The best gets better - Home Assistant Connect ZBT-2

Blog
83

@IOT7712 - Thanks for the perspective, and you’re absolutely right about Silicon Labs being caught in the middle here. I completely agree with the pragmatic approach - “if you always wait for the best, you’ll never commit.” That’s solid advice.

But let me share my thinking, especially coming from the ZWA-2 experience.

The ZWA-2 Standard

I paid 299 PLN (~€69) for the ZWA-2, and it’s genuinely the #1 Z-Wave coordinator on the market right now. Why? Because it has everything the current Z-Wave specification offers:

  • Z-Wave 800 series (latest chip)
  • Z-Wave Long Range
  • Security 2, SmartStart
  • All features from the 2024B spec

There’s literally no newer Z-Wave technology available. Silicon Labs isn’t announcing a 900 series. The ZWA-2 will remain cutting-edge for years. That’s what I call a justified investment.

The ZBT-2 Reality Check

ZBT-2 costs 199 PLN (~€45) and launches on November 19, 2025.

But here’s the timeline that bothers me:

  • November 18, 2025: CSA announces Zigbee 4.0 and Suzi
  • November 19, 2025: NabuCasa releases ZBT-2 with Zigbee 3.0

One day difference. But NabuCasa is a CSA member - they likely knew about Zigbee 4.0 and Suzi months in advance. Silicon Labs definitely knew they’d need new hardware for Suzi support (Sub-GHz requires different radio capabilities).

The Product Cycle Concern

Let’s compare release cycles:

Philips Hue Bridge:

  • Gen 1 (round): October 2012
  • Gen 2 (square): 2016 (4 years later)
  • Bridge Pro: September 2025 (9 years later!)

NabuCasa Connect:

  • ZBT-1 (SkyConnect): October 2022
  • ZBT-2: November 2025 (3 years later)
  • ZBT-3 with Zigbee 4.0/Suzi: 2026? (1 year later?)

I guarantee Philips won’t release a v4 bridge in 2026. They just released the Pro after 9 years.

My Point

I don’t expect NabuCasa to release new coordinators every year. I expect them to keep their finger on the pulse and release products that will be #1 for years, just like the ZWA-2.

If ZBT-2 had launched in 2026 with Zigbee 4.0 and Suzi support, I’d happily pay 299 PLN for it - same as I did for the ZWA-2. Because it would be future-proof.

Instead, we’re looking at potentially spending 2×199 PLN = 398 PLN for ZBT-2 → ZBT-3 within a year or two. That’s not economically justified for a device that might only be “best for HA” for 12 months.

The Software Issues

The integration problems I mentioned (missing LED control, no status sensor) will likely get fixed - I trust NabuCasa’s software team completely. Once those are resolved, ZBT-2 will absolutely be the best Zigbee coordinator for Home Assistant.

But for how long? If Zigbee 4.0/Suzi devices start appearing in 2026 (certification opens H1 2026), we’re right back to needing new hardware.

Bottom Line

I love HA and NabuCasa. I’m building the best HA system possible, which is why I buy their hardware. But ZWA-2 set the bar incredibly high - it’s a device that will remain #1 for years.

I just wish ZBT-2 could have achieved the same standard. Waiting until 2026 to include Zigbee 4.0/Suzi would have made it truly exceptional, even at a higher price point.

Just my thoughts as someone who wants NabuCasa products to always set the industry standard.

5 Likes
84

I just received my ZBT-2 and tried following the migration path documented at go-nabucasa.com/zbt-2

I got an error during the flashing step and was told to check the logs. The logs say

Logger: homeassistant.components.homeassistant_hardware.firmware_config_flow
Source: components/homeassistant_hardware/firmware_config_flow.py:202
integration: Home Assistant Hardware (documentation, issues)
First occurred: 11:40:48 (1 occurrence)
Last logged: 11:40:48

Failed to flash firmware
Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/components/homeassistant_hardware/util.py", line 413, in async_flash_silabs_firmware
    await flasher.enter_bootloader()
  File "/usr/local/lib/python3.13/site-packages/universal_silabs_flasher/flasher.py", line 358, in enter_bootloader
    await self.probe_app_type()
  File "/usr/local/lib/python3.13/site-packages/universal_silabs_flasher/flasher.py", line 329, in probe_app_type
    raise RuntimeError("Failed to probe running application type")
RuntimeError: Failed to probe running application type

