Notes on Fixed Satellite Antennas

L. B. Cebik, W4RNL

Fig. 1 illustrates some of the complexities facing the VHF and UHF operator who wishes to do everything (except use local repeaters, a topic for another day). Initially, the operator imagines a simple fixed omni-directional antenna. But capturing the distant point-to-point station and the rapidly moving satellite in the presence of ground clutter raises questions. It seems that even to receive weather maps around 137 MHz requires an antenna that is high enough to see over the trees and buildings to catch the satellite while it is at a low elevation angle. Similar considerations apply to the distant station, within the gain limits of an omni-directional antenna. I would not disturb this blissful formula for success if it always worked. As we shall see, it does not always work, and for reasons related to basic antenna performance in the presence of the very thing that supports all of that clutter--the ground.

Fixed Omni-Directional Antennas

When amateurs first start to dream of satellite communications or of receiving weather information from satellites, they tend to think simple and inexpensive thoughts. They want to install a basic omni-directional antenna with enough gain to allow error-free reception from and a solid connection to a communications satellite. Since satellites may appear almost anywhere in the hemisphere surrounding a given location, an ideal pattern emerges, as portrayed in Fig. 2. The small square on the Z-axis is an arbitrary antenna, since the pattern is not achievable.

There are two problems that prevent us from obtaining the ideal omni-directional pattern. One difficulty is ground clutter (see Fig. 1). It absorbs, reflects, or refracts very low angle signals. That fact prompts us to raise the antenna to "see over" the clutter in order to reach the horizon. Height improvement works to some extent for point-to-point communications, although intervening clutter may still provide interfering reflections unless we place the antenna very high indeed. However, for satellite work, the technique is not only relatively futile, it can be self-defeating. Satellites are in the extreme far field of any antenna, and the very existence of the ground at any distance will prevent us from creating an ideal omni-directional pattern.

We might here go back to the textbooks and supply an array of equations by which to prove our point. However, it may be more instructive simply to model some of the most widely used omni-directional antennas and sample their patterns at various heights above ground. Each sample will use 1.5-mm diameter aluminum elements at 300 MHz (more precisely, 299.7925 MHz, where 1 meter = 1 wavelength). Each 2-element antenna will use a 90° phase-difference between elements to achieve the omni-directional pattern. There are numerous techniques to achieve turnstiling--or what engineers prefer to call quadrature feed--but most amateurs will likely use a ¼ wavelength electrical section of transmission line with the same impedance as the individual elements in the antenna. The net feedpoint impedance will be ½ the impedance of the individual element. When we place the antenna above ground, we shall use average soil (conductivity 0.005 S/m, relative permittivity 13). In this initial part of our work, we shall only be interested in the total far field of the antenna. We shall save the vertical, horizontal, and circular components for the last part of our work.

1. Turnstiled Dipoles: Perhaps the oldest quadrature-fed antenna is a pair of turnstiled dipoles. A version of this antenna has appeared in every ARRL Antenna Book since satellites first appeared, and the antenna predates that event by decades as an upper HF and lower VHF omni-directional antenna. The antenna found wide use in the days before repeaters, when virtually all VHF communications in the amateur bands used horizontally polarized antennas.

The upper portion of Fig. 3 shows the outline of the dipole turnstile. The appearance gave us the antenna name and that name eventually transferred to the simple feed method. Be certain that the feed system is properly matched at every step, because the SWR curve will remain impressively low long after the free-space patterns to the upper right have gone to pot. The beamwidth of a dipole prevents a perfect E-plane circle, but the squared circle varies only by about 1 dB from maximum to minimum. The H-plane pattern shows that the simple turnstile radiates more strongly broadside to the antenna than off the edges, an important reason why folks tried other designs.

The lower part of Fig. 3 contains the critical information about the elevation patterns of a simple dipole turnstile as we gradually raise the height of the wires above average ground. Table 1 summarizes the data.

