Dimensions for commercial antennas come from a variety of sources. Some the manufacturer provided. Some I found without provenance on websites. Others individuals measured and reported in detail. Still other antennas I measured myself. I round raw measurements, usually to the nearest 1⁄16″ for U.S. units. I resolve inconsistent dimensions from multiple sources by picking the value that seems most likely to be encountered in practice.
The designer rounds dimensions for noncommerical antennas or lists them as decimal numbers for the builder to round. Antenna files list element half-lengths since the AO Antenna Optimizer takes advantage of antenna symmetry. For best accuracy, double the values, round to the desired precision, and measure whole lengths. Multiply cubical quad parasitic half-sides by eight, round, allow for splice overlap, and measure total wire length.
To avoid error accumulation, measure element positions instead of spacings. Distance between elements or other conductors is from axis to axis unless otherwise noted.
I use a generalized notion of front-to-back ratio as a measure of pattern quality. Conventional F/B is the ratio of forward power to that in the opposite direction. I use the ratio of forward power to the peak power within a specified region to the rear of the antenna, normally the rear half-plane (90°–270° azimuth). When the rear region is 180°–180°, the measure is F/B. For any other region, I call it F/R, which stands for front-to-rear ratio. When not 90°–270°, I note the region limits. To be explicit, I entitle F/R graphs Front-to-Worst-Backlobe Ratio.
I model linearly polarized antennas in free space to maximize generality and to facilitate comparsion among designs. But since height above ground may affect the azimuth pattern of highly directive designs, I show F/R curves for various antenna heights for some models. I may also provide dimensions optimized for a particular height.
Azimuth patterns of circularly polarized designs vary with antenna height and ground quality due to differential variation of horizontal and vertical ground-reflection coefficients. I have standardized on a boom height of 20 feet for outdoor designs, a peak height of 8½ feet for indoor designs, and average ground quality. Results will differ for antennas installed at other heights or over other types of ground.
Models over ground are for flat earth. Results will differ for irregular terrain.
When the antenna feedpoint coincides with an element, I model the conductor as continuous and place a voltage source at the feedpoint. This procedure sidesteps unknown feedpoint gap detail. For noncoincident feedpoint terminals, I add a jumper wire and place a source at its center. To isolate the model from balun effects, I do not model balun leads. The leads form a transmission line whose impedance is higher than that of the antenna. This adds inductive reactance that alters SWR and mismatch loss. Avoiding this effect makes it easier to predict performance when replacing a ferrite balun with long leads with a halfwave coaxial balun to reduce ohmic loss.
Noncommercial antenna models assume isolated elements. They should not pass through a conductive boom nor use conductive mounting brackets. It is possible to compensate for these effects, but only when the boom diameter and mounting details are known. Specifying these would limit the design generality. Stauff polypropylene or polyamide clamps make excellent element mounts.
If I know their dimensions, I correct for conductive mounting brackets on commercial antennas. I use the YO Yagi Optimizer to derive either a short conductor section whose diameter is equivalent to that of the element plus bracket, or an entire element whose length and diameter are equivalent. Except at the band edges for some high-performance designs, mounting brackets little affect wideband antennas. But they can seriously disturb narrowband or spot-frequency designs.
Element numbering begins at the rear. For example, the first director is next to a driven element.
A metallic boom will not detune elements as long as they are spaced a small distance from it by nonconductive mounts. When the boom bisects an element, signal currents induced into each half-element cancel. But the boom can induce current in a phasing line whose conductors are not equidistant from it. This can happen at line crossover and may degrade the azimuth pattern.
A nonconductive boom eliminates induced current and can simplify element mounting. Fiberglass tubing is strong but may be expensive or hard to obtain. PVC or ABS pipe is cheap and readily available. Pipe with thick walls can support small antennas with little sag. Use boom guys for longer antennas. Wood works fine for noncritical designs. But unless very dry, its residual conductivity may detune elements that pass through it. It may add dielectric loss at high-voltage points if used for loop spreaders. Weather-treat wood for long-term outdoor use.
Boom length means electrical boom length, the distance between first and last elements.
An offset mast and clamp may couple to a nearby element. Locate the mast midway between elements whenever possible. Position a log-Yagi mast next to a phasing line crossover to approximately cancel any line interaction. An orthogonal conductor electrically vanishes only when it is in the antenna plane of symmetry. A centered mast attached below the boom may benefit designs with very low backlobes.
For long-boom designs, locate the mast at the point where torque is zero in steady wind. This is approximately the center of the boom for antennas with no reflector boom. If this unbalances the antenna, rebalance it by adding weight inside the light end of the boom.
Vertically polarized and circularly polarized antennas work best rear-mounted. When that's impractical, use a nonconductive mast section in the vicinity of the antenna to avoid degrading performance.
When the antenna and feedline terminal impedances differ, signal power reflects at the feedpoint, reradiates into space, and is lost. Mismatch also may degrade the receiver noise figure with a nonoptimal source impedance, an effect not modeled.
Conductor loss due to material resistivity normally is insignificant. But it may rise at a band edge where antenna currents increase for many designs.
