English
Deutsch
PolskiSharing 2.4 GHz: What a Zigbee Channel Change Costs
Contents
A Zigbee mesh that has been dropping devices is usually diagnosed at the mesh: more routers, better placement, a firmware update. The layer underneath rarely gets examined, and it is the one that decides how much of the rest can work – eighty-three megahertz of shared band, occupied by a neighbour that transmits a hundred times louder.
The two systems do not compete on equal terms and they do not compete on equal information, which is what makes the choice of channel a decision rather than a preference.
Where the Channels Actually Sit
Zigbee numbers its channels 11 to 26, and the centre frequency follows one line: 2405 MHz plus five megahertz per step. Each channel occupies about two megahertz. Wi-Fi in the same band uses 22 MHz per channel, which is why only 1, 6 and 11 can coexist – they are the widest spacing the band allows.
Zigbee centre = 2405 MHz + 5 MHz × (channel − 11) 2 MHz wide
Wi-Fi 1 2412 MHz 2401 – 2423
Wi-Fi 6 2437 MHz 2426 – 2448
Wi-Fi 11 2462 MHz 2451 – 2473
clear: 15 (2425) 20 (2450) 25 (2475) 26 (2480)Everything else lands inside one of the three Wi-Fi blocks. Channel 11 at 2405 MHz is a common default and sits squarely under Wi-Fi 1; channel 25 is the one that is clear, well supported and rarely chosen because nothing in the interface suggests it.
The Interference Only Runs One Way
Wi-Fi transmits at roughly 100 mW. Zigbee transmits at 1 to 3 mW – a factor of thirty to a hundred, before antennas are considered. That alone would be manageable if both sides took turns, and one of them does: Zigbee listens before sending and backs off when the channel is busy.
Wi-Fi does not hear Zigbee. A two-megahertz signal at one milliwatt does not register as a busy channel to a Wi-Fi radio, so the access point transmits straight through it. The result is not a fair split but a mesh that waits politely for gaps in a transmission that never acknowledges it exists – which is why the symptom is latency and lost messages rather than a clean failure.
What Actually Sits in the Band
| Source | Where | Effect on a mesh |
|---|---|---|
| Wi-Fi at 20 MHz | 1, 6 or 11 | the planned case; three blocks, four gaps |
| Wi-Fi at 40 MHz | anywhere | covers two blocks at once and leaves almost no gap |
| Bluetooth | hops across the whole band | brief collisions everywhere, no channel escapes it |
| Microwave oven | around 2450 MHz | knocks out channel 20 for the duration of a meal |
| USB 3.0 devices | broadband noise | halves the range of a coordinator plugged in beside them |
The last row is the one that looks like a mesh problem and is not. A USB 3 port and its cable radiate broadband noise centred right in this band, and a coordinator stick plugged into the socket next to an external SSD loses a large part of its range without anything else changing. A one-metre extension cable is the entire fix, and it is worth trying before any channel is touched.
The 40 MHz row deserves the same treatment in reverse: a Wi-Fi set to 40 MHz in the 2.4 GHz band takes away the gaps that the whole plan depends on. Twenty megahertz there, and the width where it belongs – on 5 GHz.
Why the Change Costs More Than the Choice
A coordinator changes channel by announcing it, and mains-powered routers follow within seconds. Battery-powered end devices are asleep when the announcement goes out – that is what makes them last two years on a coin cell – and they wake up on a channel where nobody answers.
What happens next depends on the device and is rarely good: some scan and rejoin, many simply keep trying on the old channel until the battery is flat. In practice a channel change means walking around the house and pairing every sensor again, which is why the value of choosing correctly at installation is not the twenty megahertz – it is the afternoon.
The Cheaper Side to Move
When a mesh is already in place and the band is crowded, the question is not which Zigbee channel to move to but which system is cheaper to move at all. A Wi-Fi channel change takes one setting, applies within seconds, and costs every client a brief reconnect that nobody notices. A Zigbee channel change costs a re-pairing session.
So the order is: check what the Wi-Fi is doing, set it to 20 MHz, put it on 1, 6 or 11 with the largest possible distance from the Zigbee channel in use, and only then consider moving the mesh. And when the mesh does have to move, it is worth doing once, to a channel that is clear now and will still be clear when the neighbours upgrade their router – which in this band means 15, 20 or 25.
Questions and answers
Channel 26 is clear as well. Why does the article end up naming only 15, 20 and 25?
Because channel 26 has two weaknesses that do not show up in the frequency table. At 2480 MHz it sits close to the upper edge of the band, which is why some radio chips transmit at reduced power there and not every device supports it at all. On top of that, one of the three fixed advertising channels of Bluetooth Low Energy sits at exactly 2480 MHz.
The table rightly describes Bluetooth as hopping across the whole band, but that applies to established connections. Advertisements go out on three fixed channels: 2402, 2426 and 2480 MHz. Channel 26 therefore shares its frequency with every advertising Bluetooth device in range, and channel 15 at 2425 MHz sits only one megahertz from one of these channels. Channels 20 and 25 keep their distance from all three, which supports the article’s point that 25 is clear and well supported.
Channel 20 is clear, but it sits in the microwave oven’s range. Does that count against it?
Usually not. The interference lasts only while the oven is running and mainly affects nodes close to it. If the coordinator or an important router sits in the kitchen, however, a central node hears poorly during every meal, and then 15 or 25 is the better choice.
What changes when a Thread network is also transmitting in the same house?
Thread uses the same channels 11 to 26 with the same two-megahertz width and, like Zigbee, listens before sending. Two such networks on the same channel therefore do not interfere the way Wi-Fi and Zigbee do, but they share the airtime. On separate channels even that goes away, and the band has enough gaps for it, for example Zigbee on 25 and Thread on 15 or 20.
The two differ sharply when it comes to changing channel. Thread distributes a new channel as a pending dataset with a delay, so that sleeping devices also learn about it the next time they poll their parent, before the change takes effect. A Thread network can therefore in principle be moved without pairing anything again, provided the border router offers the change; border routers built into speakers or hubs from the large manufacturers often do not allow the channel to be set.
What does it mean for channel planning when the router picks its Wi-Fi channel automatically?
That the plan does not hold. The automatic selection only weighs what it can hear, and as the article shows, a Wi-Fi radio does not register Zigbee as occupancy at all. It can therefore pick any channel, including one between 1, 6 and 11: Wi-Fi 3 covers Zigbee 15, Wi-Fi 8 covers Zigbee 20, and Wi-Fi 13, which is permitted in Europe, covers 25 and 26. Some routers also set the channel width in the 2.4 GHz band to an automatic choice between 20 and 40 MHz. Channel and width therefore need to be set by hand once it is clear what the Wi-Fi actually occupies, and both are worth checking again after the router has been reset.
Related articles
