Thermal Cycling Tests for Assemblies

What Cycling Tests

A thermal cycling test exposes an assembly to repeated changes in temperature and measures how long it survives. The mechanism it targets is the difference in expansion between the materials: the laminate, the copper, the solder and the component package all expand at different rates, and every cycle imposes a strain wherever two of them are joined. The test does not simulate a particular field condition exactly; it applies an accelerated version of the thermal stress that the assembly will see, so that a weak structure fails within a reasonable time and can be identified.

The Test Parameters

Four parameters define a cycling test: the upper and lower temperatures, the dwell time at each, the ramp rate between them and the number of cycles. Each of them changes which mechanism dominates and how quickly it appears. A test with a wide range stresses the expansion mismatch more; a test with a long dwell gives the assembly time to reach a uniform temperature, so that the whole structure sees the full excursion; a fast ramp adds a thermal gradient across the board, which produces a different stress pattern and can be harsher on a thick assembly. The published specifications define a set of conditions, and a test that departs from them is not comparable with the results of another, which is why the conditions should always be reported with the result.

Range and Dwell

The lower temperature is often the more severe end for a solder joint, because the solder becomes less able to accommodate strain as it cools and the stress at the joint rises. The upper temperature matters for the laminate and for the component, and for a lead-free assembly the excursion from a low temperature to a high one is larger than it was for a eutectic alloy, which is one reason lead-free products are often cycled to a more demanding specification. Dwell time has to be long enough for the assembly to reach the set temperature, which for a large or heavy board can be a significant fraction of the cycle. A test with too short a dwell understates the damage, because the interior of a thick board never reaches the extreme.

Ramp Rate and Its Effect

The ramp rate determines the temperature gradient within the board and therefore the stress distribution. A slow ramp gives a nearly uniform temperature and stresses the joints mainly by the difference in expansion between the materials. A fast ramp adds a transient gradient that stresses the board and the components, and it can produce failures at the edges and the corners that a slow ramp would not reveal. Both are legitimate, but they test different things, and a product that will see rapid temperature changes in service should be tested with a ramp that reflects them.

Cycle Count and Failure Distribution

The result of a cycling test is a distribution, not a number: a set of samples is cycled to failure and the distribution of the failure times is what characterises the design. In practice, the test is run to a defined number of cycles and the samples are then examined, usually by cross section, to find the damage that has started. That is a different approach from running to failure and it is what most qualification programmes do, because it gives a defined pass or fail at a defined cycle count. Where the design is being compared with an alternative, running both to failure gives a more informative comparison, at the cost of a much longer test.

What Fails First

The structures that usually fail first are the plated barrels of vias in a thick board, the interface between the barrel and the pad, the solder joints of the largest and stiffest components, and the solder joints of components at the corners of a package or at the edge of a board where the strain is greatest. On an assembly with a large ball grid array, the failure often starts at the corner balls, which is why the corner locations are the ones examined first in a cross section. On a board with a metal core or a ceramic substrate, the attach layer and the dielectric are usually the weak points rather than the solder.

Interpreting the Result

A failed sample is only useful if the failure is located and identified. The standard approach is to cross section through the failure and to describe what cracked: through the solder, along the interface between the solder and the pad, through the barrel, or through the laminate. Each points to a different cause and a different fix. A crack through the solder is usually a thermal expansion mismatch; a crack at the interface is often a plating or an intermetallic problem; a barrel crack points to the aspect ratio, the plating thickness or the laminate. Where the failure appears in the same place on every sample, the design has a specific weak point, and where it appears randomly, the process is the more likely cause.

Cycling and the Design Decisions

The results of a cycling test point back to specific design choices, which is what makes the test worth running rather than simply accepting a specification. A barrel crack points to the aspect ratio and the plating thickness; a crack at the barrel to pad interface points to the annular ring and the desmear; a joint failure at a corner ball points to the pad geometry and the package; a failure of a thick board points to the laminate and the stack. Where a design has a known weak structure, the cycling test will find it, and the corrective action is usually a change to the geometry rather than to the material. This is the reason a cycling test is most valuable early, on a prototype, when the geometry can still be changed, rather than late, when the design is fixed and the only remaining levers are expensive.

FAQ

What does a thermal cycling test show? How an assembly tolerates the strain caused by the difference in expansion between the laminate, the copper, the solder and the component.

Why does dwell time matter? Because a short dwell does not let a thick board reach the set temperature, so the interior never sees the full excursion and the test understates the damage.

Is a faster ramp harsher? It adds a temperature gradient across the board, which produces failures that a slow ramp may not reveal.

What fails first? Usually vias and their barrel to pad interfaces in a thick board, and the corner joints of large packages.

How should a failure be reported? With the test conditions, the cycle at which it occurred and the location and nature of the crack, determined by cross section.

Conclusion

A thermal cycling test is defined by its range, dwell, ramp and cycle count, and its result is only meaningful when those conditions are reported with the failure location. Run the test to a defined cycle count, examine the samples by cross section and record where the damage started. Reliability work belongs to quality management, the structures that fail are described in PCB manufacturing, and the assembly conditions that affect them in PCB assembly. Cycling is normally introduced alongside prototype PCB assembly in 2026.