SCVs Part 6: The Bruce Array: An Update
The ARRL Antenna Book has long had some interesting information on the Bruce Array, developed in recent editions (pp. 8-42 to 8-47 in the 19th Edition) by Rudy Severns, N6LF. Nonetheless, I still receive inquiries about the array, most wondering how it stacks up against half-squares and bobtail curtains and whether it is among the class of antennas that I call SCVs (self-contained vertical arrays).
The short answer to the inquiry is that it does quite nicely. An optimized half-square for 3.6 MHz shows a gain of about 3.8 dBi broadside to the plane of the array at a take-off angle of about 18 degrees when we make it about 0.46 wavelength long and place the base about 15' above ground. A bobtail curtain at the same height above ground requires a length of about 1.08 wavelengths for a gain of about 5.5 dBi at a 19-degree take-off angle, when designed for the same frequency. A 3-element Bruce array, when about 1/2 wavelength long gives the same gain as a half square. When about the same length as a bobtail curtain, it yields nearly 6.3 dBi gain.
However, for a given length, the Bruce array requires more wire, since the vertical elements are only about 1/4 wavelength apart. As well, since the Bruce array has some of its horizontal wires at the bottom, it tends to show maximum gain--the numbers cited above--when the base is about 0.15 wavelength above ground--about 41' above ground. Hence, the maximum gain placement of the antenna tends to also raise the top level of the antenna to about 0.41 wavelength or 111' at 3.6 MHz. You may wish to compare these numbers with the dimensions of half squares and bobtail curtains in Parts 4 and 5 of the SCV series. You may also wish to compare the dimensions with other forms of the SCV in earlier parts of the series, especially the side-fed rectangle and double-rectangle.
So simple answers do not tell us the entire story of the Bruce array, since the feasibility of using this array depends very much on local circumstances. However, it may be useful to explore the array to see what some of its potentials and limitations may be.
What is a Bruce Array and How Does It Work?
A Bruce array is a continuous length of wire arranged to form 1/4 wavelength vertical wires spaced about 1/4 wavelength apart. To ensure that the array is vertically polarized--at least as a starting principle--we shall feed it at the center of any vertical element. To make a complete vertical dipole, we have to count from the feedpoint 1/4 wavelength, which takes us either to the center of a bottom horizontal wire or to the center of a top horizontal wire. The end vertical require that we add about 1/8 wavelength of wire to the open end to achieve full vertical dipole length. These end wires can be pointed outward, but to save space, most builders fold them back inward. The direction makes only a little difference, since the array shows predominantly (but not exclusively) vertically polarized radiation, with the remnant currents in the horizontal wires canceling to a large degree.
Fig. 1 shows the most common forms of Bruce arrays used by amateurs, ranging from 2 to 5 vertical elements. In reality, you can build them to virtually any length for which you have the space. The figure also gives us the names of the array parts: horizontal, vertical, and end wire. In each case--at least at the start--we shall feed the antenna on either the center vertical or one of the verticals closest to the center.
In this exercise, we shall use a design frequency of 3.6 MHz so that the array is more easily comparable to other SCVs. Dimensions will be given in wavelengths or fractions of wavelengths so that you can scale the antenna within the band with no problem. All models will use AWG #12 copper wire, except for one brief set of notes down the line. My initial model of the Bruce used the 4-element version in The ARRL Antenna Book with the recommended dimensions of 1.05 times a standard fraction of a wavelength. Hence, the 1/4 wavelength sections are actually 0.26 wavelength long and the 1/8 wavelength end wires are 0.135 wavelength long. We shall eventually change that a bit.
The approximate dimensions suffice to show how the Bruce works, and Fig. 2 can help us in this regard. The curved lines show the current magnitude along each element when we feed the antenna at the center of a vertical element. The fields--as indicated by the current magnitudes on the vertical elements--add up to yield a bi-directional pattern broadside to the plane of the array. The horizontal wires show a current minimum at their centers, and of course, the end wires show the current going to zero at their ends--where we find a high voltage, as a safety note.
