Tuesday, October 13, 2015

80 Meter Tower Vertical with Short Radials

Now that some other tasks are out of the way and the weather is sufficiently cool that traffic has declined in my backyard, it is time to get to work on my new antenna for 80 meters. If you read my plans for this band you'll know that I am unimpressed with the performance of my loaded half sloper, and I want to do better in time for this season's major contests.

80 meter load half-sloper in storage with other antennas
In this article I'll walk through the process, starting with theory, going on to modelling and then to construction, tuning and an early performance assessment. The last is made difficult since it was necessary to remove the half sloper, and conditions for DX have been poor.

Model expectations

In that article I referenced a formula in the first edition of ON4UN's book Low-Band DXing to estimate the electrical length of a tower with a yagi up top. At 3.65 MHz the formula worked out to 105° for my 14 meter tall tower with a Hy-Gain Explorer 14 at 15 meters. Extrapolating to 90° the estimated resonant frequency would be 3.13 MHz.

I took my interaction model, added a set of 8 radials, each 8 meters long, and fed it at the bottom. Since the tower is mostly isolated from ground this is allowable, at least in theory. The model confirmed that the structure should be resonant (X = 0) at 3.25 MHz. So far so good, or at least the ON4UN formula is roughly consistent with the NEC2 model.

The modelled resistance component of the feed point impedance at 3.65 MHz is 80 Ω. It is high due to ground loss and being well off resonance. I chose to simply place a capacitor load in series with tower (monopole), chosen to have a reactance of equal magnitude to the inductive reactance of 44 Ω at 3.65 MHz. This requires a capacitor of 1000 pf. However, it isn't quite that simple since the capacitor transforms the antenna feed point impedance and antenna behaviour. Instead a value of 700 pf series capacitor inserted into the model is needed. The resulting excellent SWR curve is below 1.6 between 3.5 and 3.8 MHz. The resistance at the new resonant frequency of 3.65 MHz dropped to 65 Ω, which further tamed the SWR.

Notice that the resistance is still high for a λ/4 vertical (typically 37 Ω). The largest component of the additional resistance comes from the modelled ground resistance (loss), which is in series with the radiation resistance. For 8 short verticals that isn't too bad. NEC2 often underestimates ground loss, so this should be kept in mind.

Radial configuration

Since writing my planning article I experimented in the model with a variety of radial systems that could be built in my narrow and long backyard. I increased the radial count to 8 to minimize ground loss for what will be short radials, and to minimize the resonance/tuning effect of the radials on the antenna system. More radials would be better, but the added nuisance is undesirable and in any case would benefit performance very little.

What I did resolve was to make the radials of equal length. Having a mix of long and short radials (to maximize use of the available space) always resulted in low current in the short radials, making them redundant. The final configuration of 8 radials of 8 meters length best equalizes radial current and still fits nicely within the 15 meter width of my yard. Modelling indications of this arrangement were promising with respect to performance.

Preparing for construction

Continuity insurance
Before building the antenna I did some preparatory work. First, as I said above, the 80 meter half sloper was removed. Modelling showed that it would alter the behaviour of the vertical. In any case I wanted to free that coax for a planned inverted vee on 40 meters.

I next unrolled a 130' length of RG-213 that I had in storage and ran it from the shack to the tower base. Although it would be best to bury this coax for best common mode results this is undesirable for reasons that have nothing to do with performance. Instead it was run above ground along the same messenger cable used for the other cables going to the tower, then dropped from a 3 meter height to the base. I will return to this point later since this does have a performance impact.

In one trip up the tower I removed the half sloper, installed a short aluminum tube to support the planned 40 meter inverted vee, and then installed a wire from the tower to the rotating mast.

Since the mast and yagi are to be part of the 80 meter vertical it is vital to ensure electrical continuity from tower to mast, and then to boom and the yagi elements. The yagi is fine as is since all but the driven element are bonded to the boom, and the boom to the mast. The risk shows up between tower and mast.

