Thermal Cycling Qualification For Boards

A board in service sees the temperature change every time the equipment is switched on and every time the load changes. Each change makes the materials expand by different amounts, and the difference is carried by the joints, the vias and the interfaces between the layers. Thermal cycling is the test that reproduces that accumulation in a compressed time.

This article describes what the test does, the profile parameters that determine whether it is meaningful, and how the result is monitored and judged.

What The Test Reproduces

The test cycles the assembly between two temperature extremes and repeats the cycle many times. The damage accumulates because the materials have a different coefficient of thermal expansion, so the solder joints, the plated barrels and the laminate interfaces are loaded on every transition. A single cycle does nothing measurable, and the effect appears only after a large number of them.

The test is therefore an acceleration of a slow process rather than a simulation of a service condition. The number of cycles that a product will see in its life is estimated from the usage pattern, and the test applies an equivalent or greater number in a shorter time. The acceleration is achieved by using a wider temperature range than the service condition, which is why the profile has to be stated with the result.

The acceleration is legitimate only within limits. A test that uses a temperature range far wider than the service range can introduce failures that would never occur in the product, particularly where a material changes phase or softens at the higher temperature. The useful range is therefore the one that stresses the same mechanisms as the service life without introducing new ones, and the profile should be justified from the service condition rather than from the convenience of a shorter test.

The number of samples matters as much as the profile. A single assembly that survives the test demonstrates that one unit survived, while a set of samples taken across the panel and across several lots demonstrates that the process is capable. Where the result will be used to qualify a supplier or a material change, the sample size should be large enough to give a meaningful result rather than a single data point.

The Cycle Profile And Its Parameters

Four parameters define the profile: the high temperature, the low temperature, the ramp rate and the dwell at each extreme. The temperatures should bracket the service range with margin, and the extremes should be the temperatures at which the materials are most stressed rather than the temperatures that are most convenient for the chamber.

The ramp rate determines how much of the difference in expansion the assembly sees as a transient gradient. A fast ramp produces a larger gradient across the board and therefore more stress on the joints, while a slow ramp allows the assembly to reach a more uniform temperature. Two laboratories using the same temperatures and different ramp rates will produce different failure rates, which is why the ramp is part of the specification.

Dwell Time And Why It Matters

The dwell is the period at the extreme temperature before the ramp begins. Its purpose is to allow the whole assembly to reach the extreme, because a sample that is still cooling when the ramp starts has not seen the intended condition. A dwell that is too short under tests the product, and one that is too long adds time without adding damage once the assembly has stabilised.

The dwell also allows the creep and stress relaxation that occur at the high temperature to take place. A solder joint that is held at a high temperature relaxes the stress that the ramp introduced, and the relaxation is part of the damage mechanism. A profile with a very short dwell therefore removes one of the mechanisms that the test is meant to reproduce.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/rigid-flex-pcb-design-guidelines.webp" alt="Cross section of a plated via wall crack after cycling” />

Resistance Monitoring During The Test

The standard practice is to monitor the resistance of a daisy chain that runs through many joints or vias, so that a single failure is detected as a step change. The measurement is continuous or sampled at a rate high enough to catch a transient open, because a cracked joint can close again as the assembly cools and the failure would otherwise be missed.

The event criterion defines what counts as a failure, usually a rise in resistance above a threshold that lasts for a defined time. The threshold and the duration should be stated, because a very short spike and a permanent open are different events with different causes. The monitoring is what turns the test from a pass or fail into a way of locating the failure, since the chain that failed identifies the structure.

Failure Modes That Appear

The characteristic failures are a crack in the wall of a plated via, a crack in a solder joint at the interface with the pad or the component, and a delamination at an interface in the laminate. The via wall crack is driven by the difference in expansion between the copper and the laminate in the direction of the board thickness, and it appears where the plating is thinnest, which is the middle of a deep hole.

The joint crack is driven by the difference between the component and the board, and it appears at the corner of a package where the distance from the neutral point is greatest. The delamination appears where the bond between the resin and the glass or the copper is weakest, and it is often associated with moisture that was absorbed before the test. The structure of the joints that resist those mechanisms is described for via to trace clearance and in the general discussion of multilayer construction.

Setting A Qualification Requirement

The requirement should state the temperatures, the ramp rate, the dwell, the number of cycles and the failure criterion. A requirement expressed only as a number of cycles is not reproducible, because the other parameters change the damage per cycle substantially.

The number of cycles should be derived from the service life rather than chosen from a table. The temperature difference that the product actually sees in service, the number of times per day it is switched on, and the expected life in years together give an estimate that the test can be designed to cover. Where the product sees very few cycles in its life, the test is a confidence check rather than a life estimate, and the acceptance criteria for the result belong to the general quality characteristics of the product.

FAQ

Why does the dwell time change the result? A dwell that is too short does not allow the assembly to reach the extreme temperature, and it removes the stress relaxation that occurs at the high temperature. Both reduce the damage per cycle.

Why is the resistance monitored continuously? Because a cracked joint can close again as the assembly cools. A measurement taken only at the end of the test would miss an intermittent failure that would still be a defect in service.

Where does a via wall crack usually start? Where the plating is thinnest, which is normally the middle of the barrel. That is why the plating specification applies to the minimum rather than to the surface.