Building your own
2 m/70 cm Yagi antenna

The one on every roof

Everyone knows it from rooftops: a row of rods on a boom, the longest at the back, getting shorter towards the front. For half a century that’s what television antennas looked like, and that’s what most VHF/UHF directional antennas look like today, from amateur to early radars.

The idea is a hundred years old. From 1924 at the Imperial Tohoku University in Sendai, Hidetsugu Yagi and his assistant Shintaro Uda developed an antenna in which only one element is driven. The others, though not connected to the cable at all, operate from the neighbour’s field. Yagi patented it in 1926.

For a radio amateur it’s a dream homebrew project: there are no parts to buy from an electronics shop, apart from the socket, cable and ferrites. What you do have are dimensions — and that’s what everything depends on. I described the antenna’s history with a wartime anecdote on the SQ4MN dual‑bander.

Aluminium, not steel

Rods. The reflector for 2 m is over a metre long, and at Castorama the Ø 4, 6 and 8 mm rods are only 1 metre (9.98 PLN for Ø 6, 16.98 PLN for Ø 8). OBI has 2‑metre Ø 6 mm rods (GAH Alberts, 18.49 PLN), and wholesalers of aluminium and online shops sell Ø 6 and Ø 8 mm in 2 m lengths (about 13–14 PLN). For 70 cm, 1‑metre rods are enough: you get three elements out of one.

Boom. 20×20 mm square tube (for 70 cm 15×15 will do) in aluminium in 2 m lengths. Prices vary: 20×20 at OBI about 30 PLN, at Castorama 94 PLN for a thick‑walled 20×20×1.5 mm tube. Alternative without correction: an impregnated and varnished wooden batten or a plastic tube.

Fasteners. Stainless A2 or A4 screws, preferably with nylon spacers and washers. Stainless steel in aluminium is a galvanic couple: the aluminium around the screw can corrode, and water running off copper is particularly harmful to aluminium. Do not make elements from steel or bare copper.

Feed. Small IP65 sealed plastic box (the socket should end up right next to the driven element terminals), panel‑mount N socket (on 70 cm an N is more dependable than UHF/SO‑239), low‑loss cable to the radio (H‑155 for short runs, thicker for longer; RG‑58 only on a handheld) with a dense copper braid, including tinned copper, not copper‑clad aluminium (marked CCA), which solders poorly and quickly loses contact in the connector, 2–4 ferrite cores on the cable and neutral silicone, not acetoxy.

Where do these millimetres come from

The correction was measured by DL6WU, and G3SEK expressed his chart as a formula, which today’s Yagi calculators use. What matters is the boom thickness relative to wavelength, so on 70 cm the same boom gives more than twice as much correction as on 2 m. Add the correction to the entire element length, half on each side.

Sources differ by 1–2 mm (square counted as a circle or as a circle 1.18 times thicker, wall thickness, distance from the boom end). On 2 m it’s irrelevant, as the design tolerates ±5 mm. On 70 cm the correction is larger than the antenna’s entire tolerance, so there I recommend a wood or plastic boom, or checking the antenna with an analyser after assembly.

The driven element is always insulated, as its halves go to the box. Authors don’t agree here, so the safest route is to do as DK7ZB does: leave 2 mm spare on each side and trim after measurement.

Why ferrites are mandatory here

In the designs on this page the split driven element goes straight to 50‑Ω cable. The dipole is balanced and the coax isn’t. Without a choke some current returns on the outside of the braid and the cable starts to radiate as an extra element: the beam skews, the backlobe grows, and the SWR changes when you move the cable or change its length.

Ferrite cores fitted as close to the driven element as possible choke this current. Current on the braid starts to flow at the feedpoint, so every centimetre of cable before the choke can radiate, and on 70 cm 17 cm is already a quarter‑wave. Fair‑Rite specifies mix 43 for attenuation from 20–250 MHz, and 61 above 200 MHz, so on 2 m use 43, on 70 cm 61. Jim Brown K9YC advises aiming for 500–1000 Ω impedance for current on the braid. A mix 43 Fair‑Rite 2643625202 core for cable up to 7.9 mm shows 361 Ω at 100 MHz and 415 Ω at 250 MHz per the catalogue, so on 2 m two are enough, three for margin.

