Friday, April 17, 2020

Reversing the Single Wire Beverage

In my previous article on the reversible coax Beverage receiver antenna I said that my next Beverage would be the same oriented east-west, and that it would be done shortly. I lied.

Due to circumstances I've put that project aside. Purchasing a 150 meter reel of RG6 became more difficult and expensive due to the pandemic lock down with delivery uncertain. Then there were unresolved difficulties along the planned route of the east-west Beverage. Some of the trees are quite large and old with rot and other challenges that would require additional effort to prepare the route.

What I did have was half the 400 meter (¼ mile) reel of aluminum electric fence wire left over from building the 175 meter long single wire northeast Beverage. All I had to do was hang it along the same route to make an open wire transmission line Beverage so that it could be electrically reversed. The switch to select direction is slightly different than that for coax due to the use of a balanced transmission line. The critical difference is the transformers, two of which are more complex to build and test.

As you read this article you may wish to refer to the linked articles for details I am not repeating in this one, especially that for the reversible coax Beverage. This includes references to other material, links to which are largely not provided here.

How it works

As before let's start with the rough schematic of the two-wire Beverage from ON4UN's Low-Band DX'ing book. My article on the reversible coax Beverage reviews the basic operation of the reversible Beverage. Here I will largely limit the discussion to the differences between a coax and open-wire line Beverage, covering both the physical construction and the electronics.

As before signals from the right don't see T3 and signals from the left don't see T1 and T2. Due to the travelling wave behaviour of the Beverage the signal is zero at the antenna's forward edge. The signal builds as it travels along the two parallel wires to in common mode: signal amplitude and phase are identical on the two wires.

Signals from right build up along the length of the open wire line and encounter T2. This signal does not appear at the secondary winding because there is no potential difference on the two ends of the primary winding. Instead they combine at the primary centre tap and are fed through the primary of T1 to ground, making that transformer functional. The impedance is stepped down to that of the coaxial feed line. T1 is identical to that in the coax reversible Beverage since the surge impedance is approximately the same.

Signals from the left are also common mode when they encounter T3. The same behaviour occurs, But this time the common mode signal from the secondary winding centre tap flows through the primary winding to ground. Signal energy is coupled back into the open wire line in differential mode so that the two wires now behave as an open-wire transmission line. This signal power couples through T2 and is transformed to the impedance of the coax feed line.

Operation is a little tricky but is understandable when you study the schematic for a few minutes and carefully trace the signal paths. The reversible coax Beverage is easier to understand since the exterior of the coax is the Beverage antenna while the interior of the coax is the transmission line. For the two wire Beverage the two wires do both tasks.

Physical design

Twinning the single wire Beverage easier than putting in a new Beverage. It took about a day's work spread over 3 days to fabricate parts, clear brush and install the wire. A few of the tree insulators had to be moved to accommodate the second wire and the higher tension. Unrolling 175 meters of aluminum wire and routing it through the bush was probably the easier task!

To simplify construction I oriented the wires vertically. Since this can introduce imbalance (relative to ground) the wire spacing is just 10 cm (4 inches). Doing it this way saved me a lot of work since the original Beverage is sometimes supported on the left side of a tree and other times on the right. Rerouting the first wire is very difficult and dealing with rotation for wire crossover has its own challenges.

I can add rotation in its vertical orientation should I suspect an imbalance. This seemed unlikely based on the reported experience of other hams. Initial performance tests (see below) appear to confirm this.

Several methods are used to support and space the wires. There is a mix of commercial electric fence wire supports and ones I made from PVC pipe. The higher tension allowed me to eliminate a few of the tree supports. They were no longer needed or they were sufficiently out of line to develop excess lateral force.

Simple PVC spacers were used on the longer unsupported spans to keep the wire spacing uniform. Small gauge copper wires tie keep the wire from slipping out of the slots. I first used this technique several years ago to build a multi-band inverted vee.

