Please help! Set up of a small school with OpenHAB

  • Platform information:
    • Hardware: Intel x64 (i5) 3.4 Ghz /x4, 8Gb RAM, SATA 500 Gbdisk
    • OS: Linux Debian 10
    • Java Runtime Environment: OpenJDK 64-bit build 11.0.5+10-post-Debian-1deb10u1
    • openHAB version: 2.4
  • Issue of the topic: Cannot reach devices that are farther from my z-wave controller, even if I added 3 z-wave electrical switches
  • Please post configurations (if applicable):
    I’m attaching one graph generated by openhabian, z-wave devices

and the other of my physical topology

All the thermostat are Go-control GT-TBZ48 Z-wave plus connected with a battery
The three 3-way light switches are Innovelli LZW30

I connected my Go-control devices near the OpenHAB server and were successfully connected to the network, then all near (Office, Staff, G1, Kindergarten, G2) worked correctly, but the worse performing are those far away (probably around 40 meters) and in the basement. I understood that the mesh was going to take care, but this is not the case.

Please help I’m pulling the little hair I have off trying to make this system work, and winter is coming harsh here up in Alberta
Thank you very much!

I believe there is a limit of how many hops Zwave can do, I want to say 3 but I am not sure about this.

As a way to troubleshoot to validate the further devices work, is moving your OH system to them; is this an option?
Schools tend to be a lot of cinder blocks, perhaps you are lacking coverage due to wall thickness or something is reflecting the signal (vents)?

If you are able to test moving OH closer to the devices and they work, is there an option to setup another OH system with a secondary zwave? It is possible to send updates to a central OH system, but it does add to your complexity and could make it difficult to maintain.

4 hop limit according to this FAQ. They also recommend having a device every 30 feet or closer. This FAQ may be good reading material.

Can you elaborate on what is bad about performance? Initial configuring is a different story, as I expect you already know?

HI Thank you for all the responses. Great community that is amazing!!!. Yes I tried moving the main computer near the south of the school (where the worst performin devices are) and they catch up, so it seems a distance issue.
Is there a way to know where those devices are connected? OR force them to the repeater light switches?
The extra 3 switches on the hallway are not farther than 10 meters each, and there are no walls between them…

I am new to this thank you for helping me out! The issue is they start working, but then after a while they don’t update, or sometimes a command works, but they barely (or never) update temperature…
I am open to upload logs/debugs, but not sure how… my question is why the mesh is not working for me? Or are those LZW30 devices not acting as repeaters?
Thanks for your hints/ideas!!!
Tomorrow morning I’m thinking on putting a long USB cable from the servers room and hung the USB z-wave hub on the hallway and see if I can make this work at least for now… just a patch but might work…

Hi Bruce,
Thank you for this. Yes I have only 3 LZW30 Innovelli switches that I was trying to use as repeaters, but they don’t seem to make a difference on the network.
On a sidenote, since those devices are new, they are not supported on OH stable 2.4, so I needed to go the route of upgrading to OH 2.5 just for the z-wave binding, and I’m afraid this is the beginning of my trouble… but if I don’t upgrade I don’t get support for the light switches…
Not sure where to go from here. Help!

From my experiences the hardware will pick its path.

As for the 3 switches, what numbers are they in that zwave graph you posted?

Also it is possible to install a zwave binding from the 2.5 snapshots (or milestones) into a stable version of OH 2.4.

Switches are in the hall, Node 9 (north, at the left of the hall in the diagram), node 13 (middle of the graph) and node 12 (south - the right of the diagram)
Node 12 (one of the switches) in the z-wave network viewer seems to be connecting to the node 19 (grade 8, just beside it) and to node 20 (grade 9, besides the switch). But this node 12 should be connectin to the other nodes (just 10 feet downstairs) and they show offline
Good to know about the zwave binding with 2.4 OH stable, it works great in my home (but I have smaller distances so that is not an issue here!)

Version 2.5 just hit Milestone 6 and will have a release candidate in a couple of weeks. If you are running a snapshot version I would recommend moving to Milestone 6. The snapshot builds are called unstable for a reason. Everything may not function as intended.

If you are looking to add devices, you may want to consider Zooz, available on or their store, They have excellent support, reasonable prices, and are helping support OpenHAB by providing devices to update our database.

I’m not very zwave savvy, but how does the “heal” process play into this? Should you wait for a heal after pairing devices neat to the controller and the moving them? Do I recall some issues with heal in some binding versions?

Thank you for the message. I’m not running a snapshot of OpenHAB, but of the zwave binding.
I installed those Zooz devices, they work great, but I went to the other ones (inovelli) because of price only… (10 bucks less), but I guess some of my fighting is becuase of this price difference…

Hi! didn’t happen last night (the automatic healing)… this morning I tried moving the main USB stick controller to the corridor, and after working for 2 hours I discovered that my USB extender was only 1.0 and was disconnecting frequently… so I ordered a USB 2.0 extender and hope for the best…
Will keep you updated (in a month because I’m traveling for most of december)
If anybody has any other idea I can work from remote… please help!!!