The elevation pattern of the dipole turnstile has deep nulls between elevation lobes at any height above ground. Even at a height of 10 wavelengths, the null regions where a signal may drop more than 3-dB relative to the lobe value is larger than the region where the signal is less than 3-dB below peak lobe value. Note also that the zenith region (directly overhead) changes as a function of the antenna height, with a deep null at integral height multiples of ½ wavelength and a lobe at odd multiples of ¼ wavelength. In general, the basic dipole turnstile is quite unsatisfactory for general satellite communications and for weather satellite reception, even though it might be useful for local horizontal point-to-point communications.

2. Turnstiled Dipoles with a Planar Reflector Screen: The immediate amateur development to improve the dipole turnstile was to add a screen below the antenna. The impression that this maneuver leaves is that obtaining more gain cures everything. A proper planar reflector extends at least 0.5-? beyond the limits of the driven antenna elements in every direction. In this case, we may arrange the element to point toward the corners of a square screen in order to minimize excess screen size. Nevertheless, an adequate planar reflector will be about 1.4 wavelength per side, as shown at the top left of Fig. 4. The free-space patterns should suffice to forewarn us that this antenna system will have very limited utility. The antenna shows good gain and an excellent free-space front-to-back ratio. But as the data in Table 2 reveal, along with the patterns in the lower half of Fig. 4, the antenna is good only for satellites directly above us and is worse than the simple turnstile dipole system near the horizon.

The table lists the half-power beamwidth of the antenna patterns at each height, as a measure of how much of the 180° hemisphere might be missing from coverage. The patterns in Fig. 4 also show several other interesting facets of pattern development with increasing height. For example, note the pattern for a height of 10 wavelengths above ground. It shows 2 peaks that represent what amounts to a circle of maximum gain with a null at the exact zenith angle. As well, the low angle lobes and nulls, although smaller than those for the simple dipole turnstile, remain in place, since they are a result of the antenna's height and not of its specific design.

The third aspect of pattern development with increasing height for antennas that point generally upward is the development of high-angle ripples in the main forward lobe. These ripples tend to be most prominent in antennas that use crossed elements to achieve circular polarization. However, we shall encounter them to one degree or another in the patterns of all antennas raised to the vertical.

In the end, the dipole turnstile with an adequate planar reflector does not serve well as a fixed position omni-directional antenna for any service. However, in the upper UHF region (70 cm and upward), it may serve as a simple and effective aimable antenna, if 8 to 8.5 dBi is adequate gain for the operation. Do not skimp on the screen size, because gain will drop quite rapidly as we downsize the reflector dimensions.

3. Turnstiled Quad Loops: We can generally create a turnstile pair out of any element that has a linear dimension and makes no more than a flat plane. The quad loop is a variation on the dipole. In fact, it is roughly 2 dipoles spaced ¼ wavelength vertically, each bent so that the ends just touch. If we set two quad loops at right angles so that the feedpoints are close but do not touch, we can add a ¼ wavelength phasing line between them, but we must adjust the characteristic impedance of the line to match the individual self-resonant impedance of the loop--about 125-130 Ohms. The feedpoint impedance of the turnstiled quad will be half that value. It appears not to matter if the top crossing points do or do not touch.

Turnstiled quads came into play as omni-directional antennas with slightly more gain than the dipole turnstile. They come in two varieties, depending on whether we start with square quad loops or with diamond quad loops. Fig. 5 shows the general outline of both versions, along with free-space patterns. Notice that in each case, the patterns form a nearly perfect sphere, rather than showing the egg-shape of the dipole turnstile. Hence, we expect a small increase of gain in the E-plane. The square loop version unfortunately arose in a decade in which cute names were more important to antennas than performance. Hence, someone dubbed the antenna the "eggbeater," a name best forgotten, lest someone rename the turnstiled diamond loops the "whisk."

