Refining the EFHW with a Variable Transformer

[F4LEK]

Hello,

I was working on 3D-printed antenna winders for my EFHW when I set up the antenna in the garden. While analyzing the SWR curve with my NanoVNA, I was intrigued by the SWR reading on 40m, almost 2:1 at its minimum. This was unusual because I had never experienced such behavior while operating in the mountains (rocky summits).

Suspecting a ground effect, I decided to measure the impedance, at 7 MHz, of the antenna without the transformer. To my surprise, it was around 6000 Ω! I then moved the antenna to different locations in the garden, but the impedance remained in the range of 5500 to 6500 Ω. This confirmed that the issue was likely due to the interaction with the soil composition and moisture levels.

Impedence measurement:

To match 6000 Ω to 50 Ω, we need a transformation ratio of 11. With 3 turns in the primary it’s 33 turns in secondary.

Instead of making multiple transformers, I designed a single transformer with a tap at 22 turns. This provides two different transformation ratios:

• Primary to 22 turns → ( 22:3 ) ratio → ( (22/3)^2 = 54:1 ) → Matches ~2700 Ω
• Primary to 33 turns → ( 33:3 ) ratio → ( (33/3)^2 = 121:1 ) → Matches ~6000 Ω

To test the transformer, I set up the following configuration:

• Antenna: A ~20m long EFHW, with the apex at 5.5m height and both ends at 1m above ground (forming an angle of approximately 125°).
• Coaxial Feedline: 3m of RG58 coax between the transformer and the VNA.
• Common Mode Suppression: A choke was placed before the VNA (I also tested without it, with no noticeable difference).

I measured the SWR for the following transformers:

• Standalone 49:1 transformer using an FT114-43 toroid (black curve).
• Standalone 64:1 transformer using an FT114-43 toroid (dark red curve).
• New variable transformer on position 1 = 54:1 ratio (red curve).
• New variable transformer on position 2 = 121:1 ratio (yellow curve).

When testing classic EFHW transformers (49:1 and 64:1), we can immediately notice that they are not ideal across multiple bands. If the antenna is cut to resonate at 7.1 MHz, it ends up being slightly too short for 14 MHz, causing a frequency shift of approximately 350 kHz on 20m. This behavior can be attributed to end effects, which occur due to the high-voltage nature of the EFHW antenna. The last portion of the wire interacts with its surroundings (insulator, ground proximity, nearby objects), introducing stray capacitance. This effectively shortens the electrical length of the antenna, shifting its resonant frequencies higher than expected. Additionally, the impedance transformer itself introduces some inductance and capacitance, further modifying the effective resonance points. Because of these effects, it is common practice to compromise when cutting an EFHW antenna. Many operators tune it to a preferred section of a band (e.g., the CW portion or the middle of a preferred band). Additionally, a compromise must be made on the transformer ratio, as the impedance shifts from 3000 to 6000 ohms when switching from the 20m to the 40m band.

To address these limitations, I tested a variable transformation ratio approach. This allows slight impedance tuning, which in turn affects both matching efficiency and the exact resonance points of the antenna.

Results

Using the 54:1 Tap:

• The resonance on 20m is well-centered at 14.150 MHz, with an excellent SWR of 1.2:1.
• The resonance on 15m is more than acceptable, with a good SWR of 1.5:1.
• However, on 40m, resonance shifts too low, around 6.920 MHz, leading to a high SWR of ~2:1 at 7.100 MHz—less than ideal for that band.

Using the 120:1 Tap:

• The resonance on 40m is optimized, with an SWR as low as 1.0 at 7.063 MHz.
• However, this tap is not ideal for 20 and 15m, as the resonance shifts undesirably, similarly to single ratio transformers.

Not only does the ability to adjust the impedance transformation improve matching, but it also helps to fine-tune the resonance points to better align with the desired bands. To make this approach more practical, I designed and 3D-printed an all-in-one antenna winder and transformer case, integrating two banana connectors for quick antenna connection. This setup allows me to switch the matching ratio within seconds to better suit the operating band, providing greater flexibility in the field.

