Industrial PCB in SCADA Systems: Architecture and Design

A SCADA system is the layer that connects the plant floor to the people who operate it. Remote terminal units and programmable controllers gather measurements from sensors and drive actuators, and the SCADA software presents the result and issues commands. The boards inside those field devices are industrial PCBs in the fullest sense: they run continuously for ten or fifteen years, they operate in cabinets that are hot in summer and cold in winter, and they are connected to cables that run across a plant and collect every electrical disturbance the site has to offer.

What the Architecture Requires

A remote terminal unit accepts a mixture of input types. Analogue inputs carry a measurement as a current or a voltage; digital inputs report a contact state; pulse inputs count events; and analogue or relay outputs drive valves, motors and indicators. The board has to accommodate all of them, and it has to keep them independent, because a fault on one channel must not spread to the others or to the controller.

Behind the field interface sits the processing section: a microcontroller or a processor, its memory, a real time clock and the communication interfaces. It is unremarkable digital design on its own. What makes it industrial is everything around it, and the layout is dominated by the interfaces rather than by the processor.

Isolation as the Organising Principle

Isolation is the idea that structures the whole board. Every channel that connects to the plant is separated from the logic and from the other channels, so that a fault, a surge or a difference in ground potential on one of them cannot propagate. Analogue channels are isolated with amplifier or converter devices that carry the barrier through them; digital inputs are isolated optically; relay outputs are isolated by the relay itself.

That arrangement has physical consequences. Each isolated channel needs its own isolated supply, usually a small converter or a transformer driven from the logic side, and the barrier has to be continuous across the board. Creepage and clearance distances apply at every crossing, slots may be milled to increase the surface distance, and the barrier is marked on the silkscreen so that it survives every layout revision. Our component tolerance and reliability notes describe how those barriers are assessed.

Analogue Accuracy in a Noisy Cabinet

The measurements that a SCADA system reports are used to make decisions, so their accuracy has to be maintained in an environment that is electrically hostile. A four to twenty milliamp current loop is the traditional interface because a current is unaffected by the resistance of the cable and by the voltage drops along it, but the receiver still has to convert it to a digital value without picking up the noise of the cabinet.

That means a precision shunt, a low noise amplifier and a reference that does not drift with temperature or with the supply, all in a small area of the board. The reference is the component that determines long term accuracy, and its temperature coefficient matters more than its initial tolerance. Where the loop supply is derived on the board, the switching converter that produces it must be laid out so that its switching node is far from the sense resistor, and a passive filter between the converter and the measurement stage is usually necessary. Our layer assignment notes describe how the stack is planned.

Communication and Redundancy

Field devices report over a fieldbus or over Ethernet, and the interface in both cases is a differential pair that must be protected against the transients that a long cable collects. Where the device is critical, two communication paths are provided and the board must be able to switch between them without interrupting the process, which means the two paths are laid out independently all the way back to the processor so that a single fault cannot disable both.

Redundancy extends to the supply. Many industrial devices accept two power feeds and continue to operate if either is lost, which requires the two feeds to be combined without allowing one to back feed the other. The usual arrangement is a diode-or circuit, and the diodes belong near the connector with their own thermal path, because the forward drop at a few amps is a real dissipation.

Wide Temperature Operation and Thermal Design

The cabinet may be unheated in winter and unventilated in summer, so the board must be designed for the full range rather than for a comfortable room. That affects component selection — capacitors in particular have a much shorter life at high temperature — and it affects the thermal design, because the internal ambient sets the starting point for every calculation.

Heat sources on this class of board are modest but numerous: the isolated supplies, the communication transceivers, the relay drivers and the processor. None of them is individually dramatic, but together they raise the internal temperature, and the board should spread their heat across the copper rather than concentrating it. Our thermal management article describes how the copper areas are sized for that case.

Protection, Testing and Service Life

Every conductor that leaves the board is an entry point for a surge, and the protection belongs at the connector with a defined return path so that the energy is clamped before it reaches anything sensitive. Series impedance, clamping devices and a robust return are the standard combination, and the layout of those parts determines whether they work at all.

The product standard for this class of equipment requires specific immunity tests, and designing for them means planning the protection and the filtering from the beginning rather than adding components after a failed test. Functional testing then covers every channel type, with the measurements checked against a reference so that the accuracy claim can be demonstrated rather than assumed. Our design release checklist places those checks in the review sequence.

Firmware, Diagnostics and Uptime

The board exists to keep a process observable, so its ability to report its own condition is part of the design rather than an extra. Supply voltages, cabinet temperature, communication error counts and channel status are all worth monitoring, and on an industrial board the monitoring is usually built into the hardware so that the firmware can read it without disturbing the process. A device that can report that one of its isolated supplies is drifting gives an operator time to plan a replacement instead of discovering the fault when a channel stops responding.

The watchdog arrangement follows the same logic. A hardware watchdog that the firmware must service, an independent check on the communication link, and an output state that defaults to a safe condition when the controller is not running are the three elements that keep a device from failing in an unsafe direction. Designing those into the board, rather than adding them to the firmware afterwards, is what makes the product suitable for a plant that cannot be shut down for a software fix.

FAQ

Why is every channel on an RTU isolated separately? So that a fault or a surge on one channel cannot reach the others or the logic. Channel to channel isolation also prevents ground loops between sensors that are earthed at different points.

What determines the long term accuracy of an analogue input? The reference rather than the converter, because the reference drifts with temperature and time. A device with a modest converter and an excellent reference will hold its calibration longer.

How is a board like this qualified? By immunity and emission testing to the applicable standard, by a wide temperature soak, and by a long duration functional test with all the channels active.