Forward gain includes mismatch and conductor losses. Modeling results list SWR, mismatch loss, and conductor loss so you can spot unusual values. Listed results include no other losses.
Loss for 75:300Ω ferrite baluns is about 0.5 dB for small cylindrical baluns with twinlead terminated in spade lugs and intended for indoor use. It is about 0.75 dB for larger outdoor cylindrical baluns with unterminated copper leads and for indoor push-on baluns with screw terminals. Loss for a 75:300Ω RG-6 halfwave coaxial balun is 0.05 dB midband and 0.12 dB at the band edges. Loss for an RG-6 coiled-coax balun is 0.05 dB.
I have seen up to 0.5 dB loss for a ferrite power combiner, but typically it is 0.3–0.35 dB. Loss for a junction splitter followed by 22″ of 54Ω Belden 8219 RG-58A/U to transform 37.5Ω to 75Ω is about 0.1 dB.
Plastic end caps may cause dielectric loss due to the high electric field at element tips. Loss may occur at high-voltage points along phasing lines that use dielectric line separators.
Loss for RG-6 feedline is about 2 dB/100′. Check the manufacturer's data for your cable for a more exact figure.
A short wire can have significant reactance at VHF. Use the shortest possible leads for coax connections and feedpoint capacitors. If you must use two capacitors to create a nonwstandard value, parallel them to reduce inductance. Insulation or waterproofing material may add shunt capacitance. Use low-permittivity or porous dielectrics and minimize interconductor volume. Stray feedpoint reactance can degrade SWR and increase mismatch loss. Plastic end caps that detune elements by dielectric loading may shift the frequency response.
Use galvanized, cadmium-plated, or stainless steel hardware. Use an antioxidation compound such as Noalox, Ox-Gard, or Jet Lube SS-30 to inhibit corrosion. Bimetallic washers with copper and aluminum surfaces can prevent corrosion due to dissimilar electrode potential between these metals. So can weatherproofing a joint, but avoid silicone sealants that smell like vinegar. Their acetic acid will corrode metal.
Due to their specialized element mounting brackets and phasing lines, attempts to duplicate commercial antenna designs using substitute parts may not be entirely successful.
Adhere closely to the dimensions and construction guidelines for noncommercial designs. You can substitute conductor material of equal diameter, but avoid bimetallic corrosion and high resistivity (chrome plating and stainless steel are rather lossy). ⅜″ and 10 mm tubing should be interchangeable for wideband designs since their diameters differ by just 5%.
You can substitute 6063-T5 aluminum angle from Home Depot for ⅜″ or 10 mm elements. The 0.5″ × 0.5″ right-angle shape is electrically equivalent to a 0.4″ round conductor. For accurate conductor spacing, mount all elements with the angle oriented the same way.
SWR and pattern measurements can be inaccurate for antennas near ground. To virtually eliminate ground effects when measuring SWR, point a directional antenna straight up with the rear element a few feet above ground. Be sure to remove any nearby conductors when making a measurement, including yourself. The azimuth pattern of a vertically polarized antenna can be measured near ground, but that of a horizontally polarized antenna is best measured at installation height.
Though mechanically challenging, stacking horizontally polarized antennas side by side works better than stacking one above the other. Vertical stacking causes high mutual coupling that detunes elements and creates large backlobes. Moreover, elevation patterns differ for antennas at different heights, which reduces stacking gain from free-space values. Finally, horizontal stacking greatly narrows the forward lobe, which can reduce interference.
Stacking for maximum forward gain generates large secondary forward lobes. A reasonable trade-off is to stack so that these lobes are down 20 dB at 98 MHz. This yields 2.0 dB stacking gain for typical feeder and ferrite power combiner losses and approximately half the beamwidth. For this trade-off,
d = 76.3 + 0.415g²
where d is boom stacking distance in inches and g is forward gain in dBd for a single antenna at 98 MHz.
To avoid degrading performance, the crossboom should be nonconductive and the feeders should drop vertically for several feet. Beginning at the feedpoints, install current baluns at 30″ intervals to help decouple the feeders.
Unless its elements bisect and are orthogonal to the FM antenna elements, mount any antenna designed for a frequency between 50 and 150 MHz on a separate mast. Outside this range, mount the booms at least ten feet apart on a single mast. Closer spacing may degrade the pattern of designs with very low backlobes. Whenever possible, model the antenna combination instead of relying on these general guidelines.
Sensitivity analysis evaluates the robustness of a design to construction errors. To each symbol the modeling program adds and subtracts a tolerance. Listed are the largest forward gain and F/R degradations over the stated frequencies. The numbers can suggest when a particular measurement needs extra care.
I have standardized on a tolerance of 1 mm (0.04″) to facilitate design comparison. It should be possible to construct an antenna to this accuracy using ordinary hand tools. Symbols that represent element half-length use half the standard tolerance since the antenna builder can and should measure the full length.
Sensitivity analysis varies one dimension at a time and does not address accumulated performance degradation. To maximize antenna performance, measure all dimensions as accurately as possible.