What the diagrams to not show is that the current undergoes a 180-degree phase shift from one vertical center-point to the next. We shall use this fact later on in a discussion of alternative feed systems. For now, we may simply note that this continuous shift in current phase from one center to the next sets up a condition in which the radiation from the horizontal wires largely cancels itself, leaving a predominantly vertically polarized array.
In Fig. 3, we have the elevation pattern overlaid on the outline of the array, when the array is about 0.15 wavelength above ground. The pattern is very well behaved as vertical antennas go and is clearly broadside to the plane of the array.
However, there is a tendency among hasty builders to reason that the center bottom horizontal wire is much more convenient. Hence, they break the wire at its center and install a parallel feedline. When we make this move, everything changes.
At first modeling sight, we might not realize that we have change the nature of the array. Fig. 4 shows a pattern of current magnitudes that is very similar to those in Fig. 2. What these current plots do not show here is that the phase along the horizontal wires is not shifting, but is the same at each end of the horizontal wire. The result is a pattern like the one shown in Fig. 5.
Feeding the Bruce array in series at the center of a horizontal wire changes the pattern from broadside to endfire, and a very considerable decrease in gain--about 3.2 dB for the 4-element array shown. There are ways of feeding the Bruce at the bottom center that will yield the relatively high gain vertical broadside array that we want, but the simple series feed system here will not do the job.
Fig. 6 shows a more correct way to obtain our vertical array with a center feed. We connect a wire from the center of the bottom wire to ground and feed the wire at its base against ground. We shall look more closely at this feeding system further down the line. Right now, we shall stay with the basic properties of the Bruce and assume that we are feeding it at the center of a vertical wire or in a way that simulates this feed system.
Fig. 7 presents an SWR sweep for our initial Bruce design for 3.6 MHz. The 250-Ohm SWR curve is interesting, since it shows a usable under-2:1-SWR limit of slightly over 200 kHz. However, the resistance and reactance curves are even more interesting.
The resistance changes very little with frequency. The reactance changes at a greater rate, but in nearly a linear fashion with rising frequency. Together, these two facts provide the user with a method of expanding the frequency coverage of a Bruce array. First, design the array so that it shows an inductive reactance at the lowest frequency of operation. Second, place remotely tuned variable capacitors in the feedline at the antenna terminals. The capacitors will compensate for the inductive reactance, leaving essentially a resistive impedance close to 250 Ohms (for our test array). Third, add a 4:1 balun and feed the system with 50- or 75-Ohm coaxial cable. Most baluns operate most efficiently when there is little or no reactance at their load terminals, so the balun in this set-up should provide good service. Fig. 8 shows the general scheme for this feed system.
We shall examine some other feed options before we conclude this exploration of the Bruce array. However, it is now time to refine the Bruce dimensions and mounting situation.
How Big and How High?
If we review the common outlines of amateur-size Bruce arrays, something should immediate strike us as odd about the 2-element version. It resembles a side-fed quad loop, but with a gap. We may place the gap at either the top or the bottom. Fig. 9 shows the quad loop and its 2 Bruce cousins.
Note that the Bruce gap, in either version, is at a high impedance, high voltage, minimal current point. When examining the evolution of the half-square from a closed loop, such as a delta, I discovered that one can open a loop at the high-impedance point with little or no effect upon performance. There is little or no difference between a high-impedance point on the loop circumference and a very-high-impedance gap, if the gap is not too large. In fact, the following table shows the performance figures for a 2-element quad and its Bruce counterparts at a base height of 0.15 wavelength.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Modeled Performance Reports for a side-Fed Quad Loop and Bruce Arrays
Antenna Gain TO Angle Feed Impedance
dBi degrees R +/- jX Ohms
Quad 1.87 16 137.0 - j 3.7
Bottom-Gap Bruce 1.89 17 134.6 - j25.8
Top-Gap Bruce 1.86 17 135.3 - j26.1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The original 4-element Bruce model using AWG #12 copper wire and the listed dimensions (0.26 wavelength vertical and horizontal wires and 0.135 wavelength end wires) showed an impedance of 248 - j61 Ohms when fed at the center of one of the interior vertical wires. This condition obtained with a base height of 0.15 wavelength or about 41'. The next question was how to go about enlarging the structure to bring the array to resonance.
