Some ABCs of A-B-C: Notes on Triangles of Doublets 
On occasion, folks have asked me questions of this sort:
My all-band doublet doesn't seem to be doing the job. So what if I replace it with a longer one, a loop, or whatever. Or suppose that I add an antenna of a different kind to the existing doublet. What should I do?
In many cases, I recommend a second or third antenna of the same type. It struck me that perhaps some short background on why I make that recommendation on some occasions--but not all--might be useful to those who do not have much experience with all-band wire doublets.
A difference of height will make little or no difference in the azimuth patterns of a doublet as we move from band-to-band (as long as the antenna is not too close to the ground). The pattern is mostly a function of the antenna length in terms of wavelength at each operating frequency. Small changes of frequency do not materially affect the pattern, so we can use a single frequency on each band as a sample that holds true for the whole band.
Let's start with a 135' center-fed doublet. 135' is about (and "about" is plenty good enough here) 1/2 wavelength long on the 80-75-meter band. At 40 meters it is 1 wavelength. On 20, it is 2 wavelengths, and on 10 it is 4 wavelengths.
Any center-fed doublet will have only 2 lobes for any length in wavelengths up to and just beyond 1 wavelength--1 lobe on each side of the wire, broadside to the wire. When the antenna is 2 wavelengths long, there will be 4 lobes--2 on each side of the wires--and they will angle away from the wire leaving a null directly broadside to the wire. A 4 wavelength antenna will have 8 lobes--4 on each side--and the strongest ones will be angled further away from the broadside directions.
Fig. 1 provides snapshots of the azimuth patterns that reflect the notes I just gave. In addition, it shows the 15-meter situation. The antenna is 3 wavelengths long, so we get 6 lobes. For all of the patterns, the antenna runs from left to right (or right to left) across the center line of the plots. So broadside is up and down on the patterns.
The first thing we can do with these patterns is explain why a given doublet gives good results in certain directions on some bands but not on others. Assume that we set the wire in the U.S. so that broadside goes to Europe and to Australia. By the time we operate on 15 to 10 meters, our strongest lobes are no where near the headings for those two major target areas.
So our first lesson is in wire antenna orientation. By knowing the antenna length, we can roughly determine the directions of lobes on our favorite bands and set up the wire to give the strongest performance in those directions by how we orient the wire. We cannot obtain a perfect setting on all bands, but we can (assuming that we do not need to move any supporting trees) obtain good settings on our favorite bands.
Before we look further at the matter of direction, we have a few more preliminaries to note. For example, what band should I choose as the basic one for DX in my favorite directions? Here is where a small performance table may help for the bands illustrated in Fig. 1.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
135' Doublet Performance at 50' Up
Freq. MHz Max Gain dBi TO Angle deg
3.75 3.2 30 (arbitrary)
14.175 9.4 19
21.225 9.8 13
28.5 10.6 9
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The TO or take-off angle is the elevation angle of maximum radiation. It is correct for 20-10 meters, but is not for 80 meters, where the height of the antenna is low enough to direct most radiation at very high angles. So I chose a reasonable short-skip angle to take the gain reading.
Since the gain on 80 is so low and the radiation angle so high, 80 is not a good candidate for use in orienting the antenna for DX directions. The gain on 20, 15, and 10 are similar, so one of those bands is a better candidate. However, if we prefer local rag chewing, then the 80-meter broadside should aim at our target areas. Remember that the 40-meter pattern will also be a 2-lobe broadside affair, but the strength will be higher and the beamwidth narrower (since the added gain has to come from somewhere).
There are, of course, a number of other amateur bands on which the antenna is not close to an integral number of wavelengths long. What happens when the antenna is X.5 wavelengths long, where X is any integer? The answer is a function of how lobes appear. They do not pop into existence, but grow and shrink as we change the length of the antenna (or as we raise and lower the operating frequency, which achieves the same change in antenna length when measured in terms of wavelengths). At X.5 wavelengths, the azimuth pattern will show two sets of lobes in approximately equal strength: the set for wavelength X, longer than which we now are, and the set for wavelength X+1, which we are approaching.