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/components/homeassistant_hardware/firmware_config_flow.py", line 202, in _install_firmware_step
    await self.firmware_install_task
  File "/usr/src/homeassistant/homeassistant/components/homeassistant_hardware/firmware_config_flow.py", line 297, in _install_firmware
    self._probed_firmware_info = await async_flash_silabs_firmware(
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ...<9 lines>...
    )
    ^
  File "/usr/src/homeassistant/homeassistant/components/homeassistant_hardware/util.py", line 420, in async_flash_silabs_firmware
    raise HomeAssistantError("Failed to flash firmware") from err
homeassistant.exceptions.HomeAssistantError: Failed to flash firmware

and now the ZBT-2 isn’t recognized anymore. The documentation does not link to troubleshooting hints if this step fails …

EDIT: I was brave and power cycled the ZBT-2 and restarted Home Assistant. This time around no issues with the migration, most devices are up again (-:

85

@edigorn here are some facts for the record:

  • ZBT-2 firmware will for sooner or later be upgradable to Zigbee 4.0 and will then be able to fully support 2.4GHz Zigbee 4.0 devices for sure, so the argument about ”Zigbee 4.0” really is moot as that part is only software and can come in future firmware updates.
  • ZBT-2 hardware is however 2.4GHz only and does not have a dual-band radio so it will never be able to support the new ”Suzi” standard for Sub-Ghz Zigbee devices (a.k.a Zigbee Long-Range) because it does not have built-in radio transmitter and reciever for the 1-1000Mhz frequency range.
  • Silicon Labs (which is the largest manufacturer of Zigbee chips) have not yet updated their Zigbee SDK and EmberZNet Zigbee stack with Zigbee 4.0 so no one else using Silabs chips can update their firmware either, not for end-devices and not for Zigbee Coordinator apaptets. Again, as soon as that happens device manufacturers can choose to use it, and Home Assistant’s ZHA developers (who also mainatain the ZBT-2 firmware) can check to see if it is deemed mature/stable to update ZBT-2 with.
  • There are no Suzi devices available on the market today and I bet it will be long before we see any because most manufacturers of Zigbee devices want to target the largest audience to buy their devices, and since there are no other Zigbee gateways or Zigbee Coordinator adapters on the market with Suzi support they have no good reason to rush put Suzi devices when there is no buyers that can use them yet. If history shows anything it is that it took many years after the release of the Zigbee 3.0 specification before we started to see devices shipping with Zigbee 3.0 support out-of-the-box.
  • With exeption of ”Suzi” Sub-GHz frequency range Zigbee is backwards-compatible so as long as the future devices you buy support 2.4GHz then you can use them today and upgrade to a newer Zigbee Coordinator adapter at a later date.
  • You can also use ZBT-2 as a Thread radio and the Thread standard protocol only support 2.4GHz.

Remember that radio adapter migrations are now super easy today so there is nothing stopping someone to migrate their Zigbee network to ZBT-2 today and use it with every Zigbee device currently on the market and even Zigbee 4.0 devices in the future as long as they support using 2.4GHz frequency band. Later when there actually are Suzi devices on the market that only support Sub-GHz you can upgrade to a ZBT-3 or whatever other upcoming Zigbee Coordinator that will support dual-band operation on a single Zigbee adapter.

In summery, you will be able to buy and connect devices with a ”Zigbee 4.0” badge but you not be able to connect upcoming devices with the ”Suzi” badge unless those devices themselves are dual-band (which some devicesemphasized text will be) as will only ever support the 2.4GHz frequency range.

Anyway, take a deep breath and realize that you are complaining about ZBT-2 not supporting devices that do not yet exist on the market, and there is nothing you can do to change the fact that the existing ZBT-2 can not ever support Sub-GHz radio transmissions. Suggest you can step away from the keyboard now and come back when there actually is a Suzi device you can buy and is available shipping.

Please do take this the wrong way, I love to look forward to future products and features myself, but such ideas belongs in feature request for upcoming products and not in complaints in threads about a product that have already been released. Neither you or them can time-travel to change the hardware of a product that has been released. Look forward to ZBT-3 instead, so maybe start separate wish-list feature request discussion thread about that.

6 Likes
86

Counter Point: You wouldn’t have been happy at all. You’d be an early adopter of a new standard and have to face all the teething issues early adopters face, ending up frequently having to update the firmware just to get stuff working stably. That’s not to mention the fact that a year from now, there will only be a handful of Suzi devices available to purchase.
We’ve all seen this happen recently when Matter was announced. It’s taken them 3 years to support all devices at a firmware level. Hell, camera support was added in Matter 1.5 just a couple of days ago, and the actual cameras won’t be on sale until mid 2026.