As Table 3 shows, there is very little performance difference between the two versions of the same antenna. Since the heights of the antennas depend on the bottom-most point or the feedpoint, the slightly taller diamond loop antenna shows values that reflect the small increase in average height. See Fig. 6 for the patterns of the square quad loops over ground and Fig. 7 for the corresponding diamond loop quad patterns. Note that in both cases, the maximum gain applies to the lowest elevation lobe in the pattern. The maximum gain in this lobe is about 1 dB greater than for a dipole turnstile at the same feedpoint height. As the catalog of patterns shows, the gain improvement is not as significant as it might initially seem. In general, we choose between the 2 versions of the turnstiled quads most often due to a preference for certain construction methods. The models used here at 300 MHz use 1.5-mm diameter elements, but common wire sizes will do as well.

Unlike the dipole turnstile, the zenith-angle domes occur at integral multiples of a half wavelength, with upward nulls occurring at odd multiples of ¼ wavelength. Otherwise, we have no reason to consider the turnstiled quad loops further as candidates for fixed satellite communications service. They display all of the nulls shown by the dipole turnstile for all practical heights. Therefore, they suffer from the same variability of signal strength as a satellite moves across the sky.

4. Turnstiled Moxon Rectangles: The number of lobes and nulls in any turnstile horizontal antenna is a function of the height of the antenna above ground. Hence, we cannot eliminate them. However, we can go some distance in reducing the null depth by selecting the right kind of antenna to fit into the turnstile. One key factor is the inherent H-plane beamwidth of the antenna element in isolation. The Moxon rectangle is a 2-element parasitic beam with the element ends folded back toward each other. It has a very broad beamwidth in the H-plane, and has enough side radiation off the element tails to partially fill the nulls at low angles. The resulting pattern is far from perfect, but it goes a considerable distance in the right direction.

Fig. 8 shows the outline of a turnstiled pair of Moxon rectangles at the upper left. Although the driver centers require enough separation for the phase line, the reflector elements may touch at their centers. We can design the rectangle for a 50-Ohm feedpoint. The phase line will require a 50-Ohm section of coax, but the 25-Ohm net feedpoint impedance will require transformation back to 50 Ohm. Alternatively, we may design the rectangle for a 95-Ohm feedpoint and use coax of that value for the phase line. The resulting net feedpoint impedance is close enough to 50 Ohms not to require any special matching.

The free-space patterns show the wide beamwidth that results in the more modest nulls in the patterns taken above ground. As revealed in Table 4, the beamwidth of the upper dome is always at least 90°.

The turnstiled Moxon rectangles achieve much superior evenness of performance at higher elevation angles than any of the fixed antenna candidates that we have so far surveyed. The maximum gain is remarkably consistent regardless of the height of the antenna above ground, largely due to the fact that the antenna has its own highly effective reflector. Direct downward radiation is almost non-existent.

However, as shown by the elevation plots in Fig. 8, the antenna only goes part way toward smooth performance at lower elevation angles. The antenna configuration limits, but does not eliminate the nulls between elevation lobes. As we increase the height of the antenna above ground, we encounter the same growth in the numbers of lobes and nulls that we have seen in all of the fixed satellite antenna candidates. If we cannot satisfy our operating needs at angles higher then about 30° above the horizon, then we must continue the search for a better fixed-position antenna.

5. The Modified Lindenblad: In 1941, N. E. Lindenblad developed and patented a design for a circularly polarized antenna to use at the television station atop the Empire State Building. The necessities of World War II delayed the actual construction of a Lindenblad antenna until after the war, and then, the builders intended it for possible aviation use. The Lindenblad has undergone re-invention and modification in recent times without due credit to the original developers. The references should rectify this situation. See Appendix 1 for a comparison of the original Lindenblad with the modified version shown here.

