Monday, May 2, 2016

Adding 160 to the 80 Meter Vertical Parasitic Array

My earlier article describing a model for a 3-element, 4-direction vertical yagi for 80 meters includes a provision for an omni-directional 160 meter vertical. I chose not to fully develop that aspect of the antenna at that time since it is deserving of a separate and comprehensive treatment of its own. That time has now come.

You ought to read that article before proceeding since I will not repeat much about the physical layout and construction of that antenna. That background is vital to reading the present article. I will assume that you have read it.

My objectives

I am not one of the great enthusiasts for 160 meters. The DX challenge intrigues me since I have never had an opportunity to build and use a good antenna for this band, nor have I been in a sufficiently low noise environment to even want to try. Perhaps my enthusiasm will grow, or perhaps not. For the present my objectives are modest.

First, I want to enhance my contest scores with QSOs and multipliers on 160. Second, I would like to achieve DXCC on 160. Both are attainable with a modest antenna. High power would also help, however doing so with 100 watts is an attractive option, if only for the challenge. After all, 10 db of amplification can cover for a poor antenna, but I want to make an antenna that is as good as possible without extreme effort.

If you've followed this blog for any length of time you'll know that designing and building antennas is something I like to do. Getting around that with an amplifier doesn't satisfy me. Going to high power is something to do after the antenna works to my satisfaction.

Alternative approaches

There are basically 3 alternatives to extending the 80 meter antenna to 160 meters, to convert what is a λ/8 vertical to an electrical λ/4. The support structure for the 80 meter parasitic elements includes an additional 6 or 7 meters of mast and a wire top hat incorporated into the catenaries. Additional loading may also be required since the mast and top hat will not be adjustable once built.
  • Top relay: A high-voltage vacuum relay at the top of the 80 meter vertical switches in the additional mast and top hat for 160 meter operation.
  • Trap: Placed at the top of the 80 meter vertical it allows a fully passive approach to adding 160 meters. However there are costs, on both bands, that need to be considered.
  • Parallel wire: Run a separate wire parallel to the tower and attached to the extended mast. A switch at the feed point selects the tower (80 meters) or wire (160 meters). This is the one pictured in the article describing the 80 meter array. There are switching and coupling issues to be considered.
The radial field for the 80 meter array will initially serve for 160 meters. Although the diameter of the field is ~55 meter it is dense enough to allow for good, not great, efficiency on 160. The ground impedance will be higher than on 80. On 160 it will be electrically short so the radiation resistance will be reduced, which will also impact efficiency.

With all the alternatives I want an acceptable match to 50 Ω without the need for a matching network, just as for the 80 meter array in omni-directional mode. The switching system required for 80 meters direction and mode, and for band selection is as complex as I want. That is, more switching requirements for an L-network or 160 meter band segment are undesirable.

These constraints versus performance are acceptable to me, at least for the near future. I can revisit 160 after the rest of my next antenna farm has been built. This is in accord with my priorities.

I will now run through the presented options to see how they compare. All can work well with careful design. My overriding concern is to sustain excellent 80 meter performance and operational band width, and make sacrifices on 160 if necessary.

Basic model

Before going further it will help to go over the common aspects of the modelling. The 80 meter vertical is 20 meters tall, just as it was before. My objective is that the vertical in omni-directional mode require no matching network. If not for that a precise vertical height would be unnecessary since this would be compensated for in the L-network without affecting array operation.

The extension mast begin slightly above the 80 meter vertical, just enough to convince NEC2 that it is not connected to the 80 meter mast. As previously described the 80 meter vertical is a tower of 250 mm effective diameter and about 19 meters tall, with an adjustable mast to tune it to resonance. The two masts are mechanically coupled with a structural dielectric rod such as fibreglass.

A short adjustable spire projecting above 27 meters is included in the model, but may not be required. Once built the spire is too far out of reach for tuning.

The extension for 160 meters is as shown in the physical layout of the earlier article. What I've now done is make the extension mast and top hat explicit in the model. The mast extends to 27 meters height and the four top hat wires are 5 meters long. These wires should be copper-clad steel (available from many suppliers) rather than copper (hard or soft drawn) since they are under enough tension to stretch over time, and possibly break when the mast is subjected to high winds and icing. Alternatively the catenaries can be entirely rope, with copper wire running along the rope. The ends of the wires should still be looped around an egg insulator to avoid excess corona effect which could alter antenna resonance when the rope is wet.