The jumper wire shown in the photo is insurance against lack on intrinsic continuity due to the rotator bearings and mast bearing balls. These are coated in grease and could interrupt continuity since those many bearings may be the sole electrical path through these devices. The added wire is our insurance. In the photo the rotator is in its centre position, with the wire gently rotating with the mast. When you do this be careful to ensure the bonding is secure and that the wire won't snag any bolts or other protuberances when it turns.

Installing the radials

I prepared the radials early on, once I made the final decision the number and length. I have a large role of hookup wire which I had bought about 25 years ago for this very purpose, yet until now it was not used for radials. The 64 meters of wire I used made only a small dent in the roll's diameter. While the gauge is bit low for robustness it should serve well for this winter season. They can always be reused for another project.

With the lawn well trimmed I unrolled each radial and tied it to a central point on the wooden tower base. The first 4 were easily placed at 90° to each other. The remaining 4 bisected each pair so that the 8 radials were spaced 45°. These were connected to the coax outer conductor. They were stapled to the edge of the base to that they were held close to ground, in case I would need to mow the lawn before the snow flies.

The ends of the radials are anchored into the sod, and in one case into an old tree root. The anchors were chosen for expediency: 3" galvanized framing nails. These worked well enough, though I would not count on the tension being maintained for an extended length of time. As you can see in the photo the radial is stripped at the end to facilitate tying it off, but is not electrically bonded to the nail.

Feed point construction and initial on-air testing

For an initial test I wanted to compare the antenna against the ON4UN equation (3.13 MHz) and the EZNEC model (3.25 MHz). The measured value was in the vicinity of 3.15 MHz. with an SWR of 1.1. This is close enough to both that we can consider them sufficiently reliable. Or perhaps I am just lucky. Though short, the alligator clips I used to connect the coax to the radials and tower would influence the measurement. I was not concerned about that at this stage of the process.

The EZNEC model also predicted a high SWR but decent radiation resistance at 7.1 MHz and 10.9 MHz. The measured values were further off -- 6.7 MHz and 10.5 MHz -- but the SWR dipped to below 1.2. I took this as a hopeful sign that the ground loss is lower than in the model. It is also possible that something more serious is going on. I put that concern aside until I was able to test the antenna's performance.

The directly fed vertical seems hear well on 80 and 30 meters, but not on 40. Since it was daytime I did not attempt any QSOs. Instead I proceeded to installing a simple matching network

Series capacitor

The value of a series capacitor acts in an inverse manner in comparison to a series inductor. As inductance increases from 0 μH the resonant frequency declines. For a capacitor it is a value of ∞ pF which does not shift the resonant frequency of the antenna, and as the capacitance drops from infinity the resonant frequency rises. Of course no capacitor has ∞ capacitance, you can only approach it in the limit, where it is equivalent to a short; that is, no series capacitor.

In practice getting to 0 pf is easy while high values at RF are difficult. Or at least it is difficult when we want low loss (equivalent series resistance) and high power handling. These are increasingly important in a matching network where the Q and the mismatch are high. But for tuning purposes we can take a simpler approach.

At a recent hamfest I purchased a few variable capacitors with a large maximum value, just for this application. The capacitor pictured at the feedpoint measures 2,200 pf with all sections ganged, which is perfect for lots of variability around the modelled requirement of 880 pf to bring resonance up to 3.65 MHz. The plate spacing is fine for QRP but can handle 100 watts in this low impedance calculation. This permits full power on-air testing. If tuning is acceptable a fixed capacitor with low ESR and high power handling can be substituted.

Tuning did not achieve what I wanted. As the model shows, the resistance component of the impedance rises as the inductive reactance is cancelled by the capacitor. This is expected since the capacitor changes the electrical topology of the antenna. Unfortunately it rose more than predicted, resulting in a minimum SWR of 2 at the low end of the band. Resonance also moved closer in-band on 40 and 80, but with an SWR in the range of 2 to 3. I suspect, but cannot prove, that stray capacitance among the many cables and coupling to house pipes and wires (15 meters away) contributed to the measured difference.

