Thermal Management in PCB Design

Heat is the most common reason an electronic product fails early, and the board is usually the only heatsink available. Thermal management in a PCB is not a single feature but a path: from the device, through the solder joint, into the copper, across the board and out to the environment. Every step in that path has a resistance, and the design determines all of them.

Where the Heat Comes From

Every watt dissipated in a device leaves through its package, and the split between the leads, the case and the board depends on the package. A large power device with a thermal pad sends most of its heat downwards into the board, while a small package in still air loses more through its own surface.

Knowing which path dominates tells the designer where to spend effort. If most of the heat leaves through the board, the copper beneath the device is the design feature that matters. If it leaves through the case, the mechanical interface between the case and the enclosure is the important one. Our metal core PCB article describes the case where the board itself is the heatsink.

Thermal Vias and Their Limits

A thermal via carries heat from a surface pad to an internal plane or to the opposite side of the board. A single via moves very little heat, so they are used in arrays, and the array’s effectiveness depends on how well it is connected at both ends.

The via barrel is plated copper, which conducts well, but the plating is thin and the dielectric surrounding it is a poor conductor. The result is that a thermal via array is less effective than an equivalent area of solid copper, and its value comes from providing a vertical path that solid copper cannot. Filling the vias with a thermally conductive material improves them, and it also prevents solder from wicking away from the joint. Our via placement notes cover the related assembly consideration.

Copper Area and Heat Spreading

Once heat reaches the copper, it spreads laterally. The spreading resistance falls as the copper area grows, and thickness helps as well, because a thick layer has a lower resistance to lateral flow.

This is the reason a large copper pad under a power device works even without vias: the heat spreads sideways into the surrounding copper and is then lost to the air over a larger area. The practical limit is that spreading improves with distance only up to a point, so a pad far larger than necessary adds board area without adding much benefit. Our copper thickness article covers the design choices involved.

Planes as Heat Spreaders

Internal ground and power planes are large areas of copper, and they spread heat better than any surface feature. Connecting a thermal via array to an internal plane distributes the heat across the whole board rather than concentrating it near the device.

The plane also increases the thermal mass, which slows the rate at which temperature rises during a transient. That does not reduce the steady state temperature, but it does reduce the peak during a short burst of power, which is often what a datasheet limit is concerned with.

Component Placement and Thermal Coupling

Components influence each other. A temperature sensitive part placed beside a hot regulator will drift, and a hot component placed beneath another will heat it from below. The layout decides which of these happens.

The practical rules are to separate heat sources, to place temperature sensitive parts away from them and upstream in the airflow, and to avoid stacking hot components on opposite sides of the board at the same location. Where a product has a fan or a vent, the airflow direction belongs in the placement decision rather than being considered afterwards.

Estimating Junction Temperature

The junction temperature is the quantity that matters, and it is not the same as the case temperature or the board temperature. Estimating it requires the thermal resistance from junction to the point being measured, plus the power dissipated, plus the temperature of the environment the heat is ultimately leaving into.

The calculation is only as good as its assumptions, which is why the first prototype should be measured. Attaching a thermocouple to the device case and to the board beneath it, then running the product at its worst case load to steady state, gives the real figure. If it exceeds the estimate, the copper, the airflow or the mechanical path is the likely cause, and all three are easier to change before tooling is committed. Our reliability notes describe how the result feeds into the life estimate.

Thermal Management Beyond the Board

Eventually the heat has to leave the product. The path from board to enclosure, from enclosure to air, and from air to the environment all have their own resistances, and the design of the enclosure is part of the thermal solution.

A board that spreads heat perfectly into a sealed plastic box has not solved anything. Thermal management therefore belongs at product level, with the board designed to deliver heat to the surfaces that can reject it. Our layout tips article covers the placement decisions that support that transfer.

Choosing a Heat Spreader

A heat spreader is a piece of thermally conductive material bonded to a device or to the board to move heat into a larger area. It may be a copper pad on the board, a metal plate bonded to the package or a graphite sheet used where thickness is critical.

The choice depends on the path the heat will take. Where the board is the destination, a copper area and a via array are enough. Where the heat must travel to a distant part of the assembly or to a case, a spreader with a defined thermal interface to that destination is what makes the path work. The interface material between the spreader and its destination is often the largest single resistance in the chain.

Transient Versus Steady State

Thermal design has two regimes and they call for different responses. In steady state the temperature is set by the average power and by the total thermal resistance to the environment, so only the resistance matters.

In a transient the thermal mass decides how fast the temperature rises, and a large copper area or a heavy spreader will slow it. That distinction matters because a datasheet limit may refer to a peak temperature that lasts milliseconds, and the board’s thermal mass is what keeps a short burst inside that limit. Our component tolerance and reliability notes describe how the two regimes affect life.

FAQ

How many thermal vias are needed? Enough that the via array does not become the limiting resistance. Arrays are usually designed to match the area of the pad and to connect to a plane on the other side.

Does a heatsink always help? It helps when the thermal path from the device to the heatsink is good. A heatsink attached to a poorly conducting path adds cost without lowering the junction temperature much.

Is the board temperature the same as the junction temperature? No. The junction is hotter than the case, which is hotter than the board, and the differences are the thermal resistances along the path.

Can thermal problems be found by simulation alone? Simulation is useful for comparing options, but the boundary conditions are hard to model accurately. Measurement on the first prototype is what confirms the design.

What is the most common thermal design mistake? Treating the board as the only path and ignoring the enclosure, or the opposite. Both halves have to be designed together.