Antenna Characterization for GNSS Applications

Why GNSS Antennas Need Special Attention

A few months back, one of our clients brought in a batch of patch antennas meant for a fleet-tracking device. On paper, the specs looked fine. Out in the field, the units kept losing lock near buildings and under tree cover. The antenna wasn’t “broken” — it just hadn’t been characterized properly for how GNSS signals actually behave.

GNSS signals arrive from satellites low on the horizon just as often as from directly overhead, they’re weak by the time they reach the ground, and they’re right-hand circularly polarized (RHCP). An antenna that isn’t tested against these realities can pass a basic gain check and still fail in the real world. That’s the gap antenna characterization is meant to close.

At Diamond Microwave Chambers Ltd, this is a large part of what we do day to day — putting GNSS antennas through controlled, repeatable tests so engineers know exactly what they’re deploying before it ships.

The Building Blocks of GNSS Antenna Characterization

There isn’t one single measurement that tells you an antenna is “good.” It’s a set of parameters that, together, paint the full picture:

• Radiation pattern — how gain varies across azimuth and elevation, especially at low elevation angles near the horizon
• Axial ratio — how close the antenna comes to pure circular polarization; a poor axial ratio means signal loss and higher multipath sensitivity
• Phase center variation (PCV) — how the electrical center of the antenna shifts with angle of arrival, which matters a lot for survey-grade and RTK applications
• Gain-to-noise-temperature (G/T) for active antennas with built-in LNAs
• Group delay variation — inconsistent delay across the band can introduce timing errors, which is a real problem for GNSS since the whole system runs on precise timing
• Return loss / VSWR — confirms the antenna is actually radiating the power it’s given rather than reflecting it back

Each of these answers a different question, and skipping one usually means the failure shows up later — in the field, not in the lab.

Step-by-Step: How the Evaluation Actually Works

1. Mount the antenna on a low-reflectivity positioner inside an anechoic chambers, oriented so the boresight faces the source antenna.
2. Sweep frequency and angle across the GNSS bands of interest — L1, L2, L5 for GPS, or the equivalent bands for GLONASS, Galileo, and NavIC depending on the target market.
3. Record co-polar and cross-polar patterns at each elevation cut, since axial ratio is calculated directly from this data.
4. Extract the phase center by comparing phase data across multiple angles and fitting it to a reference point.
5. Repeat with the antenna in its actual housing (radome, ground plane, cabling) — a bare antenna and an integrated antenna rarely measure the same.
6. Compile the dataset into pattern plots, axial ratio curves, and a summary table for the design team.

That last step matters more than it sounds. Raw measurement data is only useful once it’s turned into something an RF engineer can act on quickly.

Chamber Testing vs Field Testing

Neither approach replaces the other. Chamber data gives you the clean baseline; field testing tells you how the antenna behaves once real-world clutter gets involved.

Pro Tip: If your GNSS antenna will sit near other RF components — a cellular modem, Wi-Fi module, or metal chassis — characterize it in that exact configuration, not standalone. De-tuning from nearby conductive parts is one of the most common reasons a “perfect” antenna underperforms once it’s installed in the final product.

Common Mistakes Teams Make

• Testing only at boresight and assuming low-elevation performance will follow the same curve
• Ignoring axial ratio degradation at wide angles, which quietly increases multipath error
• Ground plane size overlooked — GNSS patch antennas are especially sensitive to this
• Treating an active antenna’s LNA gain as a substitute for good passive antenna design
• Skipping phase center testing because it’s assumed “close enough” for the application — these backfires quickly in survey and precision-agriculture use cases