Fig. 9 shows the outline of the modified Lindenblad on the left, along with instructions on how to model it. It consists of 4 dipoles forming a square. However, each dipole is slanted 45°. Since one may easily mess up the model, the orientations of the elements and the sequence of current source phasing appear in the diagram. Note that we select one element as the prime element and progressively increase the phase angle of each succeeding source by 90° in a clockwise direction to obtain the desired "up" series. (The original Lindenblad fed each dipole in phase to achieve low-angle circular polarization. However, as the Appendix will show, the resulting antenna has severe limitations for satellite use. Hence, the use of progressive quadrature feed has advantages in satellite service.) If we leave the elements oriented as before but use a counterclockwise progression of phase angle increases, we obtain the less-desired "down" series. (Reversing both the element orientations and the progression of phase angles will result in an "up" series.)

The right side of Fig. 9 shows the differences between the up and down series. (The original Lindenblad has a free-space pattern that is symmetrical with respect to the array centerline.) The differences are vivid in the free-space pattern, but less so in the patterns above ground by 1 wavelength. However, the gain is smoother as we increase the elevation angle in the up series, and the differences between maximum and minimum gain in the azimuth pattern are smaller. The modified Lindenblad was not designed to create a high angle dome, but to provide omni-directional television transmission suitable for reception by both horizontal and vertical antennas in the early days of urban TV. Hence, the upward null was not considered a hindrance to good performance.

The optimal spacing for the dipoles in a modified Lindenblad array has come in for some discussion. The upper portion of Fig. 10 provides some guidance. The spacing values reflect a center point, so the actual spacing between facing dipoles is twice the listed value. A spacing of 0.5 wavelength provides perhaps the best obtainable pattern over average soil for elevation angles up to about 50°. 0.6 wavelength? spacing might extend the elevation angles 10° further upward, but with somewhat higher ripple in the signal strength.

The lower part of Fig. 10 provides data for the effects of soil quality on the pattern shape, using a spacing of 0.5 wavelength between facing dipoles with the feedpoints 1 wavelength above ground. Very poor soil shows some pattern degradation, but the patterns for average and better soil are remarkably consistent.

Table 5 provides data on the modified Lindenblad array set at various heights above average ground. See Fig. 11 to correlate the data with elevation patterns. In general, the Lindenblad is subject to the same variations in zenith-region radiation as the single turnstile dipole. However, the antenna excels in evenness of performance at lower elevation angles.

Even though the modified Lindenblad provides superior low elevation angle performance, the array has limits in terms of the height at which we should position it. Note that the lowest elevation lobe begins to dominate by a height of 2 wavelengths above ground. At heights of 5 wavelengths and upward, the lobes and nulls at very low angles become serious phenomena. Although 3-5 wavelengths is an appreciable height for a 2-meter antenna, the same height in wavelengths may not be sufficient at 70 cm to clear ground clutter. Hence, for all of its other virtues, the modified Lindenblad remains subject to the same pattern formation influences that informed the patterns of the other candidate antennas. (However, the lobe development is severely retarded relative to the original design.)

Of all the candidates for a fixed-position satellite antenna in our survey, the modified Lindenblad achieves the best low angle performance, where we understand "low angle" to mean a range of angles in the sky that a satellite is likely to traverse. (We must set aside some of our HF antenna ideas to grasp the needs of satellite communications.) Between 10° and about 50° elevation, no other candidate achieves an equivalent evenness of gain. For angles from about 30° up and higher across the sky, the turnstiled Moxon rectangles take honors among the candidates in the survey. Still, no single antenna type achieves the ideal of a dome across the sky with equal gain in every possible direction and at every possible height.

The survey has not been all-inclusive. It has featured some historically important first attempts at usable satellite antennas adapted from omni-directional terrestrial antennas. It finished with two more recent entries, one geared toward smooth higher-angle gain, the other designed 60 years ago for lower elevation angles. In all of the exercises, we have seen that height above ground presents the fixed-antenna user with a conundrum. Greater antenna heights clear the ground clutter, but produce considerable numbers of lobes and nulls. Lower heights tend to clean up the pattern, but may block lower elevation angles due to intervening structures, both man-made and natural. Unless we have specific interests that allow a simple fixed antenna to serve us well, eventually, we shall begin to look at antennas that we can aim and that are worth aiming. We may begin with simple hand-held antennas that we point from our patios. Even while claiming total satisfaction with simplicity, we keep on checking the prices of Az-El rotator systems and dreaming of long-boom antennas with high gain and narrow beamwidths.

