Interference Suppression in Digital Circuit and Microcontroller PCB Design

Interference problems are much cheaper to prevent than to cure. A board that is laid out without considering them will usually still work, and then fail intermittently in the field, which is the most expensive kind of failure to diagnose. The practical approach is to work through three questions in order: what is generating the disturbance, how does it reach the circuits that are affected, and how sensitive are those circuits. Answering all three before layout begins removes most of the problem at the source, where the fix costs nothing.

The Three Elements of an Interference Problem

Every interference problem has a noise source, a coupling path and a victim. The source is anything with a large rate of change of voltage or current: relays, motors, switching devices, high-frequency clocks and long cables. The path is how the disturbance travels, either conducted along a wire or radiated through the air. The victim is the circuit that misbehaves, typically an analog-to-digital converter, a microcontroller, a low-level amplifier or any high-impedance node. If any one of the three is removed, the problem disappears, which is why interference suppression is usually organised as source control, path cutting and victim hardening rather than as a single fix.

Suppressing the Source

Source control is the most effective measure and should always be tried first. Reducing the rate of change of voltage across a switching device is done by placing a capacitor across it; reducing the rate of change of current is done by adding inductance or resistance in series with it, or by providing a path for the current to continue flowing when the device turns off. A relay coil needs a freewheel diode across it to absorb the back EMF when the coil is de-energised; adding a Zener diode in series with the freewheel diode lets the relay release faster, which matters when the relay has to operate many times per second. Relay contacts benefit from a snubber across the contacts, typically a series resistor and capacitor, and motor terminals benefit from a filter placed with the shortest possible leads, because lead length adds inductance that defeats the filter. Every integrated circuit should have a local high-frequency capacitor of the order of 0.01 uF to 0.1 uF between its supply and ground, placed close to the pin so that the connection does not add equivalent series resistance. Right-angle corners in a trace are worth avoiding on fast nets for the same reason they are avoided on antennas.

Cutting the Conducted Path

Conducted interference travels along the wires that connect one part of the system to another, and it is usually removed with a filter or a barrier. The supply is the most common route, so a microcontroller supply should be filtered or regulated, for example with a ferrite bead and a capacitor arranged as a pi network, or with a resistor where the current allows it. When an input or output pin drives a noisy load such as a motor, an isolation stage between the pin and the load prevents the disturbance from travelling back into the logic. The oscillator deserves particular attention: it should sit as close to the device pins as the layout allows, its case should be grounded, and the clock area should be surrounded by ground so that it does not couple into neighbouring traces. Partitioning the board by signal level, keeping high-current switching sections away from low-level analog sections, and separating digital and analog returns so that they meet at a single point at the supply are all part of the same discipline, and the return structure is described in this article on ground routing and power trace planning.

Cutting the Radiated Path

Radiated coupling is reduced by distance, by shielding and by the geometry of the receiving circuit. Increasing the separation between a noise source and a sensitive circuit is the simplest measure, and it costs only floor area. A ground trace placed between the two can act as a partial barrier, and a metal shield over the sensitive circuit provides a much stronger one, provided the shield is well bonded to the ground plane at several points. Because the coupling depends on the area of the loop that the interference can drive current around, keeping every loop small is the most general defence, and this is where layout and interference suppression meet.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/17aa72336fa5.webp" alt="oscillator and decoupling network on a digital PCB” />

Hardening the Sensitive Circuit

A victim circuit can be made less sensitive in several ways that are independent of the source. Reducing the loop area between a signal and its return lowers the voltage induced by a magnetic field. Making the supply and ground conductors wide lowers the impedance that a disturbance has to work against, which reduces the noise voltage developed across them. Unused inputs should never be left floating: on a microcontroller the spare pins should be tied to ground or to the supply, and the same applies to unused inputs on other logic devices as long as the system function is unaffected. Power monitoring and an external watchdog keep the device from remaining in a corrupted state after a disturbance has passed, and where the application allows it, a lower oscillator frequency and slower logic family reduce both the emissions and the sensitivity. Sockets are best avoided on integrated circuits in a noisy environment, because the added contact resistance and inductance work against every other measure. Where a board carries both analog and digital circuitry, the separation rules in this article on mixed-signal PCB design guidelines apply directly.

Firmware Measures

Some of the most effective countermeasures are in the program rather than on the board. Filling unused program memory with a value that decodes as a no-operation word gives the processor a defined place to land if the program counter is corrupted. Placing a few no-operation instructions before a jump instruction has a similar effect on the instruction pipeline. A software watchdog, where a hardware watchdog is not available, monitors whether the main loop is still running. Parameters written to an external device can be re-sent periodically so that a corrupted setting is automatically restored. Communication links benefit from a checksum and from a voting scheme, where a value is accepted only if it agrees with a repeated transmission. On shared buses, holding idle lines at a defined high level rather than letting them float has proved more robust in practice than the alternative.

Verifying That the Measures Work

Interference suppression should be verified rather than assumed. The practical test is to operate the product next to the disturbance it will meet in service, whether that is a motor, a relay, a switching supply or a handheld radio, and to look for changes in behaviour. Conducted and radiated measurements confirm that the board meets the applicable limits, and the design principles behind those limits are described in this article on EMI suppression design principles. If a fix is needed, it should be applied at the source first, because a fix applied at the victim usually reduces sensitivity to one disturbance while doing nothing about the rest.

FAQ

Where should interference suppression start? At the source. Reducing the switching speed or adding a freewheel diode removes the disturbance for every victim on the board, while a fix at one victim helps only that victim.

Do unused microcontroller pins really matter? Yes. A floating input can oscillate and draw current, and it presents a high-impedance node that couples noise into the device. Tie it to ground or the supply.

Is a software watchdog a substitute for a hardware one? It is a fallback. A software watchdog cannot recover the system if the code that feeds it has itself stopped running, which is why a hardware watchdog is preferred.