Nabu did the right thing by forging ahead and releasing a stable product (which should theoretically support Zigbee 4.0 as is). If they had waited for Suzi to be commercially viable, they’d have had to wait at least until 2027.

Edit: @Hedda beat me to it.

7 Likes
87

Ordered and received mine already. There was some funnyness with the firmware update, but there was a fix online and apparently the issue is now solved with 11.3. Thereafter setup was flawless and my network is performing very well. Strong recommend and I would gladly buy a second one if I start having a lot of Thread devices, for now I am all about the Zigbee. For me, as an electronic engineer, it makes sense not combining Thread and Zigbee in one radio.

2 Likes
88

@Hedda @ShadowFist - Thank you both for the detailed technical breakdown.

You’re correct on the Zigbee 4.0/Suzi technical points:
• No Suzi devices exist today
• No Silicon Labs SDK for Zigbee 4.0 yet
• ZBT-2 will receive Zigbee 4.0 features via firmware update (for 2.4GHz band)
• Early adopter challenges are real (Matter rollout proves this)

I understand the technical limitations. However, I respectfully disagree with the conclusion that NabuCasa “did the right thing by forging ahead” with the November 2025 release.

Here’s why:

THE TIMING ARGUMENT - Still Valid

Let me clarify my position on timing, because I think it’s being misunderstood.

I’m NOT suggesting NC should have “cancelled launch last minute” or “scrapped production mid-cycle.”

I’m suggesting the launch timeline itself was poorly planned:

ZBT-1: October 2022
ZBT-2: November 2025 (3 years later)
ZBT-2 with Zigbee 4.0/Suzi: Late 2026 or H1 2027 (4-4.5 years later)

Waiting until late 2026/early 2027 would have been a perfectly reasonable product cycle - actually closer to industry standards (Philips: 4+ years between major releases).

@Hedda, you mentioned “you are complaining about ZBT-2 not supporting devices that do not yet exist” - but that’s precisely my point. If NC had planned a late 2026/H1 2027 release instead of rushing to market in Nov 2025, those devices WOULD exist by launch time, and ZBT-2 could have been a complete, future-proof product like ZWA-2.

THE ECONOMIC ARGUMENT - User Perspective

From a user value standpoint:

Software bugs → Fixed via free updates ✓
Hardware limitations (no Suzi) → Requires purchasing new coordinator ($49)

Current path:
• ZBT-2 (Nov 2025): $49 - Zigbee 3.0 only
• ZBT-3 (2026/27?): $49 - Zigbee 4.0 + Suzi
• Total cost: $98

Better path:
• ZBT-2 with Zigbee 4.0/Suzi (late 2026): $69 - Complete solution
• Total cost: $69

The “wait for better tech” argument works BOTH ways. Yes, you can always wait for next-gen tech. But when that next-gen tech is announced literally ONE DAY before your product launch, the timing becomes commercially questionable.

THE MARKET REALITY - Who’s Actually Buying?

Let’s be honest about who will purchase ZBT-2 now vs later:

Will buy ZBT-2 today:
• Enthusiasts supporting the HA project (like me)
• Users who don’t follow Zigbee spec announcements
• People prioritizing design aesthetics

Will wait for ZBT-3:
• Existing ZBT-1 users who want meaningful tech upgrade
• Users who read this forum and know Zigbee 4.0/Suzi is coming
• Smart home builders planning 5+ year installations

By releasing ZBT-2 one day after Zigbee 4.0 announcement, NC essentially created a “wait for next version” scenario for a significant portion of their potential customer base.

That’s not optimal product strategy.

THE ACTUAL VALUE PROPOSITION - What Are We Really Buying?

Let’s analyze what ZBT-2 actually offers over ZBT-1:

4x baud rate improvement (115.2k → 460.8k bps)? Yes, this is a real technical upgrade for internal communication speed. I acknowledge this improvement.

However, for end-user experience, the difference between “instant” and “4x more instant” is marginal. Zigbee device response is primarily limited by mesh hop latency and device firmware, not coordinator serial speed.

Improved range? In Zigbee, mesh quality depends on router/repeater quantity and placement, not primarily on coordinator antenna specs.

Future-proof tech? Missing Zigbee 4.0 hardware (Suzi), so no.

Better HA integration? Actually worse than ZWA-2:
• LED control entity disappears after firmware update
• No status sensor (ZWA-2 has both)
• ~12+ users experiencing firmware installation failures

Better design? Yes - genuinely excellent industrial design.

So realistically, we’re buying:

  1. 4x serial communication speed (nice-to-have, minimal user impact)
  2. Premium design aesthetics (significant value)
  3. Support for the HA/NabuCasa project (primary value)

Those are valid reasons! But they’re not the comprehensive technical superiority reasons marketed in “The best gets better” blog post.

