Horizontally Polarized Omni-Directional Antennas: Some Compact Choices

L. B. Cebik, W4RNL

In the last episode, we explored some of the more compact choices for omni-directional, horizontally polarized (HPOD) antennas. Each had some advantages and each had some limitations. In this episode, we shall continue the exploration by examining a few larger arrays using more than 2 independent elements, that is, elements fed in phase. Stacking HPOD antennas is a familiar technique of increasing the gain in all directions, so we shall also spend a little time on that question.

In the first part of this safari, we employed uniform element sizes to all models. The elements in this episode will be a bit more diverse in diameter, ranging from 1/8" to �", since the models are based on prototypes. However, we shall retain the 144.5-MHz design frequency, because on 2 meters, the first MHz is the prime territory for horizontally polarized antennas. It is possible to adapt almost any of the designs for field or hilltop service using locally available materials.

The Big Wheel

An interesting and misunderstood semi-constant-current antenna is the Big Wheel, first published in QST in September, 1961 (see "The Big Wheel on Two" by R. H. Mellen, W1IJD, and C. T. Milner, W1VFY, pp. 42-45). Originally described as three 1-wavelength loops fed in phase, the antenna is actually a complete circle fed by parallel transmission lines at three equidistant points on the circumference. The outline and patterns appear in Figure 1. Between each transmission line, we find a current peak along the circumference, simulating the constant-current loop action. The model for this antenna uses a 3/8"-diameter element with 600-Ohm NEC-TL lines from a central feedpoint. The model has a radius of 17.3" for a circumference of 108.7".

The maximum gain is about 7.3 dBi at 20' above average ground, with less than 0.4-dB variation around the horizon. Note the similarities between the elevation patterns of the constant-current loop (in the preceding episode) and the big wheel. However, the big wheel requires care in construction, because obtaining a usable feedpoint impedance for common coaxial cables involves interrelationships among the element diameter, the element radius, and the characteristic impedance of the connecting transmission lines. The goal is to obtain a pre-match impedance of about 25 - j 25 Ohms, so that the addition of a beta or hairpin transmission line stub provides the impedance transformation to 50 Ohms. Once obtained, however, the SWR should be less than 2:1 across the entire 2-meter band with very good retention of azimuth pattern circularity.

The original big wheel employed an all-tubular construction method that allowed some element warping to arrive at the desired feedpoint impedance. Nevertheless, because the connections to the rim occur at high-impedance points, the parallel or nearly parallel lines to the hub perform an impedance transformation that demands somewhat finicky adjustment. The antenna remains very popular in Europe, but is perhaps nowadays less well known in the U.S. There may be arrays of three elements with equivalent performance, but simpler matching schemes.

The Dipole Triangle and Wheel

One very straightforward array that yields a very circular horizontal azimuth pattern is a combination of three linear dipoles arranged in a triangle. Figure 2 shows the outline of such an array for 144.5 MHz using 0.5"-diameter elements. The success of the array in achieving a true HPOD far-field pattern rests on three factors: the distance of the dipole feedpoints from the assembly hub, the length of each dipole, and the method used to match the triangle to a standard coaxial cable feedline.

The modeled antenna uses a feedpoint-to-hub distance of 15.6", with dipoles that are 34.7" long. The shortness of the dipoles (relative to the 40.8" half-wavelength at the design frequency) does not result simply from the element diameter. Even though the dipole end tips are about 9.6" apart, there is considerable interaction between any one dipole and its two mates. Varying the distance of the dipoles and their individual lengths also varies the distance between element tips. However, for this example, judicious juggling of the variables produced individual dipole feedpoint impedance values very close to 50 Ohms. A 50-Ohm cable to the hub thus performs essentially no impedance transformation and therefore does not restrict the available operating bandwidth of the array.

As the patterns in Figure 2 demonstrate, the triangle is capable of producing an almost identical set of patterns to those yielded by the big wheel. In fact, the modeled deviation from perfect circularity is about 0.1-dB. (The maximum gain for the big wheel seems superior by a small amount, but the average gain around the big-wheel azimuth pattern is closer to 7.15 dBi due to a slightly greater range between maximum and minimum gain values.) Perhaps the only two disadvantages of the triangle are physical: it requires more area than a circle, and the free ends may be more susceptible to local wind and weather.

We may curve the dipole elements and form a circular version of the same array. Figure 3 shows the outline of a 3-dipole wheel and thus return us to a truly interrupted loop. With 0.5"-diameter elements, the radius is 15.7" for the 144.5-MHz antenna. The resulting circumference is 98.6". With dipole tip spacing of about 1.1", each dipole occupies 31.7" of the circumference of the circle. Like the dipoles of the triangle, the dipoles are set for close to a 50-Ohm feedpoint impedance to allow the use of 50-Ohm lines to the hub without significant impedance transformation. Note the shorter lengths of the dipole compared to those in the triangle, largely due to both the curvature and the close coupling of dipole ends.

