Board Level Vibration Testing: Fixture Design and Failure Modes

Vibration testing of a populated board is where mechanical design, material choice and soldering quality meet. A test can be run in a few hours, yet the result depends almost entirely on how the board was held, how the excitation was applied and how the response was measured. This article explains what a board level vibration test is meant to establish, how fixtures shape the result, which failure modes appear first, and how to read the data so that it changes a design rather than just producing a graph.

What Board Level Vibration Testing Establishes

A board level test applies a controlled excitation to an assembly and measures how the assembly responds. The purpose is not to reproduce the exact environment the product will see, because that environment is usually unknown and varies from one installation to the next. The purpose is to compare designs, to find the weak points of a given layout, and to confirm that a change has actually improved the behaviour.

Qualification and development tests answer different questions. A qualification test follows a written profile and produces a pass or fail against a limit, while a development test is exploratory and may sweep well beyond the expected service level to discover where the first failure occurs. Confusing the two is the most common reason a test programme produces results that nobody can act on.

Standards and Test Profiles

Most profiles are drawn from generic standards that define random vibration spectra, sine sweeps and shock pulses in terms of frequency bands and acceleration spectral density. A profile is a specification of the input, and the input alone does not determine the outcome, because the same spectrum applied at a different mounting point or in a different axis produces a different response.

The profile should be chosen for the transport and service environment the product will actually meet, and it should state the duration, the axes and the sequence. A test that runs one axis for a short time is far less demanding than one that runs all three axes to the same level, and the difference is often lost when a specification is quoted by number only.

Fixture Design and Boundary Conditions

The fixture sets the boundary conditions of the board, and the boundary conditions control the mode shapes and the natural frequencies. A board bolted at four corners and a board clamped along two edges are different structures, even when the board itself is identical. If the fixture is more flexible than the board, the test measures the fixture.

A good fixture is stiff, light and simple. Every additional bracket adds mass, damping and a resonance of its own, and those resonances appear in the measured data as if they belonged to the assembly. The fixture should be characterised on its own first, with an accelerometer on the mounting surface but no board installed, so that its own response is known before the assembly is added.

Mounting the Board in the Fixture

Mounting hardware has to reproduce the intended attachment. Screws at the design locations with the design torque, standoffs of the correct height and the correct number of fixing points together define how the board will bend. Adding extra supports to protect a board during a test invalidates the comparison with the real product, because a support that is not present in service removes a mode that the product will actually experience.

Where the product uses card guides or edge connectors, the fixture should reproduce that support rather than replacing it with a rigid clamp. The mounting detail is often the difference between a test that predicts field behaviour and one that merely passes.

Resonance and Mode Shapes

Every board has natural frequencies at which a small input produces a large response. Those frequencies depend on the board outline, the thickness, the laminate properties and the mass and stiffness of the components. A large component placed near the centre of a board lowers the first mode and increases the deflection at that location.

Finding the mode shapes is what turns a pass or fail into an understanding. A sine sweep at low level identifies the natural frequencies without damaging the assembly, and a measurement at several points across the board shows how it is bending. Our component tolerance notes describe how placement and tolerance decisions affect reliability at the design stage.

Random Vibration and Sine Sweeps

A sine sweep concentrates energy at one frequency at a time, which makes resonances easy to see and easy to excite to a damaging level. Random vibration spreads the energy across a band and is closer to what a product experiences on a vehicle or near rotating machinery. Both are useful, and a programme often uses a low level sine sweep to characterise the assembly and a random profile to accumulate fatigue damage.

Duration matters as much as level, because fatigue damage accumulates. A short test at a high level is not equivalent to a long test at a low level, and the relationship between them is not linear. The profile should therefore state the duration with the level, and the accumulated time should be tracked when a test is run in several stages.

Instrumentation and Measurement

The accelerometer placement decides what the test actually observes. A control accelerometer on the fixture table closes the loop and holds the input at the specified level, while response accelerometers on the board measure what the assembly does. Placing a response accelerometer at the predicted maximum deflection, and a second one at a component of interest, gives both the global response and the local behaviour.

Attachment matters because a poorly bonded accelerometer adds mass and can detach during the test. Cable routing matters too, since a stiff cable tied to the board changes the stiffness of the very structure being measured. Both are easy to fix and both are common sources of unusable data.

Failure Modes and Their Locations

The first failures in a vibrating assembly are usually at the solder joints of the largest and heaviest components, at the corners of a large package, and at the interface between a stiff lead and a compliant board. Cracks in the bulk solder, cracks at the intermetallic interface and pad cratering produce different appearances under inspection, and distinguishing them points at different fixes. The progression from a hairline crack to a complete solder joint fatigue failure is exactly what a vibration test is designed to accelerate in a controlled way.

Component cracks, fractured leads and damaged laminate around a fixing hole are also common. Our solder defect notes describe how these mechanical failures present on a cross section, and our laminate properties guide explains how the board material itself contributes to stiffness and damping.

Interpreting Results and Improving Design

A result becomes useful when it is compared with something. Repeating the test after a change, with the same fixture, the same mounting and the same profile, shows whether the change helped. Reducing the free span, adding a stiffener, moving a heavy component away from the centre or changing to a thicker board are all measurable interventions.

The record should include the fixture drawing, the mounting torque, the accelerometer positions, the profile and the observed failure. At gopcb the manufacturing side of that record is covered by our fabrication notes, which keep the board construction consistent between the prototype that was tested and the production units that follow.

FAQ

Should the test be run with a bare board or a populated assembly? A populated assembly, because the mass and stiffness of the components dominate the response. A bare board test is still useful as a reference point, but it cannot be used to qualify the product.

What causes a board to fail at a frequency far below the first resonance? Usually a local resonance of a heavy component on compliant leads, or a fixture resonance that was mistaken for a board mode. Characterising the fixture on its own separates the two.

How long should a vibration test run? Long enough to accumulate the intended damage at the chosen level, which is a function of the profile rather than of convenience. The duration should be stated with the level and the axes in the test specification.