The antenna is based on three closely-spaced parallel wires. As you can see in the figure below, these wires are connected to one another in a way resembling two letters: Latin L and Greek Pi. That’s why I called this antenna LPi. Please note that for better clarity, the distance between the wires in the drawing is shown much greater than it is in reality. 

The antenna is fed between “the letter” L and “the letter” Pi.

At a first glance, one can think that this is nothing else but the well-known end-fed Zepp (a.k.a. J-Pole) folded in half. But, no. The proportions are different – see below.

The electromagnetic fields generated by three closely spaced conductors interact strongly with one another, so it is very difficult to imagine the distribution of current magnitudes along the wires, not to mention the distribution of phase shifts. You need to rely on the antenna simulators. The simulation has shown that the LPi gain and radiation pattern are almost identical to half-wave dipole. However, the most amazing was the fact that you can significantly influence the antenna SWR bandwidth by changing the spacing D between the wires.

The figure above shows the SWR bandwidth as a function of distance D between the antenna wires. The bandwidth is shown for SWR≤ 3:1. It makes sense because for such SWR values, the additional loss in the feed line is still relatively low, and every modern transceiver equipped with a built-in ATU is capable to match its transmitter output to such an impedance range. As you can see in the plot, even with the wires as closely-spaced as 0.001 wavelength of the lowest operational frequency, the LPI antenna has greater bandwidth than a half-wave dipole. When D grows, the bandwidth grows too. Finally, for D=0.01 WL, we achieve the ratio of the maximum to minimum frequency equal to 1.46. With such a D, the LPi bandwidth is 4 times larger than that of a half-wave dipole! Please mind that the SWR for the LPi was calculated for Zo=200 ohms and the dipole Zo=75 ohms. 

In order to understand why this antenna has such a large bandwidth, it is worth checking out its feed point impedance versus frequency plot.

The antenna has resonance (X=0) at three different frequencies within its operational bandwidth. Its impedance resistive component R changes moderately, from about 70 ohms to about 370 ohms. At the same time, the reactance value X is quite limited too. Thanks to that, it is possible to get SWR ≤ 3:1 in a wide frequency range if only you feed the antenna with a source of internal resistance 200 ohms. 

The antenna design process starts with defining the minimum and maximum operational frequencies: Fmin and Fmax and their ratio (Fmax/Fmin). You can choose any ratio from the range 1.18 through 1.48. You find the dimension D (expressed in wavelength of Fmin) in the plot shown below.

Once you know D, you use the next plot to find A and B. Finally, you convert A, B and D expressed in wavelengths to meters. To do that, you multiply A, B and D by 300 and divide by the Fmin expressed in MHz.

The dimensions found in the described way will be accurate for the antennas designed for frequencies close to 10-15 MHz frequency range built with bare wires of diameter equal to 2 mm. They will be less accurate for other frequencies (like 3.5 MHz or 28 MHz) and for other diameter wires. Therefore, it is recommended to simulate the antenna designed in this way in one of the available antenna simulation programs (EZNEC, 4nec2, MMAGA-GAL) and fine-tune its dimensions to get the best possible SWR in the frequency range of interest. Because not all hams can smoothly use such simulators, I prepared a table showing the LPi dimensions for different frequency ranges suitable for various HF bands.

So far, we analyzed the LPi antenna simulated in free space. But as everybody knows, proximity to the earth changes not only the antenna radiation pattern but also its feed point impedance and hence the resulting SWR. I ran a number of simulation to find out how much the height of horizontally installed LPi influenced the antenna bandwidth. The results were very positive. The antenna SWR remained below 3:1 even for low hanging antennas for the 80 m and 160 m bands. You can see that on the figures below.

The impact of earth proximity was even smaller for the higher bands, as well as for the LPi antenna mounted vertically. We can state that the LPi antenna is not sensitive to earth proximity. 

I simulated the LPi with different antenna simulators and got very similar results. In such a situation, the next logical step was to build a real antenna to see if the real world measurements would agree with the simulated predictions.

The LPi can work in any orientation: horizontal, vertical or tilted. I decided to build a vertical prototype antenna for three bands: 15+12+10 m. It has been built around a fiber glass mast (a fishing rod) to which I have attached cross arms to keep the three wires in the desired distance. 

