Definition
RTU (Remote Terminal Unit) is specifically designed for remote monitoring and control. It features multiple AI/DI/DO channels and supports protocols such as Modbus, DNP3, and IEC 104, suitable for data collection and remote control at unmanned sites such as pump stations, sluice gates, pipelines, and substations, and can be connected to SCADA systems.
System Composition
An RTU typically includes: CPU and storage; multiple analog inputs (4–20mA, 0–10V); multiple digital inputs/outputs; RS485, Ethernet and other communication interfaces; 4G, fiber optic and other upstream communication; wide temperature range, IP protection industrial design. It supports Modbus, DNP3, IEC 104 and other SCADA protocols, enabling four-remote functions in communication with the master station. Compared with DTU, RTU has local I/O and protocol processing capabilities, suitable for complex site conditions.
Use Cases
Power: substations, distribution automation; Water: pump stations, sluice gates, pipelines; Oil & Gas: pipelines, wellheads; Municipal: street lights, environmental monitoring. RTUs collect data from on-site sensors and instruments, execute control commands issued by the master station, and report to the SCADA master station via 4G, fiber optics, and other channels.
The Four Point Types, and Holding Through a Broken Link
Local I/O and the protocol stack live in one unit
A remote terminal puts collection, a small amount of logic, and the uplink protocol in one field device. Analog inputs take transmitters. Discrete inputs take positions and alarms. Discrete outputs drive relays. Communication ports then attach instruments that already speak a protocol. To the master, these quantities are arranged as four types: analog measurements, discrete status, discrete commands, and setpoint changes. A measurement is a number. A status is a state. A command is a switching order. A setpoint is a target value. The point list states what each point number means, its ratio, its deadband, and whether the master may write it. Without a point list, what the protocol brings up is only a channel number.
It differs from a transparent data radio. The terminal itself knows the engineering meaning of a point. It can judge a limit locally, save through a power loss, and apply a default output after communication fails. The period on which it scans local I/O is usually shorter than the uplink poll. A change between two visits from the master has to be held by the terminal, or recorded as a sequence of events. Otherwise a narrow alarm pulse is lost.
The site has to keep itself when the master is lost
The conditions for use are power, grounding, and a clear point list on site, plus an output policy already defined for the time the master is lost. Pump stations, gates, and pipeline sites are often unattended. They still have to keep sampling and protecting equipment through an uplink outage of several hours. An output that should hold its last state must not be cleared because communication timed out. A valve that should fall to a safe position must not stay half open. Those policies are written in the terminal. They do not depend on the master being online. The uplink protocol may be any one of several master conventions, but point numbers have to match the master database. Change one side and the other side has to change too.
For analog values, check the transmitter range against the terminal's full scale. If the engineering range that corresponds to 4–20 mA is written wrong, the zero sits at a fixed offset. The shape of the trend still looks real, and dispatch that operates from the number will be wrong. Discrete inputs have to distinguish normally open from normally closed. Reversed wiring inverts every alarm. See how a 4–20 mA loop is powered, and confirm whether the terminal supplies the loop or the transmitter has its own supply. Wired the wrong way, there is no current, or the supply is shorted.
Point list, ratio, and a frozen measurement
Ratio and freeze are two sources of distorted history. If a current-transformer ratio is not written into the point list, the measurement is off by a factor and every energy and load total shifts. If a freeze samples at the moment the master polls, a communication break leaves a blank. If the terminal freezes on its own fixed period, it should backfill the freeze file after recovery, rather than filling the gap with the current value. Filling backward with the current value manufactures a flat curve that never happened.
A command without the two-step select and execute can move equipment when the wrong point number is touched. If a point-list change updates only the terminal, and the master still uses the old number, the command hits a different device. If time sync fails, the sequence of events loses its meaning and the order of an incident across sites cannot be reconstructed. Finally, test the broken link. Pull the uplink and see whether outputs hold or fall back as designed, and send communication state up as a status point, so the master does not keep displaying stale numbers from the outage as if they were live.
Retention of local storage through a power loss has to be tested as well. A terminal that keeps its books only in memory starts counting from zero after power returns. The master fills the gap with a sloping line, and flow or energy increases out of nothing. Confirm that counts and freezes are written to a medium that survives power loss, and after power returns send a status that says the data was retained or the data was lost, so the master can decide whether to continue the series or mark the gap.