On 17 meters, a wavelength is about 54' long and our 135' doublet is about 2.5 wavelengths long (give or take a little). (A wavelength is about 984/f feet long, where f is the frequency in MHz.) We shall have 4 lobes for the 2 wavelengths that we passed in length. We shall have 6 lobes for the 3 wavelengths that we are approaching. So the pattern will be composed of 10 lobes total. That sounds good, since we get lobes in so many different directions. However, for every lobe, there is also a null. So we have 10 blank directions and each lobe is narrower than its counterpart in a 2 wavelength or a 3 wavelength wire.
This note on nulls gives us the second lesson concerning wire doublets and their azimuth patterns. Study the azimuth patterns twice. First, look at the lobes that tell you where the radiation is going. Second study the nulls that tell you where performance will be very weak. Only then should you make decisions about how to orient the antenna.
What happens when a wire is X.25 or X.75 wavelengths long. Since lobes grow and shrink, the answer is almost obvious. At X.25 wavelengths, the wire is not long enough for strong lobes from the X+1 length, but will show some smaller lobes derived from that length. At X.75 wavelengths, the wire is too long to support full scale lobes from the X length, but will have smaller lobes (meaning lower gain) still present. In either case, as some lobes show low development, the larger ones are that much larger, since the radiated power is relatively constant (ignoring external variables) across all frequencies.
Now suppose that you have studied the azimuth patterns and decide that you cannot place a strong lobe every where that you want to communicate. It is precisely here that folks immediate jump to thoughts of other antennas. But virtually every other horizontal wire antenna, whether straight or looped, has a pattern of lobes and nulls. Most of them are more complex to install. So what is a solution to our quandary?
Now wire is relatively cheap. Antenna supports are not. So any solution that we come up with should have the desirable property of involving a minimum of new supports.
The drawing shows an equilateral triangle, but almost any shape will do. In fact, a better ideal is to angle each antenna so that its lobes--on your favorite bands--hit your favorite targets. Just give yourself a little separation at the wire ends--perhaps 10' on the end of a 135' doublet--to minimize interactions among the wires.
The sketch also shows two different means of feeding the antenna. You can run 3 feedlines, each the same length as the others, to a central point where you install a weather-proof relay box to switch among the wires. A single feedline runs to the shack, along with a relay power line and switching lines via an A-B-C switch to activate one of the 3 antennas at a time.
The alternative is to run three lines, again, all the same length, to the shack for use with a manual A-B-C switch at that location. Since the lines will be parallel transmission lines, follow the usual precautions about keeping them free and clear of anything that might disrupt their balance.
The point of using equal-length lines in each case is so that you can switch between antennas and determine by ear the strongest signal. If they are not the same length, you will have to do some rapid re-adjustment of the tuner settings for each switch position. With identical antennas and feedlines, you should be able to pick out the strongest one and then only do a final tweaking of the settings on the tuner.
As Fig. 3 shows, you may not always have a good choice as to exactly where you position the supports for the antenna. Fitting the triangle to available yard space is an eternal amateur antenna problem. However, as we shall see, we can effect some improvement on our operation, even if we cannot perfect it.
Notice that we were able to add two antennas for the cost of wire and with only one extra support. We may well trade any remaining imperfections in the system for that major simplification of structure.
Now the question is simply this: what do we get for our pains? The easiest way to show what we get is by overlaying azimuth patterns for each antenna in one massive plot for each band covered in Fig. 1. We shall note both the advantages that we accrue from the new arrangement and the remaining problems that we could not solve.
Fig. 4 shows the three antenna patterns at the same 30-degree angle that we used in Fig. 1. It is immediately apparent that we can cover more of the horizon with our signal (and reception) than with a single antenna. However, since the antenna is close to 1/2 wavelength, there is some interaction between the antennas so that the inactive ones act as reflectors. The 2-dB difference should not affect performance too much. You will find the same phenomenon on 40-meters, where the wires are all nearly 1 wavelength.
