English
Menu

CAN vs LIN vs Automotive Ethernet: Key Differences Explained

Inside a car, several networks are used side by side to match the speed and cost each job needs. LIN is an inexpensive network that uses a single signal wire at up to about 20 kbit/s and connects body components such as switches and motors. CAN uses a differential signal on two wires, lets ECUs communicate with priorities, and serves as a highly reliable network of up to 1 Mbit/s (several Mbit/s for the data phase in CAN FD) for controlling the engine, brakes and so on. Automotive Ethernet sends 100 Mbit/s or more over a single twisted pair and carries large volumes of data, such as from cameras, as well as the backbone that links ECUs.

AspectLINCANAutomotive Ethernet
Maximum speed (typical)About 20 kbit/sUp to 1 Mbit/s for classic CAN (500 kbit/s is commonly used in practice). CAN FD reaches several Mbit/s in the data phase100 Mbit/s (100BASE-T1) and 1 Gbit/s (1000BASE-T1). Faster standards also exist
WiringOne signal wire plus groundDifferential signal on two twisted wires (CAN_H and CAN_L), with 120 Ω termination resistors at both endsFull duplex over a single twisted pair
Node topologyOne master and multiple slavesA multi-master bus in which all nodes are peersStar or tree topologies with point-to-point links centered on switches (a 10 Mbit/s bus-type standard also exists)
How transmission order is decidedThe master transmits according to a schedule, and slaves respond to the master's requestsArbitration by ID. The smaller the ID, the higher the priority, and the winning frame is not corruptedSwitches forward the traffic. For communication that must arrive within a fixed time, TSN manages time and bandwidth
Data per transmissionUp to 8 bytesUp to 8 bytes for classic CAN, up to 64 bytes for CAN FDUp to about 1,500 bytes per frame
Error detectionChecksumCRC, acknowledgment (ACK), and disconnecting nodes based on error countsCRC. Higher-layer protocols handle retransmission and similar functions
Cost per nodeLowest. Slaves can run without a crystalModerate. Many automotive microcontrollers have a built-in controllerHigher. Requires PHYs and switches, but is cheap per bit
Main usesBody functions: power windows, mirrors, seats, switches, wipers, rain sensors and so onControl functions: engine, transmission, brakes, steering, communication between body ECUs, fault diagnosisCamera and radar data, infotainment, the backbone linking ECUs, software updates

Figures are typical orders of magnitude and vary by process generation and product.

When to use which

LIN
When connecting switches, motors and sensors that need no speed and where low cost is the top priority to a nearby ECU. Used as a subnetwork of CAN.
CAN
When you want to exchange control data reliably according to priority. When many ECUs need to be connected on one bus with noise-resistant communication. When data volume grows, CAN FD is used.
Automotive Ethernet
When bandwidth of 100 Mbit/s or more is needed, such as for large volumes of camera and sensor data, software updates, and the backbone linking the central computer to zone ECUs in a zonal architecture.

FAQ

What happens in CAN when two nodes start transmitting at the same time?

Each node monitors the bus state while sending the ID bits, and only the node sending the higher-priority frame (the smaller ID value) continues transmitting. The losing node stops and retransmits later, so the winning frame arrives intact. This is called non-destructive arbitration.

Why is LIN cheap?

It needs only one signal wire, and the master decides communication timing, so slave circuits are simple. Because slaves align their clocks to a synchronization signal sent by the master, they can run on oscillators without a crystal.

How does automotive Ethernet differ from ordinary Ethernet in offices and elsewhere?

It performs full-duplex communication over just a single twisted pair, making the wiring lighter and cheaper. It is also built to withstand the strong noise and wide temperature range inside a vehicle, and for communication that must arrive within a fixed time, time and bandwidth are managed with TSN standards.

Will CAN and LIN disappear as automotive Ethernet spreads?

Not any time soon. Ethernet is used for the high-speed backbone and for cameras and the like, but for low-volume communication where reliability matters, such as switches, motors and control data, the inexpensive and proven CAN and LIN will continue to be used.

Check your understanding

Which correctly describes a feature of the in-vehicle network LIN?

