What Is Bus Topology?

Bus topology is a network layout in which every device taps into one shared cable, called the bus or backbone. A message sent by one device travels along the cable in both directions and reaches every other device, and each one keeps only the messages addressed to it. A bus network is also called a linear bus.

Each end of the bus has a terminating resistor matched to the cable’s impedance. Without it, the signal reflects off the open end and corrupts the next bits on the line. Because the cable is shared, only one device can transmit at a time, so a bus needs a rule for taking turns: collision detection on early Ethernet, bitwise arbitration on CAN, or a master that polls each device in turn on Modbus RTU.

Early Ethernet was a bus. 10BASE5 “thick Ethernet” and 10BASE2 “thin Ethernet” ran over coaxial cable with 50 ohm terminators, and devices shared it using CSMA/CD as set out in IEEE 802.3. Twisted-pair Ethernet in a star replaced both during the 1990s.

A bus is the cheapest way to connect devices strung out along a line, but the shared cable is also its weak point.

The bus lives on in industrial fieldbuses and vehicle networks, where devices sit along a machine or a vehicle and the data rates are modest. The table gives typical figures from each standard. Reach falls as the bit rate rises.

Our explainers on how RS-485 links devices on one pair and how CAN bus arbitration works cover the two most common buses in detail, and CAN bus I/O modules bring a CAN network into a computer.

A daisy chain links devices one after another, each connected to the next. It is often confused with a bus, and on a fieldbus the two meet: RS-485 and CAN are wired in a daisy chain from device to device, because short drops from the backbone keep the bus electrically clean.

An Ethernet daisy chain is different. Each device has two ports and an internal switch that passes traffic on, so it is a chain of point-to-point links, a line topology, not a shared bus. Its weakness is that a device losing power breaks the chain for everything beyond it. Some M12 Ethernet switches for rolling stock include bypass protection that keeps the chain passing traffic when a switch loses power, and closing the chain back to the start turns it into a ring.

Compared with a star, a bus uses less cable but loses the star’s fault isolation, since a star limits a cable fault to one device. Compared with a ring, a bus has no second path, so a backbone fault splits the network. That is why office networks moved to switched stars, while buses remain where cable runs are long, devices are many and each one sends little data, as on a production line or in a vehicle.

Many plants now run a mix. Sensors and drives stay on a fieldbus at the machine, and the controllers above them talk over industrial Ethernet such as PROFINET, EtherNet/IP or Modbus TCP. A gateway or an industrial computer with serial and CAN ports sits at the boundary and translates between the two.

When extending an existing bus, keep within the device count and cable length of the standard, keep drops short, and check that there are exactly two terminators, one at each physical end. Moving a terminator to the wrong place when a device is added is one of the most common causes of an intermittent bus.

We fit the serial, CAN and Ethernet interfaces each project needs in the industrial computers we build, and configure the drivers and bus settings before the system ships.