First, we shall look into some suggestions for improving standard J-pole performance by increasing the length of the radiator. Our goal will be to understand why most of them do not hold promise of success.
Second, we shall look at a longer version of the J-pole that does work: the collinear J-pole. Along the way we shall look at the question of how well it works.
Finally, we shall examine a model of the Jagi, the J-pole-driven Yagi. Why it does work will become the final exam to see if we really do understand the J-pole.
Fig. 1 illustrates the most common suggestions for increased length that I have heard. The 5/8-wavelength radiator idea emerges from ground-plane antenna ideas. A 5/8-wavelength vertical has theoretically the highest gain of any ground-plane monopole. The 1-wavelength suggestion emerges from the idea that if 1 half-wavelength radiator is good, then 2 must be better. As well, each half-wavelength section ends in a high impedance, which is what the matching section needs to see. The 1.25-wavelength notion stems from the extended double Zepp (EDZ), which is essentially 2 half-wavelength sections at the outer ends with a phasing section in the middle.
Unfortunately, none of these ideas promises much when modeled either in free-space or over ground. As I have throughout these notes, I shall place each antenna in this final section 10' or 120" above average ground. (Whenever we compare an antenna to a J-pole, the comparator will be elevated to a height that places its region of highest current at about the same height as the equivalent region of the J-pole. For a standard J-pole at 146 MHz, that region is about 30"-40" above the antenna base.) As well, unless otherwise specified, all J-pole variants in this final set of notes will use 0.375" diameter aluminum elements, and the matching section will use a 2" spacing. This construction is both feasible and allows models in NEC-4 that result in average gain test (AGT) values close to ideal (1.00 or 0.0 dB).
To see what happens if we simply extend the length of the J-pole radiator, let's model the suggested new versions in free space. We shall be especially interested in the elevation patterns, which correspond to the E-plane patterns of vertically polarized antennas like the J-pole.
Fig. 2 contains a wealth of data, since it contains both the free-space E-plane patterns and the current distribution representations for all three antennas. However, by combining the data, we can begin to see the way in which end-fed linear elements differ from center-fed elements of the same length.
The 5/8-wavelength pattern gives us our first warning. We begin to see the emergence of multiple lobes at roughly 45-degree angles to the horizontal and vertical axes of the pattern plot. As we increase the length of the radiator to a full wavelength, the emergent lobes become fully formed. We are used to seeing such lobe formation in wires without the matching section only when the wire length was about 2 wavelengths.
The corresponding current distribution graphics to the right of Fig. 2 verify the wire length--that is, that I am not presenting models that falsify performance. The 5/8-wavelength graphic shows a single full half-wavelength rise and fall of current magnitude above the minimum close to the matching section. Of course, the matching section does not start at a current minimum. Therefore, the currents in that section show very significant imbalance--to the point that the section almost functions as a simple antenna fold-back rather than as a transmission line.
The full-wavelength current graphic shows 2 full excursions of current above the matching section, with a current null about in line with the top of the matching line pair. Since the currents are better balanced, the radiator itself largely controls the lobe formation. Nevertheless, with end feed, the pattern shows 4 distinct lobes.
The pattern for the 1.25-wavelength radiator version of the J-pole resembles the pattern we might expect from a center-fed 1.5-wavelength wire. The 6 lobes represent in a 1.5-wavelength wire the emergence of the 4 corner lobes of a 2 wavelength wire and the decrease of the horizontal axis lobes typical of a wire 1 wavelength or less. The corresponding current distribution graphic seems to confirm this analysis if we take the lowest current excursion as being completed by the unbalanced match line section.
Those interested in lobe formation for end-fed wires may wish to read the short Appendix to this episode to develop the full portrait of current magnitude and phase along a wire longer than 1/2 wavelength in order to fully appreciate the lobes in the E-plane patterns for the suggested J-pole improvements. What the models tell us is that the longer radiators are likely to produce more high-angle radiation than low-angle radiation when we place them over a ground.
Fig. 3 confirms our suspicions. Each J-pole model has its antenna base at 120" above average ground. Each elevation plot lies along the plane of the antenna legs so that a small front-to-back ratio is detectable. In each case, the strongest lobes are at much higher angles above the horizon than we would desire for point-to-point communications on 2 meters. The 5/8-wavelength pattern is usable, but at a lower level than a standard J-pole. The gain at a 6-degree elevation angle is about 2.5 dB lower than for a standard J-pole (2.6 dBi vs. 5.1 dBi). The 1.25-wavelength version does show a low angle lobe--corresponding to the free-space lobe along the horizontal plot axis. However, this lobe is about 1.5 dB weaker than the main lobe of the standard 1/2-wavelength J-pole (3.5 vs. 5.1 dB). The main lobe of the 1.25-wavelength J-pole is indeed quite strong at about 6.4 dBi, but the 42-degree elevation angle is hardly ever useful for point-to-point communications.
