EMI Shielding for GNSS Receiver Modules: Protecting L-Band Front Ends on Crowded Boards
Knowledge Base

EMI Shielding for GNSS Receiver Modules: Protecting L-Band Front Ends on Crowded Boards

The Most Fragile Signal on Your Board

A GNSS receiver listens for satellite signals arriving at roughly −130 dBm at the ante

a — power levels around 10⁻¹⁶ watts. Meanwhile, the board it sits on hosts switching regulators, digital clocks, and cellular radios radiating anywhere from −60 to −20 dBm in the near field. The desired signal sits more than 100 dB below the interference, right inside the L-band window (1575.42 MHz for GPS L1, with GLONASS G1 near 1602 MHz, Galileo E1 at the GPS center, and BeiDou B1 nearby). This asymmetry is why GNSS front ends are uniquely intolerant of board-level EMI: any spurious emission that lands in-band, or any strong out-of-band signal that desensitizes the LNA, directly erodes time-to-first-fix and position accuracy.

Where the Interference Comes From

Three board sources matter most. Clock harmonics: a 52.514 MHz reference oscillates at exactly one third of GPS L1 — deliberate in many module designs, but any oscillator whose harmonic lands near 1575.42 MHz (100 MHz PCIe clocks hit it at the 15.75th, and fractional-N synthesizers place spurs unpredictably) puts noise into the GNSS band. Real offenders are often unintentional: spurious content at 1562.5 MHz from a 125 MHz Ethernet clock (12.5×), or the dense spurious grids that cellular fractional-N synthesizers sweep across L-band during registration bursts. Switching regulators: fast switch-node edges radiate harmonics past 1 GHz, and a buck converter within 20 mm of the ante

a feedline is a classic desensitizer. Co-located transmitters: an LTE or Wi-Fi module transmitting at +23 dBm can drive the GNSS LNA into compression or generate front-end intermodulation even when its carriers sit far from L-band.

Shield Can Design for the GNSS Module

The first and most effective layer is a shield can over the GNSS section — whether it is the ca

ed module itself or your discrete front end. Three design points determine whether it works. First, isolation of the ground: the can should ideally co

ect to a local ground island dedicated to the GNSS section, tied to the main ground plane at a single controlled point (or a few stitching vias), so noisy return currents from the rest of the board do not flow across the receiver’s ground. Second, zero apertures over the sensitive area: seams between the can wall and the PCB must be closed by ground pads on a fine pitch (2–3 mm), and the can must not overhang the edge of its ground pads, where it becomes a slot ante

a. Third, cavity resonance: a shield can with wall openings behaves like a trap or re-radiator at certain frequencies. If in-band spurs appear only when the can is fitted, cavity or edge coupling is the culprit; a slightly different can size or an internal absorber sheet moves the resonance.

Filtering What Enters the Can

Every conductor crossing the shield boundary must be filtered, or the can becomes decorative. Power entry gets a pi-filter or ferrite plus capacitor combination sized to pass 3.3 V cleanly but present high impedance above 1 GHz. The RF path is handled by the front-end selectivity itself — most GNSS modules include an integrated SAW filter after the LNA, and if your design is discrete, a bandpass SAW (typical 1.2–2.5 dB insertion loss, more than 30 dB rejection of LTE bands and 40+ dB skirt below 1.5 GHz) placed before the LNA is the single best protection against blocking. The 1PPS output, UART, and I²C lines leaving the can should each carry a small ferrite bead; these lines are long, unshielded, and efficient at carrying digital noise back into the shielded volume.

Absorbers, Layout, and Verification

Inside the can, a thin ferrite absorber sheet on the can lid damps cavity resonances and suppresses near-field coupling between the module and its own shield — the same technique used in 5G modules, and cheap insurance at GNSS sensitivities. On the layout side: keep the ante

a feedline short and over a clean ground, route no digital signals parallel to it, place the GNSS section at a board corner farthest from transmitters, and give the LNA supply its own LDO with dedicated decoupling.

Verify with a staged approach. First, a near-field probe sweep across the board with the receiver muted, hunting for fields at 1559–1610 MHz. Second, a carrier-to-noise (C/N₀) comparison: run the receiver with the shield fitted and removed, and with nearby transmitters active and idle — a healthy design loses less than 1–2 dB-Hz across all four conditions. Third, the ultimate test is a sky test: outdoors with an open view of the sky, measuring time-to-first-fix and C/N₀ against the datasheet. If TTFF stretches and C/N₀ drops when a cellular module powers up, your shield-and-filter chain still has a leak — and it will almost always be at a cable, a filter component with too little blocking impedance, or a can seam with too few ground stitches.