What Is CAN Bus?

CAN bus (Controller Area Network) is a serial network that lets the controllers in a vehicle or machine talk to each other over a single pair of wires. Bosch developed it in the 1980s to replace the bundles of point-to-point wiring in cars, presented it in 1986, and it was later standardised as ISO 11898.

Every device on the bus, called a node, can transmit, and every node hears every message. A message carries no destination address. Instead it has an identifier that says what the data is, such as engine speed or door status, and each node decides whether it needs it. The identifier also sets the message’s priority: the lower the number, the higher the priority.

If two nodes start sending at once, they compare identifiers bit by bit as they transmit. A 0 bit is dominant and overrides a 1, so the node with the lower identifier carries on and the other stops and tries again. The winning message is never corrupted, which is why CAN suits real-time control: a brake or fault message always gets through ahead of routine data.

CAN uses two wires, CAN_H and CAN_L, twisted together. The receiver reads the voltage difference between them rather than either wire against ground, so interference that hits both wires equally cancels out. That differential signalling is the same idea that gives RS-485 its noise immunity.

The bus is a single line with short drops to each node, and it needs a 120 ohm resistor at each end. Measured across CAN_H and CAN_L with the power off, a correctly terminated bus reads about 60 ohm. Missing or extra terminators are one of the commonest causes of a bus that works on the bench and fails in the field.

Speed trades off against length, because every node must see each bit within the same bit time. As a rough guide, 1 Mbit/s works to around 40 m, 500 kbit/s to around 100 m, 250 kbit/s to around 250 m and 125 kbit/s to around 500 m. Truck and machinery networks often run at 250 or 500 kbit/s for that reason.

The original protocol has been extended twice to carry more data per frame and at higher rates. All three versions can share the same twisted-pair wiring, although a classical CAN controller cannot read CAN FD frames.

Classical CAN also comes in two identifier lengths: 11-bit identifiers in CAN 2.0A and 29-bit extended identifiers in CAN 2.0B. J1939 uses the 29-bit form.

CAN itself only defines how bits and frames move. What the identifiers and data bytes mean is set by a higher-layer protocol, and that is usually what an engineer means by “the CAN data” on a given machine.

CAN started in cars but now runs anywhere several controllers need to share data reliably over modest distances. Common places to find it include:

A computer joins a CAN network through a CAN controller and transceiver, either built onto the board or added as a module. Many in-vehicle computers include isolated CAN ports alongside their ignition control, so they can log vehicle data and shut down cleanly with the engine. Where a computer has no CAN port, our CAN bus I/O modules add it, from USB adapters with J1939 or CANopen firmware to rugged mPCIe cards and isolated PC/104 modules.

Isolation matters on vehicles and machines, where ground potentials differ between the computer and the equipment it connects to. For a static machine that also uses serial devices, our comparison of RS-232 and RS-485 covers the serial side, and our industrial mini PCs suit cabinet installations that need CAN, COM and digital I/O on one system.

We fit CAN ports and modules to industrial and in-vehicle computers as standard, install the drivers and test each interface before the system ships. Tell us the protocol, bit rate and connector on your network, and we will specify the interface to match.