The lesson that we might take from these models is that the standard-type J-pole's best radiator length is in the vicinity of 1/2 wavelength--as adjusted for the match section requirements and the element diameter. Hence, for a 146-MHz design frequency, the radiator will be somewhat shorter than the 40.2" true half wavelength. Still, this lesson does not mean that we cannot make longer improved-performance J-poles. It simply means that we must make them in a more nearly correct manner.
Fig. 4 shows the modeled version of the collinear array. As with many of the models in this series, it is a proof-of-principle model, not a construction blue-print. All the wire sections composing the model are 0.375" diameter, even though one might ordinarily build the 18.5"-long 2"-wide phasing section of thinner material, such as 0.1" diameter wire. Indeed, the selection of the 2" width resulted from modeling needs to keep the wires sufficiently far apart so as not to result in modeling errors.
All dimensions are in inches in the figure. If you compare the model with the corresponding 2"-wide 3/8" diameter J-pole in the preceding segment of these notes, you will discover that the short leg has grown from 19.6" to 23.6", with the feed tap moved upward by 4". The half-wavelength radiator sections are not of equal length: 41" for the upper and 35.4" for the lower. In part, the differential results from adjusting the upper section to achieve the desired 50-Ohm feedpoint impedance (in addition to the matching section adjustments) without harming the overall gain of the system.
Fig. 5 shows the current distribution along the antenna, at least in terms of magnitude. The phase-line section changes the current phase by 90.5 degrees, and its placement and length are critical to obtaining full performance from the antenna. Ideally, the current magnitudes at the two junctions of the phase-line should be equal. However, lengthening the upper radiator results in a detectable variance. The 90-degree phase shift is more significant in this application than equalization of current magnitudes.
The lower radiator current minimum coincides nicely with the top open end of the matching section. However, since the current magnitudes are not equal between the lines at the top end, considerable imbalance exists along the matching section. The result is a spurious lobe that will become evident in elevation patterns for the antenna.
In Fig. 6, we find the elevation patterns for the 102" long collinear J-pole with the base 10' above average ground. In the plane of the elements, the antenna shows a maximum gain of about 7.7 dBi, with a small (0.4 dB) front-to-back ratio. The average gain of the antenna shows up by looking at the pattern at 90-degrees to the plane of the elements, as shown in the lower half of the figure. The gain is about 7.4 dBi. This value is about 2.3-2.4 dB higher than the average gain of a single-radiator J-pole. Indeed, although many folks like to bandy the gain advantage of a collinear arrangement as 3 dB greater than a single section, we rarely obtain in real antennas more than about a 2.0-2.5 dB increase in gain.
Compared to many vertical antennas, the collinear J-pole shows a remarkable reduction in high-angle radiation. For any vertical collinear array, the only place from which to obtain energy for increased gain at lower elevation angles is from the high-angle energy of a single section. If the single section lacks high angle radiation, creating a collinear version of the antenna will rarely yield improvements in performance that justify the added structure.
The upper portion of Fig. 6 shows the spurious lobe that results from both the imbalance in currents in the matching section and from the imbalance of currents in the phasing line. For the model, the phasing section protrudes in a straight line in the direction of the open end of the matching section. In a physical implementation of the design, the phasing section would likely wrap around the main element axis in a circle.
To evaluate the collinear J-pole--especially in terms of whether the increased height and complexity of construction is warranted--we should compare its performance with some antenna or other. since the J-pole is 102" long, the vertical extended double Zepp, which is about 100" at 146 MHz, is a good comparator.
The EDZ used in this test is 0.375"-diameter aluminum and 100" long. The length allows the model to set the base at 120" above ground, since an exact correlation to the collinear J-pole regions of maximum radiation is not feasible. In the J-pole, those regions are roughly centered in each of the two radiator segments. In the EDZ, the regions of maximum current are located about 1/4 wavelength inward from each end--when we feed the antenna at the center. (End-feeding the EDZ results in a somewhat different distribution of current--enough to disrupt the anticipated lobe formation.)