To avoid scratching the fiber glass mast, I used plastic tube clamps made in accordance with DIN3015 standard. They also enabled me to attach the matching network box to the mast. The clamps are inexpensive and are available in many sizes, what makes it easy to find the right one matching the changing diameter of a tapering fiber glass mast. I made the cross arms with square rods of ABS plastic. You can see them attached to the mast in the photographs below. The clamps for the clamps mounted low on the mast are screwed with regular M6 stainless steel bolts and nuts. However, the clamp fixing the top cross arm uses polyamide (nylon) bolts and nuts. Thanks to that, I reduced the mass attached to the thinnest section of the mast. 

The LPi requires a 200 ohm source. To make it work with 50 ohm transceiver and 50 ohm coaxial cable, you need a 4:1 transformer (unun). Aside from that, it is very desirable to use a common-mode choke (1:1 current balun, a.k.a. line isolator) connected between the 50 ohm unun port and the coaxial feed line. The matching network schematic is shown below. 

The unun and the common-mode choke were both wound on toroidal cores of 1.4" outer diamter made of 43 material produced by Fair-Rite. The core part number is: 5943002701. I used RG174 coax wind the unun and the CMC. Their windings have 8 turns. Winding the unun with a coax instead of a pair of insulated wires is unconventional, but my measurements revealed that such an unun has an almost ideal 4:1 impedance transformation ratio in the frequency range 21-29.7 MHz. The photograph below shows the interior of my matching network box. 

The antenna made in accordance with the dimensions in the table above and mounted on the balcony railing was measured with an antenna analyzer. Its frequency range was shifted towards lower frequencies. Proximity of the fiber glass mast and other object in direct vicinity were most likely the reason for that. A and B dimensions had to be shortened. After a few trials, the final dimensions were A=2.29 m and B=5.59 m. Dimension D was left unchanged, that is 14.4 cm. The SWR measured directly on the SO-239 socket (without any coaxial cable) approached 3:1 at 21 and 29.7 MHz and had a minimum 1.3:1 at 22.7 MHz. So, it was similar to the simulation results. Maximum SWR measured at the transceiver end of the coax was naturally lower, about 2.5:1 (because the coax is a lossy feed line what reduces SWR), see the plot below.

In the prototype antenna system, an additional common-mode choke was inserted about 2.5 m away from the matching network. This choke reduces the likelihood of the excitation of the common-mode current in the coax due to antenna radiation. It is a good practice to apply such a precaution not just for the LPi antenna but for any other end-fed antenna as well as GP category antennas. This additional choke should ideally be connected about 1/4 wavelength away from the matching network. In this case, the wavelength is related to the maximum operational frequency (Fmax). It was about 2.5 m for Fmax=29.7 MHz.

If you would like to experiment a bit with the LPi for the same frequency range, a good idea would be to make D a little larger, say, 16 cm. You should then decrease SWR at the band edges and increase it a little in the center of the band. It may be important for the transceivers not equipped with internal ATU. I would not bother though if my transceiver had the ATU. 

Speaking about experimenting, I also built an LPi on an aluminum mast. My simulation predicted feed point impedance closer to 100 ohms instead of 200 ohms. I built a 50 ohm to 112.5 ohm step-up unun. 9:4 unun, if you will. Unfortunately, I got smaller bandwidth than expected. Perhaps the unun I built was not perfect enough and its impedance ratio was different? Or the simulation was inaccurate? It is a well-known fact that NEC-2 and NEC-4 based simulators introduce errors in the simulation results when closely spaced wires of very different diameters are used (as it was in this case). It is possible that the newest NEC-5 engine still has this flaw? Chances are, I would finally get the desired bandwidth by changing antenna geometry or/and the unun construction by trial and error. But I was not patient enough and disassembled this antenna version. I just want to let you know that if you decide to use an aluminum mast rather than a fiber glass one, be prepared to spend more time (and work) for antenna tuning.

I compared the newly built LPi with my proven GP-7DX vertical antenna installed on the other side of the house. I could not tell which one works better. They performed equally well. The LPi efficiency leaves nothing to be desired. This is an inexpensive, easy to build and easy to tune antenna. Even a beginner can build it. When installed vertically, it radiates at a low angle, just like a vertical dipole. That's why it is good for DX hunting. However, if your QTH is surrounded by other buildings, metallic fences, lamp posts, it is recommended to install it not just above the ground but a couple of meters higher or even on your house roof. Nothing will block your signal then. You can install the LPi practically everywhere because it has minimal footprint and does not require grounding or counterpoises. I have been using this antenna for almost two years now, and I really recommend it to every ham.