The easiest way to accomplish this feat is simply to lengthen the end wires somewhat until the array arrives at something close to resonance on the design frequency. However, this tack drops the gain a bit. The original design showed a gain of 5.20 dBi at a TO angle of 16 degrees, with a feedpoint impedance of 248.6 - j60.6 Ohms. Lengthening the end wires by 0.01 wavelength to 0.145 wavelength brought the array close to resonance with a feedpoint impedance of 259.2 - j 2.5 Ohms. However, the gain dropped to 5.11 dBi.
A second way is to lengthen the vertical dimension slightly. This tactic seems prima facie more promising, since it promises to add a bit to the overall height of the array. However, this route also decreases gain on the road to resonance. Increasing the vertical lengths by 0.005 wavelength to 0.265 wavelength nearly resonated the array at 261.4 - j3.3 Ohms, but showed a 0.02-dB drop in gain. Although the drop was slight, this was not the desired trend.
In fact, the best method of sustaining array gain and arriving at resonance is to increase the length of the horizontal wires. Although the increase is only slight, the wider spacing of the elements increases gain. Ideally, verticals achieve their highest gain in a phased array at a spacing of about 1/2 wavelength. We cannot achieve this in the Bruce, because the array consists of dipoles connected end-to-end by bending them around. Still, any reasonable widening that does not decrease the array height shows up as a smidgen of increased gain.
The following table lists the final values selected for the parts of the Bruce arrays. The slight decrease from the recommended vertical length and the slight increase in the horizontal length optimized gain while bringing the feedpoint impedance close to resonance.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bruce Array Dimensions
All dimensions in wavelengths
No. of Horizontal Vertical End Total
Elements Wire Wire Wire Horiz. Length
2 0.27 0.255 0.13 0.27
3 0.27 0.255 0.14 0.54
4 0.27 0.255 0.145 0.81
5 0.27 0.255 0.145 1.08
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
With the array at a base height of 0.15 wavelength (and a consequential top height of 0.405 wavelength), the arrays show interesting elevation patterns broadside to the plane of the array. Fig. 10 gives us some modeled snapshots.
Although the vertical beamwidths of all four versions of the Bruce are similar, the versions with an even number of vertical elements show a deeper null directly above the array. The versions with an odd number of vertical elements have their end wires arranged with one high and one low. Whether this factor is the key to the difference in the depth of the zenith null I have not explored, since the difference is not operationally significant. It is, simply, interesting.