Nails are galvanized 3" straight framing nails. Since the trees grow around them to repair the damage leave some of the nail shafts exposed. The force on the nails is quite low so they won't bend or split the living wood. As trees grow and die the supports may have to be moved or replaced. Although the trees can handle the injury this is wild bush where the trees are expendable.

The terminations are more difficult than with a single wire or for the coax Beverage. Ideally we want a smooth curve down to the electronics. Unfortunately mine take a sharper angle. On the left is the feed point (southwest end) and on the right is the northeast end deep in the bush alongside the swamp.

Some improvement is required though, as we'll see, performance seems unaffected. If nothing else the boxes need to be firmly anchored to prevent connection fatigue and crossing each other. Occasional tree contact doesn't seem to affect performance. Now that testing is complete I'll proceed with the "finishing" work. The ground wires are a little long because the ground rod had to be away from tree roots and the open wire line is difficult to route downward. This doesn't noticably affect performance.

Electrical design and transformers

The ON4UN circuit for this type of Beverage is reproduced here along with annotations for the impedances and transformer turns ratios for my chosen design. There is flexibility in the selection of impedances to permit finding a good balance between physical design (wire size and spacing and Beverage height) and transformer design (turns count and ratio and centre taps).

From: ON4UN's Low-Band DXing with my annotations

The size, spacing and height of the two wires determine the surge impedance. There are two of them: common mode for the Beverage antenna (both directions) and differential mode for use as a transmission in the reverse direction. Although the precise impedance values are not important in themselves the choices are critical for designing the transformers.

What we ideally want for both is an impedance that allow for an integral number of turns on the primary and secondary windings on the 3 transformers for an exact transformation ratio. This is next to impossible and is not an absolute requirement and the common mode surge impedance will vary from a theoretical calculation. We want to get close so that performance is what it should be.

The surge impedance of the open-wire transmission line is easily calculated and with a small spacing relative to height there is no variation due to ground proximity. For my antenna those distances are 10 cm and 230 cm, respectively. The transmission line impedance is 620 Ω given 10 cm spacing of AWG 17 wire. The Beverage surge impedance depends on wire gauge, spacing, height, and in the case of common mode, ground quality.

There are equations for both in ON4UN's book: equations 7-2 and 7-3 on page 7-87. To experiment with the parameters I made a spreadsheet (Open Office). Below is a screen capture. I highlighted the cell with the common mode surge impedance equation so you can see what it looks like when coded. Measurement units are centimeters.

The ground quality is estimated as "good" based on what I know of the area. Choosing 440 Ω is a reasonable guess. If in practice the value is wrong it can be measured (procedure described in the ON4UN book) and the transformers redone.

The selected transformation ratios are not exact but should be close enough. In the performance section below I'll come back to this. The turns ratios are the square roots of the impedance ratios. The centre-tapped windings have an even number of turns so that the tap is never at a half turn which can introduce loss and makes for awkward vertical mounting of the transformers on the PCB.

To minimize loss at high impedance more turns are required. That's why I chose 4:12 for T2 rather than 6:2. Other than the low impedance secondary of T1 all windings are small gauge enamel coated copper. With so many turns the insulated Cat5 wires I like to use do not fit the binocular cores. Insulated wire is more convenient since it doesn't require the PTFE protective lining.

These are the 3 transformers built and tested, and the reflection transformer in its enclosure ready to be installed. I put temporary labels on the transformers so that I don't confuse them. Later the reflection transformer was taped to the box bottom to stop wire fatigue due to shaking by the wind.

For best performance the centre-tapped transformers must be balanced. ON4UN's book discusses the importance and gives a procedure to test balance. To be frank I found that advice less than useful. The more important thing, and which is not described, is how to achieve balance in the first place. Knowing that a transformer is unbalanced tells you next to nothing about how to correct the problem.

