[Top] [All Lists]

Re: [TowerTalk] Re: radials for shunt fed tower

To: <>,"W2RU - Bud Hippisley" <>
Subject: Re: [TowerTalk] Re: radials for shunt fed tower
From: "Guy Olinger, K2AV" <>
Date: Wed, 15 Dec 2004 22:59:48 -0500
List-post: <>
I have found I can't directly measure characteristics of the ground itself with what I own so I could make a model give an absolutely accurate gain level. BUT...

If you are comparing two different configurations of the same or similar antennas, IF you make your best guess at the ground, AND hold the ground constant in your comparison of the configurations (and everything else you can hold constant), the DIFFERENCE (not the absolute value) tends to make it to reality. Obviously you have to use the same program, configuration and avoid its technical "gotcha's."

While this does not satisfy those who want to know the actual value, and correctly point out why one won't have it, a difference or comparison method is useful for using models to improve antennas.

For a 160m general antenna reference I use EZNEC/4's NEC-4 calculation of a single #12 1/4 wave vertical with 120 #12 1/4 wave radials and a Norton Sommerfeld "medium" ground. It gives me a reference for all the other stuff I model for 160 that use EZNEC/4's NEC-4 with Norton-Sommerfeld "medium" ground.

I would LOVE to have a certain $100,000 worth of test equipment, an antenna test range, and the spare time to make use of it all. But lacking that I would rather make incremental progress on antennas than be paralyzed by no means for absolute measurements.

For Bud's issue I would lay a halfwave dipole for the frequency on the ground in question, using insulated wire with known insulation thickness and characteristics. Measure its resonance frequency and impedance. Trim the dipole until the resonance is at the desired frequency. Be sure to stay away from conductors in or above the ground

Stick the trimmed length in a model with Norton-Sommerfeld ground calculations and insulation specification available (EZNEC v4 or ...) and adjust the ground characteristics until the model gives you the same Fres and Zres. Use those ground numbers in the model to decide whether the radial wire is worth it.

I have sometimes also had to drop the level of the ground in the model, down from "zero" to match the measured characteristics. This may be an indication of the non-monolithic, or "partially transparent" RF ground structure that is sometimes discussed, giving the appearance of an RF ground "below ground."


----- Original Message ----- From: "W2RU - Bud Hippisley" <>
To: <>
Sent: Wednesday, December 15, 2004 12:31 PM
Subject: [TowerTalk] Re: radials for shunt fed tower

At 17:57 2004-12-05, wrote:
Thanks for the Heads Up. Found it at:

In a cursory reading of K3LC's radial article, I found two data points that seem counter-intuitive to me.

In general throughout Tables I - IX, for a fixed amount of wire (8000 feet, say), the "peak gain" determined by EZNEC4 increases monotonically as the soil quality is varied from "very poor" to "average" to "very good". This seems reasonable to me, since the soil is effectively in parallel with the sparse radial screen.

However, this does not appear to be true on 40 meters, where the "peak gain" figures for 4000 and 8000 feet of wire are worse for the case of average soil than they are for either very good or very poor soil. According to Tables I - III, both "very poor" and "very good" soils provide higher peak gain (1.68 and 2.10 dBi, respectively) at 7.15 MHz than does average soil (1.4 dBi) when using 8000 feet of wire. A similar "dip" is reported for the case of 4000 feet of wire (1.02 and 1.88 versus 0.9 dBi).

I wonder if these are typographical errors or anomalies in the modeling.

Other comments inspired by K3LC's article:

1. One thing that jumped out at me as I scanned Tables I - IX was that, despite scaling the total wire lengths for the different bands, equivalent progressions from minimum total wire length to maximum total wire length resulted in less improvement at lower frequencies. On 40 meters, for instance, the total variation in calculated peak gain for average soil was 2.89 dB (from -1.49 dBi for 125 feet of wire to +1.40 dBi for 8000 feet of wire). On 80, the corresponding total gain variation was 2.37 dB, and on 160, it was only 1.86 dB.

2. Two other data points I would like to have seen are the peak gain (in dBi) for the same antenna model over "perfect" ground and over sea water, so I could get a sense of how far down in gain EZNEC4 says the article's finite radial fields are, compared to those two idealized situations.

3. The charts in the article make an excellent case for choosing well when siting your future vertical, and for determining how "good" your soil is *before* engaging in a major radial project. Consider 160 meters, for example. With "very good" soil (Table IX), each doubling of total wire used in the radial field results in only 0.1 dB improvement in peak gain. In "very poor" soil (Table VIII), by contrast, each doubling of wire results in about a 2/3 dB improvement -- more than 6 times as much! But in "very poor" soil, you're starting from a far worse loss, so it's going to require a whole lot more wire (and effort) to get comparable peak gain with "very poor" soil. Again comparing Tables VIII and IX, and extrapolating additional 2/3-dB increments of improvement for each doubling of radial wire in "very poor" soil, it's going to take more than 128,000 feet of radial wire for a vertical over "very poor" soil to have the same peak gain as one with only 500 feet of radial wire over "very good" soil! Forget the wire cost -- the labor differential is more than 256 times!!!

Bud, W2RU


See: for "Self Supporting Towers", "Wireless Weather Stations", and lot's more. Call Toll Free, 1-800-333-9041 with any questions and ask for Sherman, W2FLA.

TowerTalk mailing list


See: for "Self Supporting Towers", "Wireless Weather Stations", and lot's more. Call Toll Free, 1-800-333-9041 with any questions and ask for Sherman, W2FLA.

TowerTalk mailing list

<Prev in Thread] Current Thread [Next in Thread>