Choke test: connect an analyser and grab the cable by hand a few tens of centimetres from the box. If the SWR twitches, current is flowing on the braid. Add another core.

Short leads inside the box

The leads between the socket and the driven element terminals are not ‘a little wire’, they’re part of the antenna. Two wires 1 cm apart behave like a line of about 300 Ω and add inductance that shortening the driven element will no longer compensate. I ran the numbers in NEC‑2: on 70 cm as little as 2 cm of wire gives an SWR around 3, and by 8 cm the antenna stops working.

Limit: unshielded ends on 70 cm no more than a few millimetres, on 2 m about 1 cm. That’s why the box should be small and the socket right by the terminals.

Bigger box? Don’t run loose wires inside; bring the coax right to the terminals instead: the antenna cable through a gland, or a short length of coax (e.g. RG‑316) from the socket. A 50‑ohm section doesn’t spoil the match — only the unshielded end matters. You then fit the ferrites on that coax right by the terminals, because the current on the braid starts at the dipole, not at the socket.

SQ4MN variantMarian SQ4MN · this is how he feeds his antennas

Folded dipole and λ/2 loop

Marian SQ4MN feeds his Yagis differently from the designs on this page. The driven element is a folded dipole designed for about 200 Ω, and the 50‑Ω cable is connected via a balun made from a half‑wave loop, i.e. a 4:1 transformer.

The midpoint of the continuous leg of the dipole is at zero potential, so it is connected to the boom. In the same place the braids of the loop and the feed cable are tied together. The loop forces equal and opposite voltages on the two halves of the dipole, and the braid is joined where the voltage is zero, so hardly any current flows on the outside of the cable. With this feed a choke is not necessary. You can add ferrite as cheap insurance if the cable runs close to the elements.

Here length really matters — but the length of the loop, not the cable to the radio: λ/2 × velocity factor, with VF 0.66 about 682 mm on 145 MHz and 227 mm on 435 MHz. Check a loop cut to the catalogue VF with an analyser, and solder on coax with a copper braid. You cannot transfer the dimensions from the tables above straight to this variant, because a folded dipole needs different spacings than a split 50‑Ω driven element.

How to measure so the result means something

Antenna skywards. Set the boom vertical, directors pointing up, at a height of at least two metres and away from metal. VK2KU measured all antennas like this, and DG7YBN advises there should be nothing in the beam within the nearest 20 metres. Measuring in a room or on a balcony by a wall shifts the resonance.

Cable in a loop. The cable should hang in a wide curve, not run along the antenna. Calibrate the analyser at the end of the cable, or remember that each decibel of cable loss inflates the measured return loss (RL) by two decibels (Owen Duffy).

What to use. NanoVNA‑H measures from 50 kHz to 900 MHz. Above 300 MHz it works on harmonics with lower dynamic range, but that’s perfectly sufficient for SWR on 70 cm. Newer NanoVNA‑H4 units reach 1.5 GHz. An SWR meter for 144/430 MHz and a radio are also enough, as long as you use several frequencies.

What to expect: a curve similar to the one in the design plot. If it’s the same shape, just shifted, it means the elements came out too long or too short. The table below suggests what to do.

Mast and a vertical antenna

Between stations with different polarisations you lose 20 dB or more according to the ARRL. That’s a hundred times less power – more than the whole antenna gives you.

With vertical polarisation there’s a second trap: a metal mast running through the middle of the antenna is parallel to the elements and becomes one of them. I ran the numbers for the 2 m design with a 40 mm diameter metal mast: without the mast 10.7 dBi and front–to–back 22 dB; with the mast through the centre, depending on its length, from 10.4 down to just 7.8 dBi, and front–to–back drops even to 8 dB. With the mast 35 cm behind the reflector: 10.7 dBi and 23 dB, i.e. no change.

