Topband: Fw: Shortened Radial Experiments

[Brian Mattson]

----- Original Message ----- 
From: Brian Mattson 
To: topband at contesting.com 
Sent: Saturday, October 14, 2006 7:56 PM
Subject: Shortened Radial Experiments


In response to Eddy's (VE3CUI) question about anyone using "shortened radials", I have been using these for two years now, with very good results.

Coming to TopBand after decades on VHF, I was confused by the myriad of opinions on radials. Comments like "resonance disappears after a few buried radials" and "longer is better" were often seen. As a degreed electrical engineer, I was puzzled by the abandonment of the laws of physics once a radial was buried, or laid along the ground. Sure, the velocity factor & loss factors change significantly once a radial gets near, or below, ground, but basic electrical laws must still apply.

As I first got on TopBand in the dead of winter, I used the single elevated radial as discussed in "Low Band DXing". Pointed towards Europe, and about 5 feet off the ground, it worked surprisingly well. However, when it came time to upgrade the ground system, confusion set in with all the conflicting opinions I read. Fortunately, I ran across Rudy Severns' (N6LF) article on "Verticals, Ground Systems and Some History" in QST (July 2000). ( As an electrical engineer in the switching power supply industry, I have learned to listen when Rudy speaks!). One comment that really caught my attention was on page 41: "For the 0.1 wavelength high (vertical) antenna, if we have a good ground screen out to a distance of 0.1 wavelength, we'll eliminate over 90% of the ground loss!". The lightbulb came on right then. I could instantly visualize an Electrostatics Fields class representation of a ground referenced hemispheric field intensity bubble with a radius of the vertical height. I use a phased pair of inverted L's for my transmit antenna, and each has around 50 feet of vertical rise, so a system of enough 50 foot radials should suffice. But the nagging thought of resonant length still bothered me. Time to experiment (play).

The inherent beauty of a quarter wavelength radial is in it's impedance transformation properties. Basically, the higher the impedance on one end, the lower the impedance on the other end. As the far end of the radial is open circuited, the antenna end is as low as possible, and it is non-reactive. Two opposing radial elements look suspiciously like a dipole, so that's where I began. All my measuring was done at 1.83 MHz, so a free-space dipole would be about 269 feet & have an impedance around 73 ohms. All my experimenting was done with #14 solid insulated THHN copper wire. 

My first experiment was to construct a full size dipole and lay it on the ground. The resulting dipole was well below the lower operating frequency of the MFJ analyzer, so pruning was in order. I finally achieved resonance with a length of 182 feet! Rs was 130 ohms. So the velocity factor was thus: 182/269 = 0.677. So Eddy, don't take the 0.5 number from "Low Band DXing" as gospel, as it depends a lot on the type of soil you have. My soil is sandy (almost like beach sand). Note too that the ground proximity has increased Rs substantially. Next, I buried the dipole in a slit trench approximately 6" deep. Again, the dipole was way too long. To prune the buried dipole, I found it easiest to have the ends bent up so that they protrude just above ground & place a bright colored "wire nut" on the end (so I could find it again!). The resonant length of the dipole was now  107 feet! Rs was 148 ohms. The buried velocity factor was: 107/269 = 0.398. Note that burying the dipole has added even more losses to Rs. 

The result of experimenting thus far resulted in a resonant radial length (in my soil) of 53.5 feet (half of the dipole). With my 50 foot vertical inverted L's, I was ecstatic. But how many radials would I need?

I constructed another buried dipole of 107 feet length, at right angles to the first, and so their centers were coincident. This gave me four radials. I tested the second dipole as a separate entity, and it's numbers were very close to the first. Next, I connected the two dipoles together (two adjacent wires as one node/ the other two adjacent ones as the other). I was astounded when the resonant frequency plummeted!! I almost gave up at this point. As a VHFer, I knew that whether a ground plane has two or four radials shouldn't make much difference. Indeed, some conicals feature solid sheet ground screens. In any event, the quarter wavelength dimension shouldn't change much. After stewing on this for a few days, I realized that I had constructed a "Fan Dipole" which greatly increased the capacitance to ground, thus lowering the resonant frequency. I then came up with what I consider to be my only "original" contribution to this experimenting. By connecting up opposing pairs of radials as one node, and the other two opposing pair as the other node. sanity was restored. I was pleased to see that Rs dropped almost exactly in half (75 ohms), as two parallel impedances should. The basic laws of physics were still intact! For want of a better name, I refer this connection scheme as cross-connected dipoles. Realizing that with many additional radials being added, the "cross-connection" scheme could easily get lost. The solution was to have TWO connecting rings at the radial junctions. The radials are then alternated from one ring to the other, so that each ring has half the radials, but with NO adjacent ones. For operating, the two rings are both connected to the coax shield, but for testing, the two rings are separated to connect to the analyzer. One curious effect was noted when the resonant frequency dropped slightly (about 36 kHz). Pruning the radials by 6" restored 1.830MHz numbers. (Radials now 53 feet each). This slight (second order) effect is probably due to increased capacity, even with the cross connected configuration. 

I then doubled the number of dipoles to four ( 8 radials), the cross-connected dipoles again dropped to half Rs (now 38 ohms). Again I had to prune the radial length (now 51').

Then the number of dipoles was doubled to eight (16 radials). Rs was now 15 ohms. Elements again trimmed (now 49').

Finally, the number of dipoles was doubled to sixteen (32 radials). Rs was now 7 ohms. Elements now trimmed to 48'.

Please note that all the Rs readings were cross-connected dipoles in the ground and NOT the antenna impedance.

I then added my resonant vertical (50 feet vertical, rest in the top-hat).

The antenna measurements were:  56 ohms with two radials. 43 ohms with four radials. 30 ohms with eight radials. 26 ohms with 16 radials. And, 24 ohms with 32 radials.

One great feature of short radials that everyone seems to agree on is that FEWER of them are required. From the antenna measurements, you can see that doubling the amount of copper (& labor!) resulted in only 2 ohms improvement from 16 to 32 radials. My second antenna only has 16 radials.

My 48' or 49' radials are an efficient match for my 50 foot verticals, but if I were to have a full-size (135 foot) vertical, I would still go to resonant tuning. In this case, in my soil, the 3/4 wavelength radials would probably end up around 3X48' = 144'. (possibly slightly shorter due to the second order effect).

Thanks to Rudy for his inspiration!

Sorry for the long message, but I think it's sound.

Best Regards,

Brian Mattson  K8BHZ