Tuesday, January 12, 2021

Beverage Feed Line Interactions

At the back of my mind there lingered a concern about the planned path of my newest east-west reversible Beverage receive antenna. Due to its location the feed line would have to lie beneath or beside the south-north Beverage. Placing a long conductor alongside or below a Beverage will affect its performance. After a futile attempt to find a better location I am stuck. 

The alternatives are shown in the diagram. The original plan 'a' was for a coax reversible Beverage running along the treed wooden fence line between the fields, out to the edge of the eastern swamp. The density of the trees along the route are difficult to navigate, with many large trees and tree limbs at the required 2.5 meter height and the risk of deadwood damage is high.

Although route 'b' is better there are still lots of large trees and bush along the way. The required shift eastward of the west end and the minimum 150 meter length I require pushes the east end into the swamp. Maintenance will be difficult and repairs near impossible except when the swamp is dry (high summer) or frozen (late winter).

Route 'c' avoids most of the these difficulties. Half the route is free of dense foliage and I can extend the Beverage to 160 meters without delving more than a few meters into the swamp. There are large trees at both ends for support anchors. Since there are few trees on most of the route I will build a two wire reversible Beverage rather than one made from RG6 coax. The higher tension possible with wire allows for greater separation between supports. Those must be constructed. 

The problem with 'c' is the routing of the coax feed line to the Beverage remote switch. It must run parallel to the north-south reversible Beverage and interaction is certain without remedial measures. There is little point in going to the trouble of building a great antenna and have interactions ruin its performance.

For background I will note that it is safe for Beverage antennas to cross, provided the wires don't get too close, couple energy and thus affect each others' patterns. Even 30 cm (12") is more than enough in most cases. A common support at the intersection takes cares of this requirement. The other requirement is for feed points and terminations to be at least 5 to 10 meters apart so that the proximity of ground rods or radials doesn't promote common mode transference between antennas, which will degrade the directivity.

Having selected route 'c' I sat down in front on the computer to model alternative coax routes and configurations to get a handle on how to best proceed.


A Beverage requires real ground to be modelled in NEC2. Without a ground connection, doable with a MININEC ground, radials are affixed at the antenna ends to simulate ground. Just two 20 meters radials orthogonal to the antenna orientation and slightly raised are sufficient to couple to the ground for an accurate pattern calculation. These are 0.2 meters high. For this exercise the Beverage is in all cases 2.5 meters above EZNEC average ground.

The wire representing the coax is 0.1 meters high so that it passes beneath the radials without creating anomalous results. This, too, has radials at the far end where it (roughly) connects to the Beverage remote switch and its ground. The feed point to the other Beverage is 70 meters along the Beverage. There it simply terminates because the other Beverage is fed via a transformer (two actually in the reversible Beverage) that block common mode, and the other Beverage itself couples very little since it is roughly orthogonal to the one modelled and the termination grounds are very far apart.

The far end of the coax is 15 meters behind the Beverage and is therefore 85 meters long. It is broken by 1:1 common mode chokes that interrupt the common mode and divide the coax into 3 equal lengths of about 30 meters. The loads represent the capacitance between windings of the binocular core transformers. The value is adjusted per the transformer design. Transformers cannot completely block the common mode so modelling the capacitance matters.

Notice that despite the choke action of the transformer current is induced on their sections. These couple to the antenna above and can distort the pattern. That is what we are investigating with the model. The inter-winding capacitance is 10 pf in the above current plot. That is approximately what was measured by ON4UN for a 2:2 turns transformer on a Fair-Rite BN73-202 binocular core. The frequency is 3.5 MHz since at 1.8 MHz the effect of the coax is less visible.

The Beverage azimuth pattern is compared with and without the presence of the coax, and coax with and without common mode chokes. These effects are not large but would certainly be noticable. 

The main lobe is broadened while the minor lobes are smaller. The RDF is reduced by about 0.4 db which, though negligible, the minor lobe degradation can be a problem. The impact is less on 160 meters since the interaction is (roughly) inversely proportional to wavelength.

