\require{AMSmath} \require{eqn-number}


The J-Pole Antenna

There are some missing graphics on this page - I'm working on it.

The J-Pole antenna is a common omni directional antenna used in amateur radio, particularly on the VHF and UHF bands.  They look pretty weird, as can be seen in the following photo:



The picture to the right is of a 2m amateur radio band J-pole I built from an EZNEC simulation. It has weathered six or seven Maine winters. It has been up long enough for lichen to grow on the supporting 2x4. So long in fact that the simple mounting block is breaking up and needs replacement.


J-pole analysis using NEC


C. Bronson H. Crothers
September 14, 2006 - February 27, 2010



Abstract

Unlike many commercial 2m antennas the J-pole is obviously a DC short. Yet, when properly constructed, the antenna clearly makes a good match for typical 50 ohm coaxial transmission line, otherwise no one would build or use such an antenna. Obviously more is going on with a J-pole than meets the eye. This article is an attempt to explore how the J-pole antenna transforms feed-line impedance to free-space impedance, all the while providing gain found only in the more expensive commercial products. Refinements to simulations of a typical J-pole antenna, using NEC, are made so that physical antennas built from the simulations, yield “no tune” antenna designs, with predicted matching, and high probability of predicted gain. Also explored, through simulation, are the ramifications of various mounting configurations.

Background

In this article, I make use of the suffix dBi in reference to antenna gain. The “i” in dBi refers to an isotropic radiator. An isotropic radiator is defined as “a hypothetical loss less antenna having equal radiation in all directions”[1]. An isotropic radiator is a theoretical antenna and of little practical use, except that it provides a reference to which we can compare one antenna to another. Isotropic antennas, as a class, can be broken into two broad groups, hemispherical isotropic and spherical isotropic. A spherical isotropic antenna radiates equally in all directions, with a wave front appearing as a sphere of ever-increasing radius. Such a radiation patten can be achieved in free space where there are no objects, within a relevant number of wavelengths, that could effect the pattern. A hemispherical isotropic antenna has the same pattern except one half of the sphere is cut away. In this discussion of J-pole antennas I will make no distinction between the two types of isotropic antenna. This discussion will be limited to terrestrial antennas, so the hemispherical isotropic will be assumed throughout.

Another antenna that will serve well as a comparison will be the venerable dipole. Keep in mind that an ordinary dipole has about 2 dBi of gain. Also, bear in mind that the dipole is not an omni directional antenna. There are nulls along its axis. A dipole’s 2 dBi of gain is realized in directions perpendicular to its axis. There are nulls, on axis, at either end. Mounting a dipole vertically will orient these nulls in directions that are less significant if a more omni directional pattern, in azimuth, is desired. In such an orientation, feeding the dipole can be tricky, since the feed line itself may distort the radiation pattern and alter the feed point impedance.

For a true omni-directional antenna, the simplest solution is some form of asymmetric antenna with as flat a radiation pattern as possible. Flat means that the preponderance of the radiated power is oriented toward the horizon. For many of us such an antenna is purchased (rather than constructed) at considerable expense. When shopping for such an antenna sticker shock can be severe.

Partly as an insulator to sticker shock and partly do to the novelty of the J-pole antenna’s construction, I thought it might be interesting to try and see how an antenna as goofy looking as a J-pole could possibly work. Claims for the antenna’s performance, in its simplest form, are near enough to high performance commercial verticals in the 2m band, that it seemed well worth pursuing. Since the cost of materials for a 2m band J-pole is less than $20, yielding a gain of 6.5 dBi or more, it would seem silly to overlook such an efficient and cost effective antenna.

All of this begs the question as to how one might go about all this. Rather than build more J-pole antennas than I could shake a stick at, I opted for the easier path of simulating their performance on a computer using NEC. Using the simulation results I could then build specific antennas to check my results. Toward that end I purchased a license for Roy Lewallen’s EZNEC, which is a front end for Windoze of the NEC (typically NEC 2)package. Some things have shown themselves in J-pole simulations using EZNEC...

  1. A J-pole does not need to be and should not be tied to ground to radiate or tune properly.
  2. It can be tied to ground, at the shunt, if an RF choke at the shut is used.
  3. A well tuned J-pole, built from 3/4 inch hard copper pipe (thin wall preferably), designed for 146 MHz, can resonate below 1.5 SWR across the entire 2m band, with 6.78 dBi of gain when mounted 8m (26ft) above ground.
  4. Feed point impedance changes little with respect to antenna mounting height.
  5. The addition of tuning screws at the resonator and radiator are not necessary. If the elements are pre-fit and measured carefully, no tuning should be required.
  6. The antenna will require no tuning if the antenna has no RF conductive path to ground.
  7. If the antenna does have an RF conductive path to ground, that path becomes part of the antenna and will alter the radiation and feed point characteristics.
  8. EZNEC with NEC2 has some warts.  It cannot simulate feed points properly when conductors of widely different diameters are used.
  9. The maximum gain occurs on the side of the antenna with the resonator.
  10. The minimum gain occurs on the side opposite the resonator.

The experimental procedure was to simulate a J-pole antenna in EZNEC then build the antenna and test its matching using a microstripline forward/reflected power coupler (FRPC). The voltages from the FRPC were then used to calculate the SWR values listed in this article.


Table 1.

Antenna Height

Elevation
 (deg)

Max Gain

Gain (90deg)
 (dBi)

Resistance
(ohms)

Reactance
(ohms)

Asymmetric Loss
(dBi)

2m

8

4.75

4.26

49.77

-0.3116

1.43

8m

3

7.22

6.78

49.69

-0.492

1.28

10m 

3

7.34

6.9

49.0

-0.5074

1.28

30m

1

8.31

7.86

49.61

-0.44

1.26


  • Resistance: real resistance at the feed point in ohms
  • Elevation: take off angle above horizontal in degrees
  • Reactance: neg. values capacitive, pos. values inductive
  • Asymmetric Loss: difference between Maximum and Minimum gain
  • The FRPC is an uncalibrated device, built by the author on 62.5 mil (1/16 inch, 1.59mm) FR-4 board. Its use was limited to ratio metric comparisons of the two output voltages generated during a given test. Table 1 lists a summary of the simulation results of a modeled antenna at different heights above ground level.

    Before detailing the simulations and results, a quick review of NEC[3] and EZNEC should be made. In order to understand the results of the simulations, it is helpful for the reader to be familiar with the origins and limitations of NEC/EZNEC. Familiarity with NEC’s methods of antenna simulation helps to explain the approach to antenna modeling used in this evaluation of the J-pole antenna.  


    A few words about EZNEC/NEC

    NEC stands for Numerical Electromagnetics Code. The preface of the NEC-2[3] documentation reads as follows... (including spelling mistakes/typos)

    The Numerical Electromagnetics Code (NEC) has been developed at the Lawrence Livermore Laboratory, Livermore, California, under the sponsorship of the Naval Ocean Systems Center and the Air Force Weapons Laboratory. It is an advanced version of the Antenna Modeling Program (AMP) developed in the early 1970’s by MBAssociates for the Naval Research Laboratory, Naval Ship Engineering Center, U.S. Army ECOM/Communications Systems, U.S. Army Strategic Communication Command, and Rome Air Developent Center under Office of Naval Research Contract N00014-71-C-0187. The present version of NEC isthe result of efforts by G. J. Burke and A. J. Poggio of the Lawrence Livermore Laboratory...

    Other institutions and agencies are mentioned as responsible, the University of California and the U. S. Department of Energy are among them.

