It is likely one device is flooding the z-wave network making it unstable. I had an issue with a lockly lock doing that but removing it and re-including it a few times until the network was stable worked.
I would be interested in knowing how to monitor z-wave traffic over the antenna.
If they open a dedicated “help me” post it is something worth exploring.
Hi @petro , Fair to call out the ranting, but this whole thread has been me asking for help and while I REALLY appreciate all of the insightful and well-intentioned replies, it seems there are still some pretty fundamental challenges with my ZWave system that I simply do not know how to overcome.
Regarding what I have tried so far, here’s a partial list (most of which I’ve mentioned in various posts earlier in this thread):
Moved the USB Z-wave dongle away from the HA Yellow using a USB extension
Relocated the HA Yellow (+ Z-wave dongle) to a completely different floor in the house
Replaced the Z-wave dongle with a new one (same make/model)
Upgraded Z-wave dongle firmware (which bricked the dongle but then a kind member of this forum helped un-brick)
Many version upgrades to HA Core and Z-Wave JS UI
Replaced/upgraded the CM in the HA Yellow (Now a RPi CM5 w/8Gb)
Confirmed that I do not seem to be running out of RAM, disk, or CPU capacity on my HA
Installed multiple Z-wave repeaters (Zooz ZAC-38’s). (My network is also quite dense and most of my devices are mains-connected (like lightswitches)… and the problems that I have do not seem to be physically consistent. For example, several of my most chronically-unable-to-adopt devices are located about 10’ from the Z-wave dongle… and inches from other Z-wave devices that do not seem to have any problems.)
Multiple calls/emails with Zooz technical support (manufacturer of my Z-wave controller) who confirmed that they believe that the problem I’m seeing is due to a known flaw in the underlying Silicon Labs firmware for which there is no timeline for a fix.
I’ve tried buying RF sniffers to measure interference (but have not managed to get any useful insights from them).
I’ve tried multiple (seemingly reputable!) brands of devices and the problems do not seem to be specific to any one.
(I do not know what you mean by “serial by id”.)
My current status is:
I have a network with ~50 devices included plus another ~12 that have not managed to include. [Serious problem]
I periodically see several “ghost” devices of unknown type/manufacturer appear in my list but then hang on “ProtocolInfo”. [Not a problem in and of itself but listing it in case it is diagnostic.]
Attempting to apply OTA updates to Z-wave devices kills my Z-Wave controller. [Not too serious but again may be diagnostic.]
Even left alone, my controller seems to die (silently) after a few days and remains non-responsive / unreliable until my HA is shut down and physically power cycled (making it hard to even program a nightly reboot…) [Moderately serious]
The most generous interpretation I can think of for the above is that I have one or more flaky Z-wave devices (either already on the network or among those pending inclusion) and the underlying Z-wave protocols/firmware is simply not resilient to (or even able to alert about the presence of) a flaky device. (Less generous interpretations include some more-serious flaws in the meshing technology, etc.) It is also possible I’m an idiot who has done something dumb (though I must at least be a somewhat clever idiot, as nobody has managed to catch my mistake yet ). Or maybe there’s terrible RF interference that I do not know how to detect/measure and somehow affects only some devices on the network (uncorrelated with physical location or proximity to other nodes)? Regardless, I think it is accurate for me to say that the consequences have been super frustrating!
At this point, I’m willing to try pretty much anything (within reason), short of manually reconfiguring the 50+ devices currently sort-of on the network (and all of the dashboards and automations referencing them), severing power to large portions of my house, or spending untold hours of electrician time temporarily disconnecting then reconnecting switches/outlets one-by-one… I would be genuinely grateful for any help or advice anyone could offer!
(I should say as well that while I have been quite frustrated with the technical experience of using ZWave, I’ve been really impressed with the knowledge and helpfulness of the user community. Thank you!)
A couple of things I picked up over the years: If you have any devices included with S0 security, pair them unsecure. S0 is very, very chatty and the chances of the communication getting warbled because of that are high. It also asks a lot of computing power, some routing devices cannot handle it if they have a lot of traffic.
S2 security does not suffer from the same problems, or at least not nearly as much. If those still give you problem, ask yourself if you need the security and try to pair unsecure if you don’t. unsecure sounds scarier than it is. For locks and such you would want to be safe, but otherwise it isn’t a big issue.
For pairing issues: I read that some devices lower transmission power during the pairing process for security reasons. They can also refuse pairing through routers. This might explain why pairing is more of a pain than normal use.
@Edwin_D Thank you. I have tried to use S2 for all devices that support it. Dumb question but is there a way in ZWave JS-UI to see which security level each device is using? I see a column for “security” but it just shows green checks or red minus signs… I cannot seem to see which level of security the secured devices are using…
@mundschenk-at Thanks for the pointer to Zniffer! Do you know if there is any way to set up a Zniffer without a windows computer? (The instructions you linked reference SI Labs software, and my recollection from trying to use it for controller firmware updates is that doesn’t actually work on a Mac or Linux…)
It might be possible to flash the firmware on 800 series boards with the CLI tools developed for the Z-Wave JS project, but TBH, I’m not sure. I’ve used Bootcamp on an old Macbook for fiddling with proprietary hardware including Z-Wave dongles.
Green check means S2 Security. Yellow check means S0 Security. Red circle means no security. Hover the mouse over the icon for a tooltip for S2 security level(only on PC of course).
Or, look in HA’s device page, Z-Wave Info panel tells you in plain text.
As an additional pointer besides the security level: Some devices, especially power monitoring switches and plugs, are very chatty by default. If you have several of them, they can easily saturate the network. Almost all of them have several parameters to configure their reporting interval (and sometimes also what values to report at all). For example, on my ZW175 Smart Plugs I always set the standard reporting interval to 3600 seconds (1 hour), and I disable reporting of voltage and amperage (V and A) as I don’t need those measurements at all. For power and energy (W and kWh), I set up an offset that triggers a non-timed report when reached (e.g. “send a report when power changes by at least 5 W”), with the value of the offset depending on what kind of device the switch monitors (i.e. for a toaster, the offset can be 200 W, whereas for for an RPi3 I might be interested in changes as small as 1 W). Doing this systematically can dramatically improve overall mesh performance.
Funny, I have it the other way around. I use Z-Wave for power monitoring/smart plugs and miscellaneous sensors, and ZigBee for lights, motion sensors and buttons/switches.
Maybe if Zooz is telling you it’s a known problem with their controller maybe you need to try getting a different brand of controller?
I’ve been using an Aeotech Gen5 Zwave controller on the original firmware for as log as I’ve been using HA (8 years) and I don’t really remember any serious issues with my zwave network. It mostly just works.
There are issues in the SDK for 700 and 800 series (though largely mitigated and/or fixed in the latest releases), the symptoms are not what @mike15 is describing. I’m also running a ZST39LR (with 45 devices) and don’t experience anything similar.