Things used in this project

Hardware components

SO-239

×1

brass tube 4mm diameter

×1

brass tube 3mm diameter

×1

aluminium tube 4 mm diameter

×1

Hand tools and fabrication machines

ender 3 v2

Story

This project presents a 12-element Yagi antenna for the PMR446 band (446 MHz), built on a 2 m aluminum boom with 3D-printed element holders. The geometry is based on the Rothammel / DL6WU model, calculated with the Changpuak online tool.

I designed a 1:1 balun, fully compatible with a simple driven dipole. The balun is built around an SO-239 connector embedded in a 3D-printed housing that secures the coax cables from both dipole arms, protects the connector from weather exposure, and attaches to the boom using zip ties.

The antenna provides a estimated gain of about 12 dBi, a very good impedance match, and a lightweight, modular design suitable for portable or fixed installations.

Antenna presented for experimentation / reception; check your local regulations before using with PMR446 transmitters. This project is presented for educational and reception purposes.

3D printed holder

Here is a view of the three parts to be 3D printed in PLA (it would also work with ABS or PC, but the antenna shown in the photo above has been installed outdoors for two years, is still working, and has survived two summers).

The AMF files are attached in the files projet.

From left to right: passive element holder, driven element holder, and the “case” used to secure and protect the SO239 connector.

You can see the recesses for M4 countersunk screws and the matching hex slots for the nuts, which allow precise and secure positioning of the holders on the boom.

On the “case, “ you can see the two openings for cable ties, used to attach the enclosure to the boom, between the driven element and the reflector.

1:1 balun and driven elements

Measurement and Tuning of the Balun

The photo above shows the setup for feeding the radiating elements via the SO239 connector.The system uses two lengths of RG58 coaxial cable, measuring λ/4 and 3λ/4, respectively, to achieve both:

• impedance matching close to 50 Ω,
• and out-of-phase feeding of the two dipole arms.

This arrangement is a quarter-wave balun (“balanced to unbalanced”), which provides two key functions:• It transforms the unbalanced signal from the coax into a balanced signal suitable for the dipole,• and suppresses common-mode currents that could distort the radiation pattern, improving VSWR stability over time.

Quarter-wave Split Balun Case (two λ/4 lines, one per dipole arm)

Each dipole arm is fed by its own (n*)λ/4 line, grounded via the coaxial cable shield.The impedance “seen” by each line is thus half that of the full dipole:

Z L = Z dipole / 2

Discussion on Dipole Impedance

• A free-space half-wave dipole has a characteristic impedance of 73 Ω (theoretical value, ref. ARRL/RSGB).
• However: in a Yagi-Uda, proximity to directors and reflector (“accompanied dipole”) lowers this impedance, typically to around 60 Ω (even down to 50–55 Ω with strong coupling), depending on geometry.
• This reduced impedance is what is used in all real-world matching calculations.

Example for Zdipole = 60 Ω (common for a well-designed 12-element Yagi, Rothammel/DL6WU):

Z L = 60 Ω / 2 = 30 Ω

Each λ/4 line transforms this load as follows:

Z in arm = (Z0^2) / Z L = (50^2) / 30 ≈ 83 Ω

The two λ/4 lines (one per arm) are brought together in parallel at the feedpoint (coax hot pin):

Z in total = 1 / (1/Z in arm + 1/Z in arm) = 83 / 2 ≈ 41.5 Ω

In practice, the dipole’s impedance depends on coupling: The closer the dipole is to reflector/directors, the lower Z_L gets (sometimes as low as 45–50 Ω). With stronger coupling, the combined input impedance comes close to 50 Ω, which is ideal for RG58.

Thus, this setup typically achieves a VSWR of 1.1–1.2: a robust and experimentally validated compromise.

Key takeaways:

• Never trim the coax lengths after initial accurate cutting.
• Fine-tune only the dipole length (±0.5 to 1 mm) to optimize the match.
• Always leave 1–2 cm of slack at the connector end for proper soldering, then trim to precise length.