Regrettably I am no expert on RF transformers. I came up with a few steps to make the centre-tapped winding physically symmetrical with the expectation that will lead to electrical symmetry. Let's find out how I did.
  • Do the centre-tapped winding first. When there's another winding underneath you will find that it is not sitting flat or centred on the inside wall of those little holes. That will cause the wire of the second winding to slip into gaps or to one side on the first winding. The result is a small inequality of the length and shape of each turn.
  • Use an even number of turns and adjust the turns of the other winding to achieve the required turns ratio. Tapping at a half turn is awkward, can increase loss and routing the tap from the other side of the core can introduce stray inductance and capacitive coupling to other components. It's small to be sure but when you want to protect pattern mulls of over 20 db it can make a difference.
  • Do the winding with two equal length parallel wires, each enough for half the required turns. For example, for my 4:12 transformer the centre-tapped winding is done with two wires for 6 turns each. Tape the starting ends together. After each half-turn (one pass of a core's hole) ensure the two wires are bedded together on the inner side of the hole then, keeping the turn  radius equal route them down the other hole for the second half of the turn. Look inside the hole to make sure and use a small screwdriver to tamp down any "bumps". Don't squash the wire! Pull the wires taut to remove any slack that may be there. Repeat until done.
  • When the winding is complete check that the ends of the wires protrude the same length. If not something went wrong. Tiny amounts can be ignored, however defining "tiny" isn't easy! Use your best judgment.
  • When done the end of one wire and the start of the other are stripped, wound together and soldered. Don't rely on a manual connection since it can slip as it handled.
  • Build the second winding over the first.
To test the transformers I placed a load across the full centre-tapped winding that would result in an impedance not too far off 50 Ω on the other winding. Trim the leads of each winding to equal length. That should give the most accurate measurement by the antenna analyzer. Check that the transformation ratio is approximately correct and that the residual reactance is close to zero.

Do the same test for each half of the winding. You likely will not get what you expect for an impedance. That isn't too important. What we want is for the measured impedances to be equal, both R and X values. Following this process the impedances on both centre-tapped transformers was less than 1%. It may have been better but that is more than most analyzers, even the best can accomplish. Use an analyzer or VNA with excellent repeatability -- same result for all measurements of the same test setup.

Switch box

There is very little difference between the switch box for this antenna and that for the reversible coax Beverage. You can read that article for detail not covered here.

The major difference is the location of the external attachments. The studs for the two Beverage wires are symmetrically placed on opposite sides at one end of the box. Avoid putting tension on the studs since that will eventually crack the plastic. Use strain relief elsewhere and place the box wherever it can be at least protected from the elements. For our snowy winters I want the box at least 45 cm (18") off the ground. Even then it will get buried in the snow occasionally.

The coax feed line and ground connection stud are on the opposite end. All the attachments are in keeping with the electrical layout and allow the forces of those attachments to be as symmetric as possible.

The circuit board looks a little different due to the arrangement of the attachments. For example, the circuit to separate the RF and DC components is on the lower left closest to the RG6 F-connector. The default (unpowered) direction is northeast since that is the most used direction for my style of operating. Southwest relies on the reflection transformer and powering of the relays through the coax. The load for the unused direction (to dissipate rather than reflect signals) is a series-parallel threesome of 51 Ω resistors.

The quickly drawn partial schematic of the transformer section of the circuit matches the physical layout. I tried to centre T2 (lower right) between the two wire studs for maximum balance. I made the two wires the same length even tthough a difference of a few centimeters is negligible at low frequencies.

For maintenance predictability the default direction is at the top, the same as for the other reversible Beverage. For my operating preferences northeast is the sensible default. The relays are powered to select southwest, the reverse direction.


Despite extensive testing the Beverage did not work when first installed. In the northeast (default) direction the SWR wasn't bad but there was evidence of poor rejection to the southwest. In the reverse direction the SWR was extreme and there was no signal at all. The switch box was brought indoors for troubleshooting.

