Circuit Card Assembly Natural Frequency

Overview

This case study demonstrates the natural frequency prediction and dynamic suitability assessment methodology for a fictional multi-layer circuit card assembly (CCA) used inside an avionics chassis. Understanding the first natural frequency (fn1) of a PCB assembly is critical for evaluating its behavior under random vibration environments — if fn1 falls within the input PSD bandwidth, the board will resonate and components may experience fatigue failure.

The analysis uses the Steinberg simplified plate method to analytically predict fn1 based on board geometry, material properties, support conditions, and component mass loading. Predicted deflection at the board center under resonance is then compared against Steinberg's allowable deflection criteria for each component type to determine fatigue risk. IPC-2221 guidelines are referenced for component placement recommendations, and the effect of conformal coating on fatigue life improvement is quantified.

Board Properties & Configuration

The CCA is an 8-layer FR4 laminate with 1 oz copper pours on signal and power layers. Board dimensions, thickness, and edge support conditions significantly influence the natural frequency. The board is supported along both long edges by card guide rails inside the chassis, providing a simply-supported boundary condition on two edges and free conditions on the remaining two edges (SCSC configuration per Steinberg notation).

Natural Frequency Prediction

The first natural frequency is calculated using the Steinberg simplified plate equation, which accounts for board flexural rigidity, support conditions via a boundary condition coefficient (C), and total board weight including components. The equation assumes uniform mass distribution, which is conservative for boards with concentrated mass at component locations.

The predicted fn1 of 287 Hz falls within the flat portion of the input PSD (100–250 Hz boundary is nearby), meaning the board will resonate during random vibration testing. This is expected for avionics PCBs and does not by itself indicate failure — the deflection amplitude at resonance must be evaluated against allowable limits for each component.

Board Deflection & Steinberg Criteria

At resonance, the dynamic deflection at the board center is estimated using Miles' equation applied to the board's modal response. The 3-sigma peak deflection (Z) is compared against the Steinberg allowable deflection (Zallow) for each component type. The allowable deflection is a function of board length, component body length, lead wire configuration, and whether conformal coating is applied.

Conformal Coating Effect

Conformal coating applied to the PCB assembly effectively increases the allowable deflection for coated components by a factor of approximately 2.0 per Steinberg's guidelines. This is because the coating distributes stress more evenly across component leads and solder joints, increasing fatigue resistance.

Applying full-board conformal coating would bring the T1 transformer ratio from 1.23 to approximately 0.62 (pass) and the U4 BGA ratio from 1.41 to 0.71 (pass). For BGA components, however, conformal coating is not always practical due to potential underfill interference, so relocating U4 away from the board center region toward a card guide edge is the preferred corrective action.

*Coating factor for BGA may be less effective; relocation preferred.

Conclusions & Recommendations

The CCA first natural frequency is predicted at 287 Hz, which falls within the random vibration input bandwidth and will result in resonance during testing. This is typical for avionics PCBs and manageable through proper component placement and protective coating.

Two components fail the Steinberg allowable deflection criteria in the uncoated configuration: the transformer T1 (ratio 1.23) and the BGA U4 (ratio 1.41). Recommended corrective actions are to apply full-board conformal coating per IPC-CC-830 for T1, and to relocate U4 from the board center toward the supported card guide edge to reduce its local deflection amplitude. All remaining components pass with margin in both coated and uncoated configurations.

Implementation of these two changes is expected to bring all component deflection ratios below 0.80, providing adequate fatigue life margin for the required random vibration test duration plus a 3× scatter factor.