Best Practices for Wireless Hub Placement

[upstatemike]

The way things are evolving it looks like the future of home automation is likely going to involve a series of interconnected hubs of one sort or another and I'm wondering what the best practices are (or are going to be) for locating all of these hubs. I am speaking specifically of hubs that involve radios such as Lutron Caseta (or Radio RA2), Hue, Homeseer Z-Net, Elk Wireless Interface, Smartthings (or Hubitat), Wi-Fi access Point, and so on. Unlike the gear tucked away in a remote wiring/equipment closet like an ISY, Homeseer server, Blue Iris, Stargate, RasPi stuff, etc., these radio hubs need to be centrally located in the house and I'm starting to think a separate "hub closet" that is distinct from the traditional wiring closet may be warranted. 
 
I wonder if it makes sense to carve out a corner of say the front hall closet to mount all of these radio based hubs and provide them with a small UPS and network switch to feed them? Is Home Automation going to demand a two closet approach going forward to better accommodate all of the the wireless equipment involved?

 

[wkearney99]

Several structured box makers now have RF transparent (plastic) in-wall boxes.  For gear that "needs" to be centrally located one of those would likely be ideal.  But other stuff like servers and gateways probably don't all need to be crammed into one box.  

 

[wuench]

If it's a mesh protocol you shouldn't have to resort to this, zigbee, zwave, wifi (if you do a mesh).    I have very limited zwave but put a lot of aoetec repeaters around to compensate, mostly behind couches etc.   My ISY is in a metal enclosure but I used an extension cable and mounted the antenna outside the enclosure.    Wasn't sure it would work out but it seems to be working ok so far.
 
I kind of expect it to go in the other direction actually as wireless protocols advance.     I just don't see consumers going through the effort of a site survey.  Who knows eventually everything may just have a cell modem built in, I am sure the carriers would love that

 

[wkearney99]

If it's a mesh protocol you shouldn't have to resort to this...

 
Ah, hahahahaahaha.... oh, you have GOT to be kidding, right?  That was a good morning laugh.. oh, it's after lunch... but still, LOL.

"Shouldn't" and RF have never worked anywhere near like what the marketing materials claim.

I get it, you're largely correct, the rationale behind mesh should make it possible to implement without undue hassles, but reality shows that often not to be the case.  But look at what you've had to do, putting repeaters around and hanging an antenna out of the enclosure.  With an FRP or nylon box (and door) you're at least going to be able to avoid having to route anything outside of it.  You're still at the mercy of RF repeaters, which are their own set of hassles (like guests or cleaning people unplugging them for something else).

 

[upstatemike]

So I want to make a distinction between best practices and "what will work". I'm sure for retrofit and add-on situations you can depend on repeaters to get by but I think best practices, assuming you are starting fresh and can do whatever is best, would suggest centralizing RF as much as possible and then relying on extenders only to the extent necessary to cover the fringe areas and dead spots.
 
Some concerns I have would include:
 
Standards in Zigbee are erratic and not all brands/devices can be counted on to act as repeaters.
 
The routing tables for protocols like Z-Wave and Zigbee are sensitive to any device failures or location changes. To get maximum reliability with minimum maintenance it seems that configuring for as much direct to hub communication as possible would be a good thing.
 
Things like Lutron or Elk Wireless do not create a mesh network and depend on additional hubs for coverage so again starting out with a centralized design to maximize what is covered without additional hardware would be beneficial.

 

[ano]

Standards in Zigbee are erratic and not all brands/devices can be counted on to act as repeaters.

No Zigbee standards are open and very well documented. Z-Wave, on the other hand has proprietary standards controlled by one company.
 
But YES, not every device can act as a repeater. Every device falls into one of three types: an end device, a router node, and a gateway node/coordinator. (There is a Trust Center, but you can ignore that for now.)
 
A battery device is an end device, and it can not repeat. An example of this is a door lock, where long battery life is required. These are lower-power devices and need to be placed relatively close to a router or gateway.
 