A relevant question at this point is why I have consistently placed the test models at a base height of 0.15 wavelength. The answer is direct: this height yields the highest gain for the array. I explored the performance of the 4 versions of the Bruce array at base heights ranging from 0.05 wavelength up to 0.25 wavelength in 0.025 wavelength increments. The following tables provide the rationale for my selection of a base height.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bruce Array Performance at Various Base Heights
2-Element (closed loop version)
Base Ht Gain TO Angle Feedpoint Impedance
WL dBi degrees R +/- jX Ohms
0.05 1.60 21 194.3 + j 38.3
0.75 1.72 19 174.8 + j 15.7
0.1 1.80 18 159.5 + j 4.0
0.125 1.85 17 147.1 - j 1.7
0.15 1.87+ 16 137.0 - j 3.7
0.175 1.87+ 16 128.8 - j 3.4
0.2 1.84 15 122.5 - j 1.5
0.225 1.77 14 117.8 + j 1.3
0.25 1.68 14 114.6 + j 4.7
3-Element
Base Ht Gain TO Angle Feedpoint Impedance
WL dBi degrees R +/- jX Ohms
0.05 3.52 21 289.7 + j 48.8
0.75 3.64 20 261.4 + j 22.4
0.1 3.72 19 239.8 + j 10.1
0.125 3.77 18 222.5 + j 5.4
0.15 3.79+ 17 208.7 + j 5.2
0.175 3.77 16 197.8 + j 7.9
0.2 3.74 16 189.6 + j 12.3
0.225 3.68 15 183.8 + j 17.9
0.25 3.58 15 180.2 + j 23.8
4-Element
Base Ht Gain TO Angle Feedpoint Impedance
WL dBi degrees R +/- jX Ohms
0.05 4.89 21 356.2 + j 56.8
0.75 5.03 19 318.9 + j 24.9
0.1 5.13 18 290.1 + j 10.8
0.125 5.18 17 267.1 + j 6.1
0.15 5.21+ 16 248.6 + j 7.2
0.175 5.21+ 16 234.2 + j 11.9
0.2 5.17 15 223.4 + j 18.9
0.225 5.09 14 216.0 + j 27.2
0.25 4.98 14 211.7 + j 35.9
5-Element
Base Ht Gain TO Angle Feedpoint Impedance
WL dBi degrees R +/- jX Ohms
0.05 5.94 20 425.0 + j 75.8
0.75 6.10 19 380.4 + j 34.0
0.1 6.20 18 347.2 + j 14.6
0.125 6.26 17 321.2 + j 7.0
0.15 6.29+ 17 300.9 + j 7.0
0.175 6.28 16 285.5 + j 10.2
0.2 6.23 15 274.2 + j 16.6
0.225 6.14 15 266.9 + j 24.4
0.25 6.02 14 262.8 + j 32.6
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The maximum gain heights are marked with a '+' symbol. Note that the versions with an even number of elements are indifferent to heights between 0.15 wavelength and 0.175 wavelength, while the versions with an odd number of elements show one evident--if not operationally significant--maximum gain base height.
Fig. 11 shows the elevation plots for the 4-element Bruce array at base heights of 0.05, 0.15, and 0.25 wavelength. The plots suggest that selecting the base height for the array may involve more than just the maximum gain values, since the range of gain change is not exceptionally large. Both the lowest and the highest base levels result in patterns with significant high-angle radiation. The more optimal height of 0.15 wavelength results in the least high-angle radiation, which can mean some quieting of shorter-skip QRN during the noisier months of operation.
In Fig. 12, we can compare the azimuth patterns of the 4 sizes of arrays, with each azimuth pattern taken at the to angle for the antenna with a base height of 0.15 wavelength. As we increase the number of elements, the horizontal beamwidth of the array narrows, which is the main source of the added gain of the larger versions. However, two other factors should call themselves to our attention. First, the 5-element array begins to show side-lobes, which suggests that a truly large Bruce array might require some redesign to keep the growth of these side-lobes in check. Second, none of the patterns are truly symmetrical. The side-lobes of the 5-element version show this fact most vividly, due to their differing sizes. However, if you carefully trace the patterns of the other versions, you will discover that each favors one side over the other. The difference is once more not operationally significant, but interesting as a facet of the basic design of an antenna that is never completely symmetrical.
Fig. 13 provides us with a composite view of the gain changes with base height for the 4 versions of the array. In each case, the rate of change is not high with changes of base height, although the peak gain points are clear. More significant for this collective gain graph is the fact that we obtain diminishing returns with each added element to the array. At the optimum base height, the 3-element version has a gain advantage of 1.92 dB over the 2-element version. However, adding the 4th element nets us only 1.52 dB of added gain, and adding a 5th element gives us only another 1.08 dB. The progression downward in increased gain continues as we add elements. For any practical installation, of course, there will be a point at which adding elements does not warrant the added wire or other mechanical support requirements for a very large array.