The 20-meter situation appears in Fig. 5. We still have some nulls, but count the major directions that we can cover, letting overlapping lobes count as 1. We have 6 directions, not just the 4 that a single wire would give us. As well, the overlaps are not perfect, so that a signal that is on the fringe of one lobe may be centered in overlapping one. Of course, by carefully planning of your triangle, you can minimize the overlap and spread the area of coverage. The patterns shown simply use our equilateral triangle as their basis.
In Fig. 6, we find 15-meter quite well covered by the 3 antennas and their strongest lobes. Indeed, 15 meters is a band that really benefits from a triangle of 135' doublets.
If you look at Fig. 5 and Fig. 7, you may get the impression that when an antenna is an even number of wavelengths, it leaves more nulls than when it is an odd number of wavelengths, as in the 15-meter case. In general, this is a correct conclusion, although as we further increase the antenna length, the number of lobes becomes high enough to make it difficult to tell the difference. 10-meters is a good band for which to redesign the triangle to place a lobe in the direction that you want it.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
88' Doublet Performance at 50' Up
Freq. MHz Max Gain dBi TO Angle deg
3.75 2.8 30 (arbitrary)
14.175 10.1 19
21.225 9.1 13
28.5 10.3 9
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The performance figures for each band are not very much different than for the 135' doublet, but the patterns are considerably different. Remember that the 88' doublet is only about 1/3 wavelength at 3.75 MHz. Fig. 8 shows the 88' doublet azimuth patterns when we place the antenna 50' above ground.
The 80-meter pattern is similar to the one for the longer wire. However, the 20-meter pattern shows the typical "ears" of an extended double Zepp, since that is exactly what the antenna is at 14 MHz. It is 1.25 wavelengths, which means that the 2 wavelength lobes are just beginning to emerge. On 15 meters, the wire is 2 wavelengths long and shows the same sort of pattern that the 135' doublet showed on 20 meters. The 10-meter 88' doublet pattern is an example of a 10-lobe pattern for a 2.5 wavelength antenna.
Besides taking less space, the triangle of 88' doublets also shows less interaction among the wires. Hence, we can use somewhat smaller separations of the wire ends in making the triangle. However, in exchange for spatial economy, we shall encounter differences in the ability of the triangle to fill in the nulls on the single-wire patterns.
On 80 meters, as shown in Fig. 9, the absence of pattern distortions created by interactions among the wires yields almost complete horizon coverage. However, remember that this pattern is at an elevation angle of 30 degrees, and most radiation is upward. We can improve long-haul performance of the triangle by "merely" raising the supports to the 90-100 foot level.
Fig. 10 shows us the 20-meter combined patterns. The azimuth plot provides a good model for a hex symbol to embroider for good luck. More importantly, it shows some nulls that may call for careful design of the triangle to ensure the desired target-area coverage.
Like its counterpart 20-meter pattern, the 15-meter patterns in Fig. 11 add up to fairly complete coverage, but with nulls and overlapping lobes. Hence, one might wish to design the triangle to spread the lobes a bit. However, you will discover that with 4 lobes per wire, every spread in one direction increases an overlap somewhere else.
The situation grows both better and worse on 10 meters, as shown in Fig. 12. The null areas are wider, but often not as deep, since minor 10-meter lobes fill the null at about 1.5 S-units lower strength. Once more, designer triangle formation seems the order of the day.
It would be difficult in a general discussion to provide samples of designer triangles, since each would prove useful for only one region of the U.S.--and likely be useless outside the U.S. However, one can experiment most easily with altering the triangle orientations with modeling software. There are inexpensive packages, and even some free MININEC programs. It may pay to master them well enough to go with your self-study geography lessons in order to give yourself the bast chance of placing doublet supports at the correct locations.
In the end, there are no perfect solutions. However, at a cost of one extra support and a bale of wire--plus feedlines and an A-B-C switch--you cannot get much better horizontal coverage much more cheaply than with a triangle of doublets.
An additional caution or two: The doublet lengths used here place the lowest band at 80 meters. You can use a 70' doublet if you wish to cover 40-10 meters. In that case, the 80-meter and the 20-meter patterns shown in Fig. 1 become the patterns for 40 and 10 meters, respectively. As well, you can use a 44' doublet in place of the 88' version used in these notes. The same adjustments then apply to the patterns in Fig. 8. In both cases, the performance data in the tables would apply with band adjustments to an antenna at about 25' above ground.