    Over the years NEC has been revised and re-written. When last I checked NEC-4 was the latest version. Since NEC was written by university types using public funds, we as citizens can lay some claim to it. Unfortunately, the very latest versions are something of a trade/military secret. One can obtain the latest version but it is rather pricey (many hundreds dollars) and they do ask some questions about who is buying it. However, John Q. Public can download (at no cost) earlier versions which aren’t so sensitive. One of the download-able versions is NEC-2.

    NEC-2 is the version that is native to EZNEC, a Windows based implementation of NEC-2, written by Roy W. Lewallen. Roy is quick to point out that later versions of NEC can be used with his EZNEC shell.

    EZNEC can be found on the web at http://www.eznec.com/

    EZNEC uses NEC-2 as its calculation engine. EZNEC will allow you to hook together straight “wires” in just about any conceivable configuration. According to the NEC-2 documentation, NEC-2 will model surfaces directly, however EZNEC will not. If you want to simulate such things, you have to build them as grids of straight wires. EZNEC will allow you to define the length and diameter of your wires or antenna elements. You can tailor your conductor properties or select from 5 different preset values.  For the purposes of J-poles, the element or “wire” material chosen was copper. EZNEC also provides “canned” ground types as well as user defined ground. It is also possible to select “Free Space”, should you wish to see how your antenna will perform somewhere between here and Pluto.

    The approach NEC-2 uses to evaluate an antenna is called the Moment Method. From the NEC-2 documentation[3]:

    Wires are divided into short straight segments with a sample point at the center of each segment while surfaces are approximated by a set of flat patches or facets with a sample point at the center of each patch.

    Unfortunately, there is no provision in EZNEC for the “patches or facets” approach to surfaces.  On the other hand, entering wires or elements into an EZNEC model is straightforward and since the J-pole has no “surfaces”, we will not be concerned with making a grid of any kind.

    We will have to define a feed point for the antenna. This is not as simple as it sounds. However, once a better understanding of feed points in general (and EZNEC's approach to sources in particular) is obtained, feed points will move from problematic to pragmatic. A simple solution to feed point modeling of the J-pole antenna will be outlined, but it is not without its problems. In NEC, feed point modeling remains a bit of a Black Art. Balanis [1] defines the input impedance of an antenna as:

    ...the impedance presented by an antenna at its terminals or the ratio of the voltage to current at a pair of terminals or the ratio of the appropriate components of the electric to magnetic fields at a point.

    This impedance is composed of resistance and reactance. The resistance can be further broken down into the radiation resistance and the loss resistance. The loss resistance is largely a property of the material that the antenna is made from and like everything in RF work is somewhat dependent on frequency.  For the purposes of J-pole analysis, in the 2m band, it will be assumed that 3/4 inch copper pipe has little resistive loss. Therefore, any power we deliver to the antenna will be dissipated in the radiation resistance and the reactance. It is also assumed that the antenna will function in reception as it does in transmission.

    The decomposition of feed point impedance into radiation resistance and reactance, begs the question, which or do both, contribute to radiated power? Once again Balanis[1]explains:

    It is through the mechanism of the radiation resistance that power is transferred from the guided wave to the free-space wave.

    Here “guided wave” means the fields that are created on the antenna. In transmission our feed line delivers an RF current to the antenna. This current, in turn creates the electric and magnetic fields on the antenna. In reception, the antenna carries whatever fields are present around it. It is RF energy from the 2m HAM band in particular that the J-pole is designed to carry.

    In either case the free electrons in the copper pipe of our J-pole are set in motion and subsequently are detected by our receiver or driven by our transmitter. Note that it is the load of the radiation resistance not the reactive load that radiates or sources RF energy out of or onto the antenna. Any reactive load seen at the feed point will dissipate power but will not contribute to radiation or reception. This means that the reactive portion of the feed point impedance should be kept as low as possible, if the antenna is to make the most of the power delivered to it. This will be true in either receive or transmit mode.

    It should come as no surprise then, that great pains should be taken in order to model the feed point of an antenna accurately. There is no point in optimizing an antenna that can’t be fed with a practical source. For the 2m band a practical source would be some form of low loss 50 ohm coax.

    Feed Point Simulation

    In the case of the J-pole, this had been a weak spot for the transition from simulation to physical antenna. Early simulations focused on getting a feel for how the radiation pattern responded to changes in element length and spacing. Consequently, not much attention was paid to feed point issues. As the relationship of the gain vs. element length and spacing was optimized, the match to 50Ω coax suffered. J-pole gain rose to 6.97 dBi when fed from the middle of a 6cm shunt at the base of the antenna. Unfortunately, the feed point impedance was 17.25Ω plus 0.03Ω of capacitive reactance. It was nice not having much reactance, on the other hand a minimum of 3 VSWR, when driven with 50Ω coax, was not so nice. Adapting this antenna (same overall dimensions) to a mini-dipole feed (more on this later) reduced the gain to 6.43 dBi and drove the input impedance to 114.8Ω plus 47.58Ω of inductive reactance. The mini-dipole feed was located 6 cm up from the shunt. This was a more “traditional” location for the feed but the input impedance was still bad.  In addition, a good portion of the power will be going into the reactive term. Remember the reactive term consumes power but does not contribute to radiation.

    Some of the results above were originally obtained without the mini-dipole feed version for comparison. The mini-dipole feed was stumbled upon later (I had a brief moment of inspiration, it’s nice to know the gray matter is still somewhat functional), in an attempt to more realistically model the entire antenna.

    Just what is a mini-dipole feed?

    And why is it so important?

    NEC implements an RF source by distributing the source over the length of a segment of an antenna element or “wire”. In NEC a given wire is broken into one or more segments. A source can be thought of as being inserted into the middle of the segment, from which the two ends of the segment then drive the adjacent segments of the wire. If the wire is a single segment the two ends of the wire are driven directly.

    This can present a problem with some antennas, since they often are defined using only one or two wires.  In the case of an inverted-V antenna, if each wire has only one segment, setting a source on one of the wires will effectively place the source in the middle of the arm of the antenna.  This is not where most of us find it useful or effective to couple the antenna to our feed line. Among other problems a feed point at such a location forces the feed line to part of the antenna structure. This can cause some problems with the pattern. However, computers and simulations don’t care about such practical matters. They will do whatever you want no matter how impossible, stupid or dangerous the outcome may be.

    EZNEC does its best to allow the designer freedom to model as she/he pleases, but there are limitations. When the designer violates those limitations EZNEC is courteous enough to issue a warning.  This helps to keep the model from leading the user down the path of impossibility. These warnings are issued mostly because EZNEC is concerned that it will not be able to calculate the necessary currents properly. The program doesn’t “know” or “care” about antenna design and theory. It simply applies the analysis techniques written into the program by the authors.

    Simulators, like EZNEC, will allow you to put a source on any element whether it makes physical sense or not. This is a great advantage if you are trying to come up with some new arrangement of antenna elements for difficult situations like cars, boats, bicycles, snow mobiles, airplanes, jungle gyms...  whatever.  You can put your wires together any which way, pick an element to feed the mess, press go and presto, there is your radiation pattern. EZNEC couldn’t care less that, the feed point has 1000 ohms of real resistance and 200 ohms of reactance. The problem here is that you will have to hook the thing up to, as is often the case, a 50 ohm single ended transmission line (commonly referred to as coax).  The point is that distributed sources, as in EZNEC, don’t always lend themselves to such types of real feed line.

    Really, the culprit here is the antenna. In this case a J-pole. The simplest EZNEC model of a J-pole would have only three elements of a given diameter.  The three elements would be the obvious three geometric pieces of the J-pole. The three pieces will be referred to as the radiator (long vertical section), the resonator (short vertical section) and the shunt (short horizontal section connecting the other two). Following the guidelines of EZNEC, each of the three “wires” would be broken into segments.  This is done so that the technique EZNEC (method of moments in NEC-2) uses to simulate the antenna can optimize the calculation of the current distribution on each segment of each element (or wire if you like) of the design.

