How the GNSS antenna connection works

One coax does two jobs at once: RF comes down from the antenna to the receiver, and 3.3 V goes up the same cable to power the amplifier inside the antenna.

EinarBoard v6 (PCB) ZED-F9P RF_IN VCC_RF 3.3 V C143 100 nF DC block: RF passes, DC stops L21 68 nH bias tee: DC passes, RF blocked 3.3 V, ≤50 mA D27 ESD 0.5 pF J12 U.FL pigtail 10–15 cm enclosure wall SMA bulkhead nut + gasket seal (IP) takes the torque, not the PCB antenna cable (SMA macho) Active GNSS antenna (outside, clear sky) patch L1/L2 LNA the LNA runs on the 3.3 V that arrives up the coax ground plane / metal surface under the antenna kills reflections L1 / L2 from satellites RF signal → down the coax → into RF_IN 3.3 V DC → up the same coax → powers the antenna's LNA
Fig. 1 — the same piece of coax carries both directions. On the board, L21 (the bias tee) injects 3.3 V onto the antenna side of the line — an inductor passes DC but looks like a wall to 1.5 GHz RF, so the injection doesn't disturb the signal. C143 (the DC block) does the mirror job: RF passes into the receiver, DC cannot. Result: the antenna's built-in amplifier gets its power with zero extra wires, which is why the F9P requires an active antenna and why J12's little circuit exists.
HBOX cabinet (top view) EinarBoard v6 — 140 × 100 J12 GNSS J13 LTE ≈149 mm apart on the board SMA bulkhead SMA bulkhead antennas ≥ 20 cm apart — the #1 defence against the modem deafening the GNSS GNSS antenna highest point, clear sky LTE whip
Fig. 2 — where each piece lives. The board carries only the two U.FL receptacles, placed at opposite corners. Short pigtails jump to SMA bulkheads that clamp through the enclosure wall — the nut and gasket keep the seal, and any torque from screwing antennas on lands in the wall, never in the PCB's solder joints. Outside, the antennas separate further: GNSS up high with open sky, the LTE whip away from it.