The first thing that happens is a requirement for either 1 or 2 more support posts. If you have plenty of Douglas Firs handy, a 4-post system is no problem. But if you have to construct or erect your own supports, then the support work either matches or exceeds the work required by a triangle. The L-configuration is like the triangle, but only lacks 1 wire and its associate feedline. Of course, you can now get away with an A-B switch, rather than having to figure out how to make an A-B-C switch.
Let's put up 2 135' doublets that cross at the center (with a separation to keep the wires apart) and see what we can achieve with only 2 antennas on each of our sampling bands.
Fig. 14 shows the 3.75-MHz results. Since the deepest null is now only about -3 dB or about 1/2 S-unit, it is likely that performance will be satisfactory within the height limitations that we discussed earlier.
On 20 meters, as shown in Fig. 15, we have deep nulls and considerable lobe overlapping. Hence, for this band, adjusting one antenna by at least 25-30 degrees off a right angle will likely produce better coverage. For an individual antenna, the lobes are only about 35 degrees each side of a broadside tangent line relative to the wire, so some angling to enhance coverage seems in order.
Fig. 16 gives us the 15-meter story. Once more, the main lobes heavily overlap, but each is about 40 degrees off the line tangential to the antenna wire. Finding a compromise angle for both 20 and 15 meters will require some thought, especially when we add in the need to be aiming at communications target areas.
The most thorough coverage occurs on 10 meters, due to the multiplicity of lobes. See Fig. 17. Despite the gaps or nulls that remain on each of the bands, the level of coverage with just two 135' doublets is significantly greater than with a single doublet. Conclusion: if you cannot swing 3 antennas, at least try for two.
The 88' doublet does not fare quite as well in a 2-wire system as the 135' doublet. Indeed, the 88' doublet seems best suited to a triangular environment.
80 meters appears in Fig. 18. As is evident, there is no significant difference between the 135' and 88' 80-meter situation with respect to coverage.
The 20-meter patterns are in Fig. 19. The extended double Zepp patterns simply give us two different bi-directional options. Hence, careful broadside aiming of the wires seems the order of the day.
If we wish to target areas on 20 meters, they will be broadside to the antenna. However, on 15 meters, as shown in Fig. 20, we cannot target the same areas, since the lobes on that band angle away from broadside by about 35 degrees. Unlike the situation with a triangle of doublets, the 2-wire system of 88' doublets appears to force us to declare that either 20 or 15 meters is our favorite band, but not both--at least not into the same parts of the world.
Fig. 21 gives us the 10-meter picture. At 2.5 wavelengths, the antenna yields fairly solid coverage, although there likely is room for wire aiming on this band.
However we construct the doublets, the triangle provides superior coverage and more versatility than a crossed or L-ed doublet system. Obviously, 2 wires are better than 1, but 3 is significantly better than 2 without requiring any further supports.
I have omitted construction details, since they are so variable with the circumstances of the individual builder. As well, I have omitted inverted Vees, which will tend to change the situation on the lower bands more than on the higher, since the patterns will be broader ovals.
Nevertheless, the switched doublet triangle offers flexibility that takes advantage of the strongest lobes--especially on the upper bands--in the antenna pattern. A switched-doublet system is cheaper than a rotator and tower system, and repair costs are reduced usually to the cost of antenna wire and possibly some parallel transmission line. However, the performance of a wire--when you can place one of its main lobes on the desired station--can be surprisingly good.
Nothing here is intended to compare the doublet with other wire antennas--or even non-wire antennas. These notes are intended only to show some possibilities that we often overlook when thinking about wire doublets.
A love triangle is usually a disaster. However, a triangle of doublets is often a happy marriage of economy and performance.
Updated 05-15-2003. © L. B. Cebik, W4RNL. Data may be used for personal purposes, but may not be reproduced for publication in print or any other medium without permission of the author.
Go to Amateur Radio Page