    Although antennas are often made of wire the term “wires”, in EZNEC, is a misnomer.  In EZNEC, a “wire” could be 3 inches in diameter and 10 feet long.  The specifics of EZNECs modeling of “wires” isn’t important as long as we know that it can faithfully deal with copper tube as well as dipole wire. As long as we are careful to specify the proper diameter of our conductors, elements or “wires” (all the same thing as far as EZNEC is concerned) EZNEC will handle the rest.

    With the proper conductor specs loaded into the model, we can now add a source and see what the radiation pattern looks like. O.K.

    Where does the source go?

    Source modeling on the J-pole

    Typical construction articles often say “attach the coax a couple inches or so up from the shunt.” EZNEC doesn’t care. You can drive almost any point on the antenna and it will likely, although not always, simulate successfully. The drive point must be in a segment of an element though.  A source point can’t bridge two elements. Depending on where the source is placed, there may or may not be much of an impact on the radiation pattern. What will change dramatically is the source impedance. You may not get anything like 50Ω.

    This can be a problem.  You can fool around with the lengths of the antenna elements and get a super pattern but you will never be able to feed it with anything real.  Unless you have a balun that can match any impedance over infinite bandwidth, occupies no space, and has no effect on nearby electric or magnetic fields. This is not a likely prospect.

    What is good about this ability to place sources almost anywhere, is that you are free to mess around with antenna geometry. Such freedom is essential for the designer to get a feel for what matters and what may not mater in a particular design.  Once armed with that information, which is not always obvious, the model may be refined by adding some complexity.  What kind of complexity?  Complexity such as nearby power lines, other antennas, loads, reflectors, directors and of coarse feed points or driven elements.

    The problem with the J-pole is how to define the feed point more like reality and less like fiction. Most J-poles will be fed by coaxial cable. A simple implementation of the source point, in EZNEC, leads to an asymmetric feed that splits the radiator, the resonator or the shunt at the point chosen. In the constructed antenna, the element where the feed point was placed in the model, will have to be broken so that the transmission line can be attached.  This leads to something of a discontinuity on the element chosen. It is possible to select a point solely on the resonator where the impedance is 50Ω. However, such a feed point may have a large fraction of its impedance in the form of reactance.

    A feed point location that is too high on a vertical antenna has some serious tuning and radiation pattern ramifications. Remember that the antenna will be fed by coaxial cable. With the feed point so high and the coax in such close proximity, there will be a significant amount of coupling to the cable and therefore distortion in the desired fields in the U section. In effect the feed line becomes another element in the antenna. This is not desirable.

    There is another reason not to split any of the three major elements of the J-pole antenna. As the antenna is constructed, no element is interrupted by the feed point or otherwise. A test J-pole antenna will be constructed, from the results of the simulations, with 3/4 inch copper pipe.  The goal of the simulations and construction will be a continuous copper antenna with no breaks or discontinuities in its structure. Therefore construction should also yield no breaks or discontinuities in its electrical properties. In EZNEC, by definition, a source “breaks” the continuity of the element on which it is placed. While the “break” isn’t physical per se, it is logically there and in any case, it does define a phase reference for the antenna.

    3/4 Wave Vertical
    Figure 1: 3/4 Wave
    vertical simulation in
    EZNEC

    Such a phase reference can be seen in figure 1 which is a graphic depicting the current distribution on a 2m band 3/4 wave vertical antenna at 8m above ground. In the figure, the small circle denotes the feed point, the straight lines are the antenna wires, and the curved lines are the current distribution.  Note that the "wires" in this model are defined as 19 mm in diameter, which it more or less 3/4 of an inch.

    I have chosen a 3/4 wave length antenna  for this discussion because it is simpler in geometry yet has more or less the same length as a 2m J-Pole.  It also has a similar current distribution, across its entire length as the J-Pole.

    The impedance at the feed point on the 3/4 wave antenna totals (R = 331Ω and X = -398Ω) about 517Ω in magnitude.  It is positioned about a third of the way up from the bottom of the antenna.  It could just as easily be located a third of the way down from the top.  The current distributions would be largely the same.  However, the current maximum would occur on the lower portion of the antenna and the current minimum would occur on the upper portion of the antenna.  Thus, the rational for calling a source a phase reference.

    You will note that, in this simple case, the designer has little control over the current distribution, once the source is placed. Given the current distribution of figure 1, it would be tempting to “simply” drive the antenna near the current maximum. This should provide a much better match to 50Ω coaxial cable. The problem is that moving the source to another location also moves the phase reference. In turn the current maximum will also move, in this case to the bottom two thirds of the antenna. Without an additional element, load, source or other phasing device to force the current distribution to follow a specif pattern, little can be done to change the feed point location without distorting the desired current distribution.

    With the currents vanishing toward the ends of the 3/4 wave antenna, driving in those locations, with a low impedance source, is out of the question. The ends of the antenna are high impedance points. Why? Because the ends of a conductor, 19 mm thin wall copper pipe or typical antenna wire, terminate in a high impedance. Otherwise known as a discontinuity. Clearly, driving the 3/4 wave antenna with a low impedance source, like 50Ω coax, is tricky. It may be possible to add a load or two on the antenna to improve the match, but loads introduce losses as well.

    The 3/4 wave vertical and the Dipole: J-pole relatives

    Since this 3/4 wave antenna will become important when trying to understand the J-pole, a close look at its performance is warranted. Figure 2 is an elevation slice with data, from a 3/4 wave vertical at 146 MHz, 8m above ground simulated using EZNEC. The dot on the elevation slice, is the cursor position (3 degrees above horizontal) to which the values in the figure refer.

    Elevation Slice of 3/4 wave vertical
    Figure 2: Elevation slice of 3/4 wave vertical from EZNEC simulation.

    There are 9 lobes per side in the figure. The largest lobe (greatest magnitude) being at the quite low take off or launch angle of 3 degrees. The graph depicts a vertical slice down the vertical axis of the antenna. The complete pattern would be defined by rotating this graph 180º around the vertical axis. The gain of this antenna is 6.62 dBi at 3 degrees of elevation. Take note of the pattern and gain of this simple antenna and compare it to the pattern and gain of the J-pole to follow.

    This antenna would be a good performer, if we could only drive it properly.  Take note that the length of this antenna is about the same length as the radiator element of a J-pole, specifically about 3/4 of a wavelength.

    Vertical Dipole at 8 m

    A look at a vertical dipole at the same elevation is also instructive.  

    Dipole lengths can vary greatly.  Some construction articles suggest cutting a dipole back from its theoretical length to better match 50Ω coax.  The motivation for this is sometimes suggested to be that the native impedance of a dipoles center is ~72Ω.  The implication is that if one were to shorten it slightly then the impedance would drop closer to 50Ω coax.  While this is true, a shortened dipole's resistive impedance can be brought down close to 50Ω, the consequence is that the reactance goes up.  This causes the overall match of the transmission line to the antenna to become far worse than the 72Ω case.  This especially so since the reactive component, and the coaxial cable itself (far greater VSWR) saps much of the available power.

