Can a $4.90 Board Bring an Old Lamp into Apple Home?

A costed design for bridging a Bluetooth Mesh light to Matter, with a real-state acceptance test.

Tap the phone, and the lightbulb icon turns on. Unplug the table lamp, tap again, and the icon still turns on. This is an experiment worth doing when building your own gateway: what did that first tap actually turn on, the lamp on the desk or a variable in the program?

In the previous development-board validation notes, we got a Matter prototype running on Linux into Apple Home. To save money from here, there is just one thing most worth verifying: make a Bluetooth Mesh lamp you already own actually turn on and off, and let the phone know its last state. That is a much smaller goal than “build a whole-home gateway,” and its costs are easier to work out. What follows is the next round of experiments designed from official documentation; there are no results from testing with a real lamp yet.

Look at the Lamp Before the Board

The device to build here is a Matter Bridge. To the phone, it looks like a Matter device with a lamp endpoint; to the old lamp, it speaks the Bluetooth Mesh the lamp already understands. The phone finds the bridge over the home network, and the bridge finds the lamp over Bluetooth. The upstream connection uses Ethernet or Wi-Fi, and the downstream connection uses Bluetooth. This path does not go through Thread, so the shopping list does not need a Thread border router.

First, check the lamp’s documentation for whether its specific model and firmware support standard Bluetooth Mesh and include a Generic OnOff Server. Seeing “Telink chip,” “Bluetooth lamp,” or a device that appears in a phone scan is not enough to answer that question. You also need to know whether you can use your own provisioner to add a test lamp to a test network. If a lamp is already on the manufacturer’s network, and you cannot obtain valid, usable network configuration or reprovision it, do not expect a more expensive board to solve the problem. A whole set of lamps currently in use at home is not a good candidate for factory-reset trial and error either.

This check could change the answer outright: if you only want to control one lamp at home and its manufacturer already has a Matter bridge explicitly supporting that model, put its delivered price beside the DIY option. A day of debugging is a cost too. The route below suits someone who wants control over the bridging logic and plans to connect more devices later.

The First Round Can Cost Nothing

If you already have a Linux computer with a usable Bluetooth adapter, a test lamp that can be reprovisioned, and an existing Matter controller, you need not order anything for the first round. Keep the existing Matter Bridge and let Linux’s BlueZ Mesh service talk to the lamp. Ideally, connect the computer by Ethernet first so local-network problems are easier to separate from Bluetooth problems. Being able to compile a program in a virtual machine or WSL does not mean it has access to a Bluetooth controller; first confirm that the Mesh service can use the adapter and that the adapter is not also being contested by the Matter commissioning process.

Set the phone aside for now. Following the BlueZ Mesh API, create a test network, provision the lamp, and bind the same application key to the client model and the lamp’s server model. Send one On command to the lamp element’s unicast address, then read its current state. It is worth connecting this to the upstream side only when you can record a Generic OnOff Status from the target address that agrees with the light in front of you. Detecting an advertisement, completing provisioning, and submitting a message to the Bluetooth stack are three useful intermediate results. None can answer “Am I on?” on the lamp’s behalf.

Next, map this lamp to a fixed Matter endpoint. The CSA Linux Bridge example already demonstrates dynamic endpoints and a simulated lamp, making it suitable for reuse. What needs replacing is the data source behind the simulated lamp: when the phone requests On, send an explicit On; update the bridge’s confirmed state to On only when the lamp reports On. Do not translate On into Toggle, or a retry after a timeout might turn the newly lit lamp right back off.

There is an easy-to-miss detail here: “the last confirmed state was On” and “the lamp is reachable now” are two different things. If the lamp loses power and sends no response, the bridge cannot infer from silence that it is Off, nor can it present last night’s On as a freshly read state. Save the last confirmed value and its time, update the endpoint’s reachability according to a timeout policy, and restore reachability when the lamp’s status arrives again. That way, the wall switch and the phone will not keep telling different stories.

What Does the $4.90 Board Save?

