Resurrecting the Y-Doublet 
After presenting some notes on triangles of doublets of various lengths, I received more than one message recalling an old Y-configuration from the 30s and 40s. The basic scheme was designed for a given band and consisted of three 1/4 wavelength wires at 120-degree angles coming together at a center point. There, according to recollections, the old timers used a 3-wire twisted feedline to the shack. At any one time, the operator hooked up two of the 3 wires to the antenna tuner (or, in more remote past times, to the rig output terminals). The result was a steerable doublet.
Essentially, the operator was selecting the pair of feed wires that created a doublet, with the third antenna wire relatively inert. We normally think of a doublet as linear, but bending it by 30 degrees does not especially harm its performance. So that part of the system is quite sound.
More foreign to current practice is the twisted feedline. In the 1930s, many hams commonly feed their dipoles with low-impedance parallel line. A 72-Ohm transmitting parallel line used to be available, but apparently the high power version is no longer made. 72-Ohm parallel lines made from round wires are not feasible with open-wire construction, since the required center-to-center spacing would require contact between the wires. However, by using a carefully calculated thickness of an insulating material with a known dielectric constant on each wire, the desired impedance is achievable.
Amateur practice tended to rely on two factors. First, the resulting parallel line resembled ordinary line cord, sometimes called zip cord. Second, properly configured antenna tuners or even amplifier output circuits were capable of handling a fairly wide range on impedances. Therefore, amateurs used to simply twist pairs of insulated wires together to form a low impedance parallel feedline. The wires might be line cord or they might be other insulated wires twisted and taped together.
Adding a third wire to the set and leaving it disconnected from the RF source was relatively harmless. If the wire was equally spaced from the other two hot wires, it would have negligible current on it. The antenna wire would be at essentially right angles to the main pattern and hence induce a minimum of coupled antenna current into the inert feeder. Since the currents on the other two feeder wires would be equal in magnitude but opposite in phase, any induced currents in the third wire would cancel, leaving no current in the third wire.
Overall, the system effects a space savings over three doublets in a triangle. With good solid AWG #12 wire for the elements, one can use only three corner supports and let the triangle of antenna wires support the center assembly. (One can always add a center support, if convenient or necessary.) With that promise and the potential for having a steerable doublet, the idea is worth further exploration.
We shall use as our test array a Y cut for 3.6 MHz. My free-space model used 67' legs for initial checks. Hence, ignoring the necessary insulating end ropes to the support trees or posts, we get a triangle about 116' on a side and capable of fitting within a rectangular back yard that is about 101' by 116'. The figure shows the three feed wires, of which we shall use only two at a time. For modeling, that means terminating each leg short of the exact center point. Then we connect a short wire between 2 of the 3 wires. I used a separation between inner leg ends of about 3' so that I could use a 3-segment wire for the source and use segment lengths of about 1' on the antenna wire legs.
Fig. 2 shows the overlaid free-space patterns for the Y-doublet at 3.6 MHz using different pairs of legs to form each of 3 doublets. The patterns indeed promise full horizon coverage as we switch pairs of feedlines at the shack end of the feeder lines.
I also checked the antenna's performance on each of the bands above 80 meters. All of the traditional ham bands (40, 20, 15, and 10 meters) yielded very high feedpoint impedances. Since we are working with a low-impedance feedline, I set these bands aside as not especially feasible for use with the system. (We shall review this decision before we are finished.) However, 30, 17, and 12 meters showed feedpoint impedances sufficiently low to potentially allow use of the antenna on these bands using the low-impedance feeder system employed in first half of the 20th century. The following table shows the free space performance potential of the array.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Y-Doublet Modeled Performance: Free Space
Freq. Gain Feed Z
MHz dBi R+/-jX Ohms
3.6 1.70 59 - j 6
10.125 5.16 106 - j 375
18.118 4.62 134 - j 103
24.94 4.92 171 - j 291
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The free-space patterns are generally only applicable for a real horizontal antenna over ground if the height is at least 1 wavelength. 80-meter doublets at 270' or more are rare. Therefore, I remodeled the antenna at a 50' height to reflect a more realistic scenario. At that height, the maximum gain of the antenna has an elevation angle that is nearly straight up. So I chose for that band an angle of 34 degrees to reflect typical skip angles. The resulting 3.6-MHz patterns, shown in Fig. 3, are a good bit more oval than their free-pace counterparts.
On the upper bands, I used the take-off (TO) angle for gathering potential performance data. The antenna promises performance as shown in the following table, with the leg-length adjusted to 66.5' to bring the array close to resonance at 3.6 MHz. (Wire doublets tend to vary their feedpoint impedances with height in noticeable ways when the doublet is less than 1 wavelength above ground.)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Y-Doublet Modeled Performance: 50' Above Average Ground
Freq. Gain TO Angle Feed Z
MHz dBi degrees R+/-jX Ohms
3.6 3.43 34* 56 + j 8
10.125 9.98 26 110 - j 424
18.118 9.59 14 135 - j 153
24.94 10.23 11 182 - j 365
* 80-meter elevation angle arbitrary.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The patterns on the upper bands are not ovals by any means. Fig. 4 shows these patterns, but only one pattern per band for clarity. As we increase frequency, we find two especially interesting pattern properties. First, as the legs become longer in terms of wavelengths, the patterns develop growing side "wings." Eventually, by 12 meters, the main lobe has split into two forward lobes. Second, as we increase frequency, the array becomes more directional, with a growing differential between the forward and the rearward gain.
