Wednesday, May 20, 2015

Exploring the W6NL 40 Meter Moxon with a Model

W6NL (some time ago) came up with the idea of making a 2-element rotatable yagi for 40 meters that uses no loading coils yet is no larger than typical commercial designs -- such as the Cushcraft XM240 -- and has better performance. It employs capacity hats that also achieve high element coupling to make it a Moxon rectangle. There is also a step-by-step set of instructions for converting the XM240 to this design for those who do not want to start from scratch.

In my continuing ruminations on gain antennas for 40 meters I have been intrigued by the W6NL design for some time. For now it is an intellectual exercise since I am unable to install such an antenna at my current QTH. Since this may change I have to plan ahead so I can move fast when the opportunity arises.

A 2-element Moxon yagi (or the Moxon rectangle) has no more gain than a conventional 2-element yagi. The W6NL is no different, where gain bandwidth is about what one would expect. Yet its performance improvements are notable:
  • Broad SWR bandwidth: Covers 7.0 to 7.3 with an SWR below 1.5. This is an excellent operating convenience, and can help with removing the need for retuning kilowatt amplifiers when jumping around the band during contests.
  • Improved F/B: A 2-element yagi has acceptable F/B over a narrow bandwidth. The Moxon design, by way of increased mutual coupling between elements, maintains a good SWR over a larger frequency range.
  • No coil losses: The ESR (equivalent series resistance) of loading coils in commercial short 40 meter yagis varies from reasonably good to poor. It is not unlikely that power loss is as much as -2 db in some antennas. VE6WZ has a good discussion on this topic you ought to read.
  • Direct match for 50 Ω coax: No need for a matching network at the feed (e.g. transformer or hairpin) as for a conventional yagi. A common-mode choke is still required.
There were a few things I wanted to know about this antenna so I built a model of it in EZNEC. Since there is more than one version of this antenna out there I used the XM240 modification as the template. If you are interested in the general performance and construction details you can find them in the various papers linked to above.

Model peculiarities

This is a difficult antenna to model with the NEC2 engine. Those who have tried have failed. This is no surprise since element tapering creates an anomalous reactance with NEC2. This is further exacerbated by non-linear elements, which is a feature of Moxon rectangles. The W6NL antenna has these challenges in spades. For further reading on this topic I suggest reading the analysis and modelling by W8WWV.

It is therefore no surprise the model I designed is resonant well below 7 MHz. This is expected in NEC2. To keep it simple I used the exact linear measures for the XM240 modification and substituted constant tubing diameters for the 4 wires that comprise each half-element. I placed the antenna in free space to eliminate pattern complications due to height. Performance is retained, as it is for most yagis, when it is moved from free space to at least λ/2 above real ground. In this case that would be 20 meters height.

I can get away with this shortcut since my objective is to test certain aspects of the antenna's behaviour, not to create a construction template. My model is therefore a reasonable proxy of the actual antenna that is suited to the analysis in this article. For the same reason I omitted the boom and a couple of other fine details. Should you ever build this antenna you must follow W6NL's instructions exactly, or go through the pain of a careful design using NEC4.

With the preamble out of the way let's continue on to the analysis.

Wind action

The elements are about 6.6 meters apart, as they must be when using the 22' boom of the XM240. The capacity hats do double duty as the inward turned segments to achieve the high coupling Moxon requires. The lateral distance between the tips is about 40 cm, plus an outboard offset. This is close. In a typical Moxon rectangle the distance is fixed, a critical feature since small changes can have a large impact on antenna behaviour.

The W6NL antenna must contend with the action of the wind. I wanted to know how the antenna would respond when the wind moves the element ends closer together and farther apart. My approach was to rotate one half of the reflector 5° inward and then 5° outward and check for changes in SWR and pattern. Although it is common in wind that that all of the 4 half-elements are away from the nominal positions, my modelling experiment is intended to provide a starting point for a more complete analysis. I have no immediate plans to do so.

Since the model has resonance shifted downward by ~300 kHz I tested the pattern at 6.8 MHz, which would be equivalent to 7.1 MHz for the actual antenna. I then overlaid the patterns (all azimuth at 0° elevation in free space) for the depicted bending of one reflector half-element.

