Improving (maybe) the 160 Meter Shunt-fed Tower

My shunt-fed 140' tower works very well on 160 meters. But it can be made better. With the winter top band season underway, the time has come to do so. While it may seem odd to put the effort into 160 meters when the high bands are consuming the attention of most hams, there are two strong reasons: contests and DXpeditions.

I would eventually like to achieve additive gain and directivity by shunt feeding both of my big towers, that is a project for another year. Making each of the verticals more effective are beneficial on their own and for when I proceed with the phasing project. It is time to make the existing vertical better. By better I mean the following:

• Higher efficiency: lower ground loss
• Match: broaden the SWR bandwidth
• Permanence: replace the "temporary" construction with a more robust system
• Arc elimination: lower the voltage across the gamma capacitor

Lowering ground loss is perhaps the easiest: add more radials. I increased the radial count from 8 to 16. SWR bandwidth and gamma capacitor voltage are both dealt with by increasing the gamma "rod" diameter. Permanence, well, we'll see how I've done. Let's look at each item in turn.

For convenience of construction and radial placement, I previously placed the radial hub at the base of the gamma rod/wire, positioned ~2 meters from the tower base. I have now moved it to the tower base. 

Radial hub

I wrapped the tower with a band of aluminum flashing and stainless ¼" studs to attach radial wires. First I had to move the heavy aircraft cable wrapped around the concrete pillar with a small one on the tower's pier pin base. I have these on both big towers to serve as anchors for antenna work.

The new radial hub is pretty flimsy, so permanence remains elusive. It is a cheap and easy solution until I see how well the new design works. Tripping on a radial (not uncommon!) bends it out of shape, but it hasn't broken, yet. 

Assuming it works out I'll replace it with a copper band next year. I could easily wrap the base with copper wire and solder the radials on, except that the radials must be removed each spring for the farming season. Mechanical connections makes that activity more convenient.

The new radial hub consists of a 2" (5 cm) wide strip of aluminum roof flashing (~0.015" thick). It's cheap and I have a lot of it, but it is not very strong. Tripping on a radial puts quite a kink in it. 

I am not too concerned by that weakness. As I said earlier, it's a temporary measure until I am satisfied with the overall design. It'll do for this winter.

There are a dozen ¼" stainless bolts with the heads on the inside of the flashing. A nut secures each to the flashing and a set of washers and another nut are for attaching the radials. In the picture you can see that I have two radials per bolt.

To support the radial hub, there are several short lengths of scrap ½" aluminum tubes screwed to it. That keeps it above grade but not so high that the radials are not too exposed to mishaps. A narrow trench has been dug around the concrete pillar that I plan to fill with stone. That's to discourage growth so that I can trim the hay around the tower base without damaging the radials. Yes, I have accidentally cut a few over the years.

Feed point

When I first built the antenna several years ago I used a margarine tub to hold the gamma capacitor. The intent was that it would be temporary. Of course it became permanent, as these things always do. As you can see the elements were not kind to it.

For the experiment with a cage gamma rod I used another margarine tub. Of course it'll also be temporary. Or so I hope. 

The new one has one added feature: a coax connector. It beats the wire nuts covered in plastic and tape that I used to connect the coax and jumpers to the radials and gamma wire. The picture below shows the inside, with the gamma capacitor I ended with after tuning (more on this later).

Seriously though, I do have a permanent feed point system half built that uses proper components. I can go ahead and finish it now that both CQ WW contests are behind me.

One curious effect on the feed point impedance is worth mentioning. When I had 8 radials the resonant frequency measured in the shack was about 15 kHz higher than that measured at the feed point. That's because outer conduction of the buried Heliax transmission line, when connected, becomes the ninth radial.

With 16 radials the frequency shift is negligible: what I measure at the feed point is very close to what I measure in the shack. The transmission line has less effect on resonance when there are more radials. That is not a surprise since it was what I expected.

Radial wire

Each bolt on the radial hub terminates two radials: one of the 8 original and one of the new 8. All are 30 meters long. The original radials are AWG 18 insulated solid copper purchased new. Of the new radials, two are the same and one is stranded. For most of the rest I used wire that I bought at flea markets at bargain prices. They range from AWG 20 hook up wire to AWG 16 electrical wire. When I ran out of wire I used AWG 17 aluminum electric fence wire left over from Beverage antenna construction. 

It's a hodge podge approach that's cheap. Cheap matters since the radials require 500 meters of wire. I was not concerned with wire gauge since the more radials you have the thinner they can be. The reason is that the antenna current is evenly divided among them. However, do not make that assumption for a low radial count such as 4 since they are susceptible to imbalance due to variations in the ground composition and therefore the velocity factor in each. Too thin should be avoided since the wire is easily damaged.

Power lost due to ohmic loss in the radials declines faster than the radial count increases. That is true due to the power equation P = I²R. When you double the radial count, as I have, the current in each is halved and the ohmic loss is ¼ what it was. Ohmic loss declines with the square of the radial count, all else being equal.

