KM1NDY

A Secret Little Adventure Ham Radio Blog

The Extremely Off-Center Fed (EOCF) Antenna

by KM1NDY

My posting inviting criticism of this blog entry into the “Antennas, Feedlines, Towers & Rotors” section of the QRZ.com forums led to the following very interesting discussion from WA7ARK (box below). If you click on his call sign, it will take you to his QRZ biography page where he has attached a presentation on the so-called “end-fed halfwave” antenna…so-called, well, because he argues it does not exist. Rather, he has proposed “extremely off-center fed” (EOCF) as a more appropriate nomenclature for this particular antenna design. This post really is just me further exploring antenna modeling with EZNEC software. The ARRL also has a downloadable manual entitled “How to Start Modeling Antennas Using EZNEC” that has been an excellent resource.

  • “Some suggestions about modeling your antenna using EzNec.
  • If at all possible, avoid the use of the “MiniNec Ground” model with a low-mounted (< ~ 1/4wl above ground) horizontal wires. Use the “High Accuracy” model instead.
  • This antenna is designed to be fed through a 1:7 turns ratio (50:2450 Ohms ratio) transformer.
  • You are using too few wire segments to create an accurate model. You are likely using the “free” version which is limited to 20 segs. An accurate model (one which passes the “Average Gain Test” see the Help File) likely requires more than 35 segs, and multiple wires. Your placement of the “transformer” (actually its 2450 Ohm winding) in the antenna’s radiating wire is simplistic.
  • In the real antenna, the coax shield running all the way to the radio acts as the part of the antenna. It acts like the “short” wire on the “other” side of your source. You modeled it as 1.5 segs of the main wire, while in the real world, it is whatever the length of the coax shield is…
  • The correct way to model this antenna is to use four wires (not 1). Wire1 is the antenna wire with lots of segments. Wire 2 is a short continuation of Wire 1 with only three wire segments used as a place holder for the transformer secondary winding. Wire 3 + 4 represents the path followed the coax shield from the transformer to the radio.
  • To model the impedance transformations caused by the transformer and the coax, use Virtual wires (see the EzNec Help file).
  • Connect the “source” to V1 (Virtual node 1). Connect 50ft of RG-8X (or whatever you are using) from V1 to V2 (virtual node 2). Connect the 50:2450 Ohm transformer from V2 to the seg 2 of a short wire #2. Create the path representing the coax shield as Wires 3&4,… as many wire pieces as are needed to reflect the coax routing.
  • I am attaching a model which is closer to your reality. This model was created using AutoEz, but I am posting the EzNec file created by that process to show how to model the coax shield and its effect on radiation and feedpoint impedance, the method of accounting for the impedance transformation caused by 50ft of RG8X (including accounting for the Velocity Factor and loss of the coax), the method of accounting for the impedance transformation caused by a 1:7 turns ratio transformer, and the common-mode current flowing on the outside of the coax between the transformer and radio.
  • To do justice to this reality, you should be using a minimum of ~33 segments. If I reduce the total number to only 20 so that the model can run on the free, demo version of EzNec, the model looses accuracy per the AGT test. I am only putting 20 segs in the model I am posting for you. When I run that model with only 20 segs, EzNec complains that wire segments are too long. This could be fixed by running the paid-for version and increasing the segment counts…
  • The model predicts a feedpoint impedance of 39 + j30 Ohms, for a swr50 (as seen by the transmitter) of >2. This could be reduced by putting a coax CM choke (1:1 unun) on the vertical drop of coax, about 25% of the way down the vertical coax section (~4ft below the transformer). That drops the swr at the transmitter to less than 1.5. I have added the choke to EzNec model; you can run the model with and without the choke to see how it effects current distribution on the antenna wires and how it changes the feedpoint impedance.
  • I went back to AutoEz, and optimized this antenna to “improve” it at the center of the 20m band (14.175MHz). To run the optimizer, I increased the wire segmentation. I will let the optimizer vary the wire length, the transformer turns ratio, the distance from the transformer along the coax to the coax CM choke (1:1 unun), all in an attempt to improve the Swr at mid-band. The placement of the antenna is as you specified; transformer at 15ft agl and end at 20ft agl. This would change if the antenna was mounted higher relative to dirt.
  • After optimization, the feedpoint impedance is 51.8 – j3 Ohms for a swr50 = 1.06 at 14.175MHz provided the main wire is insulated #14 awg wire, is 33.4ft long, the transformer is changed to be 1:8 (50:3200), the coax CM choke is moved to 6ft below the transformer.
  • You might learn something about EFHW antennas by reading a slide show I created. It is linked to at the bottom of my QRZ page.”
  • de WA7ARK

The antenna I attempted to model was a 20M Radiowavz end-fed half wave configured as seen in this drawing (with many more details available in the original blog post):

The original model I developed was overly simple with many flaws. Given my lack of experience in antenna modeling, I knew this would be the case and welcomed any feedback on it.

WA7ARK proposed an alternative model, whereby four wires were used with a 49:1 unun attached at virtual node 1. He also include a common mode choke (1:1 unun) on the feedline which is referred to as a “load”.

I modified this as seen below in order to better represent the antenna I had erected on Mt. Wachusett. Mostly, this was just positioning the now two radiator wires above the horizontal run of coax.

I made some more modifications to the model. Removing the “Load” (which represents a 1:1 unun for a common mode choke) from Wire 3 on the my modified model reduces the max gain to 4.56 dBi in both the azimuth and elevation far field plots. It also reduces the antenna impedance at 14.3 MHz to 40.5+J27.4 ohms from 66.89825 + J 28.28487 ohms. I added another segment of wire (wire 5) to the radiator and added an LC matching circuit for 14.3 MHz. In the figure below, I switched the orientation of the antenna model for better visualization of wires 1, 2, and 5 (which essentially make up the antenna radiator), the impedence matching circuit (L), virtual node 1 with the source, and wire 2 with the 49:1 transformer. The impedance matching circuit LC values were chosen by using a LC impedance matching calculator. The impedance of the antenna model can be found in the EZNEC software.

The vSWR of the antenna from 1 to 60 mHz of the KM1NDY modification without the matching circuit but with the common mode choke on the feedline is shown below (the grayed out curve). The thin-lined trace shows vSWR for the same antenna, now with the matching unit for frequency 14.3 mHz and without the common mode choke on the feedline. The prominent dips with an vSWR less than 2 are at frequencies: 8MHz, 14.5MHz, 30MHz, 42MHz, 45.5MHz, and 57MHz. The resonance around the 40M band with the presence of the matching circuit is the most interesting feature of the frequency sweep. More modeling shows that it is the presence of the common mode choke that appears to reduce the resonance on 40M.

Well, these are really just musings and attempts to understand antenna modeling. Thanks to WA7ARK for pointing me in the right direction with this!

KM1NDY