There must be some 'magic number' of devices where the "cranky" route discovery requests becomes an issue. I'm a ZWave house and have barely dabbled in Zigbee. I have 5-6 Iris motion sensors, both v2 and v3, a Peanut plug for routing and a pair of Iris Power plugs for christmas lights, that have been powered off for weeks now. My initial foray into Zigbee was back in early November and I haven't detected any glitch. The very first v3 motion detector is stuck to a ceiling and I'd remember having to touch it
At the low end then, I have 7 active Zigbee devices, not exhibiting the "cranky" route discovery requests syndrome. 30 is enough, I think you're saying, to see it?
Until this is fixed, and I'm fearful its a Zigbee radio update * to fix, I'd like to parrot better than average responses to messages I see.
( * The SiLabs Ember 3581 used on our HUSBZB-1 Nortek supports both the Zigbee PRO Stack Profile (for thousands of devices) or the Zigbee Stack Profile coded onto/into the SOC. ** Implication being: the stack is on the stick, independent of where the stack (source) originated. )
I have 31 zigbee devices, and it's working fine, I think the problem is when you have more than X repeaters. I have now 5 Iris plugs, 2 ST, 1 Sylvania. Everything else are end devices.
I have a similar setup - mostly z-wave but some zigbee devices. The zigbee stuff has actually seemed a little faster/more responsive but that is likely due to the limited # of devices I have per hub. Certainly for my Yale lock it's been great. Z-wave has been a slightly different story - have had issues with slow or no responses from time to time. No show stoppers yet.
I have a rock solid zigbee network
I have 71 zigbee devices
Including 14 repeaters (12 ST outlets, 2 Tradfi outlets and 3 xbee devices)
I have a couple of new zigbee 'relay' type devices which I have found also route xiaomi well so will be adding these at some point to replace a couple of z-wave fibaro devices I want to relocate
The reason I included the 3 xbees is:
Office connected to pc for scans/mapping
Front of home to cover driveway and street (some distance so needed the power of the xbee)
Rear of home to cover my back garden for devices in my shed & garden
A lot of my devices (probably over 50% if you exclude the repeaters) are xiaomi
The only drop offs I get tend to be new xiaomi devices with bad batteries, as soon as I put new batteries in, they are solid.
I've just checked and, apart from the repeaters, I have only 8 zigbee devices connected directly to the hub
I've only got about 20 z-wave devices (including another 4 aeon hems soon to be added) and have no problems with the mesh there.
I will say that my usb sticks are not connected directly to the hub, they are on 5m extension cables to bring them to the centre (both vertically and horizontally) of my home.
(hub is in a comms cabinet.)
Based on my reseach so far, I would say the more repeaters you added the worse the mesh would have performed. 30 routing devices is not really very many.
Or lack of actually. When I had wireshark running I noticed the devices that with devices that were not responding the hub was not sending route discovery messages.
I'm not clear how much of the stack the hub manages vs the stick. I wasn't able to import devices into OpenHAB from the stick. OH docs mention you can do this if you're moving the stick to another system. I decided not to pursue any of the testing in OH due to the time-heavy process of setting up devices (i.e. search for device, add device, create item, link to device). I do know a couple people using OpenHAB and none of them have had any issues similar to mine, although in fairness, none of them have large systems.
Can you grab a few more routing devices? In my testing and feedback to Chuck it appears that at least in my case, 24 seems to be the upper limit before the behavior starts. I noted this on both hubs as I split my Zigbee mesh across them each.. That was the point when I started noticing random slow response times from the plugs.
Your XBees should be repeating if they're actively joined which brings you to 17 routers.
Ha!
If someone else wants to pay for them
I can't find a use for another 10 outlets and although it says 'Staff' next to my name, I buy all my own h/ware.
No freebies for me unfortunately
I bet it is 5.4... I think that is where all HUSBZB-1 combo sticks are at. At least I couldn't find any evidence on my SiLabs account/forums to the contrary. I don't have a Nortek account, though, so can't check there.
Another reason they may have to upgrade/change hardware... have to get at least to stack v5.7.x before it supports zigbee 3.0 (if I remember the versions right... Maybe it was 5.8)...
When nortek was asked about a year ago if they will support zigbee 3.0 on the husbzb-1, they wouldn't answer the question. That could mean they didn't know, or they knew but wouldn't say externally. So who knows.
I have a gut feeling that you are correct about that. Kinda makes me want to buy the Ember SDK and play around, but the boss would probably kill me at this point if I did.
The kit costs US$1,000.. It's not cheap. But then again, I remember paying about that much to get a license for the Faircom C-Tree database back in the mid-90's which I was still maintaining my Bulletin Board Software.
Now that I've dated myself, I better get back to work.