The 3-dipole wheel requires somewhat more planning than the simple triangle, since we need 3 support arms. A non-conductive arm leading to a T at the end would allow the support to fit into the dipole ends to permit a bit of tip-spacing adjustment. At the feedpoint gaps, the dipoles will also require insulated plugs, which should be as small as feasible. If we add tube-bending into the construction equation, the 3-dipole wheel may be more complex to construct than the triangle, but the final product will form a closed circle and occupy considerably less area. Despite these differences, as shown in the elevation and azimuth patterns, the performance of the wheel is virtually identical to the performance of the triangle.

The remaining question involves matching the set of 3 50-Ohm impedance values at the hub to a 50-Ohm transmission line. Figure 4 shows us two alternatives. A parallel connection of the lines will yield an impedance in the vicinity of 16 Ohms to 17 Ohms, a difficult value to match without employing a network. As well, any remnant or stray reactance will further complicate matching. Less often employed but perfectly usable under the circumstances of this array (three identical dipoles and connecting lines) is a series connection. (In fact, the models for these arrays use a series connection system, and the patterns shown are no different from those applying separate sources to each connecting line.) The resulting impedance will be in the vicinity of 150 Ohms to 155 Ohms, and any stray reactance will be too small to seriously affect the final result. A �-wavelength section of 93-Ohm RG-62 performs the final transformation of the impedance to about 55 Ohms. Even tuned to the low end of 2 meters, the SWR only reaches 1.5:1 at the highest end of the band. The system has enough broadband capability to allow adjustment of the lowest SWR value anywhere in the band simply by lengthening or shortening the matching line slightly.

The Lindenblad

In one sense, our last option is not a true HPOD, that is, a horizontally polarized omni-directional antenna. The Lindenblad is actually a circularly polarized array with equal horizontal and vertical components within the point-to-point radiation pattern. Its origins lie in the pioneering work of N. E. Lindenblad, who first proposed the antenna design almost off-hand in a broad article on television transmitting antennas. (See N. E. Lindenblad, " Antennas and Transmission Lines at the Empire State Television Station," Communications, vol. 21, April, 1941, pp. 10-14 and 24-26.) After World War II, Brown and Woodward (who made numerous contributions to VHF and UHF antenna design) developed the idea in detail from Lindenblad's patent papers. (See G. H. Brown and O. M. Woodward, "Circularly Polarized Omnidirectional Antenna," RCA Review, vol. 8, June, 1947, pp. 259-269.) They envisioned possible aviation uses for the antenna. The overall goal for the antenna was omni-directional coverage in the X-Y plane (parallel to ground) with circular polarization.

Figure 5 shows two ways of looking at the Lindenblad. The left side shows face views and an overhead view of the array. We have four slanted dipoles, each equidistant from a center point. For circular polarization in the X-Y plane, that is, for equal horizontal and vertical components to the radiation pattern, the degree of dipole slant and the distance from the center point are interdependent. At a distance of about �-wavelength, the required slant angle is 45�. (There are refinements to the calculations. See Appendix 1, "Some Overlooked Antenna Basics for DX and Off-World Communications," Proceedings of the 2006 Southeastern VHF Society Conference, pp. 250-252, for further information. See earlier portions of the article for information on the modified Lindenblad that may be more useful for lower angle satellite communications.)

The right side of the sketch shows the dipoles and their required interconnections for an effective array. Since the dipoles are fed in phase and have individual feedpoint impedances close to 105 Ohms in the arrangement shown, 4 RG-62 �-wavelength lines provide a net parallel junction impedance of about 25 Ohms. A �-wavelength length of 35-Ohm cable (usually composed of parallel sections of 70-Ohm cable) completes the final impedance transformation to 50 Ohms. The 0.125" diameter aluminum dipoles are each 40.1" long. Other feeding arrangements are certainly possible.

The single elevation pattern in Figure 6 shows a pattern quite unlike the earlier elevation patterns. The combination of vertical and horizontal lobes, even at a height as low as 20', tends to fill the outline of the total radiation pattern in the plot. At the TO angle of 4.6�, the maximum gain is 6.15 dBi, with an overall gain variation of about 0.9 dB. Note that the total pattern is a bit squared off, mostly as a result of the combined horizontal components of the slanted dipoles. However, unlike the previous antennas that we have examined, the Lindenblad's point-to-point performance is not strictly proportional to the far-field radiation pattern. For example, notice the differing strengths of the horizontal and vertical components in the far-field pattern at the lower left. The lower right corner pattern is a ground-wave plot using a distance of 1 mile and a receiving height of 20'. In this pattern, the vertical and horizontal components are nearly equal.

Since the Lindenblad maintains its pattern over a considerable bandwidth and since the antenna has a usable SWR bandwidth that is wider than the 2-meter amateur band, the array is suitable for use as an omni-directional antenna for both ends of the band. The major disadvantage is that each component of the array's radiation (and reception) pattern is weaker than most of the other antennas in our selection of options. A second Lindenblad between 0.5 wavelength and 1 wavelength above the first and turned 45� will not only improve performance (to a maximum gain of over 8 dBi), but will also circularize the pattern.