This would be the perfect time to test it on the air. To my dismay the geomagnetic activity spiked to a K index of 7 (October 7-8 UTC). DX on 80 was hard to come by. Performance on 40 and 80 seemed poor in comparison to the inverted vee, which while disappointing was not a design objective. It was also possible that low angle paths were strongly attenuated due to the geomagnetic activity this far north.

L network

I proceeded to design an L network to better match the antenna while I waited for conditions to improve. Nothing more elaborate is needed. Had the coupling of the tower to ground been greater the impedance profile would be further from the modelled value so that a gamma or omega match might be needed.

Designing an L network is quite easy with TLW. Plug in the complex impedance and frequency, select a network topology, and it designs the network. For convenience I chose the topology that was easiest for me to build using available components.

That L network can then be added to the EZNEC model to complete the model. I did that, and the model predicts an SWR below 1.5 from 3.5 to 3.8 MHz.

The L network is easy to build. I used the same variable capacitor as the shunt element. The series coil is bare copper wire wound over a ¾" form. When released from the form it expanded to the 1" diameter I wanted. Inductance can be adjusted by squeezing or stretching the coil, but I tapped the coil with an alligator clip. Coil dimensions were calculated with a standard formula for an air core. While these may have some inaccuracy it is sufficient to get within 10% or 15%, which is near enough to allow adjustment for best match. Don't simply guess at the coil dimensions or you'll fail to get a good match.

A bit of tweaking brought the SWR in line with the model: below 1.5 from 3.5 to 3.8 MHz.. I left the L network exposed as shown for the Thanksgiving long weekend since no rain was forecast. No curious cats, skunks or squirrels disturbed the setup. My run of good luck continued when I finally worked TX3X minutes after taking the above photo. Despite the ugly wiring job it withstood 100 watts without any sparking.

Unlike with the series capacitor I tried first, the L network did not permit operation on 40 or 30 meters. I chose the "low pass" configuration in TLW since it gave the easy inductance to implement quickly and efficiently.

When tuned for best match the L network components were an inductance of 2.15 μH and a capacitance of 430 pf. The coil value was calculated from its tapped dimensions and the capacitor was measured. The stray inductance and capacitance of the messy wiring job are close to negligible on 80 meters.

At 3.7 MHz where the SWR is 1.0 the calculated antenna (load) impedance would therefore be 100+j0 Ω, not the modelled 72+j55 Ω. The resistance component of the impedance is at least in reasonably close agreement. The reactance is a puzzler, probably having to do with much stray capacitance among the many cables running on and to the tower.

Performance assessment

Propagation was quite poor when I was ready to try out the antenna. A long period with disturbed geomagnetic conditions made for difficult testing of DX paths (my main objective). While I did have some luck a true picture of antenna performance will take longer. As I write this the antenna is out of action because I removed the temporary L network and sealed the coax to protect against the rain.

For now I can say that it works but is not a miracle worker. That's as expected. I worked some DX in these horrid conditions, but only a few stations that seemed to run high power and so made it through. At this point I cannot say how it compares to the loaded half sloper. It should be better, but I have to be honest and say that, so far, I don't know. Without anything truly quantitative about its performance it will have to suffice for now to review some general points on this vertical's pros and cons versus my (dismantled) half sloper.