This experiment shows that a single fixed-ratio transformer is often a compromise, while a variable matching system can improve both efficiency and usability across multiple bands.

Further research is needed to explain why the 54:1 tap causes a downward shift in resonance points and how its behavior differs from a conventional, equivalent, single-ratio transformer. When using the 54:1 tap, 11 turns on the ferrite core remain unused. This may contribute to changes in inductance, stray capacitance, and coupling efficiency, affecting the resonance behavior compared to a single transformer with the same overall ratio.

The transformer:

The folded antenna:

The links and mast attachment pieces.

Full article on my blog

Open to discussion about this

Edit : post edited for better clarity and inclusion of latest results.

73, Rémy.

[G4TGJ]

Hi Rémy

What are you using as a counterpoise? Perhaps it isn’t long enough for 40m.

73 Richard

[F4LEK]

The counterpoise is my RG58 coax. I have 3.2 m of it between the radio and the matching box. It’s around 0.075λ

[OE5JFE]

Usual recommendation is 1/20 Lambda of counterpoise of the lowest frequency used (40 m Band ?) of coax. Then use a common mode choke to “end” the influence.

Source: Multi-band end-fed

With that setup I (think) managed to remove (or reduce) the impact of the ground pretty much. Maybe give it a try.

73 Joe

[G4TGJ]

I have a 2m wire counterpoise with my EFHW. I also use a common mode choke. It appears to work fine on 40m as well as 20/30m.

EDIT: It’s like (d) in Joe’s post.

[F4LEK]

I also have a choke at the end of the coax, at the radio side. I tried with and without, no difference, in this case.

I use this efhw for more than a year now and on the summits, it works well with my normal 54:1 transformer. However, in the garden, it seems different.

[HB9BCB]

Rémy,

Interesting findings so far, thanks for reporting.

It would be useful to know some more details about the EFHW transformer used, e.g.

• Toroid size and material mix
• Reflection and transmission graphs (s11, s21) 3.5-30 MHz, measured on the fully assembled/operational EFHW transformer

73, Heinz

[F4LEK]

@HB9BCB

It would be useful to know some more details about the EFHW transformer used, e.g.

• Toroid size and material mix
• Reflection and transmission graphs (s11, s21) 3.5-30 MHz, measured on the fully assembled/operational EFHW transformer

The toroid is a Fair-Rite 2643625002 , the same as described in my blog here: Efficient EFHW transformer for SOTA | F4LEK - Amateur Radio Blog

I can do some S21 efficiency measurements with a 2700 and 6000 ohm resistor on that new version, maybe tomorrow.

[F5MOG]

I am using 1:81 ratio, 3 / 27 turns, for about a year, with my 10/20/40 efhw , with the transformer always at the same height, on top of my walking stick. Impedance varies a lot with the height of the transformer. Satisfied with it, whatever the soil I found on the summit, on the 3 bands. The bottom impedance is different on the 3 bands, The ratio I use seems to be a mid point, vs your 54-121 range.

73 de Pierre F5MOG

[HB9EAJ]

Hi Rémy,

Thanks for posting your variable transformer solution!

Concerning the transformer, I came to a similar conclusion. I didn’t measure the antenna impedance directly, but the impedance and SWR on the transformer input side.

To measure the influence of the HF bands in relation to the transformer ratios and shunt capacitor, I built a special test coupler.

In my Portable 7-Band EFHW document on page 32 and 33 you can see different SWR curves for each transformer ratio (1:36, 1:49, 1:64 and 1:81), measured three times with different shunt capacitors (0, 82 and 120 pF).

Following, as an example, the measurements with a 82 pF shunt capacitor:

Apart from ground conductivity, height above ground, antenna configuration and wire diameter, the impedance seemed also frequency dependant, i.e. it is higher on lower frequencies and lower on higher frequencies.

73 Stephan

[G8CPZ]

Unless your garden is huge and empty you probably have large objects in the near field (I have large mature trees, a summerhouse, a garden shed, a garden pond etc) which might affect the behaviour of an EFHW there compare to an open hilltop summit.