When I assembled the unit I did not retest the RF behaviour. I reasoned that since the signal paths were good and the transformers were tested this wasn't necessary. This time I placed a 660 Ω resistance across the open-wire studs (to represent the transmission line impedance). With the relays powered I measured an RF short. That explained why southwest didn't work.

The PCB was removed and the DC resistances checked. This is difficult for the transformers since they show DC connectivity. With magnifying spectacles I discovered a solder bridge between the pads for the centre-tapped T2 primary winding. I cleared the debris with a sharp tool. Retesting yielded an impedance very close to 75 Ω. That's perfect.

Happy with my repair I rushed out into the windy and snow flurries to reconnect the unit to the feed point. Back indoors I eagerly tested the antenna. To my dismay there was no improvement. Either the unit had another problem or the reflection transformer T3 was faulty.

I retrieved both units for further investigation and repair. Before the hike to the far end I shorted the two wires so that I could test wire continuity. The open-wire line was fine. All I learned was that 350 meters of #17 aluminum wire has a resistance of about 13 Ω. As a transmission line for the reverse direction this is negligible loss since the line impedance is 620 Ω.

Again the switch box worked as it should. When I tested the reflection transformer T3 I was surprised to discover an RF short. The centre-tap and connection to the primary were teased apart with a soldering gun to allow testing of each winding. A DC test showed that the two halves of the secondary (high impedance) winding were shorted together. There was no short visible on the leads and wiggling had no effect.

The windings had to be removed. I can only assume that the enamel was thin or missing on both halves of the secondary at the same place. I rebuilt the transformer with new wire and like the first time it tested good. After soldering it into its box I again tested it to be sure that soldering heat hadn't been the cause.

More hiking and fighting with wires and cable in the blustery weather saw the Beverage reassembled and ready for its next test. This time the antenna worked in both directions. Mission accomplished, or so I hoped. It was daylight and I would have to wait until dark for a proper test on the low bands.

An interesting observation is that the differential mode transformers on an open-wire line have excellent discrimination against common mode signals. If not for that I would have heard something rather than only receiver front-end thermal noise when the reflected (southwest) mode was shorted out.


As for the coax reversible Beverage I tested the SWR after checking that signals were being received in both directions. As you can see below the match is quite good.

An hour before sunset I did my first test on 40 meters since there were Europeans stations coming in quite strong. The direction switching was remarkable. F/B was at least 20 db for both Europe and US stations approximately in the opposite direction. The same test a little later on 80 meters had the same excellent results. F/S was even better when compared to the vertical.

There was little activity on 160 meters to test with that evening. I had to check at intervals for signals, and I tried FT8. In all cases the F/B performance was excellent. There was no discernible degradation of signal discrimination in the antenna's original unidirectional northeast direction.

The 175 meter length is a popular choice since the minor lobes move away from the 180° opposite direction to give the optimum F/B (on 160 meters and harmonically related bands of 80 and 40 meters). Although the RDF is little different for slightly shorter and longer Beverages many prefer to have the best signal rejection directly off the back.

That the F/B and RDF appear to be what they should be, and about the same in the northeast direction as for the original single wire Beverage, indicates that the transformers are well balanced and that hanging the two wires vertically introduces negligible imbalance.

To change direction I used the same temporary DC injector built for the coax reversible Beverage. Notice the slight peak in the SWR near 5 MHz in these SWR plots and in those of the coax reversible Beverage. This is likely due to the 100 μH choke in the injector which has a self resonance near that frequency. Picking RF chokes with a self resonance outside amateur bands is beneficial.

Finishing up for the spring

The final task this spring is the remote switch box to select Beverage direction. That work is mostly done with completion slated within the next week. Expect to see an article on that later this month.

The remainder of the Beverage plan will wait until fall or, more likely, early winter. For the rest of the spring I will focus on the towers and yagis. There is a lot to do, most of which has been delayed by the "social distancing" of my friends who help me with these big jobs. Unfortunately there is much that will have to be delayed to August after the hay has been harvested.

Stay safe out there.

No comments:

Post a Comment

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