Troubleshooting Reversible Beverages

When I closed off the article on my latest reversible Beverage antenna it was working in one direction and not the other. It is now working. The problem was nothing profound and I had every confidence that I would find the fault and fix it. Beverages are really very simple antennas. Reversible Beverages, although they may seem magical, are uncomplicated as well.

The two challenges I face fixing Beverages are the tiny transformers and the trekking. Working with fine gauge coated wire and tiny (and abrasive) ferrite cores is a test of patience. Although I have more patience now than when I younger it is never enough for these confounding creatures. I get less frustrated soldering and de-soldering SMD components!

Following best practices on building Beverage transformers never seems to be enough. I get intermittent shorts, broken wires, confused windings and when I go to test them with an analyzer or VNA I am never quite sure that I understand what I'm seeing. The latter is especially true with tapped (balanced) transformers that are used on 2-wire Beverages. It is perhaps no surprise that most hams who build Beverages prefer to buy rather than build these transformers.

The trekking is due to the location and length of these antennas. From my house, the far end of this Beverage is perhaps 400 meters away. It is by no stretch a technical hike. However the fields and bush are bug infested, full of hazards that will catch the unwary, and are either wet or snow covered. Tools have to be carried in. Multiple trips to diagnose and repair the antennas are tedious. 

Offense is the best defense

Diligence during the design and construction of a Beverage will avoid many problems down the road, and many treks into the bush to correct mistakes. I test each component, each circuit, the antenna itself and I double check. Problems still arise despite that effort, though perhaps less than otherwise. Trusting to luck or believing that ignorance is bliss are poor strategies.

Diligence includes but is not limited to the following:

• Clear all brush and trees that could fall onto or grow into the antenna
• Follow best practices when winding transformers, then test and test them again
• Measure impedance transformation and balance of transformers
  
• Check RF and DC paths through the switching electronics
• Test head end switching and transformers with a dummy load, and do it again at the antenna
• Test continuity of the antenna wire(s), ground connections and coax

Problems can and still do occur, though hopefully fewer than the "hope and pray" style of construction practice.

Understand what you're building

It is very worthwhile to understand how Beverages of all types work. Without that understanding it can be quite difficult to interpret what you hear and what you measure. For that I will refer to the article where I described how reversible Beverages work, in a manner that I hoped would be helpful to those unfamiliar with these antennas. For reading convenience, the diagram I used and that I will refer to, is replicated below.

There are characteristics of reversible Beverages to keep in mind. Each direction must have a load. That load is either a dummy load (resistor) in the head end electronics or the receiver. Without the load the Beverage is bidirectional. The Beverage wires operate in both common mode -- as an antenna -- and in differential mode -- as a transmission line. The only difference in this respect between a coax and open wire Beverage is that in common mode there is just one wire: the outside of the coax shield. The inside of the shield is part of the transmission line.

Centre tapped transformers are used to combine and separate common mode and differential mode signals on open wire reversible Beverages. Read the above linked article to see how. Understanding is key to testing and repairing these antennas. 

I am also reproducing from that article the annotated diagram copied from ON4UN's Low-Band DXing book. It will come in handy for the following discussion about transformers and testing.

Transformers

I don't enjoy winding Beverage transformers. The BN73-202 binocular ferrite cores are tiny (½" square), the holes are tinier (⅛") and fine enamel wire is fragile and hard to work with. 

After they're built they need to be tested, and they must be securely mounted in boxes or on PCBs to avoid stress and breakage. Clumsy handling can damage them. It is helpful to use long leads so that they can survive trimming when damage does occur. At low frequencies an extra centimeter (or inch) is inconsequential.

Mistakes are difficult to correct, and usually involve building a new one from scratch. The ferrite cores are robust and usually can be reused. In any case, they're inexpensive

On the left are a pair of Beverage transformers. The one with the enamel wire is the reflection transformer (T3) responsible for the malfunctioning east-west reversible Beverage. The secondary winding is tapped to pull off the common mode signal travelling from left to right (in the diagram above). 

This is not the first time I've had a faulty reflection transformer. Despite following good construction practices failure is not uncommon in my personal experience. Assuming I'm not a worse builder than others it is a task everyone should approach with care.

Too much tension when pulling the winding tight can cut through the teflon liners and abrade the enamel coating. Conductor contact with the core can result in poor performance or a short. But if you don't make the windings tight it will be difficult to fit the many turns in the holes despite the fine wire being employed. 