The official product page lists the Seeed XIAO ESP32C3 at $4.90. It has onboard Wi-Fi and Bluetooth, 4 MB of Flash, and 400 KB of SRAM. Putting both upstream and downstream connections on this board could eliminate an always-on computer and a separate Bluetooth board. It is an attractive lower bound on hardware cost among the three options. It also squeezes two parts of the program that could previously be observed separately into the same memory and radio system.

For example, the phone might send repeated On and Off commands over Wi-Fi while the lamp reports its state over Bluetooth. Espressif’s coexistence documentation explains how the C3 shares its 2.4 GHz radio resources by time division. Getting Wi-Fi and Mesh working independently does not tell you the response latency when they run together. The ESP32 Bridge example does not implement this lamp’s downstream adapter for you either. Before ordering, compile for the exact C3 target and check whether both sets of components build together, whether the application partition is large enough, and where provisioning data will be stored. If you intend to support over-the-air updates later, you must also leave space for an update image. Only after the build passes can a real board answer questions about minimum free heap and packet loss under concurrent use.

What you already have Suggested spending What it can verify
Linux, usable Bluetooth, a test lamp, and a controller $0 in additional hardware; start with BlueZ Whether the lamp accepts control and can report its real state
Linux, and a desire to use the project’s Silicon Labs serial approach After confirming firmware support, buy the xG24 Explorer Kit, officially listed at $39 Debug the host and dedicated Mesh coprocessor separately; the onboard debugger and virtual serial port can reduce the need for accessories
A plan to build a single-board bridge from scratch XIAO ESP32C3 board price: $4.90; add power supply, USB data cable, enclosure, shipping, and taxes Whether functionality already proven elsewhere can fit on a low-cost board, eliminating the always-on computer

Prices are single-board list prices checked on November 18, 2026. As an example budget, at an assumed exchange rate of 7 yuan to the US dollar, the C3 board costs about 34 yuan. Allowing 10, 20, and 10 yuan respectively for a data cable, power supply, and enclosure puts the prototype gateway itself at about 74 yuan, plus shipping and taxes. Those last three amounts are budget placeholders, not quotations. Starting with nothing also means accounting for a test lamp, use of a development computer, and a controller; 74 yuan cannot be called the price of an entire smart-home setup. Conversely, if an old computer stays on solely to run the bridge, even 10 W of additional draw amounts to 87.6 kWh a year. Using an old computer for experiments is cheap; running it indefinitely may not be.

What the $39 option buys is observability. The project already has a BGAPI serial design that can distinguish in host logs between “the message was not sent” and “the message was sent, but the lamp did not respond.” But btmesh_empty in the Silicon Labs Python examples is a node example. Seeing a serial startup event only shows that the host can hear the board. Before buying, confirm that firmware for the specific board supports a provisioner and a Generic OnOff Client, and locate the API description file matching that firmware version. Otherwise, $39 will not buy a complete connection to the lamp.

How to Tell Whether the Money Was Well Spent

I would run four acceptance checks with the same lamp. First, send “On” twice in succession from the phone; the lamp should stay lit. Second, turn it off with a paired Mesh remote without touching the phone, and observe whether the bridge updates the phone to Off through a status publication or an active read. Third, unplug the lamp and request On again; record how long the bridge takes to mark it unreachable, and confirm that it does not invent a successful response. Fourth, restart only the bridge; confirm that the lamp remains the same endpoint on the phone and does not need reprovisioning. For each check, record the request time, the lamp’s response time, and the interface change. A recording of the operation plus those three time columns tells the next developer more about where a problem lies than the words “supports Matter.”

So for this project, I would leave the purchase list empty for now. Use the computer you already have to ask the lamp once, “Are you on or off right now?” and bring its answer back to Apple Home. Once that round trip really works, use the $4.90 board to answer the second question: can the computer be removed? Running those two experiments in that order is where this gateway can save the most money.

Original protocol diagram, not a test result.

Discussion

Share a question or your experience. Comments appear after moderation. Your email is never displayed publicly. Comment privacy

Loading comments…