Per this design the full vertical is resonant at ~2.150 MHz. A coil of 20 μH at the base of the mast extension (height of 21 meters) brings resonance down to 1.85 MHz. The coil value is convenient since I have one of just about that value in my junk box, reclaimed from the half sloper I removed last year.

A smaller coil is easily possible with a taller spire or longer wires in the top hat. Those were avoided in order to minimize the risk of interaction on 80 meters; that is, the isolated 160 mast and top hat accidentally interacting with the 80 meter array, which is the more important of the two antennas. The model shows negligible interaction. I may eventually try a redesign with longer top hat wires, just to see what happens.

The radial field of the 80 meter array mostly fills a square 55 meters on a side. This is on the small size for 160 meters, which out to be closer to 40 meters length (λ/4). For this reason in the MiniNEC ground I modelled the ground resistance as 10 Ω on 160. This may be a bit higher than the actuality, though that will also depend on ground quality. I assume EZNEC medium ground. In this case the near field ground loss is -1.7 db. For a 5 Ω ground the loss is -0.9 db.


The SWR band width is reasonably good, and could be made better with an L-network (or higher ground loss!). The L-network design is straight-forward, and can be done once the antenna is built and the impedance measured. The R component of the impedance is the sum of the radiation resistance and loss resistance, so it is easy to see how the impedance is expected to vary with the ground loss. Expect seasonal changes.

Top relay

The antenna design is really just what I've already described above for the basic model, with the addition of a SPST relay at the top of the 80 meter antenna to connect it to the extension mast, loading coil and top hat for 160 meter operation. Even so there are practical considerations in building the antenna.

No ordinary relay can be used in this application. The RF voltage can be very high at the top of a vertical (or the end of a dipole) even with moderate power. That point is very high impedance, which means close to zero current and high voltage: Z = E / I. It cannot reach infinity since there is always a conduction path in air and over the contaminated (dirty) surface of the end insulator. A large conductor reduces the maximum voltage, which is why some place a metal ball at the top of a vertical to reduce corona discharges. In some circumstances it can even reduce the noise floor on receive.

A vacuum relay is a required. This is stressed in ON4UN's Low-Band Dxing book for good reason. Even so it is possible to encounter difficulty. It can help to place a small capacity hat at the top of the 80 meter vertical to reduce the voltage at the relay. This can be as simple as a 1 meter long rod centred on the vertical element.We don't want to load the vertical. just move the high voltage pont from the vertical apex to the ends of the rods.

The other consideration is the wires carrying power to the relay. Run this inside the tower, with spacers to get some separation from the metal structure, to reduce its mutual coupling to the vertical. Chokes may also be required at both ends of the cable. Do not use the tower as a DC return path; always use a separate wire pair.

Trap

This seems to be the simplest and most convenient solution. After all, no extra switching of any sort is required to operate on both 80 and 160 when the array is in omni-directional mode. Unfortunately it doesn't work as well as one might hope.

Traps are not perfect filters that block some frequencies and let other pass. No matter how they are tuned traps introduce reactance. Typically this requires lengthening of the element inside the traps (on the band the trap is designed to block) and shortening of the element outside the traps (the lower frequency band that passes through the trap). Worse, the reactance changes with frequency and therefore increases antenna Q on 80 meters, and may do so on 160 meters.

That is what I found. Whether the trap is tuned above, within or below the 80 meter band (3.5 to 3.8 MHz in this particular antenna) or the L/C ratio the SWR band width is reduced.

The 80 meter portion of the vertical must be lengthened ~2 meters. Alternatively an L-network can be inserted in omni-directional mode and the existing L-network for directional mode can be adjusted. Gain and F/B of the 80 meter array are little affected by the use of a trap: gain reduction is approximately -0.1 db and F/B is 2 to 3 db better at the low end of the segment.

The model view of the antenna shown above is for this alternative construction. Notice that on 80 meters significant current flows beyond the trap. This is by design.

SWR of the trap 80 meter vertical array in omni-directional mode, of length 23 meters (to the trap)

By careful choice of trap the loading coil for 160 can be eliminated. For example, tuning the trap for 3.9 MHz (L=8.2 μH; C=240 pf, X=200 Ω). I found these values entirely by accident during modelling. The trap exhibits inductive reactance on frequencies below trap resonance, which is typical for traps.