Fig. 7 shows the elevation pattern for the EDZ. Since nothing in the structure disrupts the circularity of the pattern, this single plot suffices for all possible axes along which we might take elevation patterns. The gain at 5.5 degrees (about the same elevation angle as for the collinear J-pole) is just under 7.6 dBi, that is, only about 0.2 dB higher than the average gain of the collinear J-pole. The high-angle lobes are reflections of the typical ears that accompany any EDZ pattern. They shrink as we reduce the length of the antenna--but so to does the overall antenna gain. If we extend the length much beyond 1.25 wavelength, the ear-lobes will grow to dominate the pattern, resulting in a pattern with predominantly high-angle radiation.
For the version shown, the feedpoint impedance is about 120 - j380 Ohms. There are many schemes for matching an EDZ to a 50-Ohm feedline. One way to do it with lowest loss is to use a section of parallel transmission line and a shorted stub. One might also place a network of fixed or variable components at the feedpoint. Finally, one can employ transmission line sections for the inner portion of the antenna to effect an impedance transformation within the antenna itself. The latter two matching methods tend to reduce overall system gain--that is, to create some loss of radiated energy--more than the match-line-and-stub system. However, the match-line-and-stub system tends to narrow the operating bandwidth of the antenna. (There are notes on feeding EDZs among the collection at this site.)
I note the matching requirements for the EDZ so that one might make a fair comparison between the EDZ and the collinear J-pole. For roughly equivalent performance, we have roughly equivalent size and construction complexity--a not too unusual situation for antennas. In which direction one goes may ultimately rest upon which antenna most closely coincides with one's favorite shop techniques.
The collinear J-pole does offer one advantage over many other types with which it might compete: a wide operating bandwidth. Fig. 8 shows the 50-Ohm SWR curve for the modeled collinear array. The band-edge SWR values are 1.41:1 at 144 MHz and 1.35:1 at 148 MHz. I suspect most users would find these values tolerable.
Since the current distribution is normal, we might wish to use the J-pole as the driven element in a vertically polarized parasitic array. In The ARRL Antenna Compendium, Vol. 5, Michael Hood, KD8JB, presented a 3 element Jagi (J-pole driven Yagi) (pp. 62-65), using his plumbing-pipe J-pole (from Vol. 4 of the Antenna Compendium) as the driven element. Rather than try to model the complex arrangement of various pipe sizes that he used, I took a standard J-pole model that used 0.375" diameter aluminum with a 2" match-leg spacing as the driven element for a model of a 3-element Jagi.
Fig. 9 shows the dimensions of the final model. I aligned the elements of the driven element, since the degree of misalignment of element centers was not severe. Each parasitic element also uses 0.375" diameter aluminum. The final dimensions resulted from juggling element spacing and lengths along with the J-pole feed to obtain the best performance and a 50-Ohm feedpoint impedance.
The dimensions for the elements emerged initially from a 3-elements Yagi of standard design, but using 0.5" diameter elements. Therefore, in checking the performance of the Jagi, I used this standard Yagi for comparison, placing the center line at about 140" above ground. The Jagi base is 120" above ground, so the center lines are in crude alignment.
Fig. 10 shows the comparative elevation patterns of the two antennas. Over ground, they both exhibit a maximum forward gain of about 10.3 dBi. As the traces show, the standard Yagi has a slightly smaller set of rearward lobes.
As shown in Fig. 11, at an elevation angle of 6.2 degrees, both antennas share the same forward gain and beamwidth (about 110 degrees), typical of Yagis turned for vertically polarized use. The front-to-back ratios of the two antennas differ by only about 3 dB (17.4 vs. 20.3 dB), with the standard Yagi having the advantage. Considering the liberties I took in modeling the Jagi, the difference is neither unexpected nor operationally significant.
The standard Yagi had a feedpoint impedance in the neighborhood of 40 Ohms at 146 MHz, suitable for direct feed, but likely not for maximum bandwidth without some form of matching. In contrast, the Jagi had a design frequency impedance of almost exactly 50 Ohms. The band-edge 50-Ohm SWR values were 1.74:1 at 144 MHz and 1.83:1 at 148 MHz. The addition of the parasitic elements does affect the operating bandwidth of a J-pole used as its driver, but not sufficiently to prevent full band coverage on 2 meters. Fig. 12 shows the full 50-Ohm SWR curve.
Yes, the Jagi is not only feasible, but is--as well--an interesting variation on the Yagi that solves the feedline dress question for many vertically polarized applications.
As well, the collinear and parasitic J-pole applications that we examined only scratch the surface of what has been done and what might be done with the J-pole as the starting point. However, my interest in the J-pole was not to form a catalog of antennas to build. Instead, the goal has been to understand a bit better than I did before what is going on in the operation of a J-pole. These notes simply formalize a bit what I learned along the way.