I noticed - whilst using an antenna analyzer to trim the lengths of my multi-band EFHWs and CF linked dipoles in the garden – how the resonant frequency (VSWR minimum) is very dependent on the angle of slope of the wire(s) (for EFHW inverted-7 or CFLD inverted-V).

BTW: as mentioned every time someone puts ‘EFHW’ and ‘counterpoise’ in the same sentence: 1) I’ve not noticed any performance benefit deploying a CP under either of my EFHWs – so no longer bother. 2) if I had to spend the extra time to deploy a CP with my EFHW, I might as well spend that time putting up the CF linked dipole.

[F4LEK]

@F5MOG

with the transformer always at the same height, on top of my walking stick

Same as you Pierre, the transformer is usually about 1m high, attached to the walking stick with a paracord.

@HB9EAJ

Apart from ground conductivity, height above ground, antenna configuration and wire diameter, the impedance seemed also frequency dependant, i.e. it is higher on lower frequencies and lower on higher frequencies.

Yes, in my garden, the impedance was around 6000 ohm at 7 MHz and around 3000 or 4000 ohm at 14 MHz if I remember correctly (I didn’t save the curves for the bare wire).

@HB9EAJ

Concerning the transformer, I came to a similar conclusion

In your results, we can also see that the winding ratio is affecting the resonant frequency of the system in the same way as I found: The higher the transformer ratio, the higher the resonant frequency.

[F4LEK]

@G8CPZ

Unless your garden is huge and empty you probably have large objects in the near field (I have large mature trees, a summerhouse, a garden shed, a garden pond etc) which might affect the behaviour of an EFHW there compare to an open hilltop summit.

It is big. Around 30 000 m² of field. Only a few trees in the middle. The test setup was not so far from the house though, maybe 20m.

@G8CPZ

how the resonant frequency (VSWR minimum) is very dependent on the angle of slope of the wire(s)

Yes this is worth to mention. However, for that particular test setup, I used my normal mast that I always use, the carbon 6, with the top of the antenna at around 5.5m high and the two ends at around 1m high, on the same XY plane.

[HB9BCB]

@F4LEK

The toroid is a Fair-Rite 2643625002 , the same as described in my blog here: Efficient EFHW transformer for SOTA | F4LEK - Amateur Radio Blog

Ok, thanks.
An efficient EFHW transformer with this small ferrite core was presented by Owen Duffy in his May 2018 publication.
Since then, this ferrite core has achieved almost cult status among QRP EFHW enthusiasts … but not all of them are aware of how installing it in a housing can reduce these good properties, measured directly on a bare transformer.

When looking at your document referenced in the link above, I noticed at first glance the very high calculated efficiency values. These are, even for 3 primary turns and the beneficial effects of the somewhat thin wire at 0.4 mm, a little too good … and would be even better by a very small margin if the attenuation by the load resistor was correctly taken into account with -10Log(50/(50+Rload)) dB, hi.

The reason for this is an error in the evaluation of the VNA measured values, in which the transmission loss (-|s21|) is incorrectly labeled as insertion loss (transmission loss+input mismatch loss).
BTW, this exact error crept in when editing the table columns in the data sheet of my 50/30 W EFHW transformer, which was uploaded from April 2024 to February 2025 on HB9SOTA … sorry if that was used as a template.

[F4LEK]

I’m not sure I undestood your message.

The calculations in my blog are derivated from Owen Duffy blog and calculators, where the compensation factor is calculated as

• Cf = 10*log(Rload/50)
  and then
• Insertion loss = -S21gain - Cf

With your approximation of the compensation factor, a 2675 ohm resistor goes from 17.28 dB to 17.36 dB. The final efficiency is increased by 2%.

@HB9BCB

n which the transmission loss (-|s21|) is incorrectly labeled as insertion loss (transmission loss+input mismatch loss).

Could you explain this ?

I think those calculations are very touchy, and more like a relative way to compare the efficiency of different transformers more than measuring an absolute efficiency. The test setup is extremely sensitive and very prone to calibration errors. When I measure, I do it in one go, with the exact same setup, and I make sure nothing is moving betwen the measurements.