THE INTEGRATION REGRESSION - Critical Issue

Speaking of ZWA-2 comparison, here’s what concerns me most:

ZWA-2 (launched October 2025):
• LED control entity ✓
• Status sensor ✓
• Tilt detection with accelerometer (LED blinks if not upright) ✓

ZBT-2 (launched November 2025):
• LED control entity: Present initially, DISAPPEARS after firmware update ✗
• Status sensor: Missing ✗
• Tilt detection: Not mentioned in specs, not available in HA integration ✗

Question: Does ZBT-2 have accelerometer hardware that’s simply not exposed in HA integration? If the hardware exists but isn’t accessible - why not? If it doesn’t exist - why the regression compared to ZWA-2?

This is a genuine question because I can’t find this information in official specs or community discussions.

The broader issue: Why does the NEWER product (ZBT-2, Nov 2025) deliver LESS complete Home Assistant integration than the OLDER product (ZWA-2, Oct 2025)?

THE DAY-ONE QUALITY PROBLEM - This Cannot Be Dismissed

@ShadowFist, you mentioned early adopter teething issues if ZBT-2 had launched with Zigbee 4.0 in 2026.

But we’re experiencing teething issues RIGHT NOW with the current Nov 2025 release:

• ~12+ users with firmware installation failures (ZBT-2 Error - Zigbee firmware failed to install, check Home Assistant logs for more information)
• LED control entity regression (present initially, disappears post-update)
• Missing status sensor (ZWA-2 has this)
• Workarounds required: web flasher, manual procedures, device power cycles

Compare to ZWA-2 (launched Oct 2025):
• Zero firmware issues
• Complete entity integration
• True plug-and-play experience
• Genuinely “rock-solid” as marketed

So the “avoid early adopter issues by shipping stable tech” argument doesn’t hold - we’re dealing with day-one stability issues anyway.

The critical difference is: ZWA-2’s issues (if any) can be fixed via software updates. ZBT-2’s missing Zigbee 4.0 hardware cannot.

The 4x speed improvement is real, but it doesn’t compensate for missing LED control, missing status sensor, and firmware installation failures affecting day-one user experience.

THE UNANSWERED QUESTION - Critical for Users

As someone experiencing these integration issues, I need clarity from NabuCasa:

Are the missing entities (LED control, status sensor, potential tilt detection) a software bug that will be patched?
Or is this a hardware/firmware limitation of current ZBT-2 units?

Should I:
a) Keep the unit and wait for software fixes?
b) Return it as defective and wait for a corrected hardware batch?

This directly impacts support decisions for users, and NC hasn’t provided clear guidance.

BOTTOM LINE - Constructive Criticism from a Supporter

I love Home Assistant and NabuCasa. I own four pieces of their hardware specifically to support the project and because ZWA-2 set an incredibly high quality bar.

My criticism comes from a place of wanting NC to maintain that ZWA-2 standard across all products.

The technical arguments about Zigbee 4.0/Suzi are valid - you’re right that the tech isn’t ready today. But the product planning timeline that led to launching ZBT-2 one day after Zigbee 4.0 announcement was, in my view, a strategic misstep.

More importantly, the day-one integration and stability issues represent a quality regression vs ZWA-2, and that’s concerning regardless of Zigbee version support or baud rate improvements.

I’m confident NC will address the software issues through updates. But I hope they’ll also reconsider their product cycle planning to avoid similar timing problems with future releases.

ZWA-2 will be #1 for years because it shipped with complete, mature spec support AND flawless HA integration. I wish ZBT-2 had been given the development time to achieve the same standard.

2 Likes
89

Is this some kind of trolling, where you let an LLM repeat the same things over and over(each time with an even longer text) with new words? I’m really irritated …

19 Likes
90

Agreed, that’s a ridiculous reply. By the time the new protocols stable and there’s actually equipment to support it in 3 to 5 years I’m sure they can make a new one. If you have the money to buy all new equipment to support that protocol you have money to buy new adapter and also support home assistant.

3 Likes
91

Yeah, given all of their experience working with similar processes like the matter standard and the device protocol lifecycle (as new device types and improvements get added to matter), it seems weird that you would assume you know more than they do with product releases and that you know what the best decision was. Given the success of their hardware products, it seems clear they know what they’re doing

1 Like
92

Not sure if I even need it, but I’m going to upgrade, even it t was only to support HA. I frikking love HA :slight_smile:

1 Like
93

I think the comparisons with ZWA-2 are very misguided. Software support for ZBT-2 in Home Assistant is multiple times more complicated, because ZBT-2 is used in multiple deployment scenarios (ZHA, Z2M, Thread), whereas ZWA-2 only needs to be concerned with Z-Wave JS.

