Is ZigStar the Best Way to Extend Zigbee Device Coverage in HA?

I recently watched a video on YouTube discussing ways to extend Zigbee device coverage in larger, multi-story houses. The video mentioned using an open-source ZigStar gateway to create a Zigbee dongle over TCP/IP. Here’s the link for reference: https://youtu.be/yY-aD1hvwr4?si=XtMr2AMnNxvE23TN

I bought one of these gateways to test it out, and I ran into two issues:

  1. If one of my rooms only has Wi-Fi (no wired connection), and I enable Wi-Fi via the ESP32, I can’t seem to successfully connect Zigbee devices. However, using a wired connection works fine. I suspect it might be due to Wi-Fi latency not meeting HA’s response time requirements for the Zigbee dongle. Has anyone else experienced this?
  2. I also tried connecting it through a router that is in the same local network but on a different subnet and passing through a secondary router. In this case, I encountered the same issues as in point 1. Could this also be related to latency?

Wondering if anyone else has faced similar challenges or has any insights on these issues. Would love to discuss this further.

Search is your friend:

TL;DR - Mesh networks are best left alone as the nodes make better routing decisions autonomously. Just add lots of mains-powered routing nodes.

I guess a TCP/IP-connected coordinator might work in a really large space, but careful as that’s perilously close to re-inventing a Matter / Thread Border Router ! :slight_smile:

Yes, this indeed competes with the concept of a Thread Border Router, but it seems there is still a long way to go before Thread Border Routers are spread throughout an entire house and Thread products reach the current richness and low cost of Zigbee devices. Also, given the current progress of Matter, it doesn’t seem to be proceeding all that smoothly. Is there a better solution during this interim period? Perhaps a low-cost Matter Bridge could be a good idea? Would love to discuss this further.

In a single Zigbee network, as long as the device compatibility among manufacturers is good and there are enough Zigbee Routers, supporting 50-100 devices is generally not an issue. However, if your house has multiple levels, it’s often challenging to ensure the condition of having “enough Routers”. In such cases, using a “bridge” that supports TCP/IP protocol to connect several scattered small Zigbee networks to a Matter Controller can often be easier than having an average user or installer learn how to optimize a Zigbee network.

I tested another solution:

Apple Home + HomePod + HA (with Matter Server installed) + 2*Matter Bridges (each bridge connected to 30 lights).

Setting this up was much easier compared to using Z2M, and it’s far more user-friendly for people who want a plug-and-play experience without spending too much time on the installation process. Additionally, this approach makes it easier to expand to more devices. More importantly, compared to methods like Ser2Net, which converts serial ports to TCP/IP, using the Matter protocol is more standardized and naturally solves transmission issues when bridges are distributed across different subnets, making control more stable and compatible with more platforms.

This is starting to sound more and more like a setup, and my bullcrap meter is going on high alert.

  • You post about a presumed issue where you quote a gateway by a company that doesn’t even seem to be mentioned in the video (he went with an SMLight in the end, not a Zigstar gateway)

  • You then propose that the solution in that video is to use Zigbee over TCP/IP. Anyone who watches the video over the 5 minute mark will hear the guy saying “It all went to shit again. The exact same thing happened…and I cried”.
    So, I’d guess that Zigbee over TCP/IP is definitely not the solution.

  • You follow this up with a tangent of enabling Wi-Fi via the ESP32. At this point, it’s unclear what you mean by that. Did you enable Wi-Fi on your Zigstar (LAN) router and expect it not to interfere with the Zigbee chip that is literally touching the board? Did you try to pass Zigbee over Wi-fi using a different subnet? At this point, it’s anyone’s guess, but both of those are bad ideas.

  • Someone bites (sorry, James) and points you in the direction you’ve been intending all along. The brief mention of matter is enough of a lead. You propose a “low-cost matter bridge” as a solution to the “problem”.

  • You follow up with some nonsense about “device compatibility among manufacturers” - something that is negated by the core reason of existence of ZHA, Z2M & Deconz, not to mention HA itself.

  • You also mention issues with routing over a multi-level house - something which I have personally not encountered despite living in such a house with cement floors (with iron mesh) and 18cm thick brick walls. Your solution? Multiple (separate) Zigbee meshes. That might be required for a detached garage, but not for inter-floor routing.

  • You present your “solution”, which, surprise surprise, leads to your company’s site, with a handy link to buy your products as the cherry on the cake. Link points to a 12-pack of lights pack at $250, and there’s no option to buy the matter bridge you mentioned by itself. You mention you need two bridges, and the cheapest option which contains the bridge is a $109 4-light pack X2. Cool, so we’re looking at a $218 option to get 2 bridges which can be resolved with $60 in the video.

Sorry, but no. That is not a solution - at best, it’s conveniently hiding the actual solution for those who stumble upon this post before making further research, and I cannot stay silent on the matter.

