Is Z-Wave just... not good?

I really want to like Z-Wave; no cloud dependency to pair devices, and it’s not fighting for airtime on the already crowded 2.4GHz band. I’ve started small with 3 ZWave devices, and the experience hasn’t been great.

  • Physical state changes barely get reported. If I flip a switch by hand, Home Assistant (via ZWave JS) seems to catch it maybe 1 out of 6 times. It’s not just the switches either, the thermostat does the same thing. Different device categories, different manufacturers, same failure to report state changes, which is a big part of why I suspect this is a Z-Wave thing and not just bad hardware.
  • Inclusion is absurdly finicky. The order has to be exact: Factory reset the device, power off the device, start inclusion in Z-Wave JS, power on the device, put device into inclusion mode. Mess up the order and it silently fails to include. No real error, it just doesn’t work.
  • Z-Wave LR has apparently been broken in a pretty fundamental way for a while. See this zwave-js writeup: https://blog.zwave-js.io/2025/08/01/zwave-not-so-long-range/. Sounds like it’s fixed as of SDK 8.0, none of the end device I purchased has actually delivered firmware with the fix baked in. I rejoined all my devices with non-LR ZWave to see if that would fix state reporting, it did not.
  • It seems like SDK bugs in general are common? I have searched online for my issues, and it turns up a lot of topics that dead-end in “SDK bug”

My setup, for reference:

  • Honeywell T6 Pro thermostat
  • Zooz Zen77 800LR dimmer
  • Zooz Zen72 800LR dimmer
  • ZWA-2 (firmware 1.2) antenna
  • Z-Wave JS software

So, is this normal for ZWave? I see plenty of ZWave negativity online but I can never tell if that’s the usual “happy users don’t post” thing or if there’s real fundamental limitations with ZWave. Curious to hear from anyone running similar hardware. Is your experience this bad, or should I be looking at what I’m doing wrong?

How far apart are the nodes. It’s a mesh network and if they’re too far apart they can’t send messages for each other when your network is only three nodes. Not a lot of fallback… The mesh gets better with more participants…

Nope.

Describe your network. There’s plenty of folks here to help.

One thing to remember: Unlike zigbee, z-wave uses different frequencies on each continent. You must purchase the correct z-wave devices that work where you are located otherwise interference will be a problem. Sometimes online stores do not clearly state this.

To check the network and specific devices stats, you can obtain from:
System>Device and Services>Devices>Your device>Device Info panel>Configure…>Statistics

This will provide signal RX TX and signal info.

Post one or two screen shots so we can see network health.

I have 19 z-wave devices, door/windows sensors, light switches, smart plugs, siren/strobe, etc.. I had several light switches and door/window switches that were very flakey, but once I added a couple of repeaters strategically placed in my house, it cured that.

I am under the impression the Z-Wave LR (Long Range) devices only connect to the controller and that they aren’t part of the mesh. Can someone explain the difference?

LR is inherently a hub-&-spoke / star topology with the controller - devices paired LR do not (and cannot) participate at all in the mesh, nor do they talk to each other.

However, any LR-capable device can instead be paired mesh if desired (switching either way requires exclusion/re-inclusion). LR requires SmartStart for pairing, but mesh does not. Assuming it is LR-capable itself, a single controller (e.g. ZWA-2) can run both LR and mesh devices perfectly fine.

I have a ~60/40 split of LR-vs-mesh ZW devices. I pair all battery stuff LR, plus a few farther-flung mains-powered devices. Most main-powered ZW devices in our house are mesh though - some are not LR-capable, and maybe because I’m old-school, I just kinda like having a healthy mesh as a safety blanket.

I have 2 honeywell T6 thermostats and one Zooz 72. both work great as exected.

Both thermostats are C wire powered.

My zwave newtwork has around 25 - 30 nodes with a mix between mains and battery powered.

I’m using an 8 year old Aeotec Gen 5 Zwave controller.

My entire Zwave network works 99.9%+ of the time.

Is the floorplan big like a 3- (or 4- or 6-) bedroom house? Or apartment? Is the place with concrete/stone/brick walls all over? Or, is your z-wave devices far from each other and/or far from the antenna?

I know it is not easy to test, but to find out, and if it is possible, maybe consider relocating those 2 dimmers so that each of your 4 z-wave devices (thermostat & dimmers & antenna) could see at least one other, no more than maybe 1-2 walls away. Yes I would keep them non-LR so that to build your mesh.

The other way to test it out, is to add a couple of z-wave plugs with strategic placements to build your mesh.

I’m going to say this: assuming all else are equal, with 3 devices, if you cannot get z-wave working in your place, you likely will find zigbee (or thread) mesh network even more challenging.

(Somewhat related & a sidebar question: How many wi-fi routers & APs do you have in your place?)

And then if the above is still not improving, you will want to talk to the tech support at zooz regarding the issue of status not reporting.

I have the exact same controller and hardware as you in various places. I do not have any of these issues you’re reporting.

You only have 1 device that can “mesh” the T6. The rest connect directly to the controller. With that being said, if you’re dropping rx or tx, that means something is wrong.

If there’s no errors in your logs, it’s going to be signal interference. Move antenna to a better location.

Just a question though, how far away are these LR devices?

I’m a long-time Z-waver and probably have 25 z-wave devices, maybe more. I always had some lag with my systems back in my MiCasaVerde days, and my Home Seer days. I’ve been running Home Assistant for a few months now and have been very pleased. I did update to the ZWA-2 while I was at it, so I can’t say which caused the improvement. I don’t have any LR devices that I can think of offhand.