Appendix 1: The Original Lindenblad Circularly Polarized Dipole Array

The modified Lindenblad dipole array is one possibility for a fixed-position satellite antenna especially suited to capturing satellites at lower elevation angles relative to the horizon. 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. They did not design the antenna for overhead coverage.

Fig. A1-1 shows on the left the fundamental principle behind the Lindenblad dipole array. To achieve circular polarization, we need vertically and horizontally polarized components--shown as currents in the wires--such that they result in exactly equal fields at any distance from the antenna in any direction. The sketch shows right-hand circular polarization. The conceptual diagram is almost impossible to realize as a physical antenna. Lindenblad reasoned that an array of tilted dipole, fed in phase, would approximate the ideal situation. The right side of Fig. A1-1 shows the solution, highlighting 1 of the 4 dipoles. If we select the proper angle for the dipole relative to the horizontal (a), then the vertical and horizontal components will be equal. The design is subject to limitations, since we have facing dipoles. The tilt angle, a, depends in part on the distance between facing dipoles. In terms better suited to calculation, the required tilt angle depends upon the radius of the circle connecting the feedpoint positions of the dipoles. Since the fields between adjacent dipoles overlap, the required tilt angle for the dipole also depends on whether we measure fields tangential to the dipole faces or at angles that bisect two dipoles. The following table shows a few of the Brown-Woodward tilt-angle calculations.

The modified Lindenblad array using progressive quadrature feed uses a radius of 0.25 wavelength. Therefore, we may simply change the phase angles of the individual feedpoints to put them all at the same phase angle. However, this seemingly simple change results in a need to change the length of the dipoles to restore resonance to each dipole. As well, we need to re-angle the dipoles in the opposite direction from those in the modified version to achieve right-hand circular polarization. Fig. A1-2 shows what we get.

In its original configuration, the free-space elevation or theta patterns for the Lindenblad array are quite symmetrical. The slight variation in the patterns facing a dipole and the patterns at 45° to any dipole reflects the requirement for optimizing the tilt angle. In these patterns, the tilt angle is constant at 45°. The phi or azimuth pattern shows a slight squaring, comparable to the pattern of an ordinary turnstile pair of dipoles.

Over ground, we obtain elevation patterns that reflect the original uses planned for the antenna. As shown in Fig. A1-2 and in Table A1-2, the strongest lobe is always the lowest lobe. As well, beamwidth of the strongest lobe is always quite narrow. The array produces a very deep null directly overhead.

As a consequence of these pattern features, the original Lindenblad array is not especially suitable for satellite work, although it may serve as an omni-directional antenna for point-to-point communications. Unless the target stations are using the same polarization, results are likely to be little better than with a simple dipole turnstile. If the target station is using a comparable antenna, but with left-hand circular polarization, then the two antennas may be blind to each other.

Nevertheless, the Lindenblad array achieves a close approximation to circular polarization relative to point-to-point targets in the X-Y plane, that is, parallel to the earth's surface. At angle facing the dipoles (0°, 90°, and 180°), the axial ratio is close to 0.9. Between dipoles (at 45° and 135°) the ratio drops below 0.7, again suggesting the need to change the tilt angle for more nearly circular polarization. In all cases, the sense of polarization is right-handed.

