Antenna Gain and Coverage Tradeoffs
Antenna Gain and Coverage Tradeoffs
Antenna gain is not free —- it is always traded against something else. Understanding what gain costs you is essential before choosing an antenna for a mesh deployment. The fundamental law of antenna physics is conservation of energy: an antenna cannot create power, only redistribute it.
How Gain Concentrates Signal
Consider a theoretical isotropic antenna radiating 1 watt equally in all directions. At 1 km, that power is spread over a sphere of area 4π(1000)² = 12.57 million square meters. A 5 dBi antenna (3.16× linear gain) compresses its radiation into a narrower cone, delivering 3.16× more power density in its peak direction. From the perspective of a receiver in the main beam, it is equivalent to the transmitter having 3.16× the power.
This is the core of EIRP (Effective Isotropic Radiated Power):
EIRP (dBm) = Transmit Power (dBm) + Antenna Gain (dBi) − Feedline Loss (dB)FCC Part 15.247 limits EIRP to +30 dBm (1 watt) for spread spectrum systems in the 902–902 - 928 MHz band when operating with a fixed, point-to-point link with directional antennas. For point-to-multipoint operation, the limit is effectively lower. Most LoRa nodes run 17–17 - 20 dBm transmit power, leaving 10–10 - 13 dB of "antenna budget" before hitting the legal limit.
Elevation Angle and Radiation Pattern Compression
As gain increases, the radiation pattern in the vertical plane becomes flatter —- more like a pancake and less like a donut. This is measured as the vertical beamwidth (the angle between the −3 dB points above and below the horizon).
| Antenna Gain | Approx. Vertical Beamwidth | Radiation Elevation Angle |
|---|---|---|
| 2 dBi (dipole) | ~75° | Broad; works at steep angles |
| 5 dBi collinear | ~ | Slightly elevated; works for nearby nodes |
| 8 dBi collinear | ~ | Near-horizontal; close nodes may be in null |
| 10 dBi collinear | ~ | Essentially horizontal; nodes must be far away to be in the beam |
Dead Zones Below High-Gain Antennas
This is the most commonly overlooked problem with high-gain omnidirectional antennas in mesh networks. When you mount a 10 dBi collinear antenna on a rooftop, the signal goes predominantly outward —- not down. Nodes directly beneath the tower, or on the same city block, may receive weaker signal than nodes kilometers away.
The approximate dead zone radius under a vertical omni antenna can be estimated as:
Dead Zone Radius ≈ h / tan(θ/2)
Where:
h = antenna height above nodes (meters)
θ = vertical beamwidth (degrees)
Example: 10 dBi antenna at 30 m height, 10° vertical beamwidth:
Dead Zone Radius ≈ 30 / tan(5°) ≈ 30 / 0.0875 ≈ 343 metersIn this example, any node within 343 meters of the tower base would be in the side lobe or null region and might receive 10–10 - 20 dB less signal than a node 2 km away. In a dense urban mesh, this is disastrous.
The 3 / 5 / 8 dBi Decision Guide
Use this framework when selecting omni antenna gain for a fixed node:
| Gain Choice | Use When | Avoid When |
|---|---|---|
(whip, dipole, GP vertical) |
Indoor node; node surrounded by other nodes at similar elevation; portable device; building where nodes are on every floor | Outdoor exposed relay where range to distant nodes is the primary goal |
| 5 dBi (short collinear) |
Outdoor rooftop node in urban/suburban area; nodes are within |
Indoor use; terrain with significant elevation variation around the node |
| 8 dBi (medium collinear) |
High hilltop or tower relay overlooking flat terrain; all served nodes are at roughly the same elevation and |
Urban environment; any situation with nodes at varying elevations; anywhere nodes might be directly below the antenna |
Rule of thumb: When in doubt, choose 5 dBi for any outdoor fixed node. It provides meaningful gain improvement over a whip without creating serious dead zone problems. Reserve 8+ dBi for well-planned backbone relay sites with known terrain profiles.
Directional antennas: When gain beyond 8 dBi is needed, switch to a directional antenna (panel or Yagi) aimed at the intended coverage direction. You gain range in the beam, and the dead zone problem is inherent to the design intent —- it only covers one sector anyway.