One might consider running a separate “satellite” ZWave gateway node off Raspberry PI - with just ZWave (and Zigbee) service preconfigured to message a central MQTT broker (HA instance) - this way a small device with 1 or 2 USB sticks can be fit into a plastic box in a central home location, connected using Ethernet or WiFi, solely to run ZWave and Zigbee networks. Something to consider for HA creators as well to make money from selling.
Just a short extension cable to get it away from the back of the NUC… That’s all, nothing crazy.
Yes, but this is another device/PC which need regular service/back-up/updates… Another time-consuming beast…
Thank you… Very clearly explained possible options…
Taking into consideration this and other opinions i just ordered 2-nd hand ‘Vera Edge’.
The price looked very low (~20EUR), so 1st I’ll give a try this option…
Hope this will work.
Maybe I’m not qualified enough because my network is pretty small. I have 2 Eurotronic thermostats and used the old zwave integration without any problems for 2 years now. A few weeks ago I got a new house, installed 9 Fibaro Roller Shutters and decided to switch to the new ozw addon. The network is incredibly slow. Sometimes it takes up to 30 seconds until a command is send to a device and every few hours my aeotec stick is unavailable. Which results in unavailable devices. I still don’t know why.
The CPU load massively increased and the log is extremely big and flooded with the same messages over and over.
Either the stick didn’t respond in the last 500ms or this stupid USB error although it’s connected to an usb hub and never had this problem before.
There’s no command to heal the network and the owz-admin is unusable on mobile devices.
I’m really thinking about switching back to the old integration. I can’t recommend it in its state now.
There is. It’s in the ozw-admin tool.
Sorry to hear you’re having trouble. It sounds as though there is something fundamentally wrong in your configuration - you have removed all traces of the previous Zwave integration, I assume? It’s no help, but I’m not seeing any of these issues in my network.
I think a hub is still the best choice at the moment.
For Z-Wave.
Zigbee is pretty solid on HA.
Z-Wave has improved since my last test but it’s not “great” yet.
The latest mainstream effort is still in beta.
True, but the second node is pretty simple.
This is my favoured approach - Main HA running on a VM in a nice secure server, and a second HA running on a Pi and located in the best RF position (I know ZWave is a mesh, but it definitely works better if the controller is well connected). The second HA node only needs to be running MQTT and OpenZwave (Beta). You then bridge the two MQTT brokers together.
This approach has the additional benefit of isolating the two domains. Given that it is single-function, there should be little or no reason to reboot the Pi, and having it’s own MQTT broker means you can fully reboot the main HA node with no impact on the ZWave network - if you have the remote OpenZwave using the main broker, a reboot of the main HA node will require a reboot of the Pi because the OpenZwave daemon will exit when it loses connectivity.
Not at all.
I have a NUC with a metal case and I have two zwave (Aeotec and HSBZB-1) and two zigbee radios (HUSBZB-1 and a zigbee2mqtt cc2531) connected directly to the USB ports and never have any issues. And the NUC is only about 3 ft from a wifi AP. And again the NUC is only about 10 ft (thru a wall) away from a 1200w microwave oven.
The RF density is pretty thick right in that area and I never have any issues with any zwave or zigbee signals.
I’m not so certain of that.
I’ve only ever seen that happen to me one time when HA got stuck on a restart and I restarted again before the zwave network had re-initialized. I waiter a few minutes and restarted again and asfter that everything came back just like usual.
Could you elaborate a bit on how you used Hubitat in this context ? Can you use it transparently as an independent ZWave stack, purely as a gateway basically, without any other functionality that could get in the way ? Is it entirely local, will it work even if it is never going to be connected to the internet ever ? How is the performance when integrated with HA ?
Sorry for all the questions Trying to gather some potential alternatives to the new OZW integration (which is a disaster for me). The old HA integration is rock solid here, but well, it’s starting to show its age.
That’s exactly how I would recommend it.
Hubitat (my older model) has low specs.
They also don’t provide support unless you disable any third party drivers or apps, so keep it simple.
It’s really fast if you keep it to just devices and the maker API app that interfaces with the home assistance via this custom component.
Interesting, thanks. Can this be used on a local network without any internet connectivity ? You mentioned a maker app, this sounds like there is a cloud based component to it ?
All local.
No cloud required.
I killed my Internet to test it out.
The integration asks for local up addresses.