Key takeaways:Never trim the coax lengths after initial accurate cutting.Fine-tune only the dipole length (±0.5 to 1 mm) to optimize the match.Always leave 1–2 cm of slack at the connector end for proper soldering, then trim to precise length.

Feeding in Opposite Phase: λ/4 and 3λ/4 Sections

In this “split balun” design, the two arms of the dipole are each fed by a separate length of coax: one λ/4 and the other 3λ/4.The λ/4 and 3λ/4 sections are wired in such a way that a signal from the transmitter reaches one dipole arm one quarter-wavelength ahead in phase compared to the other arm, and because a λ/2 line adds 180° of phase but does not affect impedance, the total phase shift between the two arms is 180°.This means that when the RF signal reaches the feedpoint, the two arms are driven in perfect opposition of phase, exactly as required for a balanced dipole.This out-of-phase feed is what produces the desired dipole radiation pattern and ensures efficient operation.

The λ/4 and 3λ/4 configuration is a robust and simple way to achieve both:

• A balanced, symmetrical current distribution,
• And the correct phase relationship, using only standard lengths of coax and without introducing additional reactive components.

Note: The impedance matching and symmetry discussed above only remain optimal if both cable sections are cut to precise electrical lengths (within ±2 mm), and the dipole is tuned close to the target impedance.

Length Calculations

The electrical wavelength depends directly on the propagation velocity in RG58, i.e., the velocity factor (VF).At 446 MHz, the free-space wavelength is:

lambda0 = c / f = (3 * 10^8) / (446 * 10^6) ≈ 0.673 m

Propagation speed in RG58 is about 0.66 × c (per datasheet and my own measurements). Thus, the cable wavelength is:

lambda_RG58 = 0.673 * 0.66 ≈ 0.444 m

Physical lengths for the two coaxial sections:

• λ/4 ≈ 0.111 m (11.1 cm)
• 3λ/4 ≈ 0.333 m (33.3 cm)

These lengths are measured from the dipole solder joint to the SO239 connector.A 3 mm error already represents a phase shift of nearly 5° at 446 MHz....

Experimental Measurement of Velocity

I did not keep the exact measurements for the final cables, but I characterized wave velocity for two different brands of RG58.The experimental protocol is straightforward:

• The function generator (GBF) outputs a square pulse of a few volts peak-to-peak, low duty cycle (to isolate the rising edge).
• The generator output is connected to an oscilloscope, and via a T-connector, to a cable under test with its far end either shorted or left open. These two conditions represent current or voltage continuity at the cable end.
• The cable should be long enough for the reflected pulse delay to be clearly measurable.

Velocity is given by:

v = 2L / Δt

where L is cable length and Δt the measured time between initial pulse and reflection.In both cases, I measured about 0.66 × c, perfectly matching theoretical values for RG58.This value can thus be used as a standard reference for balun electrical length calculations, regardless of cable brand, as long as RG58 is used.

Impedance Match Verification

Impedance match was checked, as with my previous 3×5/8 λ project, using a SWR meter and a nanoVNA.Unfortunately, I did not keep these results; I am sharing these plans in response to a request but cannot repeat the measurements at this time.

If you build the antenna and have access to a network analyzer (SWR meter, nanoVNA…), please share your results: I will add them to the project log.For the record, during initial testing, performance was excellent: VSWR around 1.1 and resonance centered near 446.5 MHz, confirming both the model and the 1:1 balun approach.