Writing this message I went back to the bench and realised my possible mistake. I used poor carbon film resistors and measured them with a normal multimeter… When measuring them with a VNA, at the target frequency, the results are different. Since they are simple carbon resistor, of course they are not linear in frequency. As a consequence, we should measure the resistor and calculate the compensation factor for each frequency. Here is the comparison of the values given in the blog, before and after modification:

[F4LEK]

New batch of measurments today:

Raw impedance of the antenna (no transformer, nothing):

---

Comparison of the variable transformer with two other transformers I had on hand (all with the same 3m of coax + choke at vna side)

• variable transformer, position 1 (54:1)
• variable transformer, position 2 (120:1)
• transformer 49:1 (on a FT114-43)
• transformer 64:1 (on a FT114-43)
  
  

---

Zoom on the 30m band (I made a 30m link)

• variable transformer, position 1 (54:1)
• variable transformer, position 2 (120:1)
  
  

[GM4EVS]

@F4LEK

To match 6000 Ω to 50 Ω, we need a transformation ratio of 11. With 3 turns in the primary it’s 33 turns in secondary.

The ARRL publication “Experimental Methods in RF DESIGN” by Wes Hayward, W7ZOI et al, has long been a well-regarded reference for ham radio work.

In the first edition, second printing, pages 3.33 onwards provide extensive information relating to the design of toroidal Ferrite Transformers.

One key passage states:

“A practical transformer will have a primary inductance with a reactance at least 5 times the terminating resistance at the low frequency limit …”

Working with a 6000 ohm impedance, this might suggest a winding reactance of some 30,000 ohms.

I am not sure how this guidance relates to the design of EFHW baluns.

Are there sufficent turns, and a suitable core material, to meet this criterion?

73 Dave

[HB9BCB]

Rémy,

It is not usual and somewhat confusing, considering the capacitor used to compensate for leakage inductance across the primary winding of the EFHW transformer, to use the term compensation factor for the load resistor in the transmission measurement. Owen Duffy chose the easy-to-understand term load resistor attenuation for this.

I use the template below for efficiency calculations of EFHW transformers based on the VNA measurement results. Maybe this will be helpful for you too.

[F4LEK]

And what is your opinion about the non negligible difference between the resistance values of resistors measured at CC and at HF frequencies ?

Edit 06/03/2025 : update initial post and blog article with recent results and conclusions

[F4LEK]

Little update:

I modified the transformer and added a third tap to reduce the gap between the 54:1 and 120:1 ratios. Now, I have three ratios:

• P1: 49:1
• P2: 75:1
• P3: 107:1

On the antenna

I also cut a new wire for the entire antenna since my previous one had become a bit messy with all the splices and links from previous adjustments.

The new wire is approximately 20m long, with two links at around (exact lengths not measured):

• 10m: The antenna is configured as a sloper, potentially useful for 14 MHz and 28 MHz, and also for restricted summits.
• 14.5m: For 10 MHz and 21 MHz.

Here are the SWR curves for each antenna configuration combined with all three transformer positions:

Efficiency tests

I reviewed and refined my VNA efficiency measurement protocol, ensuring:

• Characterization of the resistors according to frequency
• The same clean fixture for each measurement
• Short component leads (soldered)
• All measurements taken in a single session

I measured two different resistors (2500Ω and 5500Ω) for each transformer position. For comparison, I also tested four other transformers:

• A 49:1 and a 64:1, both based on an FT-114-43
• A 54:1 using another 2643625002
• An unusual 2643625102 (equivalent in size to two stacked 2643625002)

All transformers had the same winding configuration: three turns on the primary, with wires neatly wound on one side of the toroid, using 0.4mm wire—except for the FT-114-43 transformers, which used 0.8mm wire.

Conclusion

• The SWR curves look good—7, 10, 14, and 28 MHz are available with low SWR across the entire band. 21 MHz is usable but may not be ideal. I need to test the setup on different ground conditions to ensure it is as versatile as it appears.
• Efficiency is excellent when the correct tap is used for the impedance it is designed for.

• The behavior of the taps is not equivalent to the same ratio in a standalone transformer. The unused turns and leads affect the entire system.