The actual solution (which is also explained in the linked video) is to buy a networked coordinator. No extra $100+ Pi or $218 band-aid required. No interference issues. No more worrying about coordinator location. Most importantly - no need to use a separate protocol (plus Apple Homepods) to solve the “issue” of a different protocol.

@Hedda has spent years writing extremely detailed guides on this very topic. Hell, if anything, he’s too detailed. He has contributed to the ZHA documentation to the point that he’s practically written the whole thing himself.
I may not always agree with him, but his contributions have helped hundreds of people, way more than this parody of a workaround (with associated company profits) ever could.

9 Likes

It is, he has another thread trying to get people to test his matter lights and controller. I don’t necessarily have an issue with a company trying to get product into people’s hands so they can generate business, but this post just seems shady.

2 Likes

I know. :face_with_symbols_over_mouth:

All “I’d love to discuss”, and no “thanks for trying to help me with a genuine problem”.

Z-Wave Ethernet coordinators died a death outside of Las Vegas casino hotels as the protocols were closed and proprietary with NDAs incompatible with FOSS.

IMHO, all single-vendor solutions or protocols are doomed the same way.

1 Like

I want to sincerely apologize to everyone for my previous post, which used a bit of a marketing tactic: raising an issue and then presenting a solution involving our product, hoping to draw attention to our brand. I understand now that this approach wasn’t the most transparent way to engage with the community. Moving forward, I want to be more straightforward—either genuinely discussing technical issues or, if relevant, advertising clearly (provided the moderators permit it). I owe a big “Sorry” to @FloatingBoater and others who engaged honestly.

To be candid, we are a new brand in the North American market, selling Zigbee + Matter smart lights on Amazon. After certifying our Matter Bridge product last year, we hoped to leverage the Matter buzz to quickly gain market traction. Unfortunately, it hasn’t played out that way. We found that consumer awareness and acceptance of Matter aren’t where we thought they would be, and the Matter experience still has challenges that need to be addressed.

However, we did notice that many of our customers come from the Home Assistant ecosystem. As we explored HA, we became increasingly excited about its possibilities, particularly as the Matter server component evolved. It felt like we were on the right path.

Yet, we are still newcomers to HA, and a big question remained for us: Does our Matter Bridge make sense within the HA ecosystem? Our lights work seamlessly with HA, without needing any code modifications—which was a pleasant surprise (kudos to Koenkk!). This made us wonder if selling just Zigbee lights without a Matter Bridge would be enough.

Previously, I posted about a supposed issue of Get 50+ Free Zigbee Lights for Testing Whole-Home Matter-Compatible Lighting with Home Assistant! The idea behind that post was to find real users in the U.S. who might be facing such an issue and provide them with 100 lights for free, in exchange for helping us validate the problem and potential solutions. But we haven’t found many users facing such a large-scale scenario—we’ve only found one volunteer so far, and they only have 15 lights in their home.

This was a bit disappointing! Later, I saw the linked YouTube video, which showed that some people were indeed facing similar issues, but they opted for a ZigStar gateway (I should clarify here: the user actually mentioned an SMLight gateway. My experiments used a UZG-01 gateway, which I believe uses the same open-source software as ZigStar). I did encounter the exact two issues mentioned—cross-router problems and failed device addition when Wi-Fi was enabled—but I didn’t delve deeply into why, and for that, I apologize again. My comments on the ZigStar and SMLight solutions were not rigorously supported, and I should not have casually criticized them.

Regarding my remarks on ‘device compatibility,’ I was referring to prior issues we had when integrating third-party devices with our gateway. These issues involved different chips and protocol stack routing, and were not meant to imply that HA + Z2M necessarily has similar issues. This was my oversight, and I’m sincerely sorry for not making that clearer.

I have also removed the purchasing link from my original post. I realize that it was inappropriate in this context. I do, however, want to respond to ShadowFist’s comment on the pricing of ‘our solution.’ We are not currently selling the Matter Bridge separately because it does not have the versatility of Z2M in supporting a wide range of Zigbee devices, particularly non-certified ones. Our current package costs $249 and includes 12 lights, a dimmer switch, and a Matter Bridge—effectively making the Matter Bridge free when compared to similar market offerings. The current Matter Bridge is Linux-based, which contributes to its cost. We are working on a more cost-effective RTOS-based version and conducting market research to ensure we are headed in the right direction. Recent posts were not so much about pushing products but about gathering genuine user insights.

Thank you, @ShadowFist , for patiently pointing out my mistakes. Your feedback has been very objective and constructive. It helped me understand how to be more honest and effective in my communication with the HA community. I am still relatively new to HA—I’ve only been exploring it seriously for the past six months, but I’ve migrated my whole home setup to HA and find myself falling in love with the platform, especially given the supportive and knowledgeable community here. @James, I also appreciate you pointing me to @Hedda’s detailed posts—they were immensely helpful. I hope, in time, I can contribute meaningfully to HA as well.

