There are many ways to design a multi-band antenna. Each has its good and bad points, and all have been discussed on this blog multiple times. Common deficiencies of multi-band antennas are SWR bandwidth on the lowest bands, loss in the loading elements and pattern peculiarities on one or more bands (multiple lobes and nulls).
- Fans (parallel elements from a common feed point)
- Adjustable matching network, in the shack or at the feed point
- Frequency sensitive transmission line sections
- Motorized, adjustable length elements
All can be complex to home brew due to finicky design and construction. It is no surprise that most hams that need or want a multi-band antenna choose a commercial product in which all the complexities have, hopefully, been solved by someone else.
Perfection is impossible so tradeoffs are necessary. You choose an antenna where the tradeoffs are acceptable for your operating interests and what fits within your property and support structures. Many hams make their choice by price and size. Performance claims are either believed, ignored or deemed acceptable. Yet it is possible to avoid many of the deficits of multi-band antennas. That requires careful design, an understanding of antenna and network theory and practice, and test equipment.
It has been quite a long time since I last wrote an article about antenna design. I most often write about what I'm doing, of which there's an awful lot, and for the past year I spent little time designing antennas. I was recently motivated to investigate a multi-band challenge I have in my station. Although you may not have the same type of antenna in mind, the following discussion may be useful.
One of my objectives is to make the 160 meter mode of my 3-element 80 meter vertical yagi more effective without compromising 80 meter performance. Currently I switch in a loading coil and L-network at the antenna base. On air testing suggests a deficit of -6 to -7 db compared to the 160 meter shunt fed tower.
The big shunt-fed tower, which is an excellent top band antenna, is not available year round because I must roll up the radials for several months in spring and summer while the hay is growing and then harvested. Burying the radials is an option that I am unlikely to attempt this year, if at all. It isn't an easy project.
Alternatives that I considered long before I built the 80 meter array were impractical when it came time to build them. Before going further, consider skimming that article since it introduces several issues discussed in this one.
I am planning changes to the 80 meter array that will make it mechanically easier to improve its 160 meter performance, and prepare for 80 meter yagi performance improvements. Both depend on replacing the tower with a taller one. That will allow me to get rid of the long and problematic stinger.
A new stinger at the top of the taller tower would only be used for 160 meters. The tower itself, and probably with the help of a short "tuning" pipe at the top, will be a resonant ¼λ on 80 meters. Modest loading at the base of the stinger will resonate it on 160 meters. The base coil will be removed and the L-network redesigned for the measured impedance.
With an acre of radials (almost 2000 meters of wire) under the 5 vertical elements of the 80 meter yagi, the 160 mode of the array will be far more effective with negligible impact on the 80 meter array. Because it will be shorter than a ¼λ on 160 the load cannot be loss free, the
new design will likely be 2 db less effective than the shunt-fed
tower, but that's a notable improvement over the present design. That will let me be competitive on top band during the summer when the shunt fed tower is unavailable.
The drawings at right show just the essential aspects of the existing (left) and proposed (right) construction. Since the new stinger cannot extend 20 meters above the tower, loading will still be required on 160 meters. Support ropes for the parasitic wire elements would attach to the top of the taller tower rather than the top of the stinger, as in currently the case. That allows the stinger to be lighter duty. If it is 8 to 10 meters long, the 160 meter vertical will be about ⅜λ, so the loading coil can be small and low loss.
The stinger will have to be switched by relay. Use of a trap or parallel (fan) vertical have been discarded due their negative effects on 80 meter yagi performance and narrow bandwidth on both 80 and 160 meters. A switched stinger is far superior in this regard. The primary challenge is with the switching between 80 and 160 meters. 80 meter impacts must also be quantified.
The design can be done in the comfort of my shack by computer modelling. My tool of choice is EZNEC and its version of the NEC2 engine. I began by building a simple model to investigate switching methods, and refining that until I had a workable design.
This article focusses on the simple model to investigate switching behaviour. After the tower is replaced and I can take measurements of the new array, I will refine the design.