The router node is the heart of Zigbee. It not only stores and forwards changes for end device notes, but also acts as a repeater. There is no limit on the number of nodes it repeats, but there IS a limit on the number of end device nodes it can support. (I think the number is 10)  Any plug-in device is a router. This is light bulbs, switches, just about anything.
 
Coordinators also can act as repeaters, but they also only communicate with a limited number of routers (or end devices). And there can only be ONE coordinator in a network.
 
So what does all this mean. Just be careful with BATTERY Zigbee devices. You should make sure a 120v Zigbee device is not far away from each. Otherwise, the network takes care of itself.
 
Just be VERY careful of say 5 Zigbee locks on your doors but NO other Zigbee devices. That is asking for trouble.

 

[upstatemike]

My understanding of Zigbee repeaters:
 
Devices must be using ZHA profile and then they can repeat traffic from other ZHA devices. Not sure they can repeat at all if using the ZLL profile.
 
Some brands (such as Osram) have buffers that are too small to reliably repeat traffic. They show up as repeaters on a Zigbee map but the messages just don't get through.
 
GE Link lights have a known firmware bug that causes them to drop off the network every few weeks so they are not reliable repeaters.
 
I have also heard that some Zigbee brands have not implemented the full protocol and cannot act as repeaters.

 

[wuench]

Ah, hahahahaahaha.... oh, you have GOT to be kidding, right?  That was a good morning laugh.. oh, it's after lunch... but still, LOL.

"Shouldn't" and RF have never worked anywhere near like what the marketing materials claim.

I get it, you're largely correct, the rationale behind mesh should make it possible to implement without undue hassles, but reality shows that often not to be the case.  But look at what you've had to do, putting repeaters around and hanging an antenna out of the enclosure.  With an FRP or nylon box (and door) you're at least going to be able to avoid having to route anything outside of it.  You're still at the mercy of RF repeaters, which are their own set of hassles (like guests or cleaning people unplugging them for something else).

 
Well you've got a boat, pretty sure most of this stuff won't float   So there's always that solution...

 

[wkearney99]

Trouble is most of the stuff isn't even heavy enough to qualify for 'boat anchor' status.  
 
Don't get me wrong, it ought to work, but implementations by various vendors have fallen well short of expectations.  Expectations from both vendor and user points of view.  The vendors expected they'd be the only ones to sell the "whole solution" to the users.  This often led to naive implementations or outright attempts to black/whitelist only certain devices.  The users expected bits and pieces from various vendors to actually live up to being compatible since they claimed use of "standards".
 

 

[ano]

My understanding of Zigbee repeaters:
 
Devices must be using ZHA profile and then they can repeat traffic from other ZHA devices. Not sure they can repeat at all if using the ZLL profile.
 
Some brands (such as Osram) have buffers that are too small to reliably repeat traffic. They show up as repeaters on a Zigbee map but the messages just don't get through.
 
GE Link lights have a known firmware bug that causes them to drop off the network every few weeks so they are not reliable repeaters.
 
I have also heard that some Zigbee brands have not implemented the full protocol and cannot act as repeaters.

It is true that Zigbee v2 Light Link devices do not repeat Zigbee v2 Home Automation signals.  But do consider Zigee LL is a quite specialized beast for colored bulbs, most specifically Philips Hue.  I have many types of Zigbee bulbs in my house, mainly Cree and GE Link and they are not Zigbee LL and do repeat.  In fact I have 12 GE link bulbs that ALWAYS work.  Remember Zigbee devices, even bulbs, can be upgraded over-the-air automatically. Honestly I have no idea if they ever upgraded, or if they repeat but I can tell you, over many years they ALWAYS work. They turn on and off automatically every day, and I can't think of one instance where they didn't work, ever.
 