When in directional mode on 80 meters the L-network must be changed to accommodate the change in electrical length of the antenna due to the trap. The revised values are a series coil of 2.1 μH and a 830 pf capacitance shunt across the transmission line port. In practice it will be necessary to design the network after construction by measuring the impedance to be transformed.

The resulting SWR in directional mode is barely acceptable (compare to the SWR in the earlier article). Low SWR cannot be readily achieved when a trap is used.

Parallel wire

This approach is similar to the top relay option but moves the relay to the bottom. This placement is not only more convenient a conventional relay can be used since the voltage is low.

Of concern is mutual coupling between the parallel wire and the 80 meter vertical. The wire is switched at the feed point, run parallel to the tower and attached to the mast extension. Additional loading can be added as required.

In the model the wire is placed 1 meter from the tower axis, running from 1 meter height to 20 meters. From there it angles upward to connect to the bottom of the mast extension. At the bottom the wire runs tangent to the tower for 1 meter then straight down to ground through a 10 Ω resistor. The resistance is the estimated equivalent ground resistance (MiniNEC ground). The model doesn't allow the bottom wire to enter the tower area (to the switch box), which is why it is run at a tangent. The model accuracy impact of doing this is negligible.

The model view with current plot is for 1.85 MHz. Notice that there is some current induced on the tower even though it is λ/8 on 160. There is no difference whether the tower remains in-circuit or is floating. The parasitic elements carry negligible current, which is ideal. Even so it is likely best to float the tower and all parasites to simplify the switching matrix.

To my surprise there was no need for a loading coil between wire and extension mast; it resonated very nicely as is. However the radiation resistance is a little low, lower than the case for switch option, probably due to the tower coupling. An L-network is advisable for maximum low SWR band width.

The current flowing out of the main array volume
reduces directional mode performance on 80 meters
Unfortunately we now run into difficulty. On 80 meters the presence of the 160 meter wire severely affects operation, in both omni-directional and directional modes. The 160 wire, floated or grounded, combined with the extension mast and top hat carries significant current. Most of the problem is due to the amount of current carried high above the 80 meter array which reduces directional performance (gain and F/B). A change in the L-network easily tames the driven element impedance.

A relay can be used at the top of the wire to break the resonance, with the same requirement for a vacuum relay as with the top relay option. Opening the relay and floating the wire bottom breaks the resonance and restores 80 meter performance. The L-network will still require some adjustment to compensate for wire coupling. Keep that in mind if the 160 meter wire is added after the 80 meter antenna is tuned.

Or...keep it simple

An alternative not mentioned is to forgo use of the extension mast and top hat on 160 and simply use the tower as-is for 160. Obviously it is far from resonance since it is only λ/8 long on that band. An L-network at the base, with or without a loading coil at the base, can match the 20 meter tall tower.

The reasons I am not considering this alternative are twofold:
  • Efficiency: There will be matching network loss and increased ground loss due to the very low radiation resistance of ~7 Ω. Although I am not initially aiming for the ultimate in a vertical for 160 this goes too far in the wrong direction.
  • Exploit the mast: Even without considering 160 meters there is still a metal extension mast in my 80 meter array design so that the wire parasitic elements can be full height rather than top-loaded with diagonal T's. I want to push 80 meter performance rather than compromise.
That's my rationale. Others may want to consider this option, and accept the compromise. It's a matter of choosing what best fits one's circumstances.

Conclusions

For my purposes the trap option must be eliminated due to its negative impact on 80 meter performance. This leaves either the top relay or parallel wire options. Both are similar in that a relay is needed at the top of the tower. Even though the wire tip voltage is lower than in the case of the top relay option a vacuum relay is still recommended to better isolate the wire from the extension mast for any induced current on the wire.

My preliminary conclusion is to go with the parallel wire option. It is likely more reliable than a top relay, which must operate at high voltage, and has the same negligible impact on 80 meter performance when the wire is switched out at both top and bottom ends. If a loading coil can indeed be eliminated, all the better.

There is still the matter of suitably isolating the relay wires running up to the top relay so that it doesn't act as part of the antenna, on either band. Since it isn't possible to model the relay wire pair running inside the tower (one conductor within another) I will have to rely on the experience of others and on-site measurements during construction.

No comments:

Post a Comment

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