This is a third of my thread devices and they are constantly dropping IPv6 since this firmware update
Either it’s the CPU scheduler changes or Athom’s implementation of Matter 1.5 hasn’t gone to plan and is wreaking havoc on my Thread network, which I will stress has been working FLAWLESSLY for the best part of nine months.
There is nothing wrong with my router, it’s all IPv6 compatible with Homey and as I said, has been working absolutely fine, until 2am yesterday morning when v13.3 was installed.
I have 60 Thread devices and more than half of them are now unavailable.
If I restart Homey, it slowly builds the network, but then can’t maintain it and devices start dropping off
The radio is underpowered or Matter 1.5 is to blame
I understand why it looks that way, but that’s not quite how it works.
The CPU isn’t simply being “fed less power.” Modern SoCs continuously adjust both their clock speed and voltage based on the current workload. When there’s little work to do, the CPU runs at a lower frequency and voltage to save power. As soon as more performance is required, the governor increases the frequency again.
It’s also important to distinguish between CPU utilization and CPU performance. Running at a lower frequency can make the reported CPU utilization appear higher because the same amount of work takes longer to complete, but that doesn’t necessarily mean the CPU is unable to keep up.
The Zigbee, Z-Wave and Thread radios are also dedicated radio chips with their own firmware. While the main CPU is responsible for e.g. processing messages and running automations, a lower CPU frequency does not automatically translate into degraded radio performance.
If you’re experiencing actual issues such as reduced radio performance after updating to 13.3.0, I’d recommend getting in touch with our support team. They can collect diagnostics from your Homey and help determine whether there’s an underlying issue specific to your setup.
Was there any extensive testing done with a high density thread network (50+ devices) on Homey Pro 2023/2026, before this firmware was rolled out?
How can you be so sure that this isn’t related to the firmware and its my “setup” that’s to blame?
Why would running at a lower voltage be a good thing, when we’ve already seen in the past that feeding Homey less voltage has a seriously detrimental effect on the performance of the Thread/Zigbee radio.
As for testing, we of course test these releases extensively before they go out. That said, it’s also important to remember that no two smart home setups are the same.
For example, I’ve personally been running this firmware since early May with around 26 connected Thread devices, next to that I’m also running 60 or so Zigbee devices. I haven’t experienced any issues.
So please reach out to support, that way we can see what’s going on.
The voltage the CPU is using is very different from the supply voltage of Homey as a whole.
A good analogy is a car engine. When you’re accelerating hard or driving uphill, the engine needs more power, so it uses more fuel. But when you’re cruising at a steady speed or idling, it only needs a fraction of that. Supplying the same amount of fuel all the time would just waste energy and generate unnecessary heat.
Modern CPUs work in much the same way. They dynamically adjust both their clock speed and core voltage based on the workload. When the CPU has little to do, it lowers both frequency and voltage to reduce power consumption and heat. As the workload increases, both scale back up automatically. This has no relation to the fixed supply voltage powering Homey itself or the Zigbee/Thread radio.
Thread networks re-builds over the course of 10-15 minutes
Then devices start dropping off again
26 connected Thread devices isn’t the same as 50 now is it.
There is a fundamental flaw with this firmware which is directly as a result of underperformance of the Thread radio when attempting to maintain a high density mesh
So have you contacted Athom support as @Doekse suggested? It seems to me that you want one of their developers to look into this matter, especially if you believe it is caused by a change in the new firmware.
I haven’t seen similar reports by other users on this forum after the new firmware was released. Maybe something specific to your setup is contributing to the issue you are facing. I hope a technical analysis by Athom can help solve the issue.
I have a Matter over thread network with 26 devices and a Zigbee network with 17 devices working without issues running on this firmware. I do not see a regression in the stability of the networks.
What is interesting in this case, regarding your remark on the amount of paired devices and possible voltage drop: Homey uses the same radio for both Zigbee and Thread communication.
You indicate that you have about 50 Thread devices. @Hsd1965 indicates that he has a combined total of approx. 45 Tread and Zigbee devices. I personally have more than 70 Zigbee devices paired (no Thread devices).
If the issue would we purely related to the voltage available to the radio, given that both protocols use the same radio, then I would expect that also large Zigbee or combination of Zigbee & Treads networks work be affected.
Some other thoughts:
Next to the Homey OS firmware update, can you think of any recent other software update (on other 2.4GHz devices) that might affect communication?
Could it be that something changed in the 2.4GHz channels used by other devices in your home or perhaps close neighbors? Like a WiFi router/access point?
Did the location of your Homey in your home change or did another appliance/device, like an older microwave oven, a cordless phone or an older Bluetooth device or a closeby USB 3.0 device/connection, was moved closer to Homey, which could cause interference?
Homey is in the same place as it’s been for more than a year.
It’s on my 1st floor in the middle of the house, connected by ethernet (the correct way) and it’s more than 3metres away from my network stack and router, which is housed in the loft.
Microwave is on the ground floor and hasn’t been plugged in for months, because the utility room is being renovated (no devices live there)
We don’t have a landline in the house and no new bluetooth devices have entered the house in months.
I added two IKEA KLIPBOKK about a month ago now, but no new devices other than that in 3-4 months
The only thing that has changed in the past month, is the Homey firmware.
When devices are ‘online’, Thread Network Information show devices as regularly disconnecting, which previously had been working flawlessly for months
So I just ran a firmware update on my GLiNet router
Powered down for 10 minutes after and also powered down Homey for 10 minutes (which I’d already tried), then turned everything back on and for the past hour everything has been stable
Very odd, will keep monitoring.
I wonder if the Homey firmware update had some kind of IPv6 bug and needed to be cleared from the cache on the router+homey pro side?
Could it be that Homey killed and restarted one or more apps? I have noticed that in some cases Homey does that when an app is consuming too much RAM. Killing an app causes a sudden decrease in memory.