Still, the patterns may be usable for general amateur operations. The question left is why we get reasonably low impedances at 30, 17, and 12 meters. Fig. 5 shows part of the reason.
The graphics display the relative current magnitude distribution along the doublet for each of the 4 bands. On 80 meters, we have a somewhat typical dipole current distribution, with the current peak at the feedpoint. On the other bands, we approximate a 3/2-, 5/2-, and 7/2 wavelength doublet current distribution. Each of these configurations places a current peak at the doublet center, resulting in a relatively low feedpoint impedance.
We should also note that the unused wire shows a flat current line. The current on it models out (in its perfectly spaced geometry) at about 4 orders of magnitude less current than on the active wires. That is, if the current at the source is 1.0, then the current on the inert wire shows a value of 0.0001 or 1E-4 or less.
There are several questions about the feasibility of using such a system in modern times. The first quandary is whether we can build such a feeder system.
Modern insulated wire tends to use higher quality (lower loss) insulation than did the wire of yore. I would steer away from line cord, but modern wires use plastics with better RF characteristics, even if the only intend use is carrying DC. Since the system is designed for a low characteristic impedance, but with considerable SWR on the higher bands, I would recommend a heavy gauge wire, perhaps #12 or so. The actual characteristic impedance will depend on the thickness of the wire, the dielectric constant of the insulation material, and how tightly we hold the wires together. Consequently, I can give no exact figures.
However, you can make up lengths of a proposed feedline and check the impedance in a number of ways with a variety of dummy loads and a low-level signal source. Any one of the current crop of antenna analyzers will give you a fairly accurate reading. Given the relatively high dielectric constant of the insulating material, expect to find a significant velocity factor, something in the 0.6 to 0.7 region.
The next inquiry has to do with the effective inertness of the unused 3rd feeder wire. I re-created the model of the Y-doublet using parallel feedlines. Since twining the leads is not feasible in a physical model, I simply dropped the three leads straight down from the 50' level to 1' above ground. At that point, I connected two of the feeder ends with a 3-segment source wire. Again, all wires used a 1' segment length.
The resulting feedpoint impedances are not accurate to the low-Z feeder system. However, that was not the point of the tests using the feeders with something over 800 Ohms as the characteristic impedance. The question was whether the unused antenna and feeder wires would remain inert relative to the active wires.
As one measure, the following performance table shows the effects of the added copper losses of the physically modeled feedlines.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Y-Doublet Modeled Performance: 50' With Feeders
Freq. Gain TO Angle
MHz dBi degrees
3.6 3.35 34
10.125 9.74 26
18.118 9.60 14
24.94 9.85 11
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Gain remains virtually unchanged. So, too, do the patterns, and the outlines shown in Fig. 3 and Fig. 4 remain valid for the reconfigured model.
A second test is to check the current distribution along both the unused antenna wire and the ostensibly inert feeder. I actually performed two tests, one with the unused feeder simply left open and another with the feeder extended 1 foot to touch the ground. The 12-meter current distribution graphic in Fig. 7 remains valid for both.
Note that the current line on the unused feeder and antenna wires is flat. The relative current magnitude under either test condition on all of the bands remained less than 1E-4 relative to a source current of 1.0.
The modeling test, of course, has limitations relative to an actual twisted trio of wires. In the test, the modeled wires are widely separated and perfectly spaced along the entire 49' feeder run. How well the twisted trio performs may turn out to be as much a careful-construction issue as any other kind of issue.
However, the tests suggest an alternative feed system that just might open up the Y-doublet to use on all of the HF bands. Fig. 8 tells the story.
The Y-doublet on the traditional upper ham bands, 40 through 10 meters, can show feedpoint impedances in the thousands of Ohms, with considerable reactance. Indeed, shrinking or expanding the basic 80-meter legs may prove useful in reducing the high reactance levels that accompany lengths that are close to even numbers of half wavelengths. Commonly, we try to select for a doublet a feedline characteristic impedance that is about the geometric mean between the feedpoint impedance extremes that we are likely to encounter. There is a practical limit to this effort, since lines above 600-800 Ohms are difficult to produce. Hence, 600-Ohm or so open wire becomes typical for such applications.
We can create a trio of pairs by using circular spacers of the type shown in the figure. For HF work, Plexiglas or polycarbonate spacers should be satisfactory. We can cut a hole in the center of each to reduce the weight. The holes can actually be slots if we add bridge wires to hold the spacers in place. In essence, we are adapting techniques normally used to create caged elements and applying them to the feedline. Such lines might permit the use of the antenna on all bands with a wide-range antenna tuner and will go a long distance in maintaining something close to the modeled ideal geometry we used in the test cases.
Of course, should you choose to work with a system of this order, you can replace the alligator and crocodile clips of yore with an in-shack switching system to change the orientation of the pattern on all bands.
The resurrected Y-doublet as some potential of still being serviceable today. There are many variables beyond the limits of this initial feasibility check, so success is not assured. However, for some hams who are restricted to backyard wire systems but who wish some directional flexibility, the system may be worth a try.
With the wide spaced feeder system, the system may also be adaptable to 102', 88', 67', and 44' doublet lengths discussed in other notes at this site and in mountains of other literature. However, as with all horizontal doublets, the rule of thumb that calls for the maximum feasible height remains in play for effective operation.
Updated 05-16-2003. © 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