  • Wide bandwidth: I have an excellent SWR from 3.5 to 3.8 MHz, and still reasonable towards 4 MHz. This is typical of ground-mounted verticals. Resonance, as presently tuned, is around 3.7 MHz. 
  • Lower radiation angle: At least that's the theory. The reality should become clearer over time.
  • Common mode: Less RF is getting back into the shack than with the half sloper. That may be nothing more than blind luck (see below for further discussion).
  • 160: The antenna could work on 160 meters with a suitable L network. It would not be efficient with those short radials and would require a remote switch between 80 and 160 matching networks.
  • Near-field coupling: There are 2 or 3 houses within the antenna's near field, and all the long wires and pipes that will couple to the antenna. Vertical polarization is typically more susceptible to this, partly due to maximum current at ground level.
  • Radials: These are a hazard in a suburban backyard. I will likely have to remove them in the spring. They are also half the minimum length they ought to be to capture most of the return currents, therefore allowing too much current flow in the lossy, poor ground I have at this location.

To put it simply, I didn't use one. These are wonderful tools, but not always necessary. Don't let the lack of one stop you from experimenting with antennas. If you understand the fundamentals and the design constraints it is often quite easy to do without one for most antenna work. This is not meant to endorse the idea you shouldn't use one. It you do, take care not to be mesmerized by the numbers displayed by a pretty toy. Test equipment cannot substitute for knowledge and insight.

With EZNEC, TLW and a few simple measurements I was able to get within striking distance of my goal. When the resonance measurement differed from the model I used the same model to estimate the true impedance at 3.65 MHz. Since the model resonance was 100 kHz higher than the measurement I took EZNEC's impedance calculation at 3.75 MHz as the baseline for the matching network. This was close enough to the reality to enable a rapid tuning procedure.

Actually I would have liked to have one handy. I could have better assessed where the theory and the reality clashed, and tuning would have taken less time. Had the model been inadequate, as does happen with low band antennas in a messy suburban environment, an analyzer would be useful.

I have been shopping around for analyzers suitable to my needs. A two-port VNA (vector network analyzer) may work best for what I have planned for designing and maintaining my future antenna farm. Reasonably simple antennas such as the one in this article can be designed and built without one.

About common mode

As much as I (and many others) like to harp on the subject of preventing common mode currents on the outside of coax and other cables, this antenna makes for a good study of futility. Suppressing common mode current from a yagi or dipole is easy in comparison to this antenna. Let's look at that now.

The tower is the antenna. Cables running up the tower are capacitively coupled to the tower along its full length. Antenna currents can directly enter the inside of coax via the other antennas, especially where yagis use feed systems such as beta and gamma matches. For the same reason it would be foolhardy to operate SO2R in a contest with one rig on 80 and another on of those yagis, at least not without some exceptionally good filters.

An exercise in futility?
My overhead cable runs, from tower to house, allow coupling between radials and these horizontal runs. Burying cables beneath the radial system would help, but that is not an option in my present installation. Neither is it practical to place common mode chokes that are effective at 3.5 MHz on every cable where it approaches the tower and the house.

The upshot of all this is that I cannot properly control common mode. The coax to the vertical and the horizontal runs of the other cables will in some measure act as supplementary counterpoises, and couple some energy back into the shack (and allow local noise pickup on receive). Neither is wanted, but neither is going to be completely eliminated. The scramble wound choke I made from the surplus length of cable in the 130' run is poor at best at 3.5 MHz, but I did it anyway since the cable had to somewhere after all.

All of this is to say that I understand the implications of what I'm doing. It is not recommended and it is never a virtue (i.e. don't make excuses), but often it is unavoidable in a small station.


The L network will be moved into a weatherproof box that will include an SO-239 connector for the coax. The box will clamp to the tower for mechanical support, to keep it out of the snow and for the electrical connection for the tower. I haven't yet decided on whether to use a variable capacitor or a fixed capacitor.

With that done the vertical should be able to survive the winter and give me reliable performance in the busy season ahead. I'll report back on this antenna when I have a clearer idea of how it's doing.

1 comment:

  1. Hi from Brazil my friend! I am studying how to figure out an LC network for my 60ft mast. A friend recommended to study the smith chart and my knowledge about antenna impedance and matching improved to the point I can understand your explantion. It helped a lot my friend. Great blog. 73 Carlos PY1CAS


All comments are moderated, and should appear within one day of submission.