Glossary term
Data Bus
A shared communication pathway that transfers data between components, subsystems, or devices in a digital or embedded system.
Definition
deviceA shared communication pathway that transfers data between components, subsystems, or devices in a digital or embedded system.
A data bus may be a parallel set of conductors, a serial link, an on-chip interconnect, or a fieldbus. Its performance and reliability depend on width, protocol, arbitration, timing, electrical signalling, bandwidth, latency, error handling, and physical layout.
A data bus is a communication path used to move digital information between parts of a system. It may connect a processor to memory, a microcontroller to peripherals, boards within a product, sensors to a controller, or devices on an industrial network.
Engineering role
The bus determines how quickly, reliably, and predictably data can move. In embedded systems it can affect control-loop timing, sensor freshness, actuator command delay, power consumption, electromagnetic compatibility, and fault containment. In computer architecture it affects throughput, memory access, cache coherency, and system scalability.
Parallel and serial buses
Older or internal buses may use parallel data lines where several bits move at once. Serial buses transmit bits sequentially, often at much higher signalling rates and with fewer wires. Common design choices include synchronous versus asynchronous timing, single-ended versus differential signalling, point-to-point versus multi-drop topology, and master-slave versus peer arbitration.
Protocol and arbitration
A bus is more than wires. It includes rules for addressing, framing, timing, collision avoidance, acknowledgement, error detection, retries, priority, and ownership. If multiple devices can transmit, the arbitration mechanism determines who gets access. In real-time systems, bounded worst-case access time can matter more than peak throughput.
Physical design
At high speed, a data bus must be treated as a signal-integrity problem. Trace impedance, termination, skew, crosstalk, reflections, ground return, connector quality, cable length, and electromagnetic interference can corrupt data. Error-detection codes and retries help, but they do not replace physical-layer design.
Common mistakes
Common mistakes include calculating only nominal bandwidth while ignoring protocol overhead, arbitration delay, retries, and latency. Another error is adding devices to a shared bus without checking loading, address conflicts, termination, and timing margins. For safety-related systems, engineers should also verify what happens if the bus is shorted, stuck, noisy, overloaded, or partitioned.