    The real reason to cut a dipole back from its ideal length is to correct for velocity factor.  Velocity factor is a compensation factor for the change in the speed of light due to the material an antenna or transmission line is made of.  The ARRL Handbook , and Antenna Book have tables with various specifications of conductors used in RF work.  Included in such tables is the VF or Velocity Factor.  It is usually given as a percent, sometimes it is presented as a simple decimal.  When electromagnetic energy travels in or on a material it rarely does so at its free space velocity.  This slower speed is expressed as a non-negative decimal equal to or less than 1 or as a percentage equal to or less than 100%.  Typical  "open wire" antenna conductors have a velocity factor of 0.97 or 97%.  Thus the electromagnetic wave travels a shorter distance over one cycle (its wave length) than it would in free space.  For a resonant antenna like the dipole or J-Pole it must be cut shorter to account for this change in velocity.
    2mDipoleAt8mVerticalPerf
    Figure 3: 2m vertical dipole at 8m in EZNEC

    A dipole, mounted vertically also has similar performance to the 3/4 wave, and J-Pole.  Figure 3 is a graph of the radiation pattern, in vertical cross section, of a 2m dipole at 8m above ground.  Note that the maximum gain also appears at a low takeoff angle, and is only slightly lower in magnitude.   Also note that the number of lobes is similar but they are more pronounced in the dipole.

    Recall that a dipole in free space has a radiation pattern shaped like an annulus, as can be seen in figure 4.  In the figure the axis of the antenna is oriented in the vertical direction.  Most of us are more familiar with the dipole oriented horizontally in HF band applications.  Away from ground, in free space, the only thing that changes between these two cases is the orientation of the annulus axis.  Rather then the axis aligned vertically, a horizontal dipole's radiation axis would be rotated 90 degrees from what figure 4 shows.

    The maximum gain for the dipole is 2.11 dBi, and occurs around the middle.   Also note how uniform the pattern is.  It has none of the lobes that are shown in figure 4, where the same size antenna is with 8 m of a ground plane.  The difference between these two instances of the same antenna underscores the influence ground or a ground plane can have on a given antennas performance.

    Simply by moving a 2.11 dBi vertical dipole closer (8 m away) to the ground, the same antenna can improve to 6.36 dBi gain for low angle signals.  The reason for this is the signals reflected off of the ground plane mix with the signals directly radiated from the antenna.  As these signals intermix some cancel, and some reinforce.  If arranged properly they work to our benefit.  Fortunately, in this case, "arranging properly" simply means mounting the antenna at several wavelengths above ground.  At one wavelength (2 m) above ground the gain drops to 3.48 dBi.  At one half wavelength (1 m) the gain is down to 2.11 dBi.  By 2 wavelengths the gain is up to 5.1 dBi.  So how close the ground is to a vertical dipole has a strong influence on its radiative properties.  Note that at all these heights, the impedance at the feed point does not change much.

    The 3/4 wave antenna behaves in a similar fashion.  However, the gain loss is about 1 dBi less at the low mounting heights.  By 2 wavelengths (4 m)  the gains are about even.  These two relatives of the J-Pole are useful in that they offer a window into how a J-Pole functions.  The dipole for its familiarity, the 3/4 wave for its geometric similarity, and similar current distribution.

    But neither of these J-Pole relatives are practical to attach to our feed lines.  Attaching a feed line to one or the other will cause heavy distortion in the radiation pattern, as in a dipole, or a poor match, as in the 3/4 wave vertical.  There has to be a better way.  The J-Pole serves as a better way to couple to our feed lines.  But how best to model the feed such that our simulations yield valuable practical results?
    2mDipoleFreeSpaceVerticalPerf
    Figure 4: 2m vertical dipole in free Space in EZNEC

    A Realistic J-pole feed point; the mini-dipole

    Since a source, in EZNEC, is distributed across a segment of an element (wire), it is not possible to bridge two elements with the source. In effect, the source, though double ended, must be embedded in a single element.  There are ways in NEC/EZNEC to overcome this.  The "mini-dipole" is effectively one of the means suggested in the EZNEC documentation, although I happened uppon the idea independently.

    A source has current going into one end of a segment, and going out of the other end of a segment. If the current flows into one end of a source segment, then the same amount of current must flow out of the opposite end. Over a full period of whatever frequency the simulation is run at, the current will flow in and out of these two terminals accordingly.

    Segments cannot be connected at their respective mid points. This holds for normal passive segments, and source segments alike.

    These segment, and source restrictions lead to some limitations. A given antenna element (wire) cannot be broken up into an arbitrary number of segments. There is a range of segmentation that provides better fidelity of results. EZNEC has tools to help find a reasonable number of segments for the simulation in question.

    For our purposes the greater the number of segments, the more choices we have for choosing the driving location. But again we can only choose so many segments before the ability of the model to simulate the antenna suffers. It is a bit like having a measuring tape where one inch is the smallest increment. If you will be measuring things that are 100 feet long, you are probably going to be OK, you will have a good deal of precision but not too much. If you are measuring something a yard long, that's not so good.  Your resolution wount be too high.  If you have to measure something only 1 foot long, your resolution is very bad.  At the other end, if you are measuring something that is miles long, one inch isn't of much use.  The numbers are just too big.  Segmenting in NEC is a bit like that.  

    In NEC the place where tings get difficult is when the segments become "over square".  This is where the segments become so short that the diameter of the segment becomes a sizable fraction of the length of a segment.  If a segment begins to look too much like a square, and not so much like a rectangle, in a way the algorithm won't be able to figure out whether the current flows across the segment or along its length.  We want the current to flow along its length but if the segment is too square, it will be hard for the algorithm to keep track of that... sort of.

    Elevation Slice of 3/4 wave vertical
    Figure 5: Typical J-pole
    EZNEC model.

    Figure 5 is a view of a typical segmented 2m J-pole as modeled EZNEC. The green dots represent the segments of the wires in the model. In this case there are 6 wires. The small red circle indicates where the feed point is located.

    There are a wide variety of J-Pole feed types. A company called Arrow Antennas.com uses a piece of angle aluminum for their shunt in their dual band 2m/70cm J-pole. The configuration is a variation of what the ARRL Antenna Book calls an open stub feed. An open stub feed is when one or the other of the two main vertical segments is isolated from the bottom shunt, and driven by the center conductor of a coaxial cable. The shield of the coaxial cable is then tied to the shunt or "shorting bar" proper. 

    The first antenna almost every HAM builds, myself included, is a ground plane antenna.  The ground plane antenna is a variation of an open stub feed.

    In the case of Arrow Antennas.com.s OSJ 146/440 the shunt of the J-Pole is a piece of angle aluminum, which is tied to the coax cable's shield, with the center conductor tied to the longer isolated resonator. The longest element, the radiator, is tied directly to the shunt.

    Simulations, derived from Arrow Antenna's instruction manual, indicate the OSJ 146/440 has similar gain performance to a single band 2m J-pole. Its gain may be down about 0.8 dBi, in the 2m band. However, the feed point impedances for my simulations, did not agree with their published figures, so my simulation is a bit suspect.  There are ways to cope with this.  Perhaps a stepped diameter correction would help.  But as we have seen with the vertical dipole and the 3/4 wave vertical, feed point modeling has less of an effect on the gain than mounting hieght.

    The biggest problem with my simulation in the OSJ 146/440 case, is the implementation of the feed point. The feed point in this case is tightly coupled to the shunt, so I have lumped them together. How a feed point is constructed in a simulation can mean the difference between a useful simulation, one that accurately represents how an antenna constructed using the simulation results performs, and an otherwise useless set of data. Often the gain, and the pattern don't change much with modest changes in the feed point model but the calculated source impedance often does change dramatically. If the feed point modeling is poor, it will be impossible to achieve the otherwise good performance indicated. It's a tricky business. There are many ways to fool yourself into believing you have a hot performer.

    The difficult aspect of my OSJ 146/440 simulation is that the driven element is the 2m resonator itself.  In my simulation I placed the source at the lowest segment on the resonator.  There is a large discontinuity between the resonator diameter, and the shunt diameter.  Typically this will cause problems with NEC simulations.  So my OSJ 146/440 simulation/model is not what it should be.  Be that as it may, the gain results are consistent with typical J-Pole performance. 

