvania plugs he had previously found to be compatible with Xiaomi. They seem to be a match made for each other. Who knows, maybe they're actually Xiaomi plugs. It wouldn't surprise me with the newly formed partnership between IKEA and Xiaomi


#144

Do you have any Zigbee bulbs connected directly to the hub?


#145

Thank you for the technical explanation.
Understood


#146

No all are going through HUE on a different zigbee mesh (channel 25). The only devices connected to the zigbee radio are Xiaomi.


#147

Too much over my head. I need to get different parts and program it? There isn't nothing that is just plug and play for a windows PC? :joy::rofl:


#148

How many devices on the Hubitat mesh? And how many on the Iris mesh?


#149

BACK TO SQUARE 1!!! ALL battery powered devices that I connected today have all dropped again. That's over 40 offline devices and 11 hours of work made useless. My entire 2nd floor mesh consists of 11 4th gen SmartThings plugs, 2 3rd Gen SmartThings plugs, and one 2nd gen plug. No Iris Smart Plugs at all on the 1st or 2nd floors. Those were all moved to a temporary build of the Iris system that I took out of storage today.

Here are the simple facts...

  1. Zigbee was stable on Iris v2 2015-2016.
  2. Zigbee was stable on SmartThings v2 2016-2018 (Same location)
  3. Zigbee was stable on SmartThings v3 2018 (Same location)
  4. Zigbee is trash on Hubitat. Battery powered devices drop with minutes. (Same location)
  5. Zigbee is stable on Iris v2 TODAY for 8 hours, including battery powered devices. (Same location)
  6. There are still 21 SmartPlugs and a couple battery powered devices on my temp Iris deployment. They're stable.

I have taken the following steps to stabilize...

  1. Wrote custom drivers for the SmartPlugs to reduce energy reporting and also give add the ability to re-bind cluster 0x6.
  2. Purchased 3 XBees (2 Xbee3, and 1 S2C), antennas and carrier boards. All are online.
  3. Nuked the entire Zigbee mesh to start over. Ineffective
  4. Purchased 5 Iris Contact sensors, retail new to try to replace sensors dropping offline. That too was ineffective.
  5. Purchased 3 Iris motion sensors, again to try to replace devices that go offline or won't pair. Ineffective as well.
  6. Shut hub down for an hour on 4 seperate occasions to rebuild mesh. Ineffective.
  7. Re-connected Iris V2 and built up a mesh with all Iris plugs on the 2nd floor, plus some of the troublesome battery powered devices that were removed before. Mesh stable (channel 24) and no devices dropped after 8 hours.
  8. Purchased and installed 11 SmartThings 4th generation SmartPlugs. Completely ineffective.

I am purely out of ideas. Support keeps insisting I have a mesh issue. But I have to be honest, my own test scenarios pretty much prove that is not the case. As I have been saying since almost the beginning, this is a Hubitat issue.

I have decided that tomorrow will be my last day with Hubitat. Support has no ideas on what to do, I've expended hundreds of hours and even more cash on a solution that isn't happening. I know I can move back to SmartThings, have a stable mesh, but deal with their cloud issues.

But for overall stability, Iris and SmartThings have both been proven to be superior.


#150

There's a stick, but it's more expensive than the Xbee and it's just going to show you a map of the network. The Xbee doubles as a really strong repeater for everything Zigbee (including Xiaomi). Of course, I don't use mine as a repeater. It's not even plugged in right now. I only used it as a tool for viewing the network map and signal strengths.

Programming it is not hard. You enter some parameters (all documented for both Windows and Mac) and you're done. You can just save that configuration to your computer and then you start using it. Pretty simple actually.


#151

Could you answer my question about device counts?


#152

On the second floor, 14 SmartPlugs, and about 50is battery powered devices are currently on Hubitat. Nothing is on the first floor, everything else is in the basement.

Iris has 21 plugs connected at the moment, and 7 battery powered devices which were previously on Hubitat but absolutely refused to reconnect after pulling the batteries.


#153

I really understand your pain, all the effort and money you have spent is really impressive.

You have the largest setup I ever seen. I wish there was a second USB Zigbee radio that you could try, but then HE will not recognize it without support adding the code for it. So you are on a pickle.

I also can't imagine how is a mesh issue, but I'm not technical enough to have an opinion either.

Good luck


#154

Thanks! I know SmartThings can handle it. I moved the entire mesh in October when my V2 hub failed. I have absolutely no doubts that I can move the mesh back to my v3 hub and have it be stable. Why? Because I did exactly that 2 months ago.


#155

You case is so anomalous. No one else has reported problems like you have, and there are a lot of zigbee devices running on Hubitat. So, the thought that comes to mind is this: We know for a fact that Nortek, the stick manufacturer, had a batch of bad Zigbee radios. We were unaware of this until we got a few customers telling us about bad Zigbee radios. Then we undertook to test every stick, and discovered that in one batch of 1000 sticks, there were about 25 bad ones. A more recent batch has had no test failures. Nortek fessed up to this issue.

So, I wonder if you have a bad stick.


#156

What kind of testing can be done to vette that. I'm open to that. However the return period for the hub runs out on Jan 29. Putting the DWs near constant bitching aside for a minute, that's really got to be my break point.


#157

A bad USB Zigbee stick ??

Wouldn't that be the likely suspect if Hubitat Apps aren't crashing, your custom drivers aren't crashing and Zwave devices aren't crashing out ??


#158

Don't worry about a return period.

I'd be more than happy to send you a new stick for you to test with, and even a hub for that matter. I seriously don't think the problem is with our zigbee stack, as you do, because other users would have encountered problems like this. It agonizes me that you've been driven to such an extremely bad view of this, as we have ample evidence to the contrary from all other users, including our own personal use of the system.

So I'd propose that I send you these devices tomorrow, and that you do one of the experiments you did today. See if you can put the 30 or 40 devices you put on Iris on a different Hubitat system, and see what happens.


#159

This should have been easy identifies by your supplier.

Such an error ratio from a Mft is really bad. That's at least 25/1000. They would fail sigma 5 and 6 validations.

The threshold acceptable are: 233 per million and 3.4 per million respectively


#160

Nice one. @bravenel