1 Like

Picking up from the other thread…

A few (random!) thoughts that might help you…

  1. Do nodes 12 and 13 work ok? I’m not sure how they are communicating with the controller based on the network diagram (although this may not be 100% correct).
  2. Note that this diagram doesn’t show that the route from (for example) node 20 is directly to the controller. It shows that node 20 CAN hear the controller, but it doesn’t necessarily mean that it is trying to send directly. Node 20 also has routes to nodes 12 and 13, so it should be trying to send through these nodes. Same for node 19 - we see that this has nodes 12 and 13 as neighbours, so it should be able to route via these nodes even if it has a (probably poor!) direct connection to the controller.
  3. Of more concern to me is that neither node 12 or 13 seem to have a direct connection to the controller. Given node 13 is half way down the hall, I would have hoped it would have a link. Of even more concern is that node 9 (which is closer to the controller) also apparently has no connection to node 12 or 13. It’s actually unclear how node 12 and 13 can communicate to the controller since they only seem to hear non-listening nodes.
  4. I would suggest that your mesh is quite weak. From the network diagram, node 9 is the only powered node that is able to communicate with the controller.
  5. Are these powered switches in the wall (I guess so). If so, this might be part of your problem. I don’t know if this is a wooden or concrete/brick type building, but it’s easy to think there’s “only” 20m between devices, and ZWave will go through a few walls, but, if the route is down the length of a wall, then you might have a hard time of it - ie if two switches are in the same wall, then they have to communicate through 20m of wood. It might not be that bad as there could be reflective paths that work better, but I would not be surprised if this is an issue. See the green line in the annotated diagram below…
  6. I would recommend to try placing a repeater (just another powered ZWave device) or two “out of plane” of the hallway - eg at the apex of the red line below. The idea here is to offer more diverse routes that aren’t potentially trying to travel down the length of a wall. Placing a couple of nodes like this could well help (ie one on the opposite side of the hall as well).

I think one trap that people fall into is thinking that RF devices have a range of X (where X is say 10 to 30 meters) but this can be significantly reduced through walls. Similar to what I’ve said above, in the diagram below the top two nodes might only be a few meters apart and have trouble communicating due to the path their link has to take, while the bottom node might be a lot further away, but may work fine as it only has to traverse a wall or two, but each wall is say 150mm, so it’s only going through 300mm in total where the top two nodes are going through 5M of wall. Note - lots of things can influence this distance, so treat this as an example of concepts and don’t quote me on exact distances :slight_smile: .


I hope that helps - as I said - it’s really just random thoughts, but maybe it helps with some ideas to try :wink: .

If the switches are located in (grounded) metal electrical boxes that would affect their range too.

That is great info! Thank you Chris! Thank you Bruce!
One thing that happened is while I was adding the switches to the system, I needed to move temporarily the server toward the back (right of the diagram) because I couldn’t make them part of the mesh…
Maybe they learned wrong and now they still try to connect to the controller even if the controller is back on the servers room far away from them?
It will be no difficult for me to remove those two electrical switches (yes they are powered) and put them back on the z-wave network if this can help…

Yes - everything in the way will affect the range. Ultimately, it’s about link budgets and the attenuation through the path. A link budget may allow for (let’s say) 100dB of loss between the TX and RX - some of this will simply be lost through the air (free space path loss) but anything more dense will impact to a greater extent.

See an example of this below. This shows that a 6 inch brick wall has around 30dB of loss at 900MHz (the ZWave frequency) where drywall or ply is maybe 10dB or less and this will make a huge difference.

I note that some of the plots on this graph don’t make 100% sense, and I don’t remember where I got this from now (I used it in a presentation a couple of years back) so treat it as education material and don’t get too hung up on the exact numbers etc…

Hi good morning! Yesterday I started to work very early on this (3:30 am!) because I really need this working, now it seems I have a different problem.
I cannot make my 3 light switches (LZW30 innovelli) to be recognized properly.
Can any of you experts give me a hint on what is going on with those devices? (I’m attaching a log of just today on one of them)
node31.txt (50.6 KB)

Right now my zwave schema looks even worse than before… and I’m starting to get desperate because I’m flying out of Alberta tomorrow and not coming for a month back here - so whatever needs to be done while I’m here needs to be done today…
Thank you for your time! I appreciate all your comments and thank you for helping me learn this new fascinating world of OpenHAB

Filtering logs makes them basically useless for troubleshooting. If you wish, this log viewer may help.

nice! I was wondering where you get those nice filtering from…
I get just a bunch of send-data and timeouts… Does this mean that…??? Is my device not working?