The changes in take-off angle are consistent for all of the arrays, so we may by-pass a graph of that factor and turn to the feedpoint resistance, shown if Fig. 14. What is most obvious from both the tables and the graph is that the feedpoint resistance increases as the base height decreases. Depending upon the feed system that a designer selects, the change of feedpoint impedance with base height may assume considerable significance.
However, we should note another interesting aspect of the change, garnered by looking at the 4 lines. The step in resistance for any base height is larger when going from an even number of elements to an odd number than when going from an odd number to an even number. Relative to the total array, the feedpoint of versions with an odd number of elements is horizontally centered. However, when the number of elements is even, the feedpoint is always horizontally off center. Late, we shall show a way to correct for this, but first, we should also examine the reactance curves for the 4 versions of the Bruce array.
The reactance curves appear in Fig. 15. In each case, the reactance is most capacitive or least inductive at or near the optimal base height. The reactance becomes more inductive as we decrease or increase the height away from the optimum level. In addition, the reactance increases at a higher rate the more elements that we use, especially at lower base heights. However, the amounts of reactance are in step with higher resistive components of the feedpoint impedance that accompany the use of more elements. Hence, we would see very little change in SWR sweeps keyed to the resonant impedance of each version.
Some Miscellaneous Variables that Affect Performance
Different builders experience differing circumstances, so we should address at least a few questions relating to differing constructions for the Bruce array. The first question concerns what happens if one chooses a different diameter wire for the array. The answer is simple: not much, although one might be tempted to tweak the design while in the modeling phase of the work. The following table shows the results for using AWG #8 through AWG #14 wire to make the 4-element version of the array.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Affects of Different Wires sizes on a 4-Element Bruce Array
Wire Size Gain TO Angle Feedpoint Impedance
AWG Diameter (") dBi degrees R +/- jX Ohms
8 0.1286 5.27 17 243.3 - j 15.4
10 0.1019 5.24 16 245.8 - j 4.3
12 0.0808 5.21 16 248.6 + j 7.2
14 0.0641 5.18 17 251.7 + j 19.0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The differences in gain are almost wholly a function of the RF losses in the wire, since with lossless wire, there is only a 0.01-dB difference in gain between the thinnest and thickest wires. Likewise, the increases in the feedpoint resistance reflect the same phenomenon. The differences in the TO angle are likely phantoms, occasioned because the actual TO angle is close to 16.5 degrees, resulting in a rounding difference at a 1-degree resolution.
One should likely avoid the temptation to tweak the model due to a change in wire size within the limits in the table, simply because differences in the ground quality will have a much more profound affect on performance. I ran the 4-element array over the standard set of ground qualities at its 0.15 wavelength optimum base height and obtained the results in the following table.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Affects of Ground Quality on the Performance of a Bruce Array
Ground Quality Gain TO Angle Feedpoint Impedance
Type Cond/D.C dBi degrees R +/- jX Ohms
VP .001/5 3.96 20 236.7 + j 18.0
P .002/13 5.02 18 245.7 + j 12.9
Ave .005/13 5.21 16 248.6 + j 7.2
VG .0303/20 7.82 13 258.6 + j 2.3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The changes in the feedpoint impedance with changes in ground quality are far less significant than the changes in the far-field pattern for the array. The chart seems to encourage one to move to a location with better soil for vertical antenna performance. However, the degree of improvement over a standard vertical antenna, such as a ground-mounted monopole with a radial system, will be comparable for each level of soil quality.
Very often, one receives recommendation for placing extensive radial systems beneath SCV-type vertical arrays. To see what effect such a radial system might have, I placed systems of 32-radials, each 0.25 wavelength long, beneath each of the vertical elements. To keep from having modeled wires intersect at other than wire ends or segment junctions, I placed the radial systems alternately at 0.001 wavelength and 0.0015 wavelength below ground. Fig. 16 shows the resulting model in outline form. For the check, I used the same soil qualities as in the preceding table. The antenna base height remained at 0.15 wavelength above ground.