Dimensions of the passive elements

For the dimensions of the passive elements, I used the online Yagi calculator available here:

🔗 https://www.changpuak.ch/electronics/yagi_uda_antenna.php

Frequency : 446 MHz (useful range: 437.08 – 454.92 MHz)Wavelength : 673 mmRod Diameter : 4 mmBoom Diameter : 20 mmBoom Length : 1952 mmd/λ : 0.006D/λ : 0.030Elements : 12Estimated Gain: 12.75 dBd (≈ 14.9 dBi)———————————————————––Reflector Length : 324 mmReflector Position : 0 mmDipole Position : 161 mm———————————————————––Director #1 Position : 212 mm, Length : 306 mmDirector #2 Position : 333 mm, Length : 303 mmDirector #3 Position : 478 mm, Length : 300 mmDirector #4 Position : 646 mm, Length : 298 mmDirector #5 Position : 834 mm, Length : 296 mmDirector #6 Position : 1036 mm, Length : 294 mmDirector #7 Position : 1248 mm, Length : 292 mmDirector #8 Position : 1470 mm, Length : 290 mmDirector #9 Position : 1702 mm, Length : 289 mmDirector #10 Position : 1944 mm, Length : 287 mm———————————————————––Spacing between directors gradually increases(from 50 mm up to 242 mm).All parasitic elements are isolated.Use insulators thicker than 10 mm.

Dimensions of the driven elements

This length must be fine-tuned during adjustment. That said, assuming a propagation velocity of 0.975 inside the tube and a free-space wavelength of λ = 0.673m, the effective wavelength becomes λ_eff = 0.975 × 0.673 m. We aim for λ/2, so the total length of the driven element should be half of this effective wavelength, split into two equal arms, with a 5 - 10 mm gap in the center. You’ll probably need to trim both arms symmetrically to achieve the best match. A straight dipole usually has a narrower bandwidth than a folded one. But in the case of PMR446, the bandwidth is very small (around 200 kHz), so that’s not a drawback. In fact, having a narrow-band antenna can be beneficial: it acts as an initial RF selector, making the receiver’s front-end cleaner and less noisy. By the way, balun 1:1 work with straight dipole only !

Materials

Passive elements: made from aluminum TIG welding rods.

• Passive elements: made from aluminum TIG welding rods.
• 3D-printed parts: all element holders are printed in standard PLA ; The antenna has been installed outdoors for two years and has survived two summers in full sunlight without any degradation or structural issue

Driven elements: made from brass tubing, 4 mm outer diameter.

Can be paired with solid 3 mm brass rods that slide snugly inside the 4 mm tubes.

• Can be paired with solid 3 mm brass rods that slide snugly inside the 4 mm tubes.

Length is locked by gently pinching both tubes together with pliers.

• Length is locked by gently pinching both tubes together with pliers.
• Driven elements: made from brass tubing, 4 mm outer diameter.Can be paired with solid 3 mm brass rods that slide snugly inside the 4 mm tubes.Length is locked by gently pinching both tubes together with pliers.

Boom: 20 mm square aluminum tube.

• Boom: 20 mm square aluminum tube.

Coaxial feed: RG58 cable between the SO-239 connector and the driven elements.Cable length must not be changed, as it depends on the signal velocity factor inside the coax and directly impacts the balun tuning.

• Coaxial feed: RG58 cable between the SO-239 connector and the driven elements.Cable length must not be changed, as it depends on the signal velocity factor inside the coax and directly impacts the balun tuning.

Electrical connectors: small electrical terminal blocks (dominos in France) are embedded inside the 3D-printed

• Electrical connectors: small electrical terminal blocks (dominos in France) are embedded inside the 3D-printed

Connector: SO-239, sourced from AliExpress.

Transmission through SO-239 is acceptable (not ideal) for UHF.

• Transmission through SO-239 is acceptable (not ideal) for UHF.

Chosen mainly for its mechanical convenience, allowing easy soldering of the two driven element wires and solid attachment to the 3D-printed housing.

• Chosen mainly for its mechanical convenience, allowing easy soldering of the two driven element wires and solid attachment to the 3D-printed housing.
• Connector: SO-239, sourced from AliExpress.Transmission through SO-239 is acceptable (not ideal) for UHF.Chosen mainly for its mechanical convenience, allowing easy soldering of the two driven element wires and solid attachment to the 3D-printed housing.

**Read more

Custom parts and enclosures

Credits