In case you're wondering (and you should!) whether the current on the coax is from the primary signal or coupled from the Beverage above I shoved the coax 30 meters to the left. The current plot provides a clear answer. Compare it to the otherwise identical one above.

On a general note, the model calculations should be treated with care. Although NEC2 is fully capable of accurately modelling Beverage antennas there will be inaccuracies. Wires close to real ground are one. Another is the break between coax segments. A capacitance load to represent the coupling between transformer windings should be verified with measurements. There are other environment effects that may be of greater magnitude than the effect of the coax under the antenna. Perfect models are impossible.

Despite the limitations of the model we can learn a great deal about Beverage performance degradation due to the coax lying on the ground. Interaction in the model, while not exact, is a warning sign. With that said, let's continue.

Parameter space

No two installations are alike. Indeed, each of my Beverages is unique with respect to construction, length and routing of the coax feed line. Exploring the vast parameter space of all possible configurations is time consuming, and likely unnecessary. For example, when we talk about a few decibels difference for a minor side lobe, in most cases the effect of the feed line placement is less than that of construction variability, ground variability and other environmental factors.

I limited my model exploration to factors that appear to be the most significant and universal to all installations. For those who want an exceptionally reliable and good pattern it is better to use a vertical array where current forcing ensures greater immunity to environmental factors. Beverages are more affected by interactions with the environment, and those are never fully under our control.

These are the model parameters I investigated:

  • Design and placement of 1:1 transformers along the coax feed line
  • Lateral offset from the Beverage of the on-ground coax feed line
  • Beverage length and operating frequency: these two are closely related
  • Effect on both directions of a reversible Beverage

I'm sure you can think of other factors; I know I did! But these are the ones that are most relevant in general and to my particular choice of route for the east-west Beverage. The model can be extended or adjusted by others to account for different situations.

The performance metrics that most interest me and most low band operators include the following. The importance of these and other metrics is up to you. The design of your antenna should meet your needs, not mine.

  • Gain: signal level in the favoured direction
  • RDF: figure of merit for directivity
  • Beam width of the main lobe
  • Side lobe and F/B magnitudes

You can see all of these, except the RDF, in the triple azimuth plot above. The RDF is calculated by subtracting the peak gain from the average gain. For EZNEC users the latter can be found at the bottom of the main panel after calculating a 3D pattern plot.  For a Beverage of 150 meters length the RDF is typically 9 to 10 db on 160 meters and 11 to 12 db on 80 meters.


Let's begin with common mode chokes on the coax feed line. The rule of thumb for a 160 meter receive antenna is that common mode chokes on the coax should be no more than 30 meters apart. Otherwise the outer conductor becomes a Beverage itself and the antenna current can seep into the signal path if one is less than careful with termination circuitry and grounds. The separation should be less to protect performance on higher bands.

Transformers wound on the Fair-Rite (BN73-202 equivalent) binocular cores were measured for loss and capacitive coupling and reported in ON4UN's Low Band DXing book. They measured about the same for 160, 80 and 40 meters.

  • Loss: -0.75 db for 2:2 turns; half that for 3:3; and, half that again for 4:4
  • Coupling: 8 pf for 2:2 turns; double that for 3:3; and, presumably double again for 4:4

Loss can be halved by using two cores at the price of greater coupling. I don't have the capacitance figures for those, however since I won't risk the higher coupling I am happy to stick with a single core. Although the capacitive coupling is frequency independent the reactance is not. Coupling increases with frequency because capacitive reactance is inversely proportional to frequency. 

There are obvious trade offs to be considered. Using 2:2 transformers minimizes the coupling but maximizes the loss: for two of these transformers the total loss is about -1.6 db. That is noticable. Receiver pre-amps have enough gain to deal with the additional loss but it can be a nuisance. 