Modelling the relay
Relays are not perfect devices. The coil requires wires for power, wires to the contacts have inductance, closed contacts have resistance and there is capacitance between open contacts. When the relay contacts are open, as they are in this application for operation on 80 meters, the voltage across the contacts can be very high, well over 1000 volts for legal limit power. This is a case where my relay phobia may be justified.
The EZNEC model ignores the wires powering the coil, assuming they are suitably routed and choked to isolate them from the high adjacent RF field. We'll return to these challenges later.
You may have to expand the picture to read the tables. There are wires for the 80 meter vertical, the 160 meter stinger and a short connecting wire containing an RLC load. The load is modelled as a pure capacitance. A small value for open relay contacts (80 meters) and a short for closed contacts (160 meters).
Relay spec sheets may or may not show the capacitance for open contacts as measured at the terminals. The capacitance depends on relay construction, comprised of that between the contacts and the wires to the relay terminals, and the housing and other conductors if they are significant. The reactance decreases with frequency so that "leakage" is greater on the higher bands. In this instance the capacitance is only relevant on 80 meters where the contacts are open.
The SWR curves were drawn with a load capacitance of 0.01 pF. Up to 2 pf the R and X components change by no more than 2 to 3 Ω. That is negligible. With a good radial system (which I have) the impedance is low enough that a matching network may be helpful. For a poor radial system the series ground loss would improve the match at 50 Ω without need for a matching network.
I adjusted the wire lengths (all have a 40 mm diameter for simplicity in the initial models) to resonate the 80 and 160 modes where I want them. Relay lead inductance is ignored but they are almost negligible for these long wavelengths and are easy to compensate with length adjustments in the built antenna.
Conducted 80 meter current in the 160 meter stinger peaks in the centre of its length. For 1 pf of relay capacitance the peak current is 6% of that at the base of the vertical. It rises to 7% for 2 pf. That's comfortably small but may impact F/B when the array is operated in its directional modes on 80. I am deferring the exploration of that interaction to a later time.
Differences between the SWR curves and gain on both bands are negligible when compared for single-band verticals; that is, without the relay, and the stinger removed from the model on 80 meters. That is what I expected. So far so good.
You don't often see antennas like this, using a relay to switch bands. Traps are far more common in this application despite their inherent loss and increasing the antenna Q on both bands. A relay has neither of those disadvantages. This is readily apparent in the EZNEC model's load data with the relay open for operation on 80 meters.
On any antenna with an open end -- which is almost all antennas other than closed loops -- the current at the ends of elements falls to almost zero and the voltage is high. Should you attempt to feed the antenna at one of these locations -- such as an EFHW (end fed half wave) -- the impedance is very high. Matching it to 50 Ω can be done, at the expense of transformer loss and difficult to control common mode current. In our case, it is only the voltage that needs to be tamed.
Relays exist that can withstand over 1500 volts of RF but they are not ones you commonly encounter. Contact flash over voltage is misleading since the spec is typically for DC or the low frequency AC found in power systems. The coil, insulators and conductors behave differently with RF flowing across the contacts. When the contacts are open, some RF current will flow due to stray capacitance or due to humid or polluted air.
The resistance can be higher than expected when the contacts are closed, and the capacitance higher when the contacts are open. Further, the actual voltage could be higher than in the model due to voltage modelled in the wire segment rather than at the wire tip, and various environment and construction details. I would at least double the relay's voltage breakdown spec to be
safe. A properly rated relay can still be destroyed by accidental hot switching. Luckily that's unlikely since a relay for changing bands is only operated when the transmitter is idle.
There are two classes of high RF voltage mitigation: use a relay designed for the application, or design the antenna so as to reduce the voltage where the relay is placed.
The preferred choice for applications like this is a vacuum relay. They are available with breakdown voltage ratings starting at 2 kV and going much higher. Unfortunately they are expensive: starting at well over $100. There is a good market for surplus and used vacuum relays to limit the expense, if you can find those with suitable specs.