Of the 16 radials, 3 had to be bent near the end because they ran into the stone wall that surrounds my yard and house. For this many radials the impact on current balance and performance is small and is not nearly as important as having the radials.

Gamma rod

I perused ON4UN's book and I made an initial trial with two AWG 12 wires spaced 40 cm (16"). The model (see below) suggested that it would work well to broaden the 2:1 SWR bandwidth to 130 kHz. I went ahead and built it with wire scavenged from a 40 year old 40 meter delta loop. It was a great antenna but I have better ones now.

The upper cage support was made from junk box metal. The brackets that hold the aluminum tube are more than adequate to handle the dead load and the tension to keep the cage taut. I added a support rope as insurance. The pre-drilled galvanized angle is found in almost every hardware store. The multitude of holes ease experimentation with wire spacing.

Another bit of scrap tube and hose clamps keep the wires in position at the bottom. Rope and brick weights provide tension. Despite the high RF voltage at the bottom of the gamma rod there is no risk of arcing because both wires are at the same potential. This arrangement was to be temporary until tuning was completed but it works well enough that I'll keep it for at least this winter.

The top of the gamma rod is at 55', 5' lower than the previous 60'. According to ON4UN the tap point is slightly lower with a cage than with a single wire. The model confirmed that but the difference is so small that lowering it was unnecessary. During tuning (see below) I raised it back to 60'.

Model

The tower is loaded with yagis which make the modelling process difficult. Since including them results in a large and unwieldy model, I substituted a single wire that is the electrical equivalent length of about 58 meters based on an earlier measurement of the tower monopole. The measurement was partly swamped by the ground ESR via the lightning ground rod (~75Ω) but gave a clear signal of resonance at about 1200 kHz. That is, the tower plus yagis is approximately an electrical ⅜λ on 160 meters. The physical height of the tower plus mast is 43 meters.

⅜λ is an excellent height for a vertical but it is difficult to match. The impedance is quite sensitive to the frequency. For that reason the SWR bandwidth is narrow. It was 70 kHz for the original gamma match. Further, the high inductive capacitive reactance requires a low value capacitor to cancel it and that results in a high voltage at the gamma match feed point. Increasing the bandwidth and taming that high voltage go hand in hand. 

Zooming into the feed point illustrates how the cage gamma rod is modelled. The top of the rod is the same but without the loads. MININEC ground is specified to simplify the model for the purpose of impedance matching; at this point I was less concerned with efficiency. The load in wire #1 is the estimated ESR (equivalent series resistance) of the soil and radial field. The load in wire #5 is the gamma capacitor. The source (feed point) is the circle in the bottom segment of the tower.

I adjusted the model parameters until I had a 50 Ω impedance. They were closely in line with the tables published by ON4UN. The gamma rod spacing to the tower was 0.7 meters (28"), with a 20 meter high tap point and the aforementioned 40 cm spacing between the cage wires.

The SWR bandwidth closely matches the 130 kHz predicted from the tables in ON4UN's book.

At 1000 watts the gamma capacitor voltage for the cage gamma is about half of what it was with a single wire. It is still high -- that's unavoidable in this situation -- but far safer and easier to construct with components found in my junk box.

When I reached this point I was sufficiently confident to proceed with building and installing the cage gamma rod. Before we turn to that let's first discuss an alternative model of the antenna.

In addition to the basic model, I also built one that includes additional detail. The 20 and 15 meter yagis at the top were modelled as single wires that are about 30% longer than the booms. That brings the tower resonance in accord with the measurement of about 1200 kHz. The lower yagis of the stack are not included since they typically contribute little to the total capacitive loading. I may add them later to confirm that considering that the lower 20 meter yagi is not far above the gamma rod tap point.

I constructed radial systems with 8 and 16 radials over EZNEC real medium ground. To keep them out and off the ground (not allowed with NEC2) they are positioned 10 cm high. The difference is slight in comparison to on-ground radials. It is a common workaround for modelling radials with NEC2.

The efficiency comparison is interesting because it is less than 0.1 db. However, this is likely not really true. NEC2 is really not up to the task of accurately modelling ground loss and I've run into this discrepancy many times before. 

The experiment measurements by N6LF and others for 160 meter radial systems suggest that the expected gain improvement is between 0.5 and 1.0 db. That may seem tiny but it can make a difference on marginal propagation that is routine on top band.

I am not equipped to measure field strength. Instead I measured the feed point impedance. In the initial configuration for the cage gamma rod, the impedance dropped from about 40 to 30 Ω. Although significant, the true improvement in ground ESR is not 10 Ω because this is the impedance as transformed through the gamma match. I did not reverse the calculation in an attempt to pin down the actual change.

Tuning

At right is the tuning setup for the cage gamma. When the picture was taken the tap point was at 55' so the cage wires reach almost to the ground. I was using a large capacitor since, as discussed above, the model led me to expect the voltage to be about half that as before. That corresponds to a low capacitive reactance: Xc = 1/(2πfC)

After several iterations of adjusting the gamma rod spacing I achieved a 50 Ω match.

