GNSS Satellite Antenna Housing: Dynamic Qualification & DFM Redesign

A modal and steady-state dynamic qualification study of a 770 mm GNSS antenna housing against the IEC 60945 vibration sweep standard used across maritime navigation equipment. The core question wasn't just whether the structure survived the sweep, but exactly where its resonances sat relative to the qualification band — and whether the standard way of reading dynamic results was even reporting the true worst-case stress and displacement.

That dynamic analysis directly shaped a structural redesign of the housing, pushing its resonant behaviour from deep inside the test sweep to the edge of it, with dynamic stress and displacement dropping sharply as a result.

The housing was modelled as a full assembly — cover, antenna body, and maintenance door — meshed with a combination of linear and quadratic tetrahedral and hexahedral solid elements, with the antenna elements themselves represented as virtual point masses coupled to the structure rather than modelled in full detail, since their inertial contribution to the dynamic response mattered more than their internal geometry. The base was encastred and excited with a 7 m/s² acceleration input swept continuously from 2 Hz to 100 Hz, matching the IEC 60945 qualification profile, with a modal analysis run alongside the sweep to extract every structural resonance falling inside that band.

A key part of the methodology addressed a subtlety in how steady-state dynamic results are normally read: a standard frequency-response solve reports the "real part" of stress and displacement at each frequency step, referenced to a fixed excitation phase — but the true worst-case response at resonance, considering all possible phase angles, can be meaningfully higher than that real-part value. A dedicated phase-sweep envelope post-processing step was built to extract this true peak response rather than relying on the default real-part output. On the original housing geometry, this made a material difference: the default real-part reading at resonance showed a peak stress of roughly 3.3 MPa, while the true phase-envelope peak came out at roughly 4.9 MPa — about 48% higher, with displacement showing a comparable gap (0.18 mm real-part vs 0.27 mm true peak). Missing that gap would mean qualifying a design against a resonance response that understates its actual worst case by nearly half.

That dynamic picture told a clear structural story. The initial, more complex housing geometry carried its first structural modes at roughly 44–50 Hz — squarely inside the 2–100 Hz qualification sweep — and produced a sharp resonant peak in structural response between 66 Hz and 80 Hz, right in the middle of the test band where the sweep would actually excite it hardest. The redesigned, simplification-driven architecture shifted that behaviour decisively: its first structural modes moved up to roughly 98–99 Hz, pushing the resonant peak response to the very edge of the qualification sweep rather than its centre. Alongside that frequency shift, peak dynamic stress and displacement both dropped by roughly a factor of five, giving the redesigned housing a genuine dynamic margin against the qualification requirement rather than a resonance sitting inside the test envelope itself.