I did discover a short that was intermittent, which only occurred when one of the leads was under tension. I redid that winding and testing went well. Yet in the field it didn't work. The symptoms were different, as measured by an antenna analyzer, but still faulty.

I rebuilt the transformer from scratch. Plastic insulated wire is far less prone to shorts, and that is my preference. The problem is that it is thicker and few turns can be accommodated. The transformer on the left has 8 turns (2 + 6, for a 9:1 impedance transformation) of Cat5 wire (AWG 25). The one in the picture comes from my first Beverage.

Those 8 turns of insulated wire are about the best I can fit into these cores. During construction I have to tamp down the wires with a narrow jeweller's screwdriver to get that many turns. I doubt I could get 10 turns with my best efforts. Unfortunately the reflection transformer requires 11 turns (6 + 5).

I went with a hybrid design: the 6 turn winding with a centre tap made from Cat5 insulated wire and the 5 turn winding with the same small gauge enamel wire. Rather than teflon inserts I used the insulated wire winding as the protective bed for the enamel wire. I carefully tamped the insulated wire down, making sure the turns covered the inner hemispheres of the holes. The enamel wire was carefully placed on top of the other wires so that it never touched the abrasive ferrite -- ferrite is a hard ceramic.

This construction method worked out well. After testing it was reinstalled and the Beverage worked as it should in both directions.

Test before deployment

The head end with T1, T2 and the reversible electronics is quite simple to build. Testing can be tricky. You need the following:

• A way to inject 12 VDC into the connector and block the DC from reaching the RF analyzer
• Ohmmeter
• Several resistors
• Antenna analyzer or VNA -- the VNWA3 from SDR-Kits is great for plotting and measuring insertion loss

In the picture I am testing the failed reflection transformer. A small breadboard from an Arduino kit comes in handy. You must be careful not to bend or break the fine wire when inserting the winding tails. I use needle nose pliers close to the wire ends to push them in. You want no more than ⅛" of wire between the pliers and the bread board to avoid wire damage. Do it a bit at a time until the wire makes firm contact.

For transformers with a low turns ratio, use resistors not too distant from 50 Ω for best accuracy. Analyzers and VNAs become increasingly inaccurate as the impedance deviates far from 50 Ω, whether lower or higher. Since the reflection transformer has a near unity 5:6 ratio I used a 75 Ω resistor on the 6 turn winding, which is near to 50 Ω on the 5 turn winding.

For high ratio transformers such as 9:1 connect the VNA to the low impedance winding and a suitable resistor on the high impedance winding. A 470 Ω works well in this instance, or 680 Ω if you are aiming for 75 Ω on the low impedance winding.

Insertion loss measurements with a VNA (S21) are more accurate with low ratio transformers. For high transformation ratio one VNA port will have to measure either a very high or very low impedance. Insertion loss is not critical for receive antennas and most of the time I don't bother measuring it. A single port antenna analyzer is therefore suitable.

For balanced transformers it is a good idea to do two further tests. Test each half of the tapped winding to check that the impedance transformations are identical. They won't be exactly identical but they should be very close. It is also a good idea to test for balance by connecting the VNA to the non-tapped winding and tying the centre tap on the second winding to one side of the first winding. When balance is good there will be almost no difference in the measured impedance.

The test works because the centre tap is "neutral". This is similar to a power transformer where the centre tap of the secondary can be grounded. Had I been more careful I would have noticed that this test failed: the impedance dropped to nearly 1 Ω. But I was in a hurry and thought that I'd accidentally shorted the wires during the test. Instead it was an internal short in the transformer, one that did not appear when measuring the transformation ratio, other than a somewhat higher X value than I expected.

I always sweep transformers from 1 to 10 MHz to uncover any anomalies. If you use long leads expect an increasing X value at the higher frequencies.

The head end is tested in a similar fashion, except that we must be careful to understand the common and differential modes that coexist on the two wires of the Beverage. We will use resistors to represent the antenna modes.

The diagram from ON4UN's book shows coax to both directional ports. The switched head ends I built have one coax port, for the selected direction, and the other port is connected to a 75 Ω resistor. Both directions need a 75 Ω load (resistor or coax + receiver) or the Beverage will be bidirectional, just as happens in a simple unidirectional Beverage.

