Most GPS positioning problems don’t announce themselves dramatically. The receiver doesn’t stop working. The fix doesn’t disappear. What happens instead is subtler: the position drifts a few meters during a slow vehicle pass, the altitude readout bounces when you’re sitting still, the fix takes longer than it used to in the same location, or the heading jumps erratically in a system that’s supposed to report it smoothly. These symptoms get attributed to satellite geometry, signal conditions, or software bugs — when the actual cause is usually antenna placement.
I’ve spent enough time troubleshooting GNSS integrations to have a strong prior: when a GPS system behaves inconsistently across conditions but passes basic receive tests, the antenna location is the first place to look.
The Multipath Problem Nobody Visualizes
Satellite navigation systems work by measuring the time of arrival of signals from multiple satellites simultaneously. The receiver calculates position by finding the location that’s consistent with the measured arrival times from all visible satellites. The entire system assumes that signals travel from satellite to antenna in a straight line.
They often don’t. Signals also arrive at the antenna after bouncing off nearby surfaces — vehicle body panels, building facades, the ground itself. These reflected signals travel a longer path than the direct signal, arriving a few nanoseconds late. The receiver can’t distinguish the reflected signal from the direct signal, so it averages them in a way that biases the pseudorange measurement — the fundamental input to the position calculation. The result is a position error that has nothing to do with satellite geometry or atmospheric conditions, and everything to do with what’s sitting near your antenna.
Multipath errors are particularly insidious because they’re location-dependent and direction-dependent. A vehicle antenna that performs acceptably when the vehicle faces east may show significant position scatter when it faces north, because the geometry of the reflecting surfaces relative to the satellite positions changes. If you’ve ever seen a GPS system that behaves inconsistently in the same general area, multipath is a plausible explanation.
Ground Plane: The Most Under-Specified Requirement
A GPS patch antenna is designed to receive signals arriving from the upper hemisphere — from the satellites overhead — while rejecting signals arriving from below. The ground plane beneath the antenna is what makes this work. It reflects downward-arriving signals away from the antenna element and provides the reference plane that determines the antenna’s elevation cutoff angle.
The ground plane needs to be appropriately sized. For a standard patch antenna at GPS L1 frequency (1575.42 MHz), the ground plane should extend at least a wavelength in radius from the antenna element — roughly 19cm. A ground plane that’s too small allows low-angle signals and ground reflections to reach the antenna that a properly sized ground plane would reject. The elevation cutoff angle rises, and the antenna becomes more susceptible to multipath from nearby surfaces.
In practice, the ground plane is often whatever is convenient: a small metal bracket, the top of an enclosure, a PCB plane that’s sized for the board rather than for RF performance. This works well enough that most installations function acceptably, but it’s the source of the marginal performance that puzzles people when conditions get harder — urban environments, vehicles with complex body geometry, installations where the antenna can’t have a clear overhead view.
A proper gps antenna installation accounts for ground plane size explicitly. If your antenna form factor doesn’t include an integral ground plane, the mounting surface becomes the ground plane, and its dimensions matter.
Obstructions That Look Harmless
The instinct when placing a GPS antenna is to find a location with a clear view of the sky, and to assume that anything that doesn’t directly block the sky is fine. This is wrong in a few ways.
Obstructions at low elevation angles — below 15 to 20 degrees above the horizon — matter more than people realize. Modern GPS receivers use satellites at these low elevation angles to improve geometry, particularly in applications where high satellite counts improve accuracy or reliability. An obstruction that cuts off the lower portion of the sky reduces the available satellite constellation, degrades the dilution of precision (DOP), and increases sensitivity to satellite outages.
Metallic obstructions adjacent to the antenna — not above it, but beside it — act as scattering surfaces. A metal bracket, a vehicle antenna, or even an electronics enclosure within 10 to 15 centimeters can scatter incoming signals in ways that perturb the antenna’s phase center and introduce heading or position errors. The errors are small enough to be invisible in coarse applications but meaningful in precision positioning.
Cable routing near the antenna is another underappreciated source of interference. A coaxial cable running parallel and adjacent to the GPS antenna feedline for any significant length can act as a coupling path for interference from nearby electronics into the antenna circuit. The cable should route away from the antenna horizontally before dropping, not run parallel to the antenna element.
The Phase Center and Why It Moves
Every GPS antenna has a phase center — the effective electrical origin from which ranging measurements are made. It’s not a physical point on the antenna; it’s a derived quantity that depends on the direction the signal is arriving from. For most antennas, the phase center varies by a millimeter or less across the elevation angles that matter — negligible for most applications.
For precision applications — survey-grade positioning, precision agriculture guidance, autonomous vehicle localization — phase center variation becomes relevant. An antenna that has 3mm of phase center variation across the hemisphere introduces up to 3mm of systematic ranging error depending on satellite geometry. Over a full solution with multiple satellites, this doesn’t completely cancel. The result is a small but consistent position bias that changes as satellite geometry changes throughout the day.
High-performance GNSS antennas for precision applications specify phase center variation explicitly and provide phase center offset tables for post-processing. For these applications, the antenna model matters not just for its gain pattern and multipath rejection but for the phase center stability it provides.
Practical Placement Checks
Before finalizing an antenna installation, a few checks catch most placement problems early.
Look for metallic surfaces within about 15cm of the antenna on any side. If you can’t avoid them, consider whether they’re symmetric — symmetric surrounding metal changes the gain pattern uniformly and may be manageable, while asymmetric metal introduces heading-dependent errors.
Check the ground plane. If the antenna doesn’t have an integral ground plane and is mounted on a surface smaller than roughly 20cm across, expect the elevation pattern to be degraded. A larger mounting plate is usually low-cost insurance.
Survey the obstruction mask. Stand at the antenna location and note anything that breaks the skyline above 10 degrees elevation. Even a single obstruction in a commonly occupied satellite track can create consistent errors at certain times of day.
Run the receiver’s sky plot and SNR display after installation. Signal strengths from satellites at similar elevations should be roughly similar — significant variations between adjacent satellites at similar elevation angles suggests multipath or obstruction effects on specific signal paths. This is the fastest way to verify that a placement is actually working as intended before deployment.
The signals are arriving correctly. The question is whether your antenna placement is giving them a fair chance to be received that way.