I feel like there was some history of it asking for the external URL but that was addressed or something to keep it all local.
I did test it. Hue surprised me that it worked locally too.
Awesome, thanks a lot for the info. I will definitely look into this.
If I had a couple of extra bucks and they weren’t sold out (might be back in stock), I’d pick one up and re-add only the z-wave.
As I said my C4 died, but that was pretty much like their production crowdfunding version…
3 years on the odometer.
The maintainer of that app is transitioning away from the ozw backend. If you’re just starting out then use zwavejs2mqtt. The experience seems to be the same to better than the original zwave2mqtt. Right now both are active but he seems to most often reply to issues telling people to move over and the fixing them there if necessary.
@keithcroshaw I’ve considered ‘Hubitat’ but it is almost triple of the price of ‘Vera Edge’… Are the possibilities/parameters also 3-times better than ‘Vera’…?
No clue.
I came from SmartThings so there was 0 learning curve.
Whenever I researched Vera I found a very old looking forum and little info on what the experience is like. They’ve since upgraded to discourse I see. Although it’s concerning to me that this is one of their top posts:
When I research Hubitat I find an active community with people who share code for drivers (which I shared and used) and apps (which I don’t use).
Not a comparison since I haven’t used the other, but unfortunately you get what you pay for might apply.
Maybe a bit late to the party, but here goes. I used to be on Vera but it has fallen WAY behind so I went to HA. For z-wave I’m using OZW (Beta) with an Aeotec stick and ran for a few months on RPi and have since migrated to a VM on my server. I have experienced the following:
First, z-wave vs non-z-wave (such as wifi or zigbee)
Good (z-wave):
-doesn’t use the wifi spectrums so it will not be affected by wifi and vice-versa.
-great for battery devices
-built from the ground up for home automation
Bad (z-wave)
-devices are a bit expensive
-although there are a lot of devices, some specialty devices are not available (typically available on wifi)
-debugging issues may require new expertise (e.g. editing .xml files, analysing traffic, optimising speed etc.)
I have been using z-wave very happily for 8 years! If I’m looking for a device and it is available in z-wave, I will choose that. Otherwise I will choose another option (e.g. wifi) and, fortunately, HA integrates those at the user level very well.
Now for the specifics of which z-wave to use on HA. “Native” is dead so that leaves zwave2mqtt and openzwave(beta). When I started with HA (recently), I saw zwave2mqtt and openzwave (beta). Though the word beta usually turns me off, reading posts and the specifics of what people were having trouble with, I chose OZW (Beta). It seemed certain to be the future (and still does). If I understand correctly, zwave2mqtt architecture means that development of zwave was partly in the hands of HA and partly in the hands of OZW as the HA integration was non-trivial. With (Beta), there is another layer in the mix but it is actually a very simple MQTT packager and will rarely need updating. This means almost all development is in the hands of OZW. So as soon as @Fishwaldo releases an OZW update, (Beta) users can benefit without waiting for HA as they must with zwave2mqtt.
Good (OZWBeta):
-Fast. Yes, fast (contrary to comments above). Note that I’m a stickler for checking my comms and configuring correctly. I don’t poll at all and I set battery devices for 5 hour wakeups. I have not seen any noticeable delay in anything except the small delay (max 2s) for the door locks to operate from HA or to notifiy HA that they were opened manually. This is mostly because of z-wave protocol for such devices (being battery powered and using beaming). Anyone observing slow z-wave should first look at the comm stats on the OZW control UI.
-No required functionality missing. Ok, maybe stick backup online but that is missing from zwave2mqtt as well. It DOES allow you to delete a dead node online, though you must do so “manually” using MQTT service. I’d love to see a delete node service and a delete node button on the control UI, but it was three, easy service calls in Developer Tools | Services. Easy enough for impatient me and works like a charm.
-Can be run on a separate machine (or VM) so a reboot of your HA host computer doesn’t cause a reboot of z-wave. This works well!
-The core functionality is present and working well and the (Beta) moniker is unfair as it is no more Beta than most other things in HA. I’m not saying it is bug free, but neither is HA, Windows or Linux!
Bad (Beta):
-Control UI loses the cursor occasionally making it awkward to use at times
-Delete node hasn’t been integrated into the control UI yet.
I would highly recommend to use OZW (Beta) and if someone want any assistance I can give to setup or optimise their network, I’m happy to help.