I’d say try to have a few more “hops” along the way. Or just for grins, move it closer and see if it helps.

Thanks for all the debug ideas guys! Good to know this is not normal, I’ll keep digging.

How far apart are the nodes.

The ZWA-2 is on the ground floor, about 5m from the T6 thermostat, 6m from the Zen72. The Zen77 is upstairs, directly over the ZWA-2 antenna. All devices have the same problem with unreliable reporting of state changes.

Is the floorplan big like a 3- (or 4- or 6-) bedroom house? Or apartment? Is the place with concrete/stone/brick walls all over?

1000sqft small wooden house, 2 floors, it’s quite compact.

Unlike zigbee, z-wave uses different frequencies on each continent. You must purchase the correct z-wave devices that work where you are located otherwise interference will be a problem

All devices are for the CA/US/MX region (according to the product pages I purchased them from), which matches what I have my ZWA-2 set for.

Somewhat related & a sidebar question: How many wi-fi routers & APs do you have in your place?)

Two WiFi APs (2.4GHz, 5GHz, 6GHz), a ZBT-2 (thread, 2.4GHz), a ZWA-2 (~900MHz), a zigbee hub (2.4GHz), a Lutron caseta hub (434MHz). I have tried disconnecting the ones physicaly near the ZWA-2 (ZBT-2 and one WiFi AP), I still have the same problems. I should pull out my SDR and take a look at the spectrum. In theory ZWA-2 is the only device in the ~900MHz spectrum, but maybe there are some harmonics.

I’m going to say this: assuming all else are equal, with 3 devices, if you cannot get z-wave working in your place, you likely will find zigbee (or thread) mesh network even more challenging.

My zigbee (2 devices), thread (12 devices), and lutron caseta (1 device) systems all work pretty well. The thread network sometimes has downtime when I add several new devices at a time, but best I can tell that’s a bug in the cheap ikea thread devices I purchased, not signal integrity.

Post one or two screen shots so we can see network health.

Are the dropped commands a problem here? I expected many more given that 5/6 times the physical state doesn’t get reported. I don’t notice any difference in reliability between the switch with 36 dropped RX commands, and the switch with 1074 dropped RX commands.

Upstairs switch (about 5m away, through the ceiling/floor):


Downstairs switch (6m direct line-of-sight to ZWA-2):

Thermostat (5m direct line-of-sight to ZWA-2):

Other info: LR (on the switches only) made no difference.

The noise floor is -100dBm (as reported by zwave-js).

The thermostat is connected with a C wire, and I joined it without batteries to ensure it reported itself correctly.

Hmm… could not see any apparent smoking guns.

Have you tried to re-include all your devices for mesh network as opposed to long range star network? I would start with the T6 (does it have like of site to ZWA-2?), see how quickly the state reporting it would be for a couple days, and add one at a time.

We know how far they are already from the ZWA-2. But how far they are to each other? I think that info is more relevant… since we are aiming mesh style network.

Also talk to zooz tech support and see what they have to say. Maybe there is some parameter(s) they would want you to set differently, maybe there is HW issue to begin with and you need a replacement.

Oh and before you do anything, what does your Z-wave JS dashboard look like?

What type odf security were the devices included with? Avoid S0 at all cost. unsecure is fine for most things (besides locks or the like)., S2 is fine too.

In what scenario would you want to use unencrypted wireless communication exactly? Plenty of old z-wave tech is already insecure when encrypted :see_no_evil_monkey:

In 2022, researchers published several vulnerabilities in the Z-Wave chipsets up to the 700 series,[63] based on an open-source protocol-specific fuzzer.[64] As a result, depending on the chipset and device, an attacker within Z-Wave radio range can deny service, cause devices to crash, deplete batteries, intercept, observe, and replay traffic, and control vulnerable devices.

Z-Wave - Wikipedia

An attacker within ZWave range on my property has been inrange (you could say downrange) of other things and on camera for 5 minutes and is literally standing next to an exterior wall on property.

Mitigated. - not an issue. I’ve already ID him willing to accept this risk and mitigate with other means.

And don’t use s0 unless you HAVE To yes unsecured is better if you like things that actually work.

I don’t have that compromise as I don’t use any Z-wave. I like to have secured and working things - at this time mostly esphome based devices which luckily can don’t have such limitations :person_shrugging:

People thought the same about the keyless car system… They didn’t had a relay attack in mind and with it often meter long atennas to bridge long distances…

Now if you spin this game further you might find out that nobody actually needs to enter your property and that all that is needed to infiltrate your unsecured/unencrypted network is a drone with some tech :flying_saucer:

We’re not having this discussion orange. We will agree to disagree and my situation is Mitigated and it’s good enough. Not everyone holds your opinion… Got it?

Got it! Still amazed how many (specially tech savy) people often do the 180° opposite of (the so called) best practice. And this is not limited to “home tinkering” (where you mostly can harm yourself and your family with such decisions) but also in IT oriented businesses. The later case can often put customer and employee data at risk. Depending where you are located it might be even illegal to not use encryption in business related tasks :stop_sign:

My point is each situation requires its own review. And not every solution requires a hammer. All mitigation is NOT everyone should use X.

There are choices, choice is good. And not everyone who disagrees with you is wrong. Ok? Have a great day man!