Continuing the discussion from Google IoT Platform: Project "Brillo", "Weave", Nest … Google I/O Conference:
I feel your pain, seriously. That sounds awful.
Several thoughts (I worked as a network engineer).
-
if it was me, if none of the following help, then, yes, I would return everything I could. That much failure makes the system worse than useless to me.
-
let’s start by seeing what does work. "All home automation is local. " sometimes it’s about the wallpaper on the walls, or the insulation inside the walls, and you just can’t get a signal from one room to the next.
So: how’s the Wi-Fi? Put your smart phone on Wi-Fi mode and just walk around the house and see what the bars do. If you using Wi-Fi range extenders, you probably already know there are architectural problems. But let’s just begin with something simple, a Wi-Fi survey of the house.
If, without range extenders, the Wi-Fi is patchy throughout the house, then it’s pretty likely that zwave/zigbee will be also. Although there are different radio frequencies that interfere, when it comes to drywall, foil backed insulation, water pipes, tinted glass, and other physical barriers, they all have similar drop offs, except WiFi is higher power than zigbee/zwave. So where the WiFi is dimmed, the zwave may be crippled.
If you do find a lot of dead or dim zones, the only option is repeaters in almost every room. It won’t matter what hub brand you choose, you’ll run into the same thing.
-
for the next steps, begin by shutting off any smart app that is doing polling or refresh. We want minimal traffic on the net.
-
if the Wi-Fi (without extenders) is fine, then we have to look at the device network that you’re using for the home automation.
Let’s start with the Cree bulbs.
Put one bulb in a lamp One room over from where the smartthings hub is.
Does it turn on and off with 100% reliability from the smartthings mobile app?
If not, but again there was no Wi-Fi issue in that room, then either the hub antenna is defective or, there’s a local interference issue.
I don’t want to go through all the things that might cause local interference in this post (see the repeater FAQ for more on that), let’s just note the issue for now.
But if you can’t get a single simple zigbee device to work one room over from the hub (not counting the presence fob, which has a separate set of problems), then you’ve got a lot of detective work to do. You’ll probably need to try changing your Wi-Fi channels, look for local interference like fluorescent lights, even borrow a Phillips hub from someone and see if you can get one of its bulbs to work in that room.
If zigbee is reliably controllable from the Phillips bridge, but your Cree bulb is not reliably controllable from smartthings, first try a different cree bulb. If that fails in the same way in the same location you may need exchange the ST hub for a different one and try that.
- if the Wi-Fi signal was okay, and you could get a single zigbee device to work reliably from the smartthings app when that device was one room over, it’s time to consider the network layout.
For this test, for technical reasons, let’s switch to a motion detector. I’m assuming it’s Zigbee. Put the motion detector one room over from the hub, you need to unplug the hub for 15 minutes, then reconnect. Wait another 15 minutes. This rebuilds all the address tables. Tedious but easy.
Now it’s a matter of walking that device away until it fails. Try 4 or 5 tests in each location.
As soon as you find the place where reliability drops off, leave the motion detector there. Unplug the hub for 15 minutes, wait another 15 minutes. Then retry.
If the detector works reliably now, fine, it just needed the new neighbors list. Continue the walk away procedure.
Once you get to where it fails and a heal doesn’t fix the problem you’ve reached a zigbee dead zone.
Check within 30 feet line of sight from where you are plus 15 feet through walls and ceilings. What non light bulb zigbee devices do you see? Any that plug in (not battery powered)? If not, put a device that repeats nearby, unplug for 15 minutes, wait for 15 minutes, retry.
If that fixes the reliability problem for that motion sensor, OK, you probably need to increase the ratio of repeaters to battery devices per room in your network. It might mean new devices, it might mean rearranging the ones you have. It might mean more effort/money than you’re willing to budget for a solution, in which case I’d go back to returning stuff.
If you can get a replicatable distance-related failure but adding a repeater and doing a heal doesn’t fix it, then you’ve likely got a local interference problem. Way harder to track down and fix.
(I know the OP is already familiar with what repeaters are, but for later readers who aren’t, also see the following repeater FAQ:
A Guide to Wireless Range & Repeaters | SmartThings )
- if everything worked perfectly in the device tests so far, turn back on the smartapps that did polling/refresh. If the motion sensor becomes unreliable now, it’s likely a traffic jam problem, you’ve overburdened the network with too much polling. easy to do, easy to fix by reducing the polling requests.
All of that was just to check the zigbee HA devices. You need to repeat the whole process for ZLL light bulbs and Zwave, as these three things don’t repeat for the others, only themselves. Also zwave doorlocks that use beaming need the last device before them to also support beaming.
But those are the typical diagnostic steps. In each case you only go to the next step when you have connection/control success with the previous
A) general RF issues as demonstrated by wifi without extenders
B) one room over connection using two different devices
C) controller substitution (philips bridge vs ST hub) for one room over test
D) distance to failure after network heal, followed by repeater addition and network heal
E) traffic check
Like I said, tedious, but pretty easy. We start with zigbee HA just because the heals are the easiest to do.
What next steps you take if you pinpoint a failure are up to you.
If you need an alternative network protocol, insteon over powerline gets around some common architectural issues, and WeMo over WiFi has possibilities although I’ve personally run into a high percentage of WeMo device failures, including one I bought this month.
Or you can wait for a viable bluetooth system.
Good luck, let us know what you decide, even if it’s just to stick with what you have.