Thursday, May 8, 2014

Choosing a High-bands Yagi (Part 4) - Rotatable Wire Yagi

Rotatable wire yagis have some attractive advantages:
  • Light weight
  • No loads or traps
  • Moderate wind load
Unfortunately these advantages are counterbalanced by several challenges, which include:
  • Unlike aluminum tubes the wires don't support themselves, and therefore require a (usually) complex mechanical design with little to no metal in the structure
  • Elements cannot in general be parallel, resulting in reduced performance and increased complexity of electrical design and behaviour
The question is whether the trade-offs are favourable in comparison to alternative small size rotatable yagis such as the 2-element or 3-element tri-banders, covered in the previous two parts of this series.

There are 2 especially popular designs in this category of rotatable wire yagi: the Hexbeam and the Spiderbeam. The former is typically limited to 2 elements due to its structure, while the latter is more flexible with regard to element count. This article will focus on the Spiderbeam since it is the more expandable design and has all the elements on one horizontal plane. The vertical stack height of the Hexbeam design requires a shorter tower to stay under the Industry Canada exemption limit of 15 meters structure height. There are enough jurisdictions outside Canada with similar limits that this issue has more than local applicability.

In an earlier article in this series I mentioned that I had found an EZNEC file on ARRL's web site for a tri-band Spiderbeam. Some aspects of the model did not follow best practices for a NEC2 model so I made adjustments to the segmentation and elsewhere, but did not alter the wires in any way.

Although wire lengths are not exactly the same as found in the Spiderbeam construction guide I was able to verify that after compensating for use of non-insulated wire in the model that the parasitic tuning is identical. The use of 1 mm diameter wire (18 AWG) is the same.

The model uses transmission lines to connect the roughly-parallel elements of driven element fan dipole. This is a subject I covered earlier. This method is superior to modelling the element connections using wires in NEC2.

In the adjacent view of the EZNEC model the antenna points in the 'X' direction. There are reflectors for 20, 15 and 10 meters at the rear (wires 10 to 15) and the same for directors at the front (wires 16 to 23). There is a second director for 10 meters so there are in fact 4 elements on that band. The 5-band version of the commercial product (not covered here) has 2 elements on 17 and 12 meters.

There are no traps or other loads in this antenna, and therefore no associated losses. There are I²R losses in the wires which are higher than in a yagi made of aluminum tubing. This can be reduced by the use of heavier gauge wire. For this model I will use the AWG 18 wire as in the found model so that I can avoid the subsequent necessity of adjusting wire lengths.

Design Issues

The Spiderbeam is a good antenna. So when I say 'issues' I am not dismissing the design or the designer. Every antenna includes compromises. This one is no different. Let's look at them now so we can understand the (model) measured performance.
  • Non-parallel elements: I have never been able to get an antenna with parasitic elements bent in the fashion of this antenna to approach the performance of a conventional yagi with parallel elements. It may not be possible. I have experimented with shorter and longer booms with not much better luck. The Spiderbeam choice of 10 meters for a boom length (on 20 meters) looks about the best possible for this style of antenna. Alternatively, parasites that are more rectangular (straight but with the ends turned sharply inward) as they are in a Moxon 2-element antenna generally perform better.
Spiderbeam Construction Manual caution: element lengths are critical (yes they are)
  • Element tuning: The ends of the parasites are close to the driven element and, except on 20 meters, are close to other parasites. Element tuning is altered (electrically lengthened) when its ends are close to any metal or other elements. Tuning is difficult since the resonant frequency is obscured in any array and is further muddied by this coupling. Adjusting the element length is further complicated if the distance to other metal objects changes when the wire length is changed. As the above warning Spiderbeam includes in the manual says, element length and placement is critical to proper operation.
  • Element interaction: There are no traps or trap losses in this antenna but there are many more elements than in a 3-element tri-bander. More elements means more interactions. NEC2 and its cousins are very good but cumulative errors inevitably creep into any numerical model with lots of close-spaced wires. Building an antenna like this from a numerical model inevitably requires wire length adjustments during construction and tuning. Further, there are also small induced currents on the many non-resonant elements (those not active on a specific band) that complicate and slightly degrade gain and F/B performance.
  • Match vs. gain performance: As we will see the commercial version of the Spiderbeam is designed for a good match (low SWR) across all 3 bands. This is at the expense of forward gain. That is true of any yagi. It can be corrected, at a price. I address this further along in the article.
Another issue with this antenna is its mechanical structure. Its 3-dimensional "closed" format makes it difficult to raise and install on many towers. You may have to use your imagination a bit to understand what I am about to say since I haven't drawn any pictures to illustrate the following points.

Unlike a conventional yagi this antenna must be lifted and installed as if it were a solid, 1-meter thick plate. This is because there are no openings in the structure; the entire space is filled with wires, ropes tying the wire ends, fibreglass tubes and ropes that tension the tubes to a central mast. That is, the antenna must be lifted over guys and other antennas and dropped down over the top of the mast.

In fact, the mast itself is an integrated part of the structure. This antenna does not play well with more than one antenna on the same rotated mast. For many hams this is acceptable. It is not acceptable to me.

If you have a guyed tower, other yagis on the rotator mast or wire antennas hanging from the tower you will have grief installing this antenna. Even if this is the only antenna on the mast today, if you decide to add another antenna in future it will have to placed below the Spiderbeam, and do so carefully to avoid tangling and damage during installation.

