Everything interacts with everything. When you swing a hammer the moon jumps. The laws of physics tell us so. It's a matter of by how much: some interactions are stronger than others!
The same applies to antennas. Antennas interact with everything, including that hammer. The critical metrics for the interacting object include:
- Material characteristics
It is not true that non-resonant conductors and insulators don't interact with an antenna. For example, a long plastic tube 10 meters from a 20 meter dipole has negligible interaction. But wrap that tube around the dipole and its effect is very apparent. A dielectric in the near field of a radiator reduces the VF (velocity factor) and a fraction of the energy, depending on the material's dielectric properties, will be dissipated as heat. We need to take care when selecting insulated wire for our antennas.
The Earth itself interacts with our antennas. Ground proximity and quality similarly affects verticals, their radials and really any antenna. Its proximity, dielectric properties and conductance can have a profound impact on loss and resonance, often far more than interactions with other antennas.
The same is true of non-resonant conductors. A short conductor near the tip of a 20 meter yagi element will alter the resonance of the element and thereby impact performance of the yagi. It is never invisible. But move it a few meters and, again, the interaction is negligible. The conductor can be a 2 meter dipole, a segment of a broken up steel guy, a power line or a metal clothesline.
Interaction is not a binary phenomenon: it's a matter of degree. It varies in accord with the previously listed metrics. Some interactions are easy to mitigate by, say, moving an antenna or interacting object a short distance. In other cases the impact of the interaction may not be concerning and therefore not worth the trouble of taking corrective action. Only you can decide. There is no one right answer.
Analysis and mitigation of interactions has been a recurring topic for this blog. Type interaction into the search box and you'll see. Examples include this, this, this, this and this, and there are many more to be found here and elsewhere in the ham literature. No matter how small or large your station it is worthwhile to give interactions some thought. I did it when my station was small and I do it now that my station is large. The only difference is that my objectives have changed.
In this article I will focus on two type of interaction, using a reference 3-element 20 meter yagi:
- Parallel 20 meter dipole in front of the yagi
- Parallel 40 meter dipole in front of and above the yagi
These will highlight "worst case" resonant interaction and non-resonant interaction, respectively. The characteristics of these simple interactions are both interesting and illuminating. Mitigation will be touched upon but not discussed in depth.
Parallel antennas resonant at the same frequency is a worst case scenario. Interaction is guaranteed. It is so severe that its effects are clearly discernible in the model with the dipole up to 20λ (400 meters) ahead of the yagi, rippling the azimuth and elevation patterns. The yagi azimuth pattern at right is for a separation of 10λ (200 meters). Resonant interactions have a long range! On the left the separation is reduced to 1λ (20 meters) to show the induced current when the coupling is extreme.
The antennas are placed 15 meters above real ground (EZNEC medium ground). This is more realistic than in free space since the ground reflections that are responsible for the elevation pattern (lobes and nulls) distort the pattern in a particular fashion. Height and antenna separation are important factors, however this aspect will not be isolated and studied in this article. But please keep it in mind when you plan your antenna farm, be it large or a small one.
The elevation plot shows the degree of pattern distortion due to interactions, with the dipole placed 200, 100 and 40 meters directly ahead of the yagi's centre. At 200 meters (10λ) separation there is a just a slight wiggle to the yagi's pattern. At 100 meters (4λ) the wiggles get larger and the increase of side lobe magnitude (more is visible in the azimuth plot, which isn't shown) steals 1 db of forward gain. At 40 meters (2λ) the wiggles are gone but the pattern distortion is severe and gain is sharply reduced.
Conservation of energy applies. Energy that balloons those side and rear lobes and fills in the nulls is not available for the main forward lobe. Interactions degrade both the pattern and the gain.
Of further interest is the effect on the yagi's impedance. The separation is 40 meters for the SWR plot above. At the design centre of 14.125 MHz the resistance drops from 37 Ω to 30 Ω while X barely changes -- for simplicity, the model uses a transformer to take 37 Ω to 50 Ω. Another important change is the antenna's Q, which has noticably risen. The higher Q reduces the SWR bandwidth.
By modifying the transformer, the SWR can be brought back down to 1 at 14.125 MHz. However, the SWR at 14.0 MHz still increases from 1.2 to 1.4, and from 1.6 to 2.0 at 14.350 MHz. With the seemingly large 2λ separation there is enough coupling to significantly alter the yagi's impedance and Q. The situation rapidly deteriorates at closer separations.