Finally, if no one objects, I’d still like to continue the discussion on the ‘manufactured issue’ of multi-Zigbee network cascading. If there truly is a need for cascading multiple Zigbee sub-networks, setting up multiple Z2M instances might not be user-friendly for the average consumer. What are your thoughts on this?

Again, I’m open to constructive criticism and appreciate the patience everyone has shown.

Thanks for wasting the time of volunteers.

Manners maketh the man - you might want to reflect on what that means.

Radio waves are fundamentally 3-dimensional, so mesh networks are fundamentally 3D.

Creating multiple mesh networks ON THE SAME FREQUENCY can only cause interference issues.

2 Likes

Thank you, @FloatingBoater. The phrase “manners maketh the man” really struck a chord with me. I realize that in my previous post, I didn’t show enough respect for the time and effort that volunteers put in, and that was truly my mistake.

The radio interference issue you raised is indeed a fundamental one, and it cannot be solved purely through software. The only viable approach is to mitigate it by setting different mesh networks on separate frequencies. Do you think this would work?

Your mention of the historical demonstration of the Z-Wave Ethernet coordinator in Las Vegas really impressed me—it shows your deep understanding of the industry. I would also love to hear your thoughts: when it comes to a more standardized and open technology like Zigbee, do Ethernet coordinators or ‘Matter Bridge’ have potential applications?

If so, this approach might help in complex or large environments, making deployments easier to manage. I genuinely want to explore this possibility and would be grateful for any insights or experience you have in this area.

Certainly. In a virtualized environment you don’t want hardware like USB sticks locking your workload to a specfic host / server. They are also much easier to place centrally, you can have one in your living room while the server hardware resides in a closet in the basement for example.

I agree with this point as well—having a USB dongle attached to a box really doesn’t look like a mature product solution.

There are two approaches:

  1. USB Dongle on Main Host or Server: This involves installing the USB dongle directly onto the main host or server.
  2. Independent Host with Separate IoT Connection Box: Use an independent host along with a separate IoT connection box (or the box itself could be part of some device).

Personally, I prefer the latter, especially when platforms like Home Assistant (HA) integrate with AI. HA can act as the “brain” across different servers, while IoT devices, as its “hands and feet,” should ideally operate independently.

No. You are totally not getting it. There are three approaches and I thought I made it pretty clear in my previous post. Hell, you even referenced a couple of Zigbee coordinators which use this very approach.

The third approach involves buying a LAN coordinator (dongle) and plugging it into whatever network point you have handy. No “Independent Host with Separate IOT Connection Box” required AT ALL. It will communicate with HA over ethernet via your network.

Please stop trying to paint a picture where your product is absolutely required. You are trying to sell a solution to a problem which doesn’t exist - using a (your) matter hub to control (your) zigbee lights.

Well, not quite. I found the PR you someone submitted 2 months ago and it paints a clearer picture.
Seems like your zigbee offerings are simply rebadged Feibit models which already had quirks applied. All you basically had to do was to add the model names and a couple of entries for upcoming devices, and even that required a few corrections by Koenkk.

Basically, most of the “device compatibility” issues had been ironed out way before you had ever heard of HA or Z2MQTT. That is why you ran into issues when trying to integrate other zigbee devices without any quirks/converters.

I can offer criticism (constructive or otherwise) all day long. However, my patience is somewhat more limited.
I refuse to fall prey to devious marketing tactics and half-truths. It only tells me one thing about the company - if this is the extent of deception they are willing to go through to make a sale, they simply cannot be trusted to provide support after I’ve given them my money.

Your “bit of a marketing tactic” seems to have spectacularly backfired and, speaking personally, you seem to have lost any hope of trust in this community. I sincerely hope this teaches you a valuable lesson in research and transparency for any future endeavours.

3 Likes

Thank you, @ShadowFist . You have indeed taught me a valuable lesson: my Matter Bridge has no value for HA clients. I will consider new research directions and once again apologize for my marketing tactics!

I, for one, am personally grateful for all the folks on here that point out the very shady practices of a company that I would never want to do business with.

The mere fact that you even attempted a bait-and-switch tactic in this community, to me, marks you as a dishonest scheming person and simply just I don’t do business with such people. Your apologies seem perfunctory and pandering to me, it you could have gotten away with this you would have taken it to the finish line even though there is no way you didn’t know precisely what you were doing and for what reason, it was set to be a bait-and-switch from the beginning and it’s impossible you didn’t just know that but do it on purpose.

Fortunately you have caught the attention of all the mods and all the top users, the likelihood that you will ever scam HA users with your products is pretty low now.

And it’s not that your product has no value to HA clients, your company has no value to anyone whatsoever, anywhere, regardless of platform.

1 Like