[TowerTalk] Radial length

[Jim Lux]

Jim Brown wrote:
> Certainly a circularly symmetrical radial system is best, but few of us are 
> able to realize it. Mine deviates significantly from that ideal, but the 
> antenna works well.  The best and most readable description I've seen for the 
> function of a radial system is by W2DU, as a chapter of his book 
> "Reflections." It's on his website. Google on w2du to find it. N6LF has 
> written a rather detailed version that's in the Antenna Book, but there are 
> typos in the associated graphics that make it difficult to follow. 
> 
> The short answer is that the radial system functions as a return for the 
> currents and the fields associated with the antenna. The lower the resistance 
> of whatever conductors serve this function (including the earth), the closer 
> the antenna comes to full theoretical efficiency. The radial system does NOT 
> function as a reflector for the signal. The earth at significantly greater 
> distance from the antenna serves that function. 


And therefore, assuming your soil isn't bone dry and nonconductive, the 
arrangement of radials doesn't much  affect the aziumuthal radiation 
pattern.  In the limiting case (1 radial), yes, there is an effect, but 
I believe someone did some experiments (might have been N6LF) and 
certainly there have been models, where highly asymmetrical radial 
layouts (i.e. a bunch of short radials all around, with really long ones 
in one direction) and there wasn't much difference in pattern.  That is, 
the asymmetry of pattern was <0.5dB, which is probably less than the 
differences due to soil properties in the far field.  Unless you're 
someone like N6RK, who has vertical antennas at his installation in 
Galt, which is a dead level alluvial plain with pretty darn uniform soil 
properties for hundreds of meters, if not km, in all directions. (Some 
years ago, I was doing data compression on topographic data, and there 
are whole 7.5 minute (about 6x8 miles on a side)quadrangles in 
California's central valley where all the data points are the same, 
within a meter or two)..  It's one of the flattest places in the US 
(flatter than a pancake and Kansas)
> 
> I have seen many references in the literature to the effect that NEC is not 
> very good at modeling ground losses and the effect of radial systems. Count 
> me among those who 1) consider real measurement of one radial system compared 
> to a reference antenna as the only meaningful predictor; 2) consider soil 
> conditions at any given location to be an influence on how many radials it 
> takes to get there; and 3) consider Tom's advice, and the advice in the 
> Antenna Book to be solid as a rock. 

Unless you have a fair amount of data and good modeling skills, I 
wouldn't use NEC to predict behavior of an antenna in absolute terms 
where there is a significant unknown (soil properties). However, I would 
trust it to tell me differences between configurations, and the 
differences should be fairly close to measurement.

That said, NEC3 and NEC4 both have been extensively validated against 
field tests (one of the reasons for changes from NEC2 is that the 
earlier versions showed poor correspondence between modeled and measured 
data for field erected vertical antennas, specifically, the change when 
driving a ground stake in addition to radials)

jim, w6rmk
>