Solutions: mount behind the reflector (boom extended to the rear; on 2 m the boom comes out about 2.05 m or use a short extension), a side standoff, or a section of non‑metallic mast within the antenna. With horizontal polarisation the mast is perpendicular to the elements and causes little trouble. This is how the Radmor antenna from Gdynia in the photo is mounted: the boom is extended behind the reflector and its end bolted to the mast.

“Most important: antennas like free space. When I kept the antenna clear of its surroundings, it really came to life.”Marian SQ4MN

I checked in NEC‑2 how far “far” is. I added metal parallel to the elements to the models on this page, because that’s what interferes most: a single 20 mm diameter tube about 3 λ long (gutter, another antenna’s mast, post) or a 2 × 2 λ metal wall made of rods spaced every 1/10 of a wavelength (sheet metal, mesh, railing, rebar). The result depends on which side the obstacle is on.

Front, i.e. the main lobe. Close in, SWR rises: a tube half a metre in front of the last director of the 2 m antenna gives SWR 1.9, and 20 cm in front of the 70 cm antenna it’s 2.0. On 2 m the SWR returns to normal from about ½ λ; on 70 cm only about 2 λ further away. Beyond that, metal in the beam still steals gain and worsens the back: a tube 20 m in front of the 2 m antenna is still −0.3 dB and F/B 19 instead of 22 dB. There should be nothing in the beam as far as you can manage.

Side, i.e. a wall along the antenna. This is where you lose the most, and an analyser won’t show it because SWR hardly changes. At a distance of ¼ λ the antenna loses over 4 dB, i.e. half of what it gives over a dipole. At ½ λ still 1–2 dB. From about 1 λ the on‑axis gain only wiggles by ±1 dB, because the reflection from the wall sometimes adds, sometimes subtracts. With horizontal polarisation a sheet‑metal roof under the antenna is also such a “side”.

Rear, i.e. behind the reflector. The reflector shields the antenna. A wall even ¼ λ behind it doesn’t change SWR, and the gain moves by at most ±1 dB. If you have to bring the antenna close to a building, do it at the back.

On a balcony point the antenna away from the wall, outwards, not along the facade. A railing with metal bars parallel to the elements acts like the wall in the table. Metal in the model is the worst case: brickwork and wood reflect less, but some energy is lost in them.

Two bands on one boom

The 2 m elements lie in one plane and the 70 cm in a plane at right angles, all centred on the boom. The field of one band then has no component along the rods of the other, so it doesn’t induce current in them: in the NEC‑2 model the presence of the other antenna doesn’t change gain or SWR. In a real antenna that symmetry is upset by the cables and imperfect assembly, so run the cables along the boom and measure each band separately. You change polarisation with a twist of the wrist, which is an advantage with satellites because their signal changes polarisation anyway during the pass.

Two cables or a duplexer. The IC‑9700 and other radios with separate sockets for 2 m and 70 cm are connected with two cables. For a handheld radio with a single socket you need a duplexer, e.g. Diamond MX‑72N (paths 1.6–30/49–150 and 400–460 MHz, isolation at least 60 dB, losses up to 0.15/0.25 dB, up to 150/100 W) or Comet CF‑4160 (1.3–170 and 350–540 MHz). The manufacturer intends the MX‑72N for indoor and in‑car use.

Through the ISS repeater and the SO‑50 satellite you transmit on 2 m and receive on 70 cm. The satellite passes calculator shows when they pass over Poland, and I’ve described how to work them in the guide on satellite communications.

Marian SQ4MN’s Yagi

Marian SQ4MN builds antennas himself and puts them up on his balcony in Warsaw. His 70 cm has an extra support under the boom, and from the elements left over from two antennas he also put together a dual‑band antenna with a copper folded dipole.

It’s good proof that a Yagi doesn’t need a workshop: rods, a boom and patience with measuring are enough. I’ve described both builds on separate pages: 70 cm Yagi with a second support and antenna for 2 m and 70 cm.