The following simple table shows the results. For the sake of easy comparison, I have replicated the results without radials over each ground quality from the preceding table.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Affect of Adding Buried Radials to a 4-Element Bruce Array
Gain TO Angle Feedpoint Impedance
dBi degrees R +/- jX Ohms
Very Poor (conductivity 0.001; dielectric constant 5)
W/O Radials 3.96 20 236.7 + j 18.0
With Radials 4.21 20 247.6 + j 25.6
Poor (conductivity 0.002; dielectric constant 13)
W/O Radials 5.02 18 245.7 + j 12.9
With Radials 5.17 18 255.9 + j 16.3
Average (conductivity 0.005; dielectric constant 13)
W/O Radials 5.21 16 248.6 + j 7.2
With Radials 5.33 17 255.0 + j 10.1
Very Good (conductivity 0.0303; dielectric constant 20)
W/O Radials 7.82 13 258.6 + j 2.3
With Radials 7.84 13 260.3 + j 3.1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The net gain increase for adding the extensive radial field ranges from 0.25 dB over very poor soil to 0.02 dB over very good soil. The chief detectable effect centers on the feedpoint impedance, which becomes less dependent upon soil quality with the addition of the radial system. The far-field effects, however, tend to be minimal. Indeed, the basic soil quality beyond the radials has the greater affect upon far-field strength and take-off angle. The advantages of adding radials come at a cost of 8743' of AWG #12 copper wire for the total of 128 radials. Whether or not to add radials to the Bruce array is, in the end, a builder's decision.
There is one system of feeding the Bruce array that does demand a radial system, but that will be part of our concluding section on alternative feeding systems for the array.
Some Alternative Feeding Systems for the Bruce Array
One alternative feed system that we have already noted is the use of a vertical wire running from the center of a lower horizontal wire to ground, with the feedpoint placed in series with the wire at ground level. Fig. 17 sketches the full required system to feed the Bruce in this way and still obtain a vertically polarized pattern.
Since the connecting wire is a form of vertical monopole fed against ground, we require a radial system to complete the antenna subassembly. Early versions of this mode of feed used as few as 2 radials, but I have modeled the system (imperfectly) using 32 radials. The imperfections stem from the requirements for a buried radial system combined with the requirements for the segment lengths on either side of the segment on which one places the source or feedpoint. The radials are 0.001 wavelength below ground, and we require a junction at ground level. Ideally, the segment above this junction should also be 0.001 wavelength long, as well as the one above that. Beyond that height, the remaining segment lengths of the vertical wire may taper toward the segment lengths used in the horizontal wire of the Bruce array where the vertical wire joins it. The model that I used is close to meeting these requirements, but may be imperfect at lower heights. The imperfections likely inflate the gain figures by a small amount.
Over average ground, with a 32-radial system beneath the vertical feed wire, I obtained the results in the following table. The aim of the exercise was to see at what height and length of feed wire the array might approach resonance.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4-element Bruce Array with a Center-Wire Feed System
Base Ht Gain TO Angle Feedpoint Impedance
WL dBi degrees R +/- jX Ohms
0.15 5.47 16 301.8 - j297.8
0.20 5.42 15 210.0 - j156.1
0.25 4.82 13 191.4 + j 12.0
0.247 4.87 13 190.6 + j 1.5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
At the optimum height relative to array gain, the center vertical feed wire shows considerable capacitive reactance. One alternative for the builder is to simply run a length of parallel transmission line to an antenna tuner located at the transmitter location. However, a second alternative is to increase the length of the center vertical wire and raise the height of the array until the wire shows a feedpoint impedance near resonance. Hence, I have included those steps in the exercise.
The results suggest more precision than the model will bear. Nonetheless, a length just under 0.25 wavelength appears to provide the near-resonant feedpoint impedance. However, we should not only note the drop in gain, but as well the increasing rate at which gain drops as we pass through this region. A length change of 0.003 wavelength yields as much gain decrease in this region as the initial change of 0.05 wavelength shown in the table.