In the adjacent plot I divided the coax into 5 sections with 4 × 2:2 transformers, represented by 10 pf series loads. The coax is directly below the Beverage, where interaction is worst case. Even on 80 meters the effect of the coax is almost completely neutralized. It is even better on 160 meters. The price is -3.2 db or more of loss and 4 transformers to build, maintain and weatherproof.

Alternatively the coax can be wound on a ferrite core to eliminate the loss. The price is a more expensive ferrite core and a larger weatherproof enclosure. Because of the foam dielectric most RG6 varieties have too large a bend radius for tight turns on a ferrite core. RG59 or similar 70 Ω coax should be used instead.

Should you choose this method be aware that achieving the equivalent choking action (isolation) of a 1:1 transformer on 160 meters is quite a challenge. It is more easily achieved on 80 and 40 meters. Again, ON4UN's book provides details.

I assume transformers in this analysis. These are represented by a series capacitor on the fat insulated wire model of the RG6. The capacitor value is adjusted per the transformer design. I ignore the additional feed line loss although that can be included in the model as a resistance at the feed point. For a coax choke on a ferrite core it is sufficient to model it as a series resistance.

Lateral offset

The farther the coax is from the Beverage the less its effect. Burial is even better. Even 2 meters away the coupling is noticably reduced and gain improved. From my modelling 5 meters is a desirable minimum distance to adequately minimize pattern degradation with respect to gain, F/B and RDF. I have room to do this in the bush to the east of the tree line.

There are diminishing returns. This should be clear from the earlier current plot for a 30 meter separation. Of course the coax has to be extended to that offset distance, if that is possible. In my case I can do it but there is a problem with laying a coax on the ground in the denser bush. Perhaps 5 to 10 meters but no more.

To determine the pattern degradation I modelled the coax laterally offset by 5, 10 and 15 meters. I deleted the chokes and ran the model on 80 meters as the "worst" case. With chokes and on 160 meters the degradation will be less.

As you can see the degradation is not so bad with a 5 meter offset. At 15 meters it is really no worse than the 30 meter case briefly described earlier. 

At a 2 meter offset (not shown) degradation is significant. It is fair to conclude that 5 meters is the minimum desirable offset. As we saw earlier, with chokes it is possible to run the coax directly underneath and achieve a good result. With a combination of chokes and lateral offset, negligible degradation can be attained. The options you choose are a matter of preference and circumstances.


My Beverages are reversible. The north-south Beverage that the coax for the east-west would lie beneath is made from RG6. The difference between coax and two-wire Beverages is slight with respect to its common mode use as an antenna. However the feed line beneath is in one case at the rear of the Beverage and near the front when reversed. It is worth a moment to see if there's a difference.

This is easiest done in EZNEC by exchanging the locations of the source and termination resistor since the antenna is otherwise symmetrical in the model. I only tested the worst case, with the coax beneath the Beverage and without common mode chokes.

The interaction is very similar for both directions, with the pattern degradation in the reverse direction slightly less. This is enough assurance that remedial measures will protect the Beverage pattern in both directions. It was worth taking the trouble to check rather than assume that would be the case.

On to construction

This has been a useful analysis. The models are simple to build and the potential consequences are large. This approach is better than relying on vague rules of thumb and unsubstantiated opinions of how to deal with the situation. I have a few options to consider.

In some respects the results of the analysis are obvious, at least qualitatively. Obviously locating the coax further away is better! The benefit of the analysis is to quantify those qualitative factors to reduce uncertainty. Building high performance antennas is a lot of work so don't ruin it by relying on guesswork and opinions.

My next step is to build the east-west Beverage. Until that's done the routing of the coax is moot, so I have time to make a decision. Should the interaction be greater than the models predict it is not difficult to alter the coax route and the chokes. For the latter it is more about transformer design and quantity since that is more convenient to me than coax wound chokes on a large ferrite core. The small transformers are easy to wind with insulated wire salvaged from scrap Cat5 cable.

In parallel I am prototyping an ergonomic Beverage direction controller for the shack desk. I'll have more to say about that in an article later this month.

No comments:

Post a Comment

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