    A Less Complicated Feed Model

    A less complicated feed point model is what I call the mini-dipole feed.  The feed can be thought of as a very small dipole, a mini-dipole, with two short (well below a quarter wave length) elements, one element terminates on the radiator the other element terminates on the resonator. 

    The EZNEC documentation puts the minimum length of such stubs to 0.02 λ (λ is the generic symbol for wavelength).  Or rather the help file says:

    The source shouldn't be placed on a wire which is shorter than about 0.02 wavelength, particularly if adjacent wires connect at an angle.

    Which is good advice.  There are other good pieces of advice in the Source Placement Precautions of the EZNEC help file.  Unfortunately much depends on context, and how the segments of the elements will interact.  In the case of the mini-dipole feed we are less concerned with radiative issues, and more concerned with connective issues.  We want to be sure the current in the feed line enters the antenna with good fidelity, and low impact on the overall antenna model.  We are not so concerned with the radiative performance of the segments and elements in the immediate feed point area.  As long as they do not impact the overall antenna model.  Thus we bend one or two of the source placement rules of thumb in a effort to improve the overall antenna model.

    This mini-dipole can be moved closer or farther away from the shunt so as to form the best match to 50 W as may be. In the amateur literature this is often suggested as a way of tuning the antenna. While this may allow one to obtain a good match to a given J-pole design, it may also be true that a substantial fraction of the power (transmitted or received) from the antenna is lost in a reactance. The overall match may look like 50 W, but much of the signal may never radiate or be received.  

    It is better to model and simulate so that the antenna's dimensions can be matched to the feed location (or vice versa), rather than move a feed point around AFTER the antenna is built, to get a good match.

    When implementing the mini-dipole feed on a constructed antenna, a connector for a coaxial cable fits in between the two stub elements.  Note that although the transmission line will be coaxial and therefore single ended, the typical feed point of a J-pole is balanced.  The idea is similar to feeding a dipole antenna with coax.  The feed point of a dipole is technically a balanced load yet with a good match, it can be fed with an unbalanced line.

    Using this feed configuration, you are effectively driving the radiator and resonator of the antenna with a small coaxial fed dipole. Such a feed in EZNEC is trivial. Simply define a wire that bridges the two conductors of the antenna, at whatever height above the shunt you design for. Divide it into an odd number of segments (1,3,5 etc. not too many not too few... I use 3 or 5), assign the source to drive the segment in the middle and you are all set.  EZNEC can tell you when you have made the segments too small.  To double check run a Segmentation Check, in case the program doesn’t check it interactively.  EZNEC will then tell you the real resistance and the reactance of the feed point (positive or negative, inductive or capacitive) in ohms. However, there will be a difference between 1 segment and 3 segments. Note that, depending on the diameter of the elements, 5 segments may cause the Segmentation Check to fail, so values above 3 may not be possible.  Overall try to make sure that the segment length on all the antenna elements are more or less the same.


    Something strange with NEC2


    But how large a diameter should the stub elements of the mini-dipole feed be?

    Should they accurately represent the diameter of the conductors used to connect the cable connector (SO-239) to the antenna?

    Here comes some voodoo (the Black Arts I alluded to previously).  If the entire antenna, including the mini-dipole feed, is simulated using 19 mm (for 3/4 inch pipe) diameter elements, the simulation agrees with the actual antenna built using the simulation’s dimensions. An exception is the mini-dipole feed. The mini-dipole feed of the physical antenna was built using 14AWG to 12AWG wire (stranded or solid) and later heavy copper strap.  The wire was laid across the radiator and resonator at the feed point, bent half way around the two and soldered.  It was then trimmed and cut in the middle so that an SO-239 could be soldered onto the two tails. The antenna was then connected to the feed line (9913 cable), driven with as much as 50W and the forward and reflected power was recorded at the band edges, and in mid band.

    The VSWR was about 1.45 at the edges of the 2m band and 1.1 at 146MHz.

    These results conform well to the symmetric mini-dipole feed simulation.  The simulation predicted 49.69 Ω of real resistance with 0.492 Ω of capacitive reactance for the feed point impedance.

    If the diameter of the mini-dipole feed in the simulation is changed to the more realistic diameter of 3mm of the 12AWG wire, instead of 20mm, the predicted feed point impedance rises sharply to 87.19 Ω of real resistance and 8.512 Ω of inductance. All other aspects of the simulation were unchanged. Only the asymmetry of the mini-dipole feed was different, 3mm vs. 20mm for the element width.

    Since the symmetrical mini-dipole feed simulation agreed with the actual antenna, it is the author’s opinion that NEC2/
    EZNEC has some difficulty dealing with geometric element discontinuities.  At the time of this writing I have not gotten far enough through the NEC2 documentation to shed any additional light onto this problem.  The EZNEC documentation does point out that NEC does not take kindly to step discontinuities in element diameter.  EZNEC has utilities to help  smooth over such transitions.  I have yet to attempt to implement them.  Unfortunately the element that forms the mini-dipole feed is 6cm long and it may be difficult to segment 4 or six more elements for the transition.  Here is another one of those engineering puzzles where practice is much simpler than theory.  In the end the mini-dipole feed may be as good as any solution to the stepped discontinuity problem.

    Some more experimentation is needed to sort this problem out. In the mean time we seem to have a workable model.


    Antennas Are Transformers?

    With the feed point well defined and looking more like reality it should be possible to tweak the design to get a good match and pattern.  After that, an antenna built from the structure defined in the model parameters should largely load up and work as predicted.  After running many simulations with poorly defined feed points and having bad SWR on the antennas built from them, despite the use of a balun, I re-modeled the antenna using the mini-dipole feed. I built the antenna as modeled and could run it at full power (50W) no problem. Without a balun.

    But why does a J-pole work?

    Everything looks like a short circuit!  

    There are three pieces to this puzzle.  The first puzzle piece is to get yourself to stop thinking about antennas as antennas and think about them as transformers.  Transformers between your feed line impedance and free space impedance.  Once you are reasonably comfortable with that idea, you are ready for the second piece of the puzzle. The second piece of the puzzle is to break the J-pole antenna into two conceptual sections. One section is the section of the radiator above the resonator, the other section is everything else.  This is where a program like 
    EZNEC comes in handy.  The third puzzle piece we will get to after some discussion of the first two.


    A perspective of the dipole antenna


    A dipole is an old friend to those of us in radio and we often take the venerable dipole antenna for granted.  It is about the simplest and cheapest particle accelerator, radiation generator and one of the most useful scientific tools ever invented. The cost (way too inexpensive) of which the federal government would never be associated with.  The dipole does a couple of amazing things for radio communication. The first is that it takes a bunch of things none of us have ever seen (electrons... particles) and makes them zip back and forth in a specific way. This causes them (the electrons) to form some fields (electric and magnetic) on and around the antenna, which we also have never seen, that someone else can collect with another dipole, and some other gear, a long way away.

    Which brings us to the second amazing thing a dipole does.  The gear we use to make the magic of radio is built by humans for humans.  Radio waves exist in an entirely different world than the one we are accustomed to. We often create them and work with them or their related electric and magnetic fields in our 50 ohm transmission line, microstripline, cavity resonator, tank circuit devices but their home is free space. Free space is not 50 Ω land. Nor is it kilo ohms or mega ohms. Free space is about 377 Ω 
    [1] [2]. Somehow, our old friend the dipole transforms our 50 Ω feed line impedance to the 377 Ω impedance of free space. In one way or another all antennas make this transformation for us. Now I have seen many a transformer in my time, and a dipole is about the last thing I would accuse of being a transformer. But in a sense, that is exactly what it is.

