Long Range WiFi Test: Directional Antennas, the Gain Story, and a Homebuilt Yagi

It all starts with patchy wireless coverage at home. The host runs a small antenna on his router instead of a bulky one, and recalls an earlier project where he measured everyday antenna designs side by side. That project left one gap: directional units, the chunky parts often marketed as Yagi antennas, which squeeze the signal into a single beam rather than spraying it evenly.

The rig is built from two ESP32 boards: one acts as the sender beneath a shelter, the other as the receiver inside a garden shed some 20 meters away. A portable power bank feeds the sender, which emits wireless packets at steady intervals. The receiving side stays tethered to a computer and prints an RSSI reading, the received-signal-strength indicator, for every packet that lands.

With the best everyday antennas from the earlier round fitted at both ends, a link comes up but the numbers disappoint: RSSI sits around minus 65 dBm. Then three commercial directional antennas sent by electronics supplier Mouser come out of their boxes: a compact fin, a large fin, and a long pole-shaped model. Swapping the ordinary sender antenna for the small fin lifts RSSI to roughly minus 50 dBm, a jump that means the captured signal power has grown more than thirtyfold.

The large fin and the pole both beat the small fin by a wide margin, though which unit posted which score is held back as a teaser that sends curious viewers to the datasheets. This is where gain enters the picture, and the host admits the word misled even him: its transistor-era echo suggests an antenna can enlarge a signal. In reality an antenna enlarges nothing; it only manages how the available power spreads through space.

The yardstick is a fictional isotropic radiator that shines equally in all directions, defined as gain one, or zero dBi. As the signal gets packed into one heading, the figure climbs: the pole lists 14.5 dBi on its sheet, the big fin 6.8 dBi. When the pole tops the RSSI ranking in the test, that gap earns its confirmation, and the radiation plots in the sheets tell the same story, with the pole drawing a slim beam while the big fin washes a broader zone.

The real surprise waits under the covers: the two fin-shaped products are not Yagis in any strict sense. Their stepped, fir-tree silhouettes mark them as log-periodic dipole arrays, LPDAs for short. Element lengths and spacings taper down from large to small, so on any frequency only a small cluster near the resonant size carries most of the load, while neighboring elements combine their waves toward the small end and partly cancel backward, which forms the beam.

Inside the pole unit sits a genuinely radiating dipole at the base plus a long forward structure with no wired link to it: this parasitic piece picks up the dipole field, reradiates, and its waves pile up along the facing direction, which is why it earns the name director. A textbook Yagi would also carry a bulkier reflector behind the driven element to choke rearward spill and reinforce the forward lobe, and that part is absent here. A legal caveat lands in between: in Germany the 2.4 GHz ceiling is 100 mW EIRP, or 20 dBm, and a high-gain antenna can punch through it with ease, so the host deliberately throttled transmit power during the runs.

For the reasoning behind spacings and geometry, a Yagi design report from the US National Bureau of Standards becomes the blueprint for a homebrew attempt: one dipole, three directors and one reflector, all riding on an insulating frame. The frame comes off a 3D printer, the radiating pieces are snipped from 1 mm brass rod and fastened with nylon nuts and bolts. The fiddliest chore is soldering on the coaxial feed, yet the finished piece looks respectable at first glance.

A quick check on a vector network analyzer shows matching well short of ideal, but the verdict that counts comes from the field. Hooked to the same rig, the homebrew antenna lands at minus 52 dBm: behind the commercial units, clearly ahead of the best ordinary whip. Turned to face the opposite way it sinks to about minus 60, which doubles as the plainest proof that the beam truly points.

For a radiation plot like the commercial sheets carry, simulation gives way to a motorized rig called NanoFarField, soon headed for crowdfunding: the sender stays put while the homebrew unit spins a full circle and the bundled software logs everything on its own. On 2442 MHz, the channel the ESP32 link uses, a distinct beam shows up, rougher than the commercial curves but unmistakably directional. A handful of brass rods, one printed bracket and decades-old antenna literature turn out to be enough for a working WiFi beam.