  1. AA high-speed scheme communicating at up to 10 Mbit/s over two differential signal lines, used for safety-critical control such as braking and steering
  2. BA scheme communicating at 1 Gbit/s over a single twisted pair, used to transmit camera video and display data
  3. CA low-speed (up to about 20 kbit/s) and inexpensive scheme connecting one master and multiple slaves on a single signal wire, used for controlling power windows, mirrors and the like
Show answer and explanation

Answer: C. A low-speed (up to about 20 kbit/s) and inexpensive scheme connecting one master and multiple slaves on a single signal wire, used for controlling power windows, mirrors and the like

LIN (Local Interconnect Network) uses a single signal wire and has slaves transmit in an order determined by the master; it is slow, up to about 20 kbit/s. It can be built with simple communication circuits similar to a UART, and slaves can synchronize without a high-precision oscillator, making it inexpensive. It connects parts that do not need speed, such as power windows, mirrors, seats and switches, as a subnetwork of CAN. The 10 Mbit/s scheme describes FlexRay, and the 1 Gbit/s scheme describes automotive Ethernet.

In CAN, what happens when two nodes start transmitting frames at the same time?

  1. AThe two frames collide and both are corrupted, so both nodes wait a random time before retransmitting
  2. BThe frame with the smaller ID value (higher priority) continues transmitting and the losing node stops and retransmits later, so the winning frame is not corrupted
  3. CThe frame with the larger ID value takes priority, and the frame with the smaller ID value is discarded and not retransmitted
Show answer and explanation

Answer: B. The frame with the smaller ID value (higher priority) continues transmitting and the losing node stops and retransmits later, so the winning frame is not corrupted

On a CAN bus, when 0 (dominant) and 1 (recessive) are driven simultaneously, 0 wins. A transmitting node compares each bit it sends with the bus state and, if it sent a 1 but sees a 0, judges that it has lost arbitration and stops transmitting. Since the ID is sent first, smaller ID values have higher priority, and the winning frame arrives intact. This is called non-destructive arbitration. Waiting a random time after a collision is the CSMA/CD method of early Ethernet.

On a CAN bus with 120 Ω termination resistors at both ends, the resistance between CAN_H and CAN_L was measured with the power off. Which value is closest to that of a normal bus?

  1. A120 Ω
  2. B240 Ω
  3. C60 Ω
Show answer and explanation

Answer: C. 60 Ω

A CAN bus has a 120 Ω termination resistor at each end to prevent signal reflections. Seen from between CAN_H and CAN_L, the two 120 Ω resistors are in parallel, giving 120 × 120 / (120 + 120) = 60 Ω. A reading of 120 Ω suggests one termination resistor is missing, and around 40 Ω suggests a wiring error such as three termination resistors being fitted. 240 Ω is the mistake of treating the two as being in series.

  1. Which correctly describes a feature of CAN FD?
  2. Which correctly describes a feature of automotive Ethernet (100BASE-T1, 1000BASE-T1, etc.)?
  3. Which correctly describes the ongoing shift to a "zonal architecture" in automotive electronic systems?

Practice Automotive Semiconductors and Functional Safety questions

Key terms explained

CAN bus
Short for Controller Area Network, an in-vehicle network that connects ECUs using differential signaling on two wires. Lower ID values have higher priority, and when nodes start transmitting at once, non-destructive arbitration leaves a single frame intact. Classical CAN runs at up to 1 Mbit/s, while CAN FD speeds up the data phase to several Mbit/s.
ECU (electronic control unit)
An in-vehicle controller built around a microcontroller that controls the engine, brakes, doors and so on. ECUs multiplied with each new function, sometimes exceeding 100 per car, and are now being consolidated through zonal architectures and similar approaches.
Zonal architecture
A vehicle electronics layout in which zone ECUs placed in each area of the car connect nearby sensors and actuators, while control processing is concentrated in a central high-performance computer. It allows a shorter, lighter wiring harness than the traditional one-ECU-per-function approach.
Differential signaling
Signaling that sends opposite signals on two wires and represents 0 and 1 by their voltage difference. Noise picked up equally on both cancels, allowing fast transmission with small swings.

Related comparisons