Stacking Larger Omni-Directional Arrays

One popular configuration for any of the larger omni-directional antennas is a vertical stack of 2. Because we may be tempted to misapply some rules of thumb derived from other antenna types, we should devote a small space to this topic before we close. Like horizontal dipoles, the horizontally polarized arrays with circular patterns increase gain when we stack two such antennas an optimal distance apart and feed the two antennas in phase. At the design frequency, 144.5 MHz, a wavelength is about 81.7", which makes stacking fairly convenient.

We need to know what separation distance is optimal for these arrays. One popular separation value is a half wavelength. The temptation to use this value arises from and is applicable to special circumstances. On the left, in Figure 7, we find the elevation patterns of a single pair of turnstiled dipoles. Because radiation is stronger at high elevation angles, the use of �-wavelength spacing in a stack of 2 pairs of turnstiled dipoles is very productive. The use of �-wavelength spacing with horizontal antennas tends to attenuate very high angle radiation and to make the energy available at lower angles. The maximum gain of a single turnstile pair is about 5.5 dBi (with a 20' height above average ground) in the lowest lobe. With �-wavelength spacing, the lowest lobe shows better than 9-dBi gain when the lower turnstile is at 20' over the same type of ground.

On the right in the same figure, we have elevation patterns for the 3-dipole wheel. The single antenna pattern uses a 20' height. However, by nature, the 3-dipole configuration does not shows very high-angle energy levels. In fact, the lowest lobe has a gain of about 7.25 dBi. Therefore, the automatic use of a stacking space of �-wavelength is not necessary, and we are free to seek out the separation that yields maximum gain in the stack's lowest lobe, as pictured in the lower elevation plot. Table 1 provides modeled data for various stacking distances when the height of the lower 3-dipole wheel is 20' above average ground.

Gain honors go to a stack spacing of 7/8-wavelength, and the elevation plot in Figure 7 uses this value. However, stacking distances between 3/4-wavelength and 1-wavelength would not show any detectable differences in performance. Noticeable in the table is the fact that the two antennas interact so that the impedance values shift with each change in stacking height. Obtaining a closer impedance value to 50 Ohms may require us to change the lengths of the 93-Ohm match sections.

The 3-dipole wheel exhibited virtually no fluctuation in the gain around the perimeter. However, construction variations may create very small distortions in the pattern. The variations remain in a stack of 2 such antennas at any stacking distance. One way to smooth the azimuth pattern is to orient the spokes at a 60� offset between the upper and the lower antennas for a 3-element array and at a 45� offset for a 4-element array. The offset technique will smooth the azimuth patterns of stacks having up to 1-dB or greater fluctuations in gain around the horizon. For example, Figure 8 shows the azimuth pattern differences when we stack Lindenblads without and with a 45� offset. The squarish pattern of a single Lindenblad reappears in the aligned stack. However, with the offset, the pattern is perfectly circular.

The offset technique of circularizing azimuth patterns applies only to arrays using independent elements fed in phase. Phase-fed elements, such as those in a turnstile, may suffer from the same treatment in a stack of two.

In-phase feeding of two HPOD antennas in a stack uses the same general rules and procedures employed in any stacking situation. For 50-Ohm feedpoints, the most widely used procedure is to employ a pair of �-wavelength 75-Ohm lines to a parallel junction. The 100-Ohm transformed impedances together form a close match for the usual 50-Ohm main feedline in most amateur installations.

However, the 3-dipole HPODs that we have examined in these notes offer an alternative potential if the main feedline happens to be a length of surplus 75-Ohm hard-line. Since the individual dipole impedances match the 50-Ohm connecting lines, we may bring these lines all the way to a central position before we wire them in series. Figure 9 shows the general scheme. The two resulting 150-Ohm impedance values in parallel provide a close match for the hard-line. A secondary function of the sketch in Figure 9 is to suggest an alternative method of routing the support elements for the 3-dipole wheels in the stack. As a support system, the idea is less applicable to the dipole triangle.


In our voyage through the land of horizontally polarized omni-directional antennas, covering this and the preceding episode, we have encountered many schemes. One common feature of most of them is the presence of one or more features that calls for precise, if not downright finicky adjustment, with a resulting narrowing of the region in which we may obtain a nearly perfect circular azimuth pattern. The smaller the array, the more problematical some of the critical features become. Of the lot, perhaps the larger 3-dipole arrays are the least problematical: once we obtain the proper physical dimensions, the matching becomes routine. As well, the larger arrays best maintain their circular azimuth patterns over a broad bandwidth. That feature may be less important during operation than it is during construction. With a broader design bandwidth, small variations in construction precision create fewer problems in the antenna's performance.

Nonetheless, our survey has unearthed many older and newer designs for the HPOD at VHF. One or more of the options should serve almost any need.

Updated 01-12-2008 � L. B. Cebik, W4RNL. This item first appeared in QEX, Jan/Feb, 2008, pp. 40-44. Reproduced with permission. Copyright ARRL (2008), all rights reserved. This material originally appeared in QEX: Forum for Communications Experimenters (www.arrl.org/qex).

Return to series index page