2 Likes
95

So edigorn, your post reads to me like fleshed out AI slop, a thread summary gone wrong. This is typical of an AI model that has insufficient training and has no knowledge/understanding of technical product development cycles. It cannot pretend to be intelligent as it has no awareness. Embarrassingly so. If you didn’t use AI, then it reflects poorly on you, rather than the Nabu Casa product. Time will tell, in probably just a few weeks. The other option is you have a hifden commercial interest in this product failing, something I would decidedly frown on, especially in this community.

Are you are saying I should delay my future purchase of a just announced ZigBee 4 protocol compatible product to save $29 for a new device that does not have a release date and has not even hit the design phase yet? Deprive myself now for the promise of something better at some unknown time in the future? You think the vaporware product will be priced the same as the just announced product, some time in the future? A very long bow to draw.

I think that is the definition of vaporware.

The issues you are touting as major problems (the LED on top, seemingly vital for operations the way you are putting it), will most likely be remedied within days with another firmware update. We know ithe LED hardware works already, so there is no design flaw. Just a quick software update.

If you get one for Christmas, I bet it will have the LED working, and playing along with Jingle Bells with a custom HomeAssistant addon someone has written, just for the lulz!

7 Likes
96

@Thyraz @orange-assistant @IOT7712 - Point taken regarding the post length.

I realize my analytical approach can be a bit heavy to read, but there is a practical reason for the formatting: I am visually impaired and rely on screen readers. Structuring posts with headers and lists makes them readable for me and others with accessibility needs.

I just want this product to work as flawlessly as the ZWA-2, hence the detailed feedback. Sorry if the style annoyed you, but being meticulous is just how I tackle technical edge cases.

To avoid adding more noise to this thread, I’ve moved the specific technical questions (about missing entities and hardware limitations) to a dedicated bug report:

Let’s focus on the technical solution there.

Thanks.

4 Likes
97

Fair point on being visually impaired, and admirable that you have tools to assist and empower you, hence my query on why having a functional LED on top is so important when it will be remedied with a simple firmware and software update very soon. We know it worked with a previous version of firmware, it doesn’t with the one you are reporting on as being an issue to avoid purchase, and the remedy, based on previous experience will be here before your article has a wider audience by being slurped up by AI bots and (to my growing alarm) offered as an ongoing fact, forever, ad nauseum by so-called Artificial Intelligence helpers to the unsuspecting masses, looking for a device that exists right now, solves their immediate need elegantly, uses current technology, is priced competitively, and supports the HomeAssistant ecosystem through Nabu Casa hardware sales.

You know - you bought one, and have documented your experience. The other thread outlines workarounds currently attempted, and the [bug discovered/ report bug/review cause /implement workarounds/ update to fix /test fix actually works/release fix to all, rinse and repeat] cycle, which is currently only part way through, but you make it out as a major flaw that should affect purchase decisions. By the very nature of the bug and software release cycles of Nabu Casa vs HomeAssistant not being lockstep, there will be that short window that you will have to wait for the fix to be achieved.

For purposes of completeness, be sure to document here and in the other thread on the solution when it arrives, which may be as close as the next weekly/ monthly release.

Can somebody point me to the GitHub report to the developers so I can follow the software changes and fixes for my interest? I’m sure it is buried somewhere in the other thread, but cannot find it here. Thanks.

2 Likes
98

To futher reduce noise to this thread I started a dedicated hardware feature request for new or updated adapter with “Suzi” Sub-GHz Zigbee support to collect suggesed hardware additions:

PS: Suggest anyone who like to see some PoE variants submit yet another new feature request.

1 Like
99

Hi all,

I received my ZBT-2 today. Currently, I run Z2M with about 50 devices. Reading all of the migration docs of the ZBT-2, sounds like there is no path to migrate from Z2M to ZHA on ZBT-2 that allows you to keep entity IDs and device names. Is that right? I’m not keen to have to rename hundreds of entities and curious if I missed a workaround.

Cheers!
Matt

100

The workaround would be to use the ZBT-2 with Zigbee2mqtt

1 Like
101

Yeah, I didn’t see an explicit migration for that in the FAQs - but maybe it’s easier than I think. Thanks.

102

You mean like this one? You’ll still lose your device names though, bummer.

103

But that one migrates Z2M to ZHA, right?