10BASE-T1S is starting to look less like a specialist Ethernet option and more like a practical machine-edge interface. Microchip announced LAN8679 and LAN8680 transceivers alongside integrated endpoint devices on 29 September 2026; the new Microchip products are in limited sampling, while NXP’s industrial TJF1410 is already production-released. For embedded and automation engineers, the interesting part is not the headline that Ethernet can now run over one pair. It is that short-range, multidrop Ethernet is becoming available in parts intended for real industrial designs.
That puts 10BASE-T1S into a space traditionally occupied by CAN, RS-485 and proprietary sensor buses: distributed sensors, actuators, cabinet I/O, small controllers and embedded nodes where conventional switched Ethernet is awkward or unnecessarily expensive.
What 10BASE-T1S actually changes
10BASE-T1S is the short-reach branch of Single-Pair Ethernet defined within IEEE 802.3. It runs at 10 Mbit/s, uses one twisted pair and supports half-duplex multidrop operation. In the basic IEEE model, a segment supports at least eight nodes over 25 metres. Commercial implementations are beginning to push beyond that reference envelope; NXP, for example, states that its TJF1410 can support applications with up to 32 nodes and cable lengths up to 100 metres. I would treat those larger figures as device and application claims to be validated in the actual EMC and topology conditions, not as a blanket Ethernet guarantee.
The feature that makes the shared bus workable is Physical Layer Collision Avoidance, or PLCA. Instead of letting nodes contend randomly for the medium, PLCA assigns transmission opportunities in a repeating cycle. That gives the designer bounded and much more predictable media access than ordinary half-duplex Ethernet contention.
This matters in control applications. A 10 Mbit/s bus is not especially impressive by modern Ethernet standards, but deterministic access, native Ethernet framing and a simple multidrop cable can be more valuable than raw bandwidth for a distributed machine subsystem.
Do not confuse T1S with T1L
CANS covered 10BASE-T1L recently. The two technologies share the Single-Pair Ethernet family name but solve different wiring problems.
| Feature | 10BASE-T1S | 10BASE-T1L |
|---|---|---|
| Typical topology | Multidrop or point-to-point | Point-to-point |
| Data rate | 10 Mbit/s | 10 Mbit/s |
| Duplex | Half-duplex | Full-duplex |
| Reference reach | Short range; at least 8 nodes over 25 m in the IEEE model | Long reach, up to about 1 km |
| Natural application | Machine nodes, sensors, actuators, cabinet devices | Process instruments and long plant-floor links |
If the design problem is a transmitter hundreds of metres away, T1L is the relevant technology. If the problem is ten or twenty small nodes around a machine or control cabinet, T1S deserves a look.
Why the 2026 hardware releases matter
Standards only become useful when the implementation choices stop being exotic. Microchip’s September 2026 launch added PMD transceivers and integrated endpoint devices for 10BASE-T1S. NXP’s TJF1410 is an industrial PMD transceiver intended for 10 Mbit/s Ethernet and supports power-over-data-line compatibility, wake signalling and topology diagnostics.
There is an architectural detail here that is easy to miss. A PMD-only device such as the TJF1410 is not a complete Ethernet PHY in the way many engineers are used to thinking about Ethernet transceivers. It handles the physical-medium-dependent section and expects the host MCU or switch to provide the digital 10BASE-T1S PHY functions. Other devices integrate the MAC and PHY more tightly and may connect to a microcontroller through SPI or another serial host interface.
That distinction affects BOM cost, MCU selection, software effort and how much of the Ethernet stack is pushed into the endpoint. For a new embedded product I would decide the host architecture first, then choose whether a PMD transceiver, full PHY or MAC-PHY arrangement makes sense.
Where I would consider it in industrial equipment
The strongest applications are not places where ordinary industrial Ethernet already works perfectly. They are the awkward low-level segments where we currently accept a protocol gateway because the field interface is cheap and simple.
- Distributed I/O inside a large machine: several small nodes can share one pair without adding a switch port for every endpoint.
- Smart sensors and actuators: Ethernet addressing and higher-layer protocols can extend further towards the device without immediately translating through CAN or RS-485.
- Control cabinets: short multidrop runs can simplify wiring between breakers, monitoring modules, motor-control subassemblies and embedded controllers.
- Robotics and mobile equipment: weight, connector count and harness complexity matter, while 10 Mbit/s is often ample for supervisory data and many sensor classes.
- New embedded platforms: products designed from the start around IP-based diagnostics, secure update and service interfaces may benefit from keeping Ethernet semantics down to the endpoint.