    The point is that somehow we need to transform the RF energy in our radio gear to RF energy in free space. From there, it can radiate in all directions (or a specific direction) such that someone else can receive it.  For obvious reasons, this transformation needs to be efficient, practical, and cost effective. Antennas are transformers and with some careful thought to their design, they are efficient, practical to construct, and cost effective.

    The J-pole is an antenna and therefore must be a transformer. If we carefully dissect the J-pole, we can see just where this transformer action takes place.  So here we go with step two. Lets break the J-pole into a couple pieces and see if we can get an idea as to how it works.  But where should the break be? 
    Our simulation offers us a clue.

    One of the many useful features of 
    EZNEC is that it will display a picture of your antenna with the RF currents on it. Figure 6 is a picture of our J-pole EZNEC antenna model similar to figure 5 but with some funny lines on it. 

    Current distribution on 2m J-Pole
    Figure 6: Current distribution
    on 2m J-Pole.


    The feed point is the apparent shunt (mini-dipole), between the radiator and the resonator, just above the bottom shunt. The little circle is the location of the feed point.  The little dots along the elements indicate the borders between each segment of the model.  Remember, segments are the sections that EZNEC (NEC-2) will try to calculate the currents of.  This wire frame EZNEC model looks not unlike a real antenna.

    Figure 6 shows the same antenna model as figure 5 but adds where the RF currents will be when it is driven by our feed line. The purple curved lines are representative of the current distribution on the antenna, when driven at its design frequency of 146 MHz.

    The most important feature of figure 6 is the current dip or null on the radiator at about the same height as the top of the resonator. At this point, some alarms may be going off in your head. The upper section of the radiator, above the top of the resonator, has the same current distribution as can be found on a dipole! This makes sense because that is about how long this section of the antennas is, about 1/2
    l.

    The current lines in figure 6 only show magnitude, they do not show any phase information. EZNEC will show the phase information as well. Figure 7 shows the same simulation with the phase info in the current trace. As shown in figure 6 you can see that the currents in the U section are nearly the same. It's a bit tricky to see, since the two elements are offset from one another but if you allow for the spacing between the resonator (short right vertical section) and the radiator (long left vertical section), the currents are a close match. Since the U of the antenna is a 1/4 wave section the currents must also be out of phase so they should, for the most part, cancel.
    figure05
    Figure 7: 2m J-Pole current distribution with phase.

    In figure 6 the same current information is shown except the phase information has been retained. The plot has been rotated so that the
    current densities in the 1/4 wave U section are shown at their maximum difference.  Figure 6 and figure 7 show the near cancellation of the current distribution in the lower 1/4 wave section of the antenna.   Figure 6, shows the cancellation, in so much as the magnitudes (taking care to compensate for the physical offset in the graph), when subtracted, would largely cancel.  Figure 5 in so much as the excursions in the opposite direction on the X axis would cancel when added together.  Each has about the same excursion but in opposite directions. The other small traces are for the currents in the small elements that make up the bottom of the antenna.

    The important current traces are the large opposing 1/4 wave section currents and the upper 1/2 wave current trace. If you look closely at the figures, you will note that the resonator’s curve has a slightly larger displacement than the opposing radiator’s curve. This means there is a phase reversal on the radiator at the height of the resonator.

    A quick inspection of the dimensions of the antenna will confirm that the overall radiator length is about 3/4 of a wavelength long (1.44m/56.7in) at 146 MHz.  Now take another look at figure 1. Notice that the current below the feed point of figure 1 is low, of opposite phase as the radiator, and drops to zero at the bottom.

    Since the currents in the U of figure 7 are in opposite directions the fields they generate will largely cancel. This should come as no surprise because the radiator and resonator are so close together, their fields interact in the same way as a balanced transmission line’s fields do. In this way, the net current, in this region of the J-pole, is small, just as in the 3/4-wave antenna.  This isn’t true of coarse since current is clearly shown in figure 7, but the fields created by those currents will cancel since the charges are moving in opposite directions. Whether the low field strength is due to low current as in the 3/4 wave antenna or due to field cancellation as in the J-pole U section, doesn’t matter. The fact remains that as far as the net fields are concerned the two antennas are quite similar. What is good about the J-pole is that we get the field and current distribution we want in the upper and lower sections of the antenna without the high impedance feed problems of the 3/4 wave antenna.

    The J-pole radiates as a 3/4 wave or dipole vertical with one important benefit, and one modest benefit.  The modest benefit is that like the 3/4 wave antenna the J-Pole has about 1 dBi more gain than a vertical dipole when mounted close to ground. The important benefit is that the J-Pole feed point, unlike the vertical dipole, will not interfere with the pattern, and unlike the 3/4 wave antenna the J-Pole can present a good match with low loss to typical 50 Ω coaxial cable.


    What about the real feed point?


    The mini-dipole feed point has been simulated at distances from 6 cm up to 14 cm (2.362 in, 5.512 in) up from the center of the shunt.  The shunt length and radiator-resonator spacing, has been simulated from 6 cm to 12 cm (
    2.362 in, 4.724 in).  The resonator is about one quarter of a wavelength long, which leaves half a wavelength of radiator above the resonator.  This half wavelength section is of coarse how long a typical dipole would be. Therefore, it should come as no surprise that the current distribution on the 1/2 wave section is similar to a dipole’s. The only strange thing about this instance of a dipole is that we aren’t feeding it at the center. We are end feeding it!

    You can’t feed a dipole from one of its ends!... Can you?

    The J-pole antenna proves that you can and efficiently to boot.

    But just how does this work?

    Here is the third and final piece to the puzzle.  The U section of the antenna, below the half wave radiator section serves two purposes. It is part phasing section and part matching section.

    Here is where I am going to go a little crazy on you. To complete the picture you need to break up this antenna puzzle a bit more, in order to put the pieces together. If you remove the half wave radiator, you are left with a big copper U. Think about this in the context of the same length of twin lead or balanced transmission line. If it were a piece of twin lead you could remove a small amount of the insulation from one end and solder the two wires together.  What do you have?  A 1/4 wavelength U shaped resonator/matching section.

    Why is this a matching section? Dig out that dusty Smith chart and recall that a full circle is half a wavelength, so a half circle is a quarter wavelength. The prime center of the chart is commonly normalized to 50Ω , but it could be 450Ω if ladder line were being used or any other reference resistance. The horizontal x-axis is the real resistance axis or real axis. Above the x-axis is inductive reactance. Below the x-axis is capacitive reactance. The extreme right, on the real axis is infinite resistance, to the extreme left is zero resistance. With a separation of 6cm and a diameter of 19mm, the U section of a J-pole has a characteristic impedance of 218.7Ω[2]. With the prime center normalized to 218.7Ω, 50Ω gets mapped onto the real axis at about 0.23 l.

    Remember that at resonance, a quarter wavelength section of transmission line will reflect RF energy off the impedance discontinuities at either end.  On one end is a short on the other end is an open circuit.  Over the coarse of one quarter wave length you transition/transform from a dead short to an open circuit/infinite impedance. This will only work at resonance. You can’t use an arbitrary length of transmission line!

    The matching section of the J-pole is one quarter of a wavelength. At one end is a short circuit, at the other end is an open circuit.  In the middle is the characteristic impedance of the line (218.7Ω), at a distance of one eighth of a wave from the shunt. A simple ratio tells us that 50Ω appears at about 0.23 or an eighth of a wavelength. Multiplying the ratio by an eighth of a wavelength gives 5.9 cm. This calculated distance is about the same as where our mini-dipole feed point is located, 6cm up from the shunt.


    Driving the dipole of a J-pole


    The center of a dipole has most of its energy in the form of current and less in the form of voltage. Usually we trim the ends of the dipole so that we obtain a ratio of voltage to current that gets us close to 50 ohms. Just as the smith chart calculations suggest, fifty ohms is much closer to a short than it is to an open circuit. Look at where the feed is on a J-pole, much closer to the shorted end than to the open end of the U.