There are two possible reactions to this beautiful SWR curve:

1. Wow! Mission accomplished.
2. No way! The antenna physics don't permit this.

I was suspicious but enough of an optimist that I was leaning towards the first reaction. I constructed the prototype high voltage gamma capacitor in the margarine carton and hooked it up. That evening I gave it a try. None of the European stations I called could hear me or could not hear me well enough to copy my call. Yet they were not weak and I usually have no trouble working them. It was then that my reaction changed to the second one and I began searching for answers.

It didn't take long to discern the likely problem. The next morning I looked over the antenna and confirmed my suspicion. When I connected the radial hub to the ground rod I forgot to reconnect the wire from the tower to the ground rod. The only path for antenna currents to complete the circuit from the base of the tower to the radials and coax shield was via the soil. The loss was therefore excessive. In other words, the ground loss dominated the feed point resistance.

When I restored the missing connection the beautiful SWR curve vanished. To cancel the inductive reactance the capacitance had to be greatly lowered. The impedance was approximately 21 + j0 Ω. That's far lower than what's acceptable.

Unfortunately I could not raise the impedance to 50 Ω. The best that I achieved was in the vicinity of 30 Ω. It was after several fruitless trips up the tower to adjust the spacing between the tower and gamma rod that I restored the tap point to its original 60'. That, too, helped very little. Worse, the further outboard I placed the gamma cage the capacitor value declined and the SWR bandwidth narrowed. What I ended up with was little better than what I had before.

Defeat

The impedance at 1850 kHz was 41 + j0 Ω and 2:1 SWR bandwidth of 80 kHz. With contests coming up and many other projects on my list I reluctantly stopped work on the antenna. It's good enough for this winter. The extra radials give my signal a boost and the SWR is easily dealt with by the amplifier. The rig's ATU is not needed across the DX segment of the band. It is needed when the I go outside that narrow range during contests when I don't use the amp.

I returned to the model to try and discover what might be happening. I can't say for certain despite learning a few new things about the antenna's behaviour. 

The first thing I learned is that a gamma match on a ⅜λ vertical does not react the same way as on a vertical that is closer to being ¼λ. The reason is that the impedance decreases as you move upward because the current node is ⅛λ above ground. That is about 20 meters in this case. With the tap point at that height, moving up or down increases the impedance. It is then transformed by the transmission line formed by the tower and gamma rod to what is measured at the tower base.

I confirmed the impedance behaviour with the simple model shown at right. The vertical wire is 57 meters long and is directly connected to MININEC ground. Peak R of more than 140 Ω occurs at 20 meters height. It is about 100 Ω at the base. I then performed a sensitivity analysis by varying the height in 1 meter steps. A 1 meter change in either direction changed R by ~10%. That's a lot!

Changing the antenna length by 1 meter in the model with the gamma match and cage caused large swings in both R and X components. Clearly the tuning is critical. It is possible there's a tap point further up the tower that will result in a 50 Ω match. Unfortunately that will bring the cage gamma rod close to the lower 20 meter yagi, and the possibility of the gamma rod threading between the elements. I'd like to avoid that if possible even though interaction ought to be minimal.

Clearly this antenna is more difficult to match with a gamma match than I expected. What I measured is not what I discovered in my cage gamma rod model or what I read in ON4UN's book.

That's all the time I have for the 160 meter antenna this season. I was already irritated that I missed the W8S Swains DXpedition appearances on 160 meters while I was in the midst of trying to resolve this mess. I likely would have been able to work them.

Planning for the next round

I am not so committed to the gamma match that I wouldn't throw it away and try something different. My options are limited since the tower is grounded. 

The only one that might work is an omega match. An extra capacitor is needed, and the best it can do is shift the impedance match from 40 Ω to 50 Ω. The bandwidth will not be improved and the voltage will remain high.

The alternative is to insert a switched L-network for the higher frequencies. I could then optimize the gamma match for 1810 to 1860 kHz (SWR better than 1.5). That addresses 90% of my needs. For contests where the activity runs up to 1900 kHz or higher, a switched L-network can lower the SWR up to at least 1900 kHz. The enclosure I plan to use is large enough to accommodate the gamma capacitor, an L-network and more.

To be completed

With the ARRL 160 meter contest coming up fast, I limited myself to an improved connection between the gamma rod and capacitor. It's cumbersome due to the 3 meter gamma rod spacing at the base (it's 2 meters at the top of the rod). 

Using an aluminum tube makes it easy to adjust the spacing without have to cut or add wire. The ABS pipe on the tower isolates the high gamma match voltage while allowing easy adjustment of the gamma rod spacing.

After the contest I'll take the antenna offline to move the gamma capacitor to its new enclosure. I will also take the opportunity to "rough in" the components for an L-network to lower the SWR higher in the band. Although I have no plan to add it this year, I want to make it easy to do next year.

I'll continue to contemplate and research the conundrum I ran into with gamma tap point and gamma rod spacing. I'd like to understand the problem better regardless of whether I install a switchable L-network.