Thanks for the help! Any prerequisites for that to work? (WiFi and Zigbee)
Also, how do I make sure I have the “right” firmware?
I have the red gen4 mini 1pm, and I tried zigbee mode (you can switch between the firmware from the management portal of the shelly device itself), but at this very moment I would not recommend switching to zigbee - it is buggy, with unexpected behaviors here and there, had to try a couple times to pair, and the feature set under zigbee (from Z2M) or Matter is not the same as from the rather stable / solid Shelly integration under HA.
If you have to go zigbee, please reference the thread above with your eyes wide open before going in. It would work - if you try hard enough or if you are lucky.
Not saying their zigbee features would not improve over time (it likely would), but for now, I’m staying with wifi only, using the (HA internal) Shelly integration - meaning no Matter over wifi, nor Zigbee.
Thanks for the perspective. I’ve successfully paired a Shelly cover switch with Z2M, and it seems to be working fine (for the very few hours it’s been in this mode). My gripes, in descending order:
- Right now Shelly in Z2M reports that my cover has a “tilt” capability, it doesn’t and that bothers me.
- I’m not really sure whether I should or shouldn’t add the device also via the Shelly integration when I’m in Zigbee mode, it is kind of nice to be able to see firmware updates, but it feels like this creates many duplicate entries.
- Pairing took 2 tries: it failed the Z2M interview once, but succeeded quickly on the 2nd try.
- Having to recalibrate after switching to Zigbee, which I guess is kinda understandable, I could’ve just not calibrated on WiFi.
@k8gg I’m just curious, can you elaborate about what instability and unexpected behavior you are still experiencing? In my experience, the 1.7 firmware is stable, but indeed still missing feature parity. Especially the detached inputs not showing up via Zigbee annoys me. I’ve heard that this will be addressed in upcoming firmware 1.8.
@shibel Thanks for reporting back. I’m glad you didn’t have too much trouble with pairing.
I actually take that back unfortunately. The Shelly 4Gen on Zigbee suddenly started being unresponsive. I reluctantly opened the electric box and tried resetting the device, but now it won’t respond to anything at all - not even reset to factory settings. I first tried the trick with cutting off mains and pressing the cover switch quickly a few times, then when that didn’t work I opened the box, pressed the reset to factory button for 10 seconds and nothing. Multiple times. The LED light does blink when I do that, but I hear no clicks signifying a restart or reset, and there’s no Shelly AP afterwards. It seems this module is bricked now.
Oh no! That does sound like a hardware failure to me. If I were you, I’d contact Shelly support and see if this can be solved under warranty.
I’m using several 1PM Gen4 and 2PM Gen4 devices, and they all show the same issue. They’re set to Zigbee mode and connected through Z2M. Sometimes when I toggle them on/off from the mobile app, there’s a delay of 2–3 seconds before they react. After that initial lag, they respond instantly to the next commands. Additionally, if I toggle them on and off repeatedly within a short period, the state sometimes gets stuck for a few seconds before correcting itself. When I use the Shelly app to toggle them rapidly, the issue doesn’t occur — they respond immediately every time.
All devices are on the latest firmware 1.7.1
Does anyone have the same experience?
@filipvlasic That sounds like a generic Zigbee issue to me. I experience that behavior with many Zigbee devices on my network, not just Shelly ones.
Not anymore; at least for me that has changed with the released 1.7.x firmware. Maybe you should try updating.
I don’t think this is a Zigbee problem. I have a number of 1gen4 which are dropping off the WLAN (while the device 10cm to the left is bathing in the strong light of the AP) while their Zigbee connection to the next hop (they chose the worst partner they could…) is stable.
I have spent some time now with Zigbee analysis trying to teach myself the basics of finding out what is broadcast on 2.4GHz using an SDR. A lot of the cheap Chinese Zigbee implementations are not that great on following the standards and the Shelly devices seem to be a bit more picky. So while Sonoff and even aged Heiman power plugs are working great the cheap Ali-trash in my house is not.
I just got some of the 2PM and 1PM Gen4s. After learning through trial-and-error that you MUST disable BT for Zigbee to work at all, I have run into the problem that these don’t seem to expose a switch cluster for the output (ZHA) for detached mode. These Gen4 devices feel a bit half-baked at the moment.
Hey @Wonko,
I just wanted to drop a quick note to confirm your “trial-and-error” discovery. I’ve been running 10 of the Shelly 1 Mini Gen4s (Model: S4SW-001X8EU) and was running into the exact same headache.