As well, elevating the antenna to a base height in the 0.24 wavelength region creates undesirable pattern effects. Fig. 18 shows the difference in the elevation patterns broadside to the array for 0.15 and 0.24 wavelength base heights. Excepting some possible special operational requirements, I shall assume that the left-hand pattern is the more desirable.
There will always be array builders who prefer to use coaxial cable from the transmitter location to the antenna. We have already seen that we can accommodate such users with a reactance-compensation and 4:1-balun system. However, we have other choices. Fig. 19 shows one of them that is especially applicable to the 4-element Bruce array.
The modeled transmission lines from the two center verticals go to a center junction, with one line reversed to achieve a 180-degree out-of-phase current situation between the two elements. With out-of-phase feeding of the two elements, each shows a feedpoint impedance of 125.0 + j 5.4 Ohms. The physical length between elements is 0.27 wavelength, requiring a physical lengths for each feedline of 0.135 wavelength. If we use RG-83 coax (characteristic impedance: 125 Ohms), the electrical length will be about 0.16 wavelength or so. When we reverse one line, the net impedance is close to 65 Ohms with virtually no reactance.
Fig. 20 shows the 65-Ohm SWR curve derived from the model of this system. Below it is the SWR curve is we use 50-Ohm coax at the junction. The use of 75-Ohm coax would show an SWR curve canted in the opposite direction from the 50-Ohm curve. Although this system has promise, it depends upon the ability of the builder to place a cable support post for the junction of the RG-83 coax length with the main feed cable. As well, the success of this system also depends upon obtaining close to the ideal feedpoint impedances on the two center vertical wires, and this figure depends upon both array base height and the soil quality below the array.
Since most Bruce arrays of modest proportions (up to 5 elements) will likely have only end supports, with non-conductive cables between them to support the array wires, we should also mention the potential for feeding the array at one end. Let's stay with the 4-element Bruce and see what happens when we give this system a try.
Fig. 21 shows the general scheme for such an end-vertical feed system. In general, the end support also supports the feedline, which then goes to the center of the vertical. For our 4-element Bruce array using AWG #12 copper wire and placed 0.15 wavelength above ground at its base, we obtain a maximum gain of about 5.3 dBi at a 17-degree TO angle--in other words, essentially the same performance we obtained feeding a center wire. The modeled feedpoint impedance is 254.0 + j22.7 Ohms, very close to what we obtained when feeding a center wire. A 300-Ohm feedline would be a close match for this system and minimize voltage, current, and impedance excursions along the line on its way to the transmitter location.
In Fig. 22, we see the modeled azimuth pattern for the array. The only notable feature is the slightly greater pattern tilt toward the fed end, relative to feeding one of the center wires. This tilt results from the slight losses in the wire from one end of the array to the other, but has virtually no operational significance.
We can also end-vertical feed the array at the lower corner. The performance does not change, but the impedance rises to 356.7 - 82.8 Ohms, which is natural as we off-center-feed a dipole. The impedance will increase relative to the center-feed impedance, and the length may no longer be resonant. Such a feedpoint is amenable either to standard parallel transmission line feeding techniques or to adjustment of the array dimensions. Once more, the exact value obtained for the corner feedpoint impedance will depend upon array base height and soil quality.
Although the length of these notes precludes coverage for other bands, the 80-meter data should provide sufficient guidance for scaling the antenna either to 160 meters or to 40 meters. The Bruce array has good potential among the SCVs in providing higher gain in a shorter overall horizontal length than the bobtail curtain. However, the cost is the need to use a higher base height than required for the open-end array. Ultimately, the choice of vertical array (rectangle, bobtail curtain, or Bruce array) for the low bands may depend upon the layout of user's land both horizontally and vertically.
Updated 01-13-2003. © 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.
Return to SCV Index