When the 160 meter shunt fed tower was completed I discussed the need to connect the tower to the radials so that the path was through wire. Allowing the current to flow through the ground greatly reduces antenna efficiency because of its high loss.
The loss, although greatly reduced, is not eliminated because the tower is connected to the lightning ground rod. Lightning protection should never be viewed as optional! Therefore I accepted the small amount of remaining ground loss. Transitioning to a shunt fed tower from the previous wire vertical was driven by the desire to minimize loss via the tower ground due to the high mutual coupling of the nearby wire onto the non-resonant tower. On 160 meters "far away" can be quite close electrically.
It is also the case that ground rods should be avoided on a ground mounted vertical. The objective of the radials is to reduce -- as much as the budget for wire, time and money allow -- the antenna's near field penetration of the ground, and the consequent loss. A ground rod (resistance) in parallel with the radial field is never a good idea, even when there are only a couple of short radials, because it routes a portion of the antenna current through the ground.
But it's winter. In this climate lightning is very rare during the coldest months of the year. It is also peak 160 meter season. As I pondered the upcoming Stew Perry Top Band Distance Challenge and my intention to once again operate QRP, I ran to the computer and quickly modelled a few options to squeeze a little more out of the antenna. With QRP on top band even a little boost can help a lot.
In the articles referenced in the first two of the above links you will see how even non-resonant towers can have a substantial mutual impedance with a vertical antenna. They are not as far away as they may seem when you consider a wavelength of 160 meters. Part of the induced current distorts the vertical's pattern -- which can be good or bad depending on your needs -- and part is dissipated in the ground via the ground rod and, if the tower is connected to the steel within the concrete base, via the Ufer ground.
My tall guyed towers are not connected to the rebar. They sit flush to the concrete surface of the base pillar and over a pier pin that is also not connected to the rebar. It is therefore possible to lift the tower off ground by disconnecting the lightning ground. So I did. It was a 5 minute job. I moved quickly because it was cold and raining.
This is obviously a hack job. The black wire is the ground connection to the tower and the green wire goes to the radial hub under the gamma match. The 68 kΩ resistor drains static charge on the tower. The resistor value is not critical and although carbon composition is the best choice, at 1.8 MHz a 2 watt metal film or other non-wire wound resistor will serve for temporary use. As I said, lightning is very rare during our long and cold winters.direct electrical connection between the tower and rotating mast, measurement with an analyzer indicates that the mast and the yagis it supports are indeed connected and extending the electrical length of the tower. I will correct this lack when the warm weather returns.
Now I must back up and explain my technical motivation. Before venturing out into the rain with my toolbox I modelled the effect of the revised ground configuration.
Ground loss is dependent on the effective resistance of the ground rod(s). Close to 0 Ω is good because there is no loss. Unless you are on salt water it will never come near that low value. Very high resistance is also good because current won't flow through ground. It's in between where the concern lies.
From the earlier articles you will find that I estimated the ESR of the ground rod to be in the vicinity of 35 Ω, which is well within that intermediate range of concern. That's the value I used in the model.
The 68 kΩ resistor I placed between the tower and ground rod is effectively infinite in a low impedance system like this one. That pushes the direct loss due to the ground rod to near 0. The high resistance makes it safe to put a kilowatt into the antenna and not damage the resistor.
The ground loss reduction is quite small. The modelled efficiency improvement is about 0.4 db, which is close to negligible. But as I often tell people that when you add up those decibel fractions pretty soon you have a significant effect. Other fractions may come from improving the SWR, using better coax or increasing the Q of coils in matching networks.
Since I'd be using only 5 watts in the contest I shrugged and figured, why not do the experiment. I'll never know but maybe it made a difference. Conditions were not good and perhaps a quarter of my 200 QSOs required repeats of my call or the exchange. Only 3 Europeans were worked, and those were all very difficult. There is no reliable method to do an A-B comparison to confirm this small of an effect.
Notice that there is a 1 db deficit on one side of the antenna. That is roughly in the direction of Europe, where the other big tower is located. The tower is electrically long and non-resonant so its behaviour as a parasitic reflector is modest.
I ran a model with the ground on the other tower similarly modified and the pattern became almost perfectly circular. The gain difference in the favoured direction was less than 0.1 db, and that is truly negligible. Well, as I said it was cold and raining, so I didn't take the trouble to do it before the contest. Maybe I should have. While we can quibble over 0.4 db there is really no doubt that 1 db does make a difference in the weak signal conditions that generally apply on 160 meters.
I will improve the modified ground so that it is protected from the weather. Perhaps I'll do the other tower as well to get that extra boost towards Europe. It is probably safe until early April when the risk of lightning returns. By May the radials will have to be rolled up anyway for haying season.