EFHW Transformers: 64:1 vs 49:1, Turns, Cores and Tuning

For most end-fed half-wave antennas, start with a 49:1 transformer (a 7:1 turns ratio). It matches a 2,450 Ω wire end to 50 Ω. Pick 64:1 (8:1 turns) when your own measurement shows the wire end sitting nearer 3,200 Ω. Operators disagree about which is closer to a real EFHW, and the measured range runs from about 1,800 to 5,000 Ω, so both ratios land inside it. In my view, core choice and winding quality matter more than the ratio.

Below is the math for both ratios, how to wind either one, and how to measure which one your wire wants. For the whole antenna, start with the end-fed half-wave build guide.

What each ratio does, and when to choose it

An impedance transformer multiplies impedance by the square of its turns ratio. So the turns ratio is the square root of the impedance ratio:

• 49:1: √49 = 7, so 7 turns on the antenna side for every 1 turn on the coax side. 50 Ω × 49 = 2,450 Ω.
• 64:1: √64 = 8, so an 8:1 turns ratio. 50 Ω × 64 = 3,200 Ω.

On an EFHW the transformer is an autotransformer: one tapped winding. The ratio is total turns divided by primary turns. A 3-turn primary that continues to 21 turns in total gives 21 / 3 = 7, so 7² = 49:1. Taken to 24 turns, it gives 24 / 3 = 8, so 8² = 64:1. The notation 2:14 means the same 7:1 as 3:21.

Ratios between those two also get used. A 3:23 winding is 23 / 3 = 7.67, so 7.67² ≈ 58.8:1. One commercial 80-10 m EFHW transformer is sold at about 56:1 (about 2,800 Ω to 50 Ω), which is √56 ≈ 7.5 turns per turn.

Operators disagree on which ratio is closer to a real EFHW:

• The case for 49:1. It is the most common published design, and it fits the middle of the 1,800-5,000 Ω range. Wound 2:14 on a single FT240-43, a 49:1 tends to give its best match around 20 m.
• The case for 64:1. Modeling an EFHW across different counterpoise lengths shows that the 8:1 turns ratio gives the most stable SWR. Some builds move from 7:1 to 8:1 because 49:1 leaves the impedance slightly too high. Even then a three-band EFHW is a compromise: high on 10 m, right on 20 m, low on 40 m.
• The measured middle. A fixed-tuned 64:1 coupler set up on a 3,200 Ω bench load read about 35 Ω on a 40 m EFHW. That works back to 35 × 64 = 2,240 Ω at the wire. A 49:1 would have shown that same wire as 2,240 / 49 = 45.7 Ω.

My rule: 49:1 for a first build and for the usual sloper or inverted-L at modest height. Try 64:1 if your bench-tested 49:1 reads consistently above 50 Ω on the fundamental with no reactance. That means the wire end is above 2,450 Ω. Neither ratio belongs on a random-length wire. That job and its 9:1 cousin are covered in how a random wire differs from a half-wave.

Dimensions with the math shown

Wire length

I cut from the dipole formula because the EFHW is electrically a half-wave:

• Length (ft) = 468 / f (MHz)
• Length (m) = 142.5 / f (MHz)

For 7.1 MHz: 468 / 7.1 = 65.9 ft, and 142.5 / 7.1 = 20.07 m. For an 80-10 m wire at 3.55 MHz: 468 / 3.55 = 131.8 ft, or 142.5 / 3.55 = 40.1 m.

Why this formula and not the free-space half-wave (491.8 / f ft, or 69.3 ft at 7.1 MHz)? A real wire resonates a little short of a free-space half-wave. Modeling an EFHW puts resonance at about 0.47 free-space wavelengths, which is 462 / f ft. 468 / f lands within about 1.3 % of that. Insulation, the counterpoise and the transformer itself all move resonance, so I cut the wire longer than the formula. Start about 4 ft over (roughly 70 ft for 40 m) and trim down. Folding the tail back on itself shortens the wire without cutting it. The pillar guide's 66 ft starting point is the same formula at 7.15 MHz (468 / 7.15 = 65.5 ft).