There are other considerations: they can be fragile, difficult to mount and protect on a tower, detailed specs may be difficult to locate (e.g. capacitance for power and signal relays), and have inconvenient coil voltage (24 to 28 VDC is most common on the used market). Choose carefully.
There are ways to reduce the stress on the relay to reduce the risk of failure, and in some cases it may be possible to use a conventional open-frame or sealed relay. These are the ones I modelled, and all work, though not necessarily very well:
- Capacitance hat below the relay
- Leakage capacitor across the relay contacts
- Large diameter wire (tower) below the relay
The models I developed to test these methods are solely intended for the purpose of exploration. I will make no recommendations or provide dimensions for a real antenna. NEC2 and pretty well all modelling engines are not highly reliable with respect to voltage, current and impedance at the open ends of wires. Real antennas and relay terminal voltages will never exactly match the models. But they can come close, and that makes the modelling experiment worthwhile.
The capacitance hat option (left diagram) was a disappointment. The voltage across the relay only dropped by 25% with two arms that are 5 meters long. For an antenna that is 20 meters long the effect is severe and must be corrected by shortening the vertical quite a lot. That's unacceptable since the shorter length would degrade performance of the 80 meter array. I did not bother to dig deeper to quantify the effect because a 25% voltage reduction isn't enough to eliminate the vacuum relay.
The reason I expected better from the capacitance hat is that it partially mimics a large diameter wire (as in the right diagram) which is known to reduce corona effects found with sharper antenna tips. Instead it behaved as if the voltage was measured inward of an ordinary element or T-top vertical.
To model the "fat" lower wire (right diagram) I increased the diameter of the 40 mm wire to 200 mm (8"). This is quite close to the top (#1) Delhi DMX tower section I am currently using, and will be again when two larger bottom sections are added to the tower. Tower taper can be ignored for the experimental model since we are interested in the voltage at the top of the tower and not its exact height. The model's wire containing the relay (again, a low-value series capacitance load) was made 2 mm (AWG 12), while the 160 meter stinger diameter remained at 40 mm (1.6").
This option is promising. The voltage across the simulated relay contacts dropped from 1500 to 850 volts, which is more than 40%. The resonant frequency on 80 meters barely changed. The voltage reduction is enough to consider using an inexpensive relay with contact and wiring isolation voltage of 2000 volts or more. The model is not definitive since there are factors to be considered in a physical antenna. For example, the relay is likely not close to the tower top plate because a tuning stinger may be required for height adjustment. There are also the effects of humidity, pollution (dirt particles in the air) and precipitation.
Despite the concerns, it may be worth the experiment when the antenna is rebuilt. Flea market open-frame relays are inexpensive enough to risk destroying a few! Since I am designing for high power it is highly recommended to do the experiment using an amplifier with fast-acting fault protection. Luckily I have one of those.
A capacitor across the relay contacts (middle diagram, above) may be an unusual option since until now we've been trying to minimize stray capacitance due to the relay. The trick is to increase the capacitance to pass enough current to cause the voltage to fall to a value where a conventional relay can be used. Reactance decreases with increasing capacitance.
The 80 meter resonant frequency drops since the series capacitor electrically lengthens the anteanna, and that can be a problem. There is no effect on 160 meters because the capacitor is shorted by the closed relay contacts. We need to know how much capacitance is needed to substantially lower the voltage across the relay while avoiding excess lowering of the 80 meter resonant frequency, as we saw for the capacitance hat.
The result is not good. It took 30 pf to reduce the voltage by 25%, and the resonant frequency dropped ~10% to 3.2 MHz. This is very similar to the 5 meter long capacitance hat described above. I suppose that should be expected since a capacitor and a capacitance hat are close relatives. Current in the 160 meter stinger peak at just under 25% of that in the 80 meter wire (tower). Gain of the vertical on 80 meters dropped by -0.1 db, which is negligible. However, the stinger current could be a problem in the array's yagi modes, but I have not run the model as yet.
The only promising mitigation measure is the "fat tower" option; the others have too many deficits. I may play with the model further to see if I can improve it beyond what I did for this article. It would only be for curiosity since I now know enough to proceed.