Lindenblad Radiation Pattern Reports in Free-Space for Phi Angles 0 to 180°

  - - ANGLES - -          - POWER GAINS -        - - - POLARIZATION - - -    - - - E(THETA) - - -    - - - E(PHI) - - -
 DEGREES  DEGREES       DB      DB      DB        RATIO     DEG.               VOLTS     DEGREES       VOLTS     DEGREES
   90.00     0.00      -3.37   -2.33    0.19    0.88674   -88.70  RIGHT     1.07851E+02    85.59    1.21611E+02    -4.72
   90.00    10.00      -3.43   -2.13    0.28    0.86140   -89.08  RIGHT     1.07138E+02    85.57    1.24368E+02    -4.70
   90.00    20.00      -3.58   -1.66    0.50    0.80191   -89.60  RIGHT     1.05333E+02    85.53    1.31350E+02    -4.65
   90.00    30.00      -3.75   -1.15    0.75    0.74144   -89.88  RIGHT     1.03279E+02    85.47    1.39295E+02    -4.60
   90.00    40.00      -3.86   -0.83    0.92    0.70553   -89.99  RIGHT     1.01937E+02    85.44    1.44483E+02    -4.57
   90.00    50.00      -3.86   -0.83    0.92    0.70553   -89.99  RIGHT     1.01937E+02    85.44    1.44483E+02    -4.57
   90.00    60.00      -3.75   -1.15    0.75    0.74144   -89.88  RIGHT     1.03279E+02    85.47    1.39295E+02    -4.60
   90.00    70.00      -3.58   -1.66    0.50    0.80191   -89.60  RIGHT     1.05333E+02    85.53    1.31350E+02    -4.65
   90.00    80.00      -3.43   -2.13    0.28    0.86140   -89.08  RIGHT     1.07138E+02    85.57    1.24368E+02    -4.70
   90.00    90.00      -3.37   -2.33    0.19    0.88674   -88.70  RIGHT     1.07851E+02    85.59    1.21611E+02    -4.72
   90.00   100.00      -3.43   -2.13    0.28    0.86140   -89.08  RIGHT     1.07138E+02    85.57    1.24368E+02    -4.70
   90.00   110.00      -3.58   -1.66    0.50    0.80191   -89.60  RIGHT     1.05333E+02    85.53    1.31350E+02    -4.65
   90.00   120.00      -3.75   -1.15    0.75    0.74144   -89.88  RIGHT     1.03279E+02    85.47    1.39295E+02    -4.60
   90.00   130.00      -3.86   -0.83    0.92    0.70553   -89.99  RIGHT     1.01937E+02    85.44    1.44483E+02    -4.57
   90.00   140.00      -3.86   -0.83    0.92    0.70553   -89.99  RIGHT     1.01937E+02    85.44    1.44483E+02    -4.57
   90.00   150.00      -3.75   -1.15    0.75    0.74144   -89.88  RIGHT     1.03279E+02    85.47    1.39295E+02    -4.60
   90.00   160.00      -3.58   -1.66    0.50    0.80191   -89.60  RIGHT     1.05333E+02    85.53    1.31350E+02    -4.65
   90.00   170.00      -3.43   -2.13    0.28    0.86140   -89.08  RIGHT     1.07138E+02    85.57    1.24368E+02    -4.70
   90.00   180.00      -3.37   -2.33    0.19    0.88674   -88.70  RIGHT     1.07851E+02    85.59    1.21611E+02    -4.72
   90.00   190.00      -3.43   -2.13    0.28    0.86140   -89.08  RIGHT     1.07138E+02    85.57    1.24368E+02    -4.70

Fig. A1-3 shows the elevation patterns for the original Lindenblad at a height of 7.5 wavelengths above average ground. Although the right-hand circular polarization component dominates the pattern at the right, the multiplicity of lobes and nulls tends to disqualify the original array from satellite service. Compare these patterns to corresponding patterns for the modified Lindenblad with progressive quadrature feed. Although the modified version of the antenna has slightly less gain, the evenness of its pattern suggests more satisfactory satellite service at lower elevation angles.

The exploration of original Lindenblad literature is not either idle or merely historical. Some antenna builders have tried closer dipole spacing than the value recommended in the main text. To obtain satisfactory patterns, they may wish to attend to the calculations of the more nearly correct tilt angle for these closer spacings. We may yet learn much more about the modified Lindenblad array.

Updated 05-01-2006. © 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.

Return to Amateur Radio Page