    What happens at the ends of a dipole depends on which end you look at. One side will be positively charged, the other will be negatively charged.  Let’s focus on the negative end with respect to electrons.  We have some current going into the tip, but once it reaches the end, it can’t go any further. There isn’t any conductor left. Electrons are lazy and a bit republican in this case, so they just turn around and head back the other way. There are lots of electrons hanging around on one end of a dipole.

    Note: There is a proportional lack of electrons on the other end of a dipole. Don’t worry about the other end right now because it will get confusing trying to sort out which is coming and going when and where. The opposite side of a dipole behaves as the complement of the electron end. Think about the electron end for now and don’t worry about the other side of the dipole.

    If this end (either end really) of your dipole got too close to a conductor at ground potential all the electrons (whatever their direction) bunched up near the tip, like good confused democrats, would lash out in a blinding flash of RF (lightening bolt) energy and make a mess of everything... Anyway, the going and coming of electrons at the tip of the dipole, have most of their RF energy in the form of voltage. This is because on average as many electrons bounce back off the tip of the antenna (heading back down the dipole) as arrive at the tip.  We have one current going one way another current going in the opposite direction, the currents cancel so there is a current null at each tip. The energy has to go somewhere so it goes into the electric field of the antenna and therefore voltage.

    Putting these ideas together with Ohms law, we have some voltage divided by big current in the center of a dipole. That means the center has low resistance. Near the ends of the dipole we have big voltage divided by some current. That means the ends have big resistance. In terms of resistance, the ends of a dipole look more like the open end of our J-pole’s U shaped matching section.

    The U section is made of BALANCED transmission line. The separation between the conductors is larger than we are accustomed to but it is, never the less, a piece of balanced transmission line. Previously we calculated the characteristic impedance of the U section to be about 218Ω.  Cycle to cycle one conductor is pulling and the other conductor is pushing electrons. If you think of current, one conductor has current flowing in one direction the other conductor has a current of equal strength flowing in the opposite direction. The magnetic fields due to the current, cancel. If you think of voltage, one potential is positive the other is equally negative and the electric fields cancel.

    The fields don’t cancel completely though. There is an additional half wave length section above the U which is generating another set of fields. A 3/4 wave length vertical, at the same height, has about 6.62dBi of gain, when driven at its current null, about 1/4 of a wavelength up.  The feed point impedance (at the null) of such an antenna is very high.  
    EZNEC calculated 517Ω (magnitude), although it is likely to be much higher. Clearly the gain of the J-pole antenna is due to its 3/4 wavelength radiator, but we can’t drive it with 50Ω coax. The geometry of the resonator and shunt with respect to the 3/4-wavelength radiator, transform this high impedance load to something we can drive with 50Ω coax.

    Tapping near the shorted end of the J-pole matching section is 50Ω feed line friendly.

    The open end of the J-pole matching section is dipole end friendly.

    The entire 3/4 wavelength (part 1/2 wave dipole) radiator, part of which forms one side of the J-pole U, is free space friendly.

    As a result we have a high gain, inexpensive and easy to build antenna.

    There are some interesting things about this antenna.  The drive point is close to the shunt at the base of the antenna. This is very convenient. The current distribution of figure 4 shows that the highest currents in this region are down at the shunt. This means there is little ELECTRIC field to couple into our feed line as it falls from the feed point past the shunt and down the rest of our antenna tower or mast. There is a large MAGNETIC field in this region of the antenna.  This magnetic field could couple into our feed line. Happily our feed line can enter this area of the antenna perpendicular to the magnetic field and thus minimize any such coupling.  It is possible to bring the feed line down below the base of the antenna in a loop.  Doing so further minimizes any coupling that might occur in the feed line.

    I took the dimensions in the model description and did my best to faithfully cut the elements to those dimensions. That can be a bit tricky when using 3/4 inch copper pipe but I did my best.  The dimensions of my 
    EZNEC model and those of other articles on J-pole construction are not the same!  They differ slightly, but none the less, they do differ. My simulations were for good gain but not at the expense of a good match to 50Ω coax. For the feed point I soldered some 14AWG to 12AWG stranded antenna wire to the radiator and resonator 6cm(2.36in) up from the center of the shunt.  I cut this wire in the middle and soldered the two ends to a Teflon SO-239 panel mount connector. The radiator end went to the center and the resonator end went to the shield. I don’t think it would matter which is soldered to which, the simulation showed no difference and in theory it shouldn’t matter which goes where. When I get a chance I
    will swap the ends and see.

    How well does this antenna tune up? According to the model, I should see about 50.59Ω of resistance and well below 0.1Ω of reactance. When I drove the antenna from the low side to the high side of the band, the SWR was below 1.5 everywhere.


    Performance


    But how does this plumber’s nightmare perform? It performs about as good as many fixed antennas.  
    EZNEC can give us some numbers though.  It can also show us the pattern.

    The pattern of figure ? is an elevation pattern. It shows a cross section of where the strongest radiation will be above the horizontal ground plane. The antenna in this case is 8m (26ft) above medium, real ground. This is one of the canned ground types and probably is representative of ground in Maine, where my QTH is located. If you have better ground, the performance will be better. The gain peaks at 6.78 dBi, at 3 degrees above horizontal. The gain of this antenna rivals that of fancy base station antennas in the $150 or more price range! The program will also predict the SWR over a range of frequencies. For the 2m band 
    EZNEC calculated the SWR to be below 1.5 across the entire band.

    I built the antenna as mentioned above, except I added about 10cm(4in) of pipe below the radiator side of the shunt to make mounting easier. I mounted it outside on a 2x4 off my back porch. The antenna is about 12 feet above ground level. I fed the antenna with a rather long (70ft anyway) piece of 9913, without a balun. I drove the feed line, with my Forward/Reflected power coupler about a foot down the transmission line from my Yaesu FT-847.

    The VSWR results for the real antenna were below the predicted values. At the low end of the band the VSWR was 1.44. At the design frequency of 146 MHz the VSWR was 1.1. At the high end of the band the VSWR was 1.004. This implies that I ended up cutting the antenna a little short.

    At this time I have no real way of evaluating the pattern. I’ll have to work on that.


    Antenna height and grounding


    Given the close agreement of the simulation with the actual antenna, it may not be much of a stretch to believe some of the other results of the simulations.

    Antenna height above ground level, anywhere from 2m to 10m, had little effect on the feed point impedance.  How we get the antenna up that high is important.  As long as the tower was non-conductive, the following would be true.

    Height did have a great effect on the gain of the antenna. The higher up you mount it the greater the gain will be. Again, the tower or mounting pole must be non-conductive for this to be true.

    At 2m above ground, car or truck mounted, the gain is 4.26dBi and the takeoff angle creeps up to 8 degrees.

    At a height of 10m you get about 6.9dBi at 3 degrees.

    At 30m (100ft) you get 7.86dBi again at 3 degrees.

    These are local heights, not altitude above average terrain or sea level.  If you live on top of a mountain your radiation pattern won’t change much, but there also won’t be much in the way. That means better and more signal all around.

    Another interesting thing that came out of the simulations was some results if the antenna was electrically connected to a metal tower or the tower was there but the antenna wasn’t bonded to it. With a 10m(30ft) grounded tower, 9cm(4in) in diameter, not connected to the base of the antenna, the gain drops to 6.55dBi and there is little change in the feed point impedance. If the antenna is connected to the tower, the gain drops to 5.33dBi and the drive point impedance takes a hit as well. It drops to 37.26 ohms plus 1.5ohms of capacitive reactance.  Without the electrical connection, the amount of RF coupled to the tower is far less and the change in antenna performance is correspondingly smaller.

