s we have discussed in a group of articles in antenneX, the Prismatic Polyhedron, or Pn, is a derivative of the simple dipole, albeit with a more complex form and feed arrangement. See, Prismatic Polyhedron Measurements by David Jefferies in antenneX, November, 2002.^[i]

The principal characteristic of this antenna, well documented in the above article, is its extremely wide bandwidth. This parameter has been extended, with NEC modeling and subsequent construction of prototypes, to a 5:1 frequency range. That is the range over which the SWR is equal to or less than 2:1.

---

---

In order to try to glean why the Pns perform as they do, we investigated a class of antennas, the Plate Dipoles, which were found to be most analogous to the Pns. The results of the studies were published in two articles in antenneX.^[ii]  These antennas were dipoles, having very wide bandwidths, where each monopole half is constructed with a rectangular sheet of metal instead of a rod. Jefferies surmises that this wide-bandwidth phenomenon is largely the result of a “coupled parallel dipole effect" which, due to the distributed C and L over parallel plates of large areas, results in low stored electrical and magnetic energy and, in turn, results in the narrow, cyclic, excursions in the Rin and Xin which are characteristic of the Pn class of antennas. The plate dipoles have wide parallel conductive edges separating the two halves of the dipole. The Pns may be thought of as being “skeletonized” plate dipoles with most of the metal etched out, leaving only a thin outer matrix.

Recently, in an exchange of e-mail with Dave Cutberth, another one of antenneX’s contributors, Dave proposed that the Pn was simply a form of the well-known “cage dipole”. This is understandable due to the superficial similarity in form between the two antennas. David Jefferies thought this was a reasonable conjecture, and set a project student to construct a 1-GHz cage dipole. To test this conjecture further, David suggested that I investigate the similarities and differences between them by NEC simulation, especially in the parameter which I feel is most important, that is, in bandwidth.

My belief, based on the previous work with Jefferies, was that the cage dipoles lacked the basic design elements necessary in obtaining the very wide bandwidths found with the Pn class of antennas; namely the individual radiators which are fed in phase by common wide, parallel, wires arrayed as transmission lines from the central feedpoint to each of the radiators. These are well illustrated in the photographs found in Jefferies’ recent article where he shows the evolution of these antennas from a dipole to a P2, a P3 and a P4. 

 All of the antennas that were studied were modeled over a frequency range of 300-900 MHz. There is no doubt as to the ability of NEC to deal with these antennas since many such have been built by myself and Jefferies’ students and, if anything, their performance exceeds that predicted by NEC.

Prismatics:

Table 1 - Dimensions are in meters, frequency is in MHz

The two-radiator Prismatic or P2, as shown in Figure 1, has its SWR frequency sweep illustrated in Figure 5. Note the cyclic variation in the R and X in Figure 6 which is characteristic of the Pn class. The dimensions, Rin, and bandwidth data are found in Table 1. The simple P2, composed of relatively thin wire (0.015 wl), has a bandwidth of 2.6:1.

Figure 5 - P2 - SWR sweep

Figure 6 - P2 - R/X sweep

The three radiator or triangular Prismatic or P3, illustrated in Figure 2, performed as in Figures 7 and 8. The antenna with the dimensions in Table 1 had a bandwidth of 2.8:1 or 300-850 MHz. Note that with thicker (in terms of wavelength) wire at higher frequencies, this antenna has a measured bandwidth which is somewhere around 4:1.

Figure 7 - P3 - SWR sweep

Figure 8 - P3 - R/X sweep

Cage Dipoles:

Table 2 - Dimensions in meters, frequency in MHz. Overall height includes the heights of the two monopoles, the feed segment of 1.5 cm and the separation distance between them

In addition to the lack of parallelism, the feed arrangement increases the distance between the monopole wires in the cage or widens their feedgaps. This, in turn, reduces the capacitive coupling between monopole ends, a factor which also appears to be significant in wideband performance.

I optimized each cage design by varying the overall perimeter size, the aspect ratio (AR) which is the ratio between the long vertical dimension and the narrower horizontal one, and the separation between each of the monopole halves. I targeted whatever Rin gave the widest bandwidth. The sizes and the performance data can be seen in Table 2. The 2 wire cage’s performance curves are in Figures 9 and 10  while the 3 wire cage’s are in Figures 11 and 12.

Figure 9 - 2-wire Cage Dipole - SWR sweep

Figure 10 - 2-wire Cage Dipole - R/X sweep

Figure 11 - 3-wire Cage Dipole - SWR sweep

Figure 12 - 3-wire Cage Dipole - R/X sweep

While the bandwidths are quite wide, they do not compare favorably with those of the Ps. The    3-wire cage dipole’s bandwidth figure of 2.1:1 does not approach that of the simple P2.

Reverse engineering:
After I had finished modeling the 3-wire cage dipole, I changed its dimensions to those of the P3 and made the feedwires parallel - as in a typical P. To clarify this further, the wires were not brought to a conical apex but were orthogonal to the radiators and parallel to each other. The only change I made was to the 6 wires coming out of the feedpoint to the radiators. These I made 1.5 mm in diameter to make them comparable to those found in the transmission lines in the P3.

Figure 13 - 3-wire Cage Dipole modified into a P3 - SWR sweep

Figure 13 clearly shows that, when the cage dipole is transformed into a Pn, the bandwidth far exceeds that of any cage conformation. This study further shows that modeling the Pn with real wires as transmission lines yields the same results as with the TL (transmission line) facility in NEC.

Some more comparative observations:
The cage dipoles have gains which exceed those of the comparable Ps by about 0.5 dB. But note in Tables 1 and 2 that the “cages” have overall heights which are greater than the P’s. The 3-wire cage dipole is 25% taller than the P3, and this is reflected in the frequency at which it becomes electrically “too long” and begins to show an elevation pattern that squints upward. The elevation pattern at 900 MHz is shown in Figure 14 and indicates that, above this frequency, most of the radiation will be at the higher secondary elevation lobes and not at the primary one at the horizon. So, even if we can find a design which will have a useful SWR bandwidth of 3:1, it will become useless, from a pattern point of view, at that point.