Microchip specifically positions the technology for automotive, industrial and robotics architectures. Analog Devices also lists industrial cabinets, building automation, energy control panels and robotics among 10BASE-T1S applications.
It does not automatically replace CAN or RS-485
I would not treat 10BASE-T1S as a reason to remove CAN, CAN FD or RS-485 from an existing machine. Those interfaces remain very good at what they do. CAN gives robust arbitration, simple controllers and a mature industrial ecosystem. RS-485 remains inexpensive and flexible for Modbus RTU and proprietary serial protocols. In many brownfield systems, replacing a stable fieldbus simply to obtain Ethernet is poor engineering.
The choice becomes more interesting on a new platform. If the device already needs IP-based diagnostics, secure remote update, standard Ethernet tooling or direct integration with an edge computer, T1S can reduce the number of protocol boundaries. Fewer gateways can mean fewer data models, fewer translation bugs and a simpler path from endpoint diagnostics to plant software.
For higher bandwidth or more demanding real-time control, conventional switched industrial Ethernet may still be the better answer. For long cable runs, T1L is the more natural SPE option. For compact deterministic control with very small payloads, CAN FD can remain simpler. CAN XL is also moving into commercial hardware and sits in a different part of the design space; our recent CAN XL article covers that transition.
PLCA helps predictability, but system timing still needs engineering
PLCA is often described as making 10BASE-T1S deterministic. That needs qualification. PLCA makes access to the shared medium predictable, which is important, but end-to-end control timing still includes queueing, protocol processing, MCU scheduling, application software and whatever higher-layer stack is used.
If a machine function has a hard timing requirement, I would measure the complete transaction under worst-case bus loading and CPU loading. Do not infer a safety or motion-control timing guarantee from the physical-layer access method alone.
The same applies to functional safety. A shared Ethernet bus can carry safety-related communications if the overall architecture and protocol are designed and certified for it, but 10BASE-T1S itself is not a safety function. Safety interlocks, emergency stops and hazardous-motion protection should remain inside the appropriate validated safety architecture.
What I would check before putting it on a PCB
- Required node count, trunk length and stub lengths under the real harness topology.
- Whether the selected device is PMD-only, PHY or MAC-PHY, and what the host MCU must provide.
- PLCA configuration, coordinator behaviour and worst-case access latency.
- EMC performance with the intended cable, connectors, grounding and motor-drive environment.
- Power-over-data-line requirements if power and communications share the same pair.
- Wake and sleep behaviour if the node must support low-power operation.
- Driver and operating-system support, especially when Linux or a third-party TCP/IP stack is involved.
- Cybersecurity architecture once Ethernet and IP reach closer to the sensor or actuator.
That last point is easy to underestimate. Moving Ethernet deeper into a machine can simplify integration, but it also moves routable protocols closer to devices that were previously hidden behind a fieldbus gateway. Network segmentation, device identity, secure firmware update and service access have to be designed at the same time as the physical interface.
For machine builders, this is now a design option rather than a research topic
The release cadence in 2026 is the part I find significant. There are now multiple semiconductor vendors supporting 10BASE-T1S in forms aimed at industrial and embedded use, not just demonstration boards. That is normally the point where a communications technology starts appearing in real product roadmaps.
I would still prototype it before committing a machine platform. Put the intended node count on the actual cable, run the motor drives, inject the expected EMC noise, load the bus and measure latency. The attraction is straightforward: native Ethernet semantics on a cheap shared pair at the edge. Whether that is better than CAN or RS-485 depends on the machine, not on the standard name.
CANS works on embedded electronics, industrial communications and automation integration across machine-level and plant-level systems. If you are designing a new controller, distributed I/O platform, industrial gateway or embedded communications architecture, send CANS an enquiry or WhatsApp CANS on +60 12-295 9602.
References
- Microchip Technology — Microchip Expands Single Pair Ethernet Portfolio with New 10BASE-T1S Devices Enabling Scalable Edge Connectivity, 29 September 2026.
- NXP Semiconductors — Industry-First 10BASE-T1S PMD Transceiver Delivers Scalable, Cost-Effective Ethernet to the Intelligent Edge, 24 February 2026.
- NXP Semiconductors — TJF1410 10BASE-T1S PMD Industrial Transceiver.
- Analog Devices — IEEE 802.3cg / 10BASE-T1S and 10BASE-T1L overview.
- Featured image: Jan Helebrant, Wikimedia Commons / Flickr, CC0.
Recent Comments