The result is a modest change in the F/B, evidenced by a different shapes of the rear lobes. Gain is effectively unchanged. SWR (not shown) improves when the reflector bends inward, and is about the same when it bends outward.

This is a promising result. In winds that are below storm level there is little concern about poor behaviour. The same is not true of stronger winds. It is all too easy for the tips to touch, even if only momentarily. I modeled this by connecting the ends of wires #4 and #11 for the inward bending case. The pattern changes to one that is equivalent to a dipole: a loss of -4 to -5 db of gain and a total loss of F/B. While undesirable this intermittent performance loss is no a deal breaker.

The more serious problem is that if contact occurs during transmissions, especially high power, arcing and element damage are likely. Also concerning is that the SWR jumps to well over 3, which can place stress on amplifiers and can cause some transmitters to shut down or roll back the power.

I should note that I have never heard evidence of this occurring in practice. Yet it must happen. Ways to deal with this, if one wishes protection, include the following:
  • Longer boom: Mutual coupling is reduced with some degradation of F/B. However even 50 cm could eliminate the problem. For example using the boom of a defunct TH6 (24') rather than that of the XM240.
  • Tying the tips together: This is not easy since the capacity hats are not in a plane. However I suspect what with some modelling they could be mated with a solid dielectric (e.g. fibreglass rod) with only a small performance impact.
  • Lateral guying: Rope or Phillistran guys can be used to hold the element steady in both the horizontal and vertical planes. The negatives are reduced visual appeal and increased difficultly raising and lowering the antenna.
  • Insulation: Wrap or cage the last foot of the inward tips of one element with a high-quality insulator. An open cage is better, though more difficult to construct, so that the dielectric effect has little impact on antenna tuning.
Out-of-band resonance

It is rare for a 40 meter yagi to stand proud and alone atop a tower. At least one more yagi on the same mast and rotator is most common. Interactions must be found and dealt with if all the antennas are to perform to their potential.

Mutual coupling between antennas and their individual elements will at a minimum cause the appearance of an unwanted reactance that will first be noticable in a degraded F/B, then SWR and gain as coupling increases.

There are basically two types of interaction: non-resonant and resonant. Non-resonant coupling can be reduced and largely eliminated by increasing antenna separation. As we increase wavelength this becomes more difficult since a large separation (in wavelength units, which is what matters most) may be physically difficult. Any 40 meter yagi is susceptible to this effect since other yagis are typically less than 0.1λ above or below it.

On the positive side for our 40 meter yagi non-resonant coupling is greatest when the element tips are close to other metal. Since the 40 meter yagi is almost certainly the largest in the stack its ends are quite isolated. However the same is not true for higher band yagis. This may need to be addressed if, say, a 20 meter yagi is closer than 3 meters to the 40 meter yagi. Modelling can help.

Resonant interaction is a greater challenge, one that no reasonable stacking distance can entirely solve. We need to characterize the behaviour of the 40 meters yagi across the higher bands to discover potential problems.

It helps that the W6NL yagi is less than full length. As a general rule any loaded antenna is not resonant on its harmonics. Therefore we may be saved from potential destructive interaction with a 15 meter or tri-band yagi. However it is no guarantee so we must do the work. The first step is an impedance scan up to 30 MHz.

We are mostly in luck: the only out of band resonance is centred on 28.8 MHz. Be skeptical about this value since as we've already noted the NEC2 resonance calculations are suspect. That does not mean that the resonance is a mirage, only that it could be elsewhere in the 10 meter band.

As it turns out the 28.8 MHz resonance is an oddity. The element radiates broadside and transports energy to the hat which radiates end-fire. The short element tips contribute nothing. The reflector and its hat couple poorly and thus contribute little to the pattern. The radiation pattern at 28.8 MHz is a 4-leaved clover with no gain (pattern not shown).

With this analysis in hand we can move on to stack another yagi near the W6NL yagi and see what happens.