Primary turns and the lowest band

The primary must show the 50 Ω source a reactance several times higher than 50 Ω on the lowest band. One published rule of thumb is 5 to 10 times, so at least 250 Ω. Inductance is L = AL × N². The FT240-43 has an AL of 1,075 nH per turn², so its reactance is X = 2πfL:

Primary turnsL on one FT240-43X at 3.5 MHzX at 7 MHzClears 250 Ω from 21.075 × 4 = 4.3 µH95 Ω189 Ωabout 9.3 MHz (30 m up) 31.075 × 9 = 9.7 µH213 Ω426 Ωabout 4.1 MHz (40 m up, close on 80 m) 41.075 × 16 = 17.2 µH378 Ω756 Ωabout 2.3 MHz (80 m up)

The table uses the datasheet AL. Wound cores often measure a little higher, near 1.2 µH per turn², which is why 3 turns on one FT240-43 is commonly used down to 80 m and why 160 m designs go to 4 turns or stacked cores. More primary turns raise winding capacitance and leakage inductance, though, and that costs you on the high bands. On one FT240-43 at 49:1, a 2:14 winding has its leakage-limited upper corner near 24.9 MHz, while 3:21 drops to about 14.9 MHz. The compensation capacitor (below) exists to push that limit back up.

Materials

PartWhat to useWhy Core, 100 W classFT240-43 (Fair-Rite 5943003801): 61 mm OD, 35.55 mm ID, 12.7 mm tallAL 1,075 nH, 1.58 cm² cross-section, µi 800 Core, QRP or light dutyFT140-43 (5943002701): 35.55 mm OD, 23 mm IDAL 885 nH, half the cross-section (0.78 cm²) Low-loss optionFT240-52 (5952003801), usually stackedAL only 325 nH and µi 250, but loss factor 45 vs 100 (×10⁻⁶) for mix 43 and a Curie point above 250 °C vs 130 °C Winding wireEnamelled copper, about 1 mm (18 AWG), 40 in or moreTight turns cut loss Capacitor100-150 pF ceramic, 500 V or moreHigh-band compensation across the 50 Ω side HardwareSO-239, antenna post, ground/counterpoise post, sealed box, PVC or PTFE tape for the coreWeatherproofing and insulation Test loadNon-inductive resistor: 2,450 Ω for 49:1, 3,200 Ω for 64:1Bench check before the wire goes up

Power ratings for these cores are not settled, so read them as ranges. Common ratings put the FT140-43 at 100 W PEP and a single FT240-43 at 400 W PEP SSB or 100 W CW/RTTY, while more cautious figures for one FT240-43 sit near 250 W. Mode matters more than peak power. Run FT8 or RTTY at the low end of any rating. Calculated core loss for a single FT240-43 at 2:14 comes out at 2.3 dB on 80 m, 1.9 dB on 40 m, 1.6 dB on 20 m and 1.5 dB on 10 m. That is roughly 30 % of your power turned into core heat. Two stacked FT240-43 cores at 2:14 calculate to about 0.6-0.8 dB, and that stack has carried about 400 W PEP on SSB.

Build steps

1. Pick the ratio and turns. For 49:1 on one FT240-43, use 3 primary turns and 21 in total for more low-band inductance, or 2 and 14 for better high bands. For 64:1, use 3 and 24, or 2 and 16.
2. Insulate the core. Wrap the FT240-43 in PVC or PTFE tape so the enamel never rides on a sharp ferrite edge.
3. Wind the primary. Leave a 3 in tail and wind the primary turns. Use a single tapped winding, or a twisted pair for the primary as many published designs do.
4. Continue the secondary in the same direction. Carry on to the total turn count. Count passes through the hole, not wraps around the outside. Keep turns tight to the core and close together.
5. Wire the connections. Start of the winding to the SO-239 shell and the ground/counterpoise post. Tap (end of the primary) to the SO-239 centre pin. Far end of the winding to the antenna post.
6. Fit the capacitor. Solder 100-150 pF across the SO-239, the 50 Ω side. Leave it off at first if you plan to compare.
7. Bench-test into a resistor. Put 2,450 Ω (49:1) or 3,200 Ω (64:1) from the antenna post to ground. Sweep 3-30 MHz with an analyzer and note the SWR on each band.
8. Box it and seal it. Mount it in a sealed enclosure with a drip loop in the coax.
9. Cut the wire long. Use 468 / f plus about 4 ft, then fit a choke on the coax near the radio, or at the feedpoint only with a short counterpoise wire behind it.