The fact that this antenna is a good broadband match to 50 Ω coax is a danger sign. That is, with respect to gain. If gain is the primary reason you choose a yagi for an antenna this should concern you. It certainly concerns me. Let's look at the performance charts produced from the EZNEC model. Keep in mind that the model has been verified against the manufacturer's construction guide so it should be a good reflection of reality.

On 20 meters the gain is a full -2 db below that of the reference 3-element yagi introduced in Part 1, never rising above 7 dbi. The F/B is, however, quite good across the band, even better than the reference yagi in many respects. Although I can't say that it is generally true, I have noticed from my various modelling activities that I can never get good gain from an antenna with inverted vee elements (which these effectively are) but I usually do get good F/B performance. This is on a boom that is 35% (2.6 meters) longer than the reference yagi.

In contrast the SWR bandwidth is definitely better. The good match comes with a cost in performance.

This is not an immaterial concern. Most hams will spend a lot of money in coax or tower height to gain another 2 db, or even less, of forward gain. To then throw it away at the antenna makes little sense to me.

On 15 and 10 meters the performance is a little better. While still not rising above 7 dbi gain on 15 meters it is ruler straight across the band at just a hair under 7 dbi. F/B on 15 is also very good, as is the broadband match. Again, the gain is -2 db worse than an equivalent reference 3-element yagi. However in this case the difference is not so bad since the reference yagi has a broad gain peak such that over much of the band the difference is no worse than -1.5 db.

The SWR bandwidth on 10 meters is narrower than on the other bands but still very good. Gain is also better, but peaks at 8.5 dbi very high in the band (29.5 MHz). The gain is still disappointing since there are 4 elements on 10 meters (2 equal-length directors), yet doesn't quite reach the gain performance of a reference 3-element yagi. F/B performance is, again, very good.

Using 1 mm wire as specified the typical copper wire I²R loss is -0.3 db. This is included in the modelled performance.

Designing for gain

I think it is worthwhile to see how far this style of antenna can be pushed in the pursuit of gain. I created a mono-band 20 meters model by eliminating all the other elements. It was then a simple matter of altering the ratio of reflector to director lengths to optimize the gain.

The simplest way of proceeding is to increase the length of the director in small steps. To minimize the impact of varying element interactions I did this by moving the director's endpoints (wires 6 and 7 above) outward on the 'Y' axis. Changing the distance between the ends of the director and the driven element would complicate matters since this would add a second variable to the tuning process. The boom length was fixed at 10 meters (distance between parasite apexes).

Once I found the optimum director length (maximum gain) I adjusted the reflector and driven element to compensate for the longer director and the absence of coupling with 15 and 10 meters elements.

The gain was increased by ~0.7 db at a centre frequency of 14.075 MHz. This includes copper wire loss of -0.7 db due to the lower radiation resistance. Heavier gauge wire can tame this loss. For example with 2 mm wire (12 AWG) the I²R loss is reduced to -0.3 db.

Gain and F/B remain very good up to perhaps 14.250 MHz, above which the F/B in particular rapidly degrades. Even when optimized in this way, and using 12 AWG wire, the gain is still -1.1 db worse than the reference 0.35λ 3-element yagi made from parallel aluminum tubing.

The ratio of reflector-to-director length in the gain-optimized antenna is 1.038. While this is quite narrow for a conventional yagi (±1.9%) it is less so in this case. To determine the actual reactance of the parasites I isolated each of them in the model and fed them directly to find their impedance curves. Their reactances are equal and opposite (±29 Ω) at 14.350 MHz. The resonant frequencies of each separate element are 14.100, 14.525 and 14.600 MHz for the reflector, driven element and director, respectively. From this the actual ratio of parasite resonant frequencies is ±1.8%. Compare this to the ±4% ratio of the 3-element reference yagi.

The frequency of maximum gain is (as shown above) 14.075 MHz, far from where the frequency where the reactances cross at 14.350 MHz. Either the use of bent parasites causes this difference from standard formulae for yagis or (far more likely) the nearness of the element ends to the driven element electrically lengthens those elements by ~2%. This can be a difficult antenna to design from scratch. I also know this from trying a similar approach several months ago for an unrelated antenna concept.

In the gain-optimized yagi the match is, as expected, more challenging. Feed point impedance drops to ~25 Ω. The antenna requires a 2:1 balun or a beta match. Either should work well since the radiation resistance changes slowly across the band.

It should be possible to do the same optimization on all 3 bands at once while continuing the use of a common feed and matching system. To do so would require paying attention to feed point impedance. It is a personal decision whether to undertake this change or to go with the manufacturer's design choice for broad bandwidth performance and simple match to 50 Ω coax.

An alternative approach for improved performance is a pseudo-Moxon design where the central section of the parasites is parallel to the driven element. I may try this in the future. If I do I'll write an article about it.


I am holding off on selecting from among the various type of small, high-bands yagis until the end of this series of posts. For now I will say that I am disappointed with the Spiderbeam's performance. The very good match and F/B performances are nice but, for me, do not compensate for disappointing gain.

As I showed it is certainly possible to change the antenna so that it has higher gain on all 3 bands. It would require tighter parasite tuning, 14 AWG wire and a 2:1 balun or beta match.

Even so the mechanical issues concern me. This is not an antenna that plays well with other antennas on the same mast or on a guyed tower, which is my situation for the near term. Your criteria may be different from mine. Many hams have bought and love this antenna. It's an especially popular choice for DXpeditions.

No comments:

Post a Comment

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