I test the head end switching before T1 and T2 are installed. After the transformers are installed both ports are shorted to ground at DC and the switching cannot be tested with an ohmmeter. After the transformers are installed they are tested separately for continuity, and only then are T1 and T2 connected (see the earlier diagram).

You'll have to inject 12 VDC, as mentioned earlier, to test the non-default mode. In this antenna the west direction is more likely to be used so I made that the default mode (reverse, or differential mode) and east is the powered mode (normal, or common mode). The DPDT reed relays allow either direction to be wired as the default. I stuck temporary labels to the relays so I wouldn't make mistakes

Testing the differential mode (reverse direction) is quite easy. Place a resistor (red in the diagram) across the wire terminals of approximately 670 Ω (I used 470 Ω and 220 Ω resistors in series) and measure the impedance at the coax port across the frequency range of interest. The SWR should be a flat line close to 1.5, up to at least 10 MHz. At higher frequencies the sloppy internal wiring will exhibit an increasing inductive reactance and the SWR will rise.

Testing the common mode (normal direction) is less easy. I've never bothered. In principle you need to tie together the Beverage wire terminals -- there is no easy access to the centre tap of T2 -- put the resistor (as above, but green) between the wire terminals and the ground terminal. Since there is no effective antenna ground in the workshop setting it is most expedient to tie together coax and earth grounds. Well, that should work, I think. 

I prefer to do the common mode test in the field on the actual antenna. So let's do that.

In the field with an antenna analyzer

A VNA is inconvenient in the field so I rely on my Rig Expert AA54. It requires no computer, it has large keys you can punch wearing gloves and the plastic body withstands abuse. A single port analyzer is perfectly adequate for the following tests.

All tests are done from the head end. Since Beverage antennas are so long that can require walking back and forth a few times to diagnose and resolve problems. That trekking is inadvisable for my situation because the bush grows thick and the ticks and flies are everywhere and the uneven ground is a hazard because you can't see it through the high vegetation. I assigned a high priority to repairing the misbehaving antenna so that it would be complete before the work became hazardous. It was that or wait for September or October.

We need to test both common mode (normal) and differential mode (reverse). These are east and west, respectively, for this antenna. In common mode both wires are in phase with the primary antenna current. The wires are tied together and terminated in the normal fashion with a resistor to ground on the other side of the transformer primary winding. The resistor for the test is the analyzer, on the transformer secondary winding.

Since the nominal impedance is over 600 Ω, it is a little high for accurate measurement with the antenna analyzer. I am using a 9:1 transformer to bring it down to about 75 Ω, which is an SWR of 1.5 on a 50 Ω analyzer. The container with the transformer and UHF connector was saved from an earlier Beverage project and conveniently saved for future use.

If you look closely at the left panel you can see the SWR plot oscillating around the 1.5 line. The oscillations are large because the far end is not terminated (wires shorted in this case, and no reflection transformer) making the Beverage bidirectional. Impedance oscillations are greater in bidirectional mode than in unidirectional mode for any Beverage. The test is successful.

For the differential mode test the open wire is a transmission line bringing the west common mode signal back to the head end via the reflection transformer. The SWR plot should look similar to that for the previous test since the common mode signal is not terminated at the analyzer, which again makes the antenna bidirectional during the test. For the two failed reflection transformer trials, in one case the impedance was very low and in the other it was too high.

It's amusing that this latest Beverage was completed right before the close of the 160 meter season. Although there is activity through our summer, it is a poor time for DX. For me the deadline is more concrete since the growing hay requires that the radials be rolled up in the next week or two. The short vertical will suffice until fall.

The growing hay is contrasted by the above picture taken the morning of May 1. Yes, snow in May. This is normal and the vegetation is well adapted to our climate. The flowers shake it off and continue their spring growth spurt. So does the rhubarb patch at the tower base. Rhubarb muffins were baking in the over soon after the May snowfall departed.

Soon the bush will soon be too wet and overgrown to traverse except for an emergency repair. I don't anticipate the need. With the new Beverage I have 6 receive directions, and it is very nice to have. They continue to be useful on 80 and 40 meters during the summer months when I am largely absent from top band.

With this article I am officially transitioning to summer mode. That means towers and high band antennas. Since summer is also a time to plan it is quite likely there will be an article or two about low band antennas.