Tri-bander stacking

I simplified the interaction model by importing the small 3-element tri-band yagi I modelled in 2014 and placed it above the W6NL antenna. I tried two vertical spacings: 2 and 3 meters, which brackets the range most hams would most often attempt on their towers. Full-sized mono-banders and longer-boom tri-banders are also commonly stacked though I expect the interactions to be similar enough for 20, 15 and 10 meters that this model suffices for a first step.

Current plot at 28.8 MHz for the small tri-bander 3 meters above a W6NL Moxon
When spaced 3 meters the interactions were modest, even on 10 meters where there is a resonance on the 40 meter yagi. Gain on the tri-bander is typically reduced less than -0.1 db on 20 and 15. F/B does degrade though not by enough to be a concern. On 10 meters the gain reduction is closer to -0.2 db and F/B degrades more. Even so the performance is effectively unchanged.

When spaced 2 meters the interaction is more pronounced. SWR increases enough within some band segments to be a concern. Exactly characterizing the impedance change is difficult since the tri-bander model is itself imperfect (discussed in the referenced article above). Many hams choose 2 meter spacing since it is man-height, allowing installion by standing on the tower top plate. Getting a little higher is advisable when mounting above the W6NL or similar 40 meter yagi.


Many hams choose to stack short 2-element 40 meter yagis rather than deal with the challenging undertaking of a full-sized 3-element yagi. That allows for equivalent gain at less expense and difficulty. It also gives greater operating flexibility with the ability to point the yagis in different directions.

The question is just how well it plays in practice for modest sized towers and stacking separations. A full analysis is not my objective at present so I chose one configuration to model. The intent is to gauge what to expect from a full analysis and an optimization. I placed one yagi at 20 meters height and a second at 40 meters height. That's a big tower yet stacking separation is a modest λ/2, about the minimum that can be expected to work well for this antenna.

To correctly model stacking the model must use real ground, not free space. In the model I first I measure the lower antenna alone as a baseline for comparison. Then I put the higher antenna into the model and measure performance for the lower, upper and both in phase (BIP). The overlaid elevation patterns of all four cases is at right.

The primary trace is the standalone yagi at 20 meters height. Its 10° gain is 5.7 dbi. This places it about -0.5 db below a full size 2-element yagi. This is quite good for a small yagi that has great F/B and SWR.

When the upper yagi is installed, but not fed, the gain drops -2 db. The upper yagi is coupling to the lower yagi and lowering its performance. This is about as one should expect. When the yagis are pointed in different directions the negative impact will often be reduced.

The upper yagi alone has two main lobes, with the lower one showing excellent low angle radiation. SWR curves for these three cases are similar and very good.

Where the array shines is when both yagis are fed in phase (BIP). Its gain is about 3 db better at low angles than the upper yagi alone, and 6 db better at 10° elevation than a single yagi up 20 meters. F/B is a bit worse at low angles. SWR is very low, even better than a single W6NL yagi up 20 meters. This assumes proper choice of phasing and power splitting systems. These are discussed in my introductory article on stacking.

Unlike my 40 meter stacking example in that article this 40 meter stack works well and can be recommended. The difference is that these smaller antennas have a broad main lobe that allows for better lobe "matching" when stacked at modest heights (in terms of wavelength). Long boom, multi-element yagis are more negatively affected by low height and separation due to their narrower main lobes.


Commercial short 40 meter yagis like the XM240 have comparatively lower bandwidth and greater loss (due to loading coils) along with poorer F/B. While it won't beat the gain of a full-sized 2-element yagi the W6NL Moxon does come close. That it plays well with a stacked tri-bander is an added bonus for those with one tower and want good contest performance on 40 through 10.

Since the W6NL 40 meter Moxon is difficult to model in NEC2 interaction testing and tuning with software is challenging. I have no intention of purchasing a NEC4 license. With the approximate NEC2 model I built I feel reasonably confident that this antenna can be effective and effectively stacked with one more antennas for the high bands. It is an antenna I will mark down for future consideration when I have the tower and space for it. However I do have some worries about the elements touching in a high wind. I would do something about that if I built one.

1 comment:

  1. Hello, the loads interact with 15m antennas if they are parallel to their elements.
    73, Felipe - PY1NB CT7ANO


All comments are moderated, and should appear within one day of submission.