Whether a vacuum relay or the ordinary kind is placed atop the top, wiring it to the switching system at the base of the tower must be done with care or the wiring itself will modify the antenna's behaviour. The usual way of doing this is to run the cable inside the tower. Skin effect is our friend in since the antenna currents primarily run on the outside of the conductor. It is more complicated with a lattice tower than a solid cylinder (wire or tube) but the current flowing along the inside of the tower should be much lower than on the outside.
Spacers must be used to keep the wires several inches from the tower surfaces. The tower is not grounded so two wires are needed, one being DC ground. RF chokes should be placed on both wires at both ends of the cable and another set midway. That will detune the relay wiring on 80 and 160 meters and keep RF out of the switching electronics. The cable will have to be carefully routed around the top tower plate. Check the relay specs to ensure that the minimum voltage breakdown between the contact wires and the coil wires is at least as good as we need between the relay contacts and wires.
If the stray capacitance of the relay is very low and the loading elements on the stinger keep it far from resonance on 80 meters, there should be little enough current on the stinger to keep it from disturbing the 80 meter yagi modes. This was modelled and discussed earlier, but I have yet to model the full 80 meter yagi to confirm that the stinger current does not degrade the yagi pattern. It probably won't but it would be foolish not to check.
That said, there is reason to add a capacitor across across the relay contacts. We want a small value that does not appreciably load the 80 meter vertical. Perhaps no more than 5 pf. Its purpose is to damp corona effects at the relay that can amplify the voltage across the relay contacts when the humidity is low. A high value resistor should be added in parallel to bleed static charge on the tower and stinger when it rains or snows. Vacuum relays can't do the impossible so we should do what we can to reduce the stress on this valuable device.
The top of the tower can be made wider with a wire cage between the top plate and the base of the relay to lower the voltage across the relay's open contacts. A capacitance hat won't do that, as we've seen, because it is too thin. The cage complicates construction and may not be worth the trouble just to avoid a vacuum relay. The structure would also make working at the top of the tower awkward and possibly dangerous. Out of curiosity I may model it regardless.
Attaching the 160 meter stinger to the tower has its own challenges. It will be quite tall with substantial bending stress at the bottom. There are a couple of ways it can be done: A) inline at the top of the 80 meter vertical's short tuning pipe, or B) bracketed to the tower. The loading coil and relay should be close to the bottom of the stinger. High quality and mechanically strong insulators (red) are needed since, as we know, the voltage can be very high.
I believe the best approach is to bracket the stinger to the tower. The mechanical demands on the insulation are far less than placing it inline, especially for a long stinger. For example, one or two layers of PVC pipe can be placed over the stinger where the tower bracket clamps to it. One layer may be enough if excess voltage due corona and static are managed as see above.
While this article is about my particular antenna, it is applicable to other vertical multi-band antennas of the same design. The technical challenge to incorporate a relay is modest and may be well worth the effort to achieve maximum performance. This can be particularly welcome on the low HF bands where a higher Q method such as traps significantly reduces the SWR bandwidth.
I am undecided whether to begin working on the 80 meter array this year. I am behind with other projects and I am trying to take a rest from major projects this year. The earliest it will happen is this autumn after the insects die off. The project will proceed in stages to avoid finding myself without a good 80 meter antenna when contest season begins in earnest.
The 160 meter change described in this article is more likely to be undertaken in 2024. Replacing the tower and guy anchors must be done first, and that requires taking the tower down and doing a lot of digging and concrete work. After the tower is replaced I can attach the wire elements to the top of the new tower and not have to change any of the existing switching electronics and matching networks.
Unlike the models in this article, the physical stinger height for 160 meters will not be full size (20 meters). That will lower the SWR bandwidth on 160 but not on 80 meters. Loading is irrelevant to the characteristics of the relay and mitigation methods.
One item I'd like to revisit for the blog is a survey of the alternative methods for multi-banding vertical antennas that I listed but did not delve into at the start of the article. It may be worthwhile to compare and contrast them in more detail. I've done all this work in the past but my perspective has changed. Perhaps this winter.