Switching Power Supply Layout: Current Loops, Placement and Routing

A switching converter lives or dies by its layout. The schematic can be textbook correct and the component values perfect, yet the finished board can still radiate badly, run hot, or refuse to regulate under load. The reason is that a switching supply is not a set of connections but a set of current loops, and the physical area of those loops determines how much energy escapes into the surrounding space.

Why Layout Dominates Converter Performance

Unlike a linear regulator, a switching supply moves energy in pulses. Those pulses contain harmonics far above the switching frequency, and any loop carrying them behaves as a small antenna. Even a trace that carries only direct current is not safe, because it can couple energy from a neighbouring trace and then re-radiate it somewhere else on the board.

This is why converters that perform well on a bench prototype sometimes fail electromagnetic interference limits in production. The difference is usually not the circuit but the loop area, the placement of the filter capacitors and the return path. Good layout reduces loop area first and worries about elegance later.

The Four Current Loops in Every Converter

Every switching supply contains four distinct current paths: the power switch alternating current loop, the output rectifier alternating current loop, the input source current loop and the output load current loop. The input loop charges the input capacitor with an approximately direct current, making that capacitor a broadband energy store.

The output capacitor performs the same role on the other side, holding the high frequency energy delivered by the rectifier while supplying the load. Because of this, the capacitor terminals are the reference point for the whole design. Each loop should connect to the supply only at those terminals. If a loop cannot reach the capacitor terminal directly, alternating energy will flow through the input or output capacitor and radiate into the environment.

Placing the Critical Loop Components First

The switch loop and the rectifier loop are the worst offenders. They carry high amplitude trapezoidal current with steep edges, transition times on the order of fifty nanoseconds, and peaks that can reach five times the steady direct current. Their harmonic content extends far above the switching frequency, so they must be laid out before any other trace is drawn.

Within each loop, the three key components, the filter capacitor, the switching device or rectifier, and the inductor or transformer, should sit adjacent to one another with the shortest possible current path between them. The recommended placement order runs as follows: transformer first, then the switch current loop, then the output rectifier loop, then the control circuitry referenced to the alternating side, and finally the input and output filter loops.

Trace Width, Spacing and Safe Clearances

Every trace carrying alternating current should be as short and as wide as the layout allows, and the components connected to it should be placed close together. Length increases both inductance and impedance, while width reduces them, so a wide short trace is always preferable to a thin long one even if the current is modest.

Spacing is governed by electrical safety rather than by convenience. Adjacent conductors must be separated enough to withstand the working voltage, and where density permits, wider spacing is better for both manufacturing and reliability. A common starting point is around eight mils between traces, with signals of very different levels kept short and separated. Pad holes should sit more than a millimetre from the board edge to avoid damaged pads during machining, and thin traces entering a pad benefit from a teardrop transition to prevent lifting.

Grounding: Pours, Planes and Common Reference

Ground strategy follows from the loop analysis rather than from habit. Copper pours on both sides of the board are typical, but the critical requirement is that each loop returns to its capacitor terminal by the shortest available route. A ground plane helps only when it provides that return; a plane that forces the return current to detour has made the problem worse.

Isolated topologies need special attention. In an isolated converter with feedback from the secondary back to the primary, both sides must share a common reference, so the two ground pours have to be joined deliberately after each is poured. Ignoring that connection leaves the feedback signal referencing a floating node and produces instability that is difficult to diagnose from the schematic.

Routing the Control and Feedback Paths

Once the power loops are placed, the control section follows. Decoupling capacitors belong immediately adjacent to the supply pins of the controller, with the shortest possible connection to the ground reference. Feedback dividers should be placed near the controller input, and their traces kept away from the switch node, which is the highest dv/dt node on the board.

Loop area reduction is the governing principle throughout. Minimising the area enclosed by every high current path suppresses radiated interference far more effectively than adding shielding later. Techniques for taming the remaining emissions are described in EMI suppression design principles and in the notes on DC-DC converter layout and routing.

Review Checklist Before Release

The final review should verify clearances between traces, between traces and pads, between traces and vias, between pads and vias, and between via pairs. Power and ground widths should be checked against the current they carry, using a trace width current calculation rather than estimation, and any location where the ground can be widened should be widened.

It is also worth remembering that every routing or via change invalidates the previous copper pour. Re-pouring after each modification, and regenerating the artwork and drill files from the final database, prevents the familiar situation where the manufactured board differs subtly from the reviewed one. Checking the design rules against the fabricator capability closes the loop.

Simulation, Prototyping and Measurement

A switching power supply layout can be reviewed on paper, but it still has to be verified on hardware. Thermal imaging at full load shows whether the switch and the inductor have adequate copper. Near field probing around the switch node identifies which loop is radiating, and conducted measurements at the input reveal whether the filter is doing its job.

Good practice is to build the first prototype on the intended stackup rather than on a convenient two layer board, because the ground plane and the layer arrangement are part of the design. Probing points for switch node, output ripple and feedback should be designed in from the start, with short ground returns for the probe itself. A converter that measures cleanly on the bench with proper probing is far easier to pass through compliance testing than one that was never instrumented.

FAQ

Why does loop area matter more than trace resistance? Resistance affects efficiency and heating, but loop area determines how much high frequency energy is radiated or picked up. Compliance failures and noise coupling are almost always geometry problems rather than resistance problems.

Should the input capacitor be electrolytic or ceramic? Both, usually in parallel. The bulk capacitor stores energy for the switching cycle, while a small ceramic device close to the switch handles the high frequency component that the bulk part cannot supply.

How wide should power traces be? Wide enough to carry the current with acceptable temperature rise, and as short as the layout permits. Compute the width from current and copper thickness instead of copying a rule of thumb from another design.