High Reliability PCB Design: Thermal and Component Rules

Boards that have to work in demanding environments are designed to a stricter set of rules than consumer products, and the difference is not one dramatic requirement but a large number of small ones applied consistently. The rules cover thermal placement, the way pads connect to copper, the expansion behaviour of the components chosen, and the way the design documents what the fabricator and assembler are expected to do.

Thermal Placement

The rules begin with where the heat-producing parts are put.

Devices that dissipate significant power belong where the air can reach them, which in a forced-convection enclosure means near the outlet rather than the inlet, so that their heat does not pass over the rest of the board on its way out. They should not obstruct the flow path, and heatsinks should be arranged so that they increase the surface exposed to the airflow rather than shielding the components behind them.

Where measurement or validation is not available, spacing rules are used to protect the parts that are sensitive to heat. A common practice is to require that components with significant temperature sensitivity — electrolytic capacitors being the usual example — are kept a defined distance from any heat source whose own rise exceeds a stated amount, with a larger separation required under natural convection than under forced airflow, because still air removes heat far less effectively. Where the space does not allow the required separation, the gap is closed by measurement: the component’s temperature is verified under load, and the design proceeds on the basis of the result rather than the rule.

Where the power density on a board area becomes high — the kind of figure produced by several dissipating devices in close proximity — additional measures are needed rather than more spacing. Thermal bars, added copper structures and conduction paths into the chassis are the usual answers, and they have to be planned while the layout is open.

Copper and Thermal Relief

Connecting a pad directly to a large area of copper makes soldering difficult, because the copper conducts heat away faster than the soldering process can supply it. The usual remedy is a thermal relief: the pad is joined to the copper by several narrow spokes rather than by a solid connection, which limits the rate at which heat leaves the pad during soldering while still providing electrical continuity.

The exception matters as much as the rule. A pad that has to carry substantial current should not be relieved, because the spokes are a restriction in the current path and they heat up. Above a defined current the connection should be solid, and the design should state where that boundary lies so that the two requirements are not applied to the same pad by mistake.

The pattern of the spokes is also specified. A cross or star arrangement distributes the connection evenly around the pad, which keeps the thermal behaviour symmetric — the same requirement that prevents small chip components from standing on one end during reflow.

Materials and Expansion

Material choices in this class of design are made for reliability rather than for electrical performance.

Laminates with a higher transition temperature are normal, because the board has to survive both the process temperatures and the operating environment without losing its dimensional stability. Where such materials are used, the thickness tolerance should be stated in the design data, since the material’s behaviour during pressing depends on it and the stack-up assumes a specific result.

The coefficient of thermal expansion is the property that decides whether a component can be used at all in some positions. A leadless ceramic package and a laminate board expand at very different rates, and the solder joints beneath such a package carry the difference as mechanical strain on every temperature cycle. Unless the combination has been validated — by test or by a design that includes a compliant structure — the mismatch is a reason to choose a different package rather than to accept the risk.

Component Selection Rules

Several rules concern the components themselves and the footprints they sit on.

Footprints must match the parts. Where a new device is introduced, the footprint is created from its own specification rather than adapted from a similar package, and the silkscreen outline is checked against the actual body so that assembly and inspection see what they expect.

Where two different components share a footprint — a common situation with relays and connectors offered in several pin configurations — each variant needs its own holes rather than a compromise pattern, and where the variants require it, the pads are electrically connected so that either version completes the circuit.

Certain components are prohibited in certain roles. A jumper or shunt used for current measurement may be specified on a non-plated hole, so that the connection is made by the component alone and the board does not provide a parallel path that would corrupt the measurement. And a surface mount part should not be used where it would have to be hand soldered, because the thermal shock of hand soldering is outside what the part was qualified for.

Similarly, a part designed for through-hole assembly should not be used as a surface mount part, and a package that has not been validated for the process it will see should not be adopted merely because it fits the layout.

Documentation and Verification

What separates a design that meets a demanding requirement from one that merely looks like it does is the documentation.

The fabrication data should state the material and its key properties including the thickness tolerance, the surface finish, and any feature that is unusual enough that a fabricator might otherwise substitute a default. The stack-up should state the constructions and the impedance requirements where they exist.

The assembly data should state the process, the temperature the components will see, and any component with a specific handling requirement such as a moisture sensitivity level that dictates drying before use.

The verification plan should state what will be measured and to what limits: the temperature of the components that were placed with a spacing rule, the impedance of the controlled traces, the barrel quality of the holes, and the performance of the board after the environmental exposure the product is expected to survive.

Most of these items are the standard content of a well-run design review rather than additions specific to high reliability. The thermal rules that place components relative to heat sources are covered further in the article on thermal management, the sizing of conductors that carry heavy current in the note on heavy copper, and the acceptance testing that confirms the result in our overview of quality management.

FAQ

When should a pad not have a thermal relief? When it carries substantial current. The spokes that restrict heat flow also restrict current, so pads above a defined current are connected solidly instead.

How far should a heat-sensitive part be from a heat source? The required separation is larger under natural convection than under forced airflow, and where the space does not permit it, the temperature has to be verified by measurement rather than assumed.

Why avoid a large expansion mismatch? Because the solder joints beneath a leadless package carry the difference between the package and the board on every temperature cycle, and that strain eventually cracks them.