    For the car or truck installation, it is vital that the antenna be isolated from any RF conductor that is RF connected to vehicle ground. The addition of a short mounting stub off the base of the radiator, no longer than 10cm(4in) actually improves the match to 50Ω coax. The gain also improves but only slightly. The antenna must be isolated from any RF ground though. The closer this antenna is to ground the more important it is that it be electrically isolated from whatever it is attached to, especially if that something is tied to ground. Since the vehicle body may or may not be at RF ground, how it affects the pattern and feed point impedance is an open question.

    A good mobile radio installation should include high quality RF grounding, where appropriate, to the car body of most of your radio gear.  This is not always the case. I often just throw my hand-held and a power amp inside, place a mag-mount on the roof and go.  Trying to use a J-pole in the face of such haphazard installations could be problematic. The ARRL Antenna Book shows some more reasonable approaches to J-pole antenna mobile mounting.

    Since your car or RV should be on four rubber tiers, it seems likely that there wouldn’t be an RF path to earth ground. On the other hand, if your RV is hooked up at a campground and your J-pole isn’t isolated, you may have a problem.  In mobile installations, make sure the antenna is not connected to RF ground at the shunt and be sure to use a coax choke.


    Element diameter and length vs. gain


    The relationship between element (in this case copper pipe) diameter and shunt length, and therefore resonator-radiator spacing, is important. There seems to be a sweet spot, where maximum gain will be obtained. Any longer or shorter and the gain drops off. At this time, I don’t know what that ratio is. Clearly if the shunt is too long you loose the matching section properties of the U section. As the shunt gets really long and the resonator gets really short, the antenna starts to look more like an L and less like a J-pole. If the shunt gets too short, gain starts to drop off as well. A possible cause may be that the fields in the resonator and radiator are too closely coupled. In such a case the lower 1/4 wave section of the radiator may not be capable of contributing to the radiation pattern, the way it would in a lone 3/4
    wave vertical context.

    Also as the spacing between the two halves of the U gets smaller, the impedance of the resultant "transmission line" equivalent drops.  Like any transmission line based matching section, the greater the disparity between the transmission lines and the load, the less efficient the match will be.  If you will be matching between 50 Ω coax and 100 Ω coax, it is best to use some 75 Ω coax when building the matching sections.  When matching  50 Ω coax to 450 Ω ladder line, it is better to use 300 Ω twinlead.

    A possible explanation as to why the gain increases with U spacing may be that the resonator begins to act like an element in a phased array.  Since the resonator is a driven element, its similarity to a phased array system is appealing.  Alternatively the resonator could be thought of as a variation on a director element in a Uda-Yagi.  The problem with that analogy is that the directors in a Uda-Yagi are not driven, and they are about twice as long as the resonator.


    Don’t tie the shunt to RF ground


    In general, it is a myth that this antenna can be bonded to ground at its shunt.  Any metal below the shunt will be influenced by the RF fields and charges in the antenna.  RF will propagate down the conductor and destroy the tuning of the antenna. On the other hand if there is a separation of as little as 5cm(2in) between the antenna and a grounded tower, there will be little impact on the antenna’s performance. A separation of 0.6m(2ft) can even improve the gain.  Again, source impedance and gain remain relatively unaffected as long as the antenna is not RF connected to ground. The antenna can be tied directly to ground, at the shunt, as long as there is enough inductance to isolate the shunt from RF ground.

    To counteract a related problem in some J-Pole designs, this would be the motivation for a coax choke in the feed line. The outside of a coaxial cable shouldn’t have any RF on it. It is a good RF conductor though. It is running up and down your tower, so it can cause the same sort of performance problems as the tower itself.  The problem is doubly important since the shield of the cable is RF connected to one side of the antenna.  The RF currents on the outside of the coax could easily ruin the performance of the antenna. You can get around this by coiling the coax.  The coiled coax won’t have any effect on the fields inside the coax but it will act as a choke to any fields that form on the outside of the coax. This choke effectively isolates the antenna from the outside of the coax. This isn’t so much of a problem for other antennas since the shield is often electrically isolated from the radiating element.


    Lightening isn’t the issue


    Lightening protection is important in any antenna installation.  However, the J-pole has no advantage over any other antenna in the context of a lightening strike.  When lightening hits, everything in the immediate area, your house included, jumps up to very high potential. It is as if the lightening charges up a leaky capacitor the size of a football field to thousands, perhaps tens of thousands of volts. All that charge has to go somewhere!

    The best you can hope for is that the grounding straps you have installed will keep localize potentials low enough, as the overall charge dissipates into the surrounding earth, so as not to destroy your gear, start a fire, or kill you. There is no harm in you and everything else being at 10,000 volts for a short time, as long as nothing nearby is at ground.  With proper grounding, you have a good chance that the energy of the strike will avoid putting uncontrolled potentials across your gear, and dissipate into the earth through your grounding systems. They may be vaporized (better them than you or your equipment) in the process but the damage will hopefully be localized.

    There is little to nothing that can be done to save your antenna or gear from a direct lightening strike, other than to have a quality ground system on your tower.  If your antennas are mounted on your house or porch, then they are going to look like lightening rods no mater what you do. Pay careful attention to where the currents are likely to go and use arresters. Make sure the arresters have high quality paths to ground. Inspect your ground system regularly.

    The kind of ground needed to protect the J-pole from lightening strikes will ruin its ability to radiate.  In any case, a properly installed tower should be a much better ground path than any antenna mounted on it. The advantage of tying the shunt of the J-pole to ground, with enough inductance to isolate the RF, is that static charge can’t build up on the antenna. Antenna’s with isolated radiators do not have this feature. It is possible for a charge to build up on an antenna (whether it is connected to your radio or not) and blow your finals or your receiver’s front end. If you use an antenna switch that does not ground out your antenna when you aren’t using it, this could happen to you. With a choked ground bonded to the shunt of a J-pole, it isn’t likely that any static charge will build up.


    What’s a Super-J?


    In the ARRL Antenna book, 16th edition, chapter 16, “Mobile and Maritime Antennas”, Steve Cerwin, WA5FRF, describes a variant of the J-pole he calls the Super-J. The Super-J is a J-pole with an additional 1/2 wave radiator attached to the end of a standard J-pole through a 1/4 wavelength phasing stub. This arrangement turns a J-pole into a phased array. Phased arrays are often used in commercial and military installations. A planar phased array is used on AEGIS class guided missile cruisers of the US Navy. The array allows the computers to steer the radar beam, much like a searchlight, to a single point within the aperture of the antenna. High gain, the pinnacle of directivity. In the case of the Super-J we get another 3dBi, bringing the grand total to about 9 dBi in an omni directional antenna.  This is serious gain for an omni.  A quick look at HRO yielded no commercial antenna with that kind of gain on 2m. Could you add another section onto a Super-J? Yes. But you will suffer the law of diminishing returns. An antenna that long also becomes a bit unwieldy.

    Besides, isn’t 9dBi enough?!

    For an omni?!!!

    I hope this article has cleared some of the fog around how J-poles work.

    Thanks for reading.

    73 de AA1ZB

    References


    [1]  Constantine A. Balanis, Antenna Theory Analysis and Design John Wiley & Sons, Inc. Copyright  1982, 1997.

    [2]  Simon Ramo, John R. Whinnery, Theodore Van Duzer. Field and Waves in Communication Electronics. John Wiley & Sons, Inc. 3rd ed. Copyright  1965, 1984, 1994.

    [3]  G. J. Burke, A. J. Poggio Numerical Electromagnetics Code (NEC)  Method of Moments Part II: Program Description  Code Lawrence Livermore Laboratory. January 1981.