This conclusion does not mean that I shall never add to this series of notes. It simply means that for the moment, other antennas are exerting a stronger call.
However, I know of no other antenna that has its own limerick:
I know a wonderful aerial: the J-pole.
You can build one without stealing a payroll.
When you tune it to peak,
You'll hear signals so weak,
That from work you'll likely go AWOL.
The results of the test were interesting. The J-pole with a 5/8-wavelength radiator has a free space elevation (E-plane) pattern very similar to a linear element 3/4-wavelength long and fed about 5-10% up from the bottom, as shown in the top pattern of Fig. 14. The J-Pole with a 1-wavelength radiator shows a free-space elevation pattern very similar to that of a 1.25-wavelength linear element fed about 15-20% upward from the bottom. The middle pattern of Fig. 14 shows the result. Finally, the J-pole with a 1.25-wavelength radiator has a free-space elevation pattern very close to that of a 1.75-wavelength linear element fed 10-15% up from the bottom, as shown in the bottom plot.
Compare the plots in Fig. 14 to those in Fig. 2. As well, compare the current distribution graphics in both figures. In all cases, the elements show zero current at the top end, with typical 1/2-wavelength current curves proceeding downward from the top. The variation in the normal curve occurs at or near the bottom end. In the linear elements, the current below the feedpoint describes a sharply tapering curve toward zero. The curves in this region and just above the feedpoint represent the equivalent linear element behavior of the corresponding J-pole within the matching-section region, given the specific current imbalance for each of the elongated models.
The lessons from this exercise are two. First, the imbalance of currents in the J-pole matching section make this section part of the total radiating system. Only in the case of the 1/2-wavelength radiator are the currents within the matching section sufficiently balanced to yield elevation patterns that are close enough to those of a center-fed linear element the same length as the radiator to allow us to generally ignore the matching section. As we saw in Fig. 2 of Part 3, if the wires are not very closely spaced--as with a twinlead J-pole--the resulting free-space pattern is offset from the horizontal plotting axis to be useless in determining antenna behavior. Elements as close together as 1" at 146 MHz are far enough apart to require that we perform all analyses above a ground. Hence, in the end, we cannot ignore the radiation from the matching section even for the classic J-pole with a 1/2-wavelength radiator.
Second, we cannot expect J-poles to perform like rough analogs of their radiator sections alone. The version with a 5/8-wavelength radiator does not perform like a 5/8-wavelength monopole with a ground plane system. The 1-wavelength and 1.25-wavelength radiator versions of the J-pole do not perform like elements of similar length in free-space when fed at their centers. The only correct equivalents of J-poles are ones that take into account the radiation from the matching section of the antenna.
All of the test models of long-radiator J-poles used standard J-pole configurations in the notes thus far. It is possible to develop a non-standard J-pole--that is, a J-pole with no shorting strap for the feed point, but instead a single feed point at the base strap--using longer than a 1/2 wavelength radiator. Consider the following published design for a 5/8-wavelength J-pole: The elements are 3/8" tube or rod, with the long radiator about 59.5" long and the short rod about 14.25" long. The rod spacing is about 6.375". As with all of our J-pole models, we shall place them about a wavelength above real ground for deriving patterns.
The resulting antenna will show a 50-Ohm SWR well under 2:1 across all of 2 meters and the adjacent CAP frequencies as well. However, as we saw in the last episode, very wide spacing tends to yield significant pattern distortion relative to the ideal of a circular azimuth pattern. See Fig. 15.
The pattern distortion ends up producing a pattern with a significant departure from an azimuth circle, and the higher angle of much of the radiation gives no more gain at the lowest lobe than we might obtain from a shorter non-standard J-pole. The gain at a 3.1-degree angle averages about 5.1 dBi, with peaks that are 70 degrees off a line made by the long and short radiator rods. The peak gain is 5.8 dBi. The gain in the direction of the short rod is 5.4 dBi, with a gain of only 3.7 dBi in the opposite direction.
The upshot is that the possibility of making a longer J-pole in non-standard form does not yield the benefits of lengthening a vertical monopole over a buried ground plane. The J-pole configuration tends to increase high angle radiation so as to negate the value of the added radiator length. The very wide spacing may give us some ease of construction, but it contributes as well to pattern distortion. A closer spaced standard or non-standard J-pole with a 1/2-wavelength radiator will perform as well with respect to low-angle gain and better with respect to pattern circularity.
Go to Main Index