Vibration and Shock Testing for Assemblies

What Vibration Does to a Board

A board mounted in a product is a plate supported at its edges or at a few points, and a plate has resonances. When the excitation matches a resonance, the displacement is amplified far beyond the input, and the components at the point of maximum deflection experience acceleration that is many times what the specification states. That amplification is why a board can survive a modest vibration input and still fail at a specific frequency, and why the test has to sweep the frequency range rather than apply a single tone. The damage appears as a fatigue crack in a solder joint, a broken component, a cracked termination, a loosened fastener or a worn connector contact.

Test Types

The common vibration tests are sinusoidal sweep, random vibration and a combination of both. A sinusoidal sweep applies a single frequency that moves through a range, and it is used to find resonances and to test a product against a known excitation, such as an engine’s rotation. Random vibration applies energy across a band simultaneously, which is closer to the environment of a vehicle or a machine, and it is specified by a power spectral density. A shock test applies a short, high amplitude pulse, which excites the higher modes and is used to represent an impact or a drop. The choice depends on the environment the product will see, and the specification should name the type, the level and the duration.

Finding the Resonances

Before a full test, the assembly’s resonances are usually identified with a low level sweep and an accelerometer, or with a modal analysis. Knowing where the resonances lie allows the test to be designed so that the dwell or the amplification is applied where it matters, and it also allows the design to be changed. Where a resonance falls in the range of the excitation, the options are to stiffen the board by adding material or a stiffener, to add mass to move the resonance, to change the mounting points so that the span is shorter, or to damp the assembly. Design changes are cheaper than an exemption from the specification.

Fixtures and What They Change

The fixture is part of the test. A fixture that is not stiff or that has its own resonances will modify the excitation that reaches the assembly, and the result will describe the fixture rather than the product. The mounting points on the fixture should reproduce the way the board is mounted in the product, since a board held at four corners behaves very differently from one supported along an edge. The accelerometer should be placed on the board at the point of interest and, where possible, at the point of maximum deflection, so that the amplification is measured rather than inferred. The fixture and the mounting should be recorded with the test.

Failure Modes and Their Location

The failures appear where the stress is highest, which is usually at the points of maximum curvature and at the interface between a rigid and a compliant part. Large components with a high centre of gravity, such as an electrolytic capacitor or a heavy inductor, are the most likely to crack a joint or a termination, because their mass loads the joint through the distance. A connector that is not latched can work loose, and the wear on its contacts appears later as an intermittent fault. A heavy heatsink attached to a package can transmit its own load into the solder balls. The location of the failures across a sample is the data that points to the design change.

Design and Assembly Measures

The effective measures are structural. Stiffening the board, shortening the span between mounts, adding a stiffener or a metal frame, and moving heavy components away from the areas of maximum deflection all reduce the stress. On the assembly side, the joint quality matters: a joint with adequate solder volume and correct wetting resists fatigue better than a starved one, and an underfilled package resists the shear load better than one without. A component bonded to the board with an adhesive reduces its ability to move, at the cost of difficulty in rework. The choice among these follows the failure that the test found, which is the reason to test before the product is released rather than after a field failure.

Interpreting the Result

A vibration test produces failures, and the question is whether the number and the type are acceptable. A single failure among several samples is evidence of a weak link rather than a random event, and it should be analysed rather than repeated. The analysis includes the location of the failure relative to the measured deflection, the condition of the joint in a microsection, and whether the failure is a fatigue crack or an overload. Where the test passes, the margin should be recorded, because a product that passes at the specification level with little margin may fail in the field where the excitation is higher or where the product is mounted differently.

FAQ

Why does a board fail at one frequency? Because the excitation matches a resonance, which amplifies the displacement well beyond the input level.

What is the difference between sinusoidal and random vibration? A sweep applies one frequency at a time; random vibration applies energy across a band and is closer to a real environment.

Why does the fixture matter? A flexible fixture or its own resonances modify the excitation, so the test measures the fixture rather than the product.

Which components fail first? Heavy components with a high centre of gravity, which load their joints through the distance from the board.

How is the margin recorded? By noting the level at which the first failure occurs relative to the specified level, so that a small margin is visible.

Conclusion

Vibration testing finds the resonances that a static test cannot, and the response is structural: stiffen, shorten the span and move the mass. Measure the deflection rather than trusting the input. Reliability testing belongs to PCBA testing, the assembly under test sits in PCB assembly, and the mechanical design is part of PCB design and layout. Vibration requirements for a new product are set during prototype PCB assembly in 2026.