There is a lesson here with regard to resonant interactions. As the interaction increases (separation decreased in this example), we see the measured deterioration occur in the following order:
- Side lobes and nulls (or RDF if you prefer)
- Forward gain
Which is to say, trying to detect interaction by measuring the impedance or SWR is ineffective except in the very worst cases. Antenna performance can degrade quite a lot without any change in SWR. In this example, the SWR barely changes at separations of 60 meters (3λ) and greater. I had to reduce the separation to 40 meters to see a significant change in the SWR.
The metrics for non-resonant interactions are different. Let's briefly explore that interaction by replacing the 20 meter dipole with one for 40 meters. Although 14 MHz is the second harmonic of 7 MHz, you have to go to the odd harmonics, starting with the third at 21 MHz, to have a resonant interaction.
As expected, the interaction with a non-resonant antenna is weaker. It is so weak that we have to bring the 40 meter dipole closer than 1λ from the 20 meter yagi's centre to see appreciable pattern degradation.
The yagi's impedance is similarly less affected by close separation. The SWR plot is for a separation of only 10 meters (½λ) from the yagi's centre. Unlike the case with a resonant interaction the effects on pattern, gain and SWR for a non-resonant interaction like this one tend to occur together while varying the separation. When stacking the dipoles on the same mast or tower (above rather than in front) the interaction is lower for both cases since the yagi has a weaker field in that direction. That's been discussed in previous articles on interactions so I won't say more about it in this article.
If the interaction is concerning and the antennas must remain in their position there are methods for reducing the interaction to an acceptable level. The obvious one is to make a resonant interaction non-resonant by loading the antenna elements for the lower band to shift the harmonic resonance out of band. This is what I did to my 3-element 40 meter yagi to remove its affect on the 15 meter stack.
If the non-resonant interaction is with a guy wire segment or other part of the near environment you may have fewer options to reduce the interaction. You may have to live with it since those "fixtures" can't be moved. But do measure the impedance and build models if possible to remove the uncertainty. The degradation may not be as bad as you fear.
Let's consider another alternative that some hams have chosen. I will make a simple change to both the 20 and 40 meter dipoles at their closest separations of 40 meters and 10 meters, respectively. The interaction with the modified 40 meter dipole is on the left and on the right is with the modified 20 meter dipole.
The change to both is that the dipole has been cut open at the centre. This is done with a high resistance load in the EZNEC model. That's signified by the open box at the dipole centres in the antenna plots above. For those runs the resistance was 0 Ω, which leaves the dipole unaltered. Notice that the 40 meter dipole interaction worsened and the 20 meter dipole interaction has almost vanished. The SWR curves are not shown but show a similar response: better for the modified 20 meter dipole and worse for the modified 40 meter dipole.
In the case of the 20 meter dipole, opening the centre has converted the dipole into a 2 element collinear 10 meter array. When the same is done with the 40 meter dipole it becomes a 20 meter collinear array. At lower resistance values, including 50 Ω, the interaction is intermediate between these patterns and SWRs and the worst case scenarios described earlier.
I am not including those additional plots in the article since they add little to the discussion. Modellers can easily explore those and more complex scenarios, including transmission lines and complex loads.
There has been some misunderstanding about whether interaction due to an antenna not currently is use can be mitigated by shorting or opening the transmission line or the antenna feed point. Opening the element is best but this is rarely done since it is difficult except, perhaps, for verticals by opening the connection to its radials with a relay.
Doing it at the end of a transmission line -- most commonly at the antenna switch -- is not usually effective since the electrical length of the transmission must be considered, and it is rarely a multiple of ½λ. At other lengths, a short, open or resistor termination appears as a complex load at the antenna feed point. In any case, as stated above, any load on the unused antenna may only modestly reduce the interaction, and can make it worse. This includes a 50 Ω resistor, which some favour. This is easy to test by inserting various loads in the model.
A further difficulty is that this mitigation technique must be done for every element of an interacting yagi. Insertion of a load at the driven element or on the transmission line that would be effective for a single element antenna will be ineffective because the other yagi elements will still interact. You have to do all of them. For severe cases it is better to do what I did for my 40 meter yagi or, if you can, position your towers so that antennas that can interact are not pointing at each other most of the time.
Deal with interactions directly rather than by implementing token measures such as switchable loads in your antenna switch. Use those switches to ground unused antennas for lightning protection instead. We go to a lot of time and expense to design, build and raise high performance antennas so it makes good sense to effectively deal with interactions that degrade the performance. Anything else is wishful thinking.