Winding style matters too. On stacked FT240-43 cores at 2:14, a single tapped, tightly wound coil measures the least loss, and the popular twisted primary with a crossover secondary the most, mainly above 20 m. I would wind the plain tapped version. Our 49:1 unun build walkthrough shows the same tapped-winding idea step by step.

Tuning and measurements

A transformer that cannot match a resistor will not match a wire. Do the bench test first. A NanoVNA used as an antenna analyzer can read R and X directly. The SWR plot alone hides too much.

• Read the resistive part at the fundamental. Adjust length until X is near 0, then read R. Multiply by the ratio to get the wire impedance. With 49:1, R = 42 Ω means 42 × 49 = 2,058 Ω. With 64:1, R = 42 Ω means 2,688 Ω.
• Use the mismatch table. The table below shows what each ratio does to the same wire. SWR here is from the resistive mismatch only.
• Adjust length for frequency, the capacitor for the high bands. Trim to set resonance on the fundamental. Then change the capacitor between about 100 and 200 pF to bring 10 m and 15 m down. Harmonic dips sit a little off each band, and a tuner covers that.
• Measure loss if you can. Build two identical transformers, join their high-impedance sides and pass power through into a 50 Ω load. Half the drop is roughly the loss of one unit.

Wire end impedanceSeen through 49:1SWRSeen through 64:1SWR 1,800 Ω36.7 Ω1.3628.1 Ω1.78 2,240 Ω45.7 Ω1.0935.0 Ω1.43 2,450 Ω50.0 Ω1.0038.3 Ω1.31 3,200 Ω65.3 Ω1.3150.0 Ω1.00 5,000 Ω102.0 Ω2.0478.1 Ω1.56

Across the realistic range neither ratio goes much above 2:1. The 49:1 wins at the low end, the 64:1 at the high end. For reading those sweeps in detail, see what an SWR sweep graph tells you.

The compensation capacitor value is another point where operators disagree. Many designs use 150 pF on the input for 10 m. Others use 100 pF, with the biggest effect above 20 m, and some tune about 150 pF for the best SWR on 20 m. Loss measured on stacked FT240-43 cores is about the same with or without the capacitor, so it changes the match more than the efficiency. I would start at 100 pF, try 150 pF, and keep whichever gives flatter high bands.

Mistakes that cost you

• Trusting a flat SWR. A lossy core turns power into heat, and heat looks like a good match. A perfect match on every band without a tuner is a reason to suspect core loss.
• Miscounting turns. One turn too many on a 2-turn primary, 15 / 2 = 7.5, gives 56.25:1 instead of 49:1. Count passes through the centre hole.
• Too few primary turns for the low band. Two turns on one FT240-43 gives 95 Ω of reactance at 3.5 MHz, well below the 250 Ω target. Use 3 turns or stack cores for 80 m.
• Using the wrong transformer. A 9:1 (3:1 turns, 450 Ω) is a random-wire unun. On a half-wave it misses by a factor of 5 or more.
• Skipping the choke. The coax shield carries current on an EFHW unless something stops it. The next question, and the most argued one, is the counterpoise. That is the subject of does an EFHW need a counterpoise.
• Using a carbon-fibre pole. Hang the wire from a fibreglass pole, not a carbon one.

If you would rather buy than wind, our roundup of end-fed wire antennas for HF lists finished options. For park activations, where a single FT140-43 at QRP power makes a light kit, the Parks on the Air guide covers the rest of the setup. On the bands the 468 / f wire does not cover (30, 17 and 12 m on a 40 m wire), an antenna tuner is what fills the gap.