Now above I say Zigbee v2, because there is also Zigbee v3, which does away with the term "Light Link" and "Home Automation."  ALL Zigbee v3 devices can communicate with any other. Basically all the Zigbee protocols are combined. Zigbee v3 devices are just appearing, which yes I know doesn't help you if you don't have them. (Zigbee v2 is backward compatible with Zigbee v3)
 
Here is more info on Zigbee v3 if interested. https://www.zigbee.org/tag/zigbee-3-0/

 

[upstatemike]

Wow that article is 5 years old! I guess it took awhile to translate the new standard into actual products.
 
I am currently using a lot of HUE stuff for areas where motion sensors are needed or as a cheap way to update rooms where the fixtures are not controlled by wall switches, just little tiny switches built directly into wall fixtures or ceiling fixtures with pull chains. My only real gripe with HUE is they all come on every time the power flickers which becomes more of a PITA as I expand... especially in bedrooms. Strengths with HUE are their stick up switches and candelabra bulbs which let me easily address situations that other product families can't handle.
 
Outside I am using Lifx Wi-Fi bulbs and using other technology switches to switch them on and off. The idea is I can set a color scene and then the bulbs will hold it through on/off cycles at the switch. I only have to worry about flaky communication when I actually want to change to a different scene. I have 13 so far and will probably top out at 20. I don't have enough Zigbee in the house to extend signal to all parts of the outside perimeter but there is plenty of Wi-Fi coverage out there. Once I get them all working I plan to use Homeseer to turn them all on and set them to strobe if the Fire Alarm goes off.

 

[upstatemike]

Am now thinking of adding ConBee/deCONZ to Homeseer (one more Hub for the RF hub closet) but I am concerned that most of the Smartthings gadgets like water detector and button are not yet supported. Am I just looking at out of date information or are these really not an option yet?

 

[ano]

Wow that article is 5 years old! I guess it took awhile to translate the new standard into actual products.

Yup, it takes a long time.
 
By the way, Philips has announced it will support Zigbee 3...
https://developers.meethue.com/zigbee-3-0-support-in-hue-ecosystem/
 
LIFX bulbs are quite interesting. I use those also.  So the endpoint of the bulb speaks W-Fi, but between bulbs it uses 802.15.4 which is the physical layer that Zigbee uses.

 

[Dean Roddey]

If Zigbee 3 starts getting some traction, I'd consider supporting it in CQC. Zigbee is something that's much needed in the  HA world, and it's really sad that it's been caught in chicken and egg state forever. Hopefully the new scheme will change that.
 
But, as always, it's sort of an issue of systems vs gadgets. Most of the folks out there aren't doing serious automation, they are living in the IoT's world (whatever that is, I define it as low end mostly.) The companies making those things have no concept of systems generally, they are just selling things. The only existing backbone that they know will be there is Wifi, so they just target that, horrible a choice as it typically is. And of course most of them look no further than a phone app, and most of them probably want to tie you to their cloud service as well, and all those things lead to Wifi.
 
So it's a really bad chicken and egg holding pattern. Until it's almost a given that a Zigbee compatible system will be in place, all these folks who don't sell systems will not tend to support it. And the folks who sell systems aren't going to be too happy letting other people make money off their backs so they'll implement proprietary Zigbee systems.
 
 
If folks like Hue support non-proprietary Zigbee 3 devices, that'll be the sort of step that's required to get the cycle moving upwards, so that there's a reasonably broad market to target. If that means there are non-proprietary devices that people have and want to use, then there's a much more reasonable argument for folks like us to do the hard work required to support Zigbee 3 in order to accommodate them.
 
I'd love nothing more than for Z-Wave to die off quickly, and Zigbee is the only likely virus that could kill it.

 

[upstatemike]

I don't know the history of Zigbee vs. Z-Wave or why Z-Wave got more popular but I'm sure it has nothing to do with the technology. The only technology advantage I see with Z-Wave is that it runs at 900 MHz with all of the better penetration and less interference that goes with that band. 2.4G is way too crowded and not that great at going through walls etc. (by comparison) Unless there are plans for Zigbee to support some high data load devices I don't see the need to operate that high on the spectrum.