I used the Web GUI to switch all of mine from Matter to Zigbee mode (running firmware 1.7.1-gd336f31), and while they joined the network easily (so Zigbee did work initially), they were a nightmare after that. I was getting constant “No network route (255)” errors in ZHA, and they’d randomly go unavailable or just ignore commands entirely.
After seeing your post, I went into the local Web UI for all 10 of them and manually disabled Bluetooth. Since then, they’ve been rock solid. Not a single dropout or “No route” error since. See also: Shelly support group (English Version) | Recently, many of you reported poor or completely non-functional Zigbee routing | Facebook
It’s pretty wild that the firmware even lets you keep Bluetooth on if it’s going to tank the Zigbee radio like that. Like you said, it definitely feels like this part of the Gen4 experience is “half-baked” and undocumented.
Thanks for sharing your fix
1 Like
Sorry to interject in the conversation as I’ve been following this thread closely. I just installed the Gen4 1PM device a few days ago with 1.71 firmware and had Zigbee activated. I was working through Chatgpt setting up the mqtt bindings for this device so I could control directly. In the process of getting things working I asked Chatgpt if I could deactivate the wireless and just control the device only through Zigbee. Chatgpt directly said this is not possible. It does not have a zigbee radio in the device. In layman’s terms Chatgpt said the current implementation is really an abstraction layer and everything for this device relies on wifi for control. The direct quote from chatgpt:
Why Shelly even exposes Zigbee/Matter hooks (the design intent)
Shelly Gen4 devices are built to:
- participate in ecosystems
- without rewriting firmware for each protocol
So they expose:
- Matter bridges
- Zigbee abstractions
- Cloud integrations
All of which are translation layers, not native radios.
Think of it as:
“Shelly speaks Wi-Fi natively, and pretends to speak Zigbee via translators.”
I can’t say I’m disappointed and maybe others in this thread actually knew this information. I thought however with the zigbee sticker on the side of the box this device actually contained a zigbee radio. If using this device with Home Assistant, what advantage is Zigbee here providing since ultimately everything seems to funnel through wifi? Here is yet another explanation via ChatGPT:
- Z2M (or a bridge) creates a virtual Zigbee endpoint
- That endpoint maps:
- Zigbee on/off → MQTT command
- Zigbee state → MQTT status
- Shelly receives the MQTT command over Wi-Fi
- Shelly publishes state over Wi-Fi
- Z2M reflects that back as Zigbee state
I’m really fact checking Chatgpt here as sometimes it’s either wrong or I don’t really understand it’s explanation.
ChatGPT is clearly hallucinating here. The Gen4 devices do have real Zigbee radios. The only thing is that the functionality is somewhat limited at the moment.
1 Like
Agreed with the above. You do not have to fact checking ChatGPT, and really could just assume that it is wrong - especially so for factual outputs such as specs.
Think about it: how would ChatGPT even know about any specs of any latest and greatest gadgets? Did ChatGPT ever train on the spec sheet from Shelly or from some 2nd hand / 3rd hand data, or data from other AI outputs? Even if so, how would any of those AI models ever know that they should split out results based on datasheet for Gen4 1PM only, and not letting any Gen3 or Gen2 datasheets polluting its results? Furthermore, how would ChatGPT know anything about Shelly’s “design intent”…?
Trying to say that, for what you are trying to get, this is not one of those scenarios where you should ask any of those AI models.
What sometimes works, is asking an LLM to “look for other people’s experiences online”. But then again, you could also just Google that. As @k8gg says, this is a use case where a good old search engine is better than an LLM.
Hey thanks for replying. I didn’t mean to stir the pot with the LLM stuff, just kind of amazed how wrong it got things. Anyway – just wanting to confirm some information I gathered through reddit conversation. As of yet, it seems device firmware can not be updated via zigbee mode. I was told this feature is coming which is great. I’ve seen recommendations that the device ideally shouldn’t be used with zigbee and wifi running simultaneously however if wifi is turned off, wifi can not be turned back on without a device reset. I was told that trying to activate or turn back on the wifi module via z2m entities would not work. I’m not sure if wifi reactivation via zigbee is a future feature or not as I did not receive any information regarding this.
I have 2 gen4 devices running with both WiFi and Zigbee enabled, without any problems. Bluetooth should be turned off though.
Thanks for this small but gugely important info. Regarding the firmware, for running on zigbee in HA under Z2M, besides disabling the BT of the Shelly device, do we still need to install the alternative firmware 1.7.1 or the official one ? Do we need to disable WiFi as well or only the BT ?