When I last visited this antenna the central full size vertical was built, with enough radials (34, with 700 meters of wire) that it easily became my antenna of choice for DX on 80. During September I did the following:
- Laid radials for the 4 parasitic elements.
- Designed, built and installed the T-loaded wire elements.
- Rough tuned the parasitic elements.
- Hard wired the northeast element as a director and the southwest element as a reflector.
- Put it on the air and compared it with the inverted vee.
A set of 8 radials for each parasite was laid down before the elements were raised. They are each 15 meters long except for the one that ties the radial hubs for the parasite and the driven element. In earlier articles I described my radial system topology as overlapping rather than connected at busses placed between the 5 elements in order to reduce the work of soldering radials to busses at the expense of more radial wire.
After the parasitic elements were installed the impedance was measured and the radials increased to 16. The impedance was again measured to determine the trend toward non-resonance in the radial system, which strongly depends on the ground's dielectric constant and thus the velocity factor.
The EZNEC medium ground model for 20 meter long radials is a very close fit to what I measured: ~40 kHz increase in resonant frequency when the radial count increased from 8 to 16. I can now adjust the elements confident that I can predict what will happen should I add more radials later. Apparently the velocity factor with on-ground and shallowly buried radials is substantially lower than in the model. This isn't surprising since it is difficult to accurately model radials of this type with NEC2. In part because they must be perched a fraction of a wavelength above ground, where less of the field flows through the ground.
Before proceeding with tuning I added 2 more radials to each parasitic element, for a total of 18. These are wires running the ~15 meters distance between each of the 4 parasitic element radial hubs. Although the change in impedance was negligible this should slightly reduce ground loss for the entire array, no matter the direction chosen, including the array's omni-directional mode.
I chose 15 meters as the radial length for two reasons: limit the overall area of the antenna to minimize the land taken from the haying operation, and; limit swings in parasite resonance as radials are added.
Parasitic wire elements
Designing the T-top wire elements was a challenge. This is an element topology that NEC2 is not able to model accurately. I knew this from the start, as I learned when I put up my 160 meter antenna last year. The second factor influencing resonance is the number and length of radials. The 8 radials in that antenna are 30 meters long, which is approximately the same as 15 meter radials on 80 meters, relative to wavelength. The resonant frequency is pulled downward.
For a single element antenna with a low feed point impedance the correction is as simple as an L-network. It doesn't matter that the antenna is not resonant since the network easily accommodates a non-zero reactance, and the efficiency is high if the resonant frequency isn't far off. For a yagi there is no forgiveness in tuning the elements; you must get it right.
As a guide I used the K3LR dimensions for a 160 meter version of this antenna which you'll find in ON4UN's Low-Band DXing book, version 5. With a bit of geometry and adjustment to fit the physical layout of my antenna I came up with dimensions for the T and the vertical. A little was then added to both to allow room to snip wire to reach resonance. The initial length of each leg of the T is 5.5 meters. The vertical is 12 meters, which includes an extra 0.5 meters for tuning purposes.
Two wire elements are visible in the photo above. Unfortunately you'll have to squint to see them. I didn't properly compensate for sag so the vertical segment drags on the ground. The centre junction is simply a stainless steel bolt which the wires wrap around and then snugged tight. It's lighter than an insulator and more secure than solder alone.The ropes that form the top end of the catenaries for the wire elements were cut to length and installed when the stinger and top section were lifted. The elements were walked up the tower and connected and the bottom ropes tied to the anchors, located 25.5 meters from the tower. That's right at the edge of the radial system (10.5 meter element spacing plus 15 meter radial length) to minimize land use.
It took a couple of hours to adjust the tension of each rope to achieve the best compromise among element sag, force on the stinger and centering the stinger. Eventually I was done and ready to proceed to the next step: tuning the elements and lighting up the yagi.
Measure, measure, measure
As I said above, precision is of the utmost importance in getting a yagi working properly. To this end I was fastidious about measurements. A marked template was set up on the ground to ensure the wire for the 4 elements were cut to the exact same length. The wire elements are therefore identical. Exactly 3" of wire are wrapped around the insulator at each end of the T. When I tied the catenary ropes up the tower I made a mark on the tower where each T should terminate, which helped achieve symmetry. All element ground anchors are exactly the same distance from the tower.Although opposite elements are, and must be 180° apart, adjacent elements are not 90° apart. This symmetry is necessary in a 4-square but not in the yagi. In yagi mode only two parasitic elements are in use. The other two are floated (disconnected from ground) making them non-resonant and effectively invisible. I used this feature to optimize array aiming for my geographic location. Floating works, as demonstrated by no change to driven element impedance when the parasitic elements were raised and left floating.
The feed point of each wire element is similarly identical with respect to radial hub and its support, switching system anchor, back plate and enclosure. Before tuning the elements I measured the resonant frequency of all 4 of them, once with 8 radials and again with 16 radials.
The resonant frequency was below the design objective of 3.680 MHz so that they can be cut to resonance. More important is that my fastidiousness resulted in a resonance spread among the 4 elements of only 20 kHz, or 0.6%. Notice that the resistance part of the impedance dropped 7 Ω with 16 radials. There is obviously more ground loss to be eliminated. I can probably bring the resistance closer to 20 Ω by doubling the radials. But that's another 1,000 meters of wire! Efficiency doesn't come easy.
To measure any element of the array, including the driven element (tower), all other elements must be floated. That is, the monopole and radial systems must be disconnected. Otherwise the mutual impedance will mask what is really going on. One of the jobs of the switching system will be to float elements that are neither directors nor reflectors for the selected direction. All parasitic elements are floated when the array is in its omni-directional mode.
Rough wiring the 3-element yagi
The next step is the fun part: lighting up the yagi. Since the most useful DX heading is towards Europe I rough wired the northeast element as a director and the southwest element as a reflector. Since I don't want to cut wire at this time I calculated the amount to shorten the element to resonate the director to 3.680 MHz and shaved the insulation of the wire at the appropriate position to expose a tap point. The ~40 cm of wire beyond the connection has negligible effect on element behaviour.
For the reflector I used a rough wound coil of AWG 14 wire from my junk box since it looked to be about the right size. I attached it and alternately squeezed and stretched the coil until it resonated at 3.450 MHz. This only took 30 seconds since the analyzer continuously recalculates the impedance. All I had to do was move my hands out of the way after each adjustment.
With the element tuned I temporarily completed the connections with wire nuts. I move the analyzer to the driven element and measured the impedance of my improvised 3-element yagi at 25 kHz intervals.The resistance part of the impedance is about as expected. This clearly shows how important it is to achieve the lowest possible ground loss with the biggest radial system you can manage. Even with a ground loss as low as 5 Ω the loss can be -2 db. That's a lot.
The X value is not terribly important, provided its absolute value is not much higher than R. That helps to minimize matching network loss. However for now I am leaving the 4:1 SWR as is and correcting it with the rig's ATU. The estimated total transmission line loss is -2 db despite being almost 300' long. It isn't difficult to keep transmission line loss low at 3.5 MHz. Once the matching network is in place the loss should be better than -1 db.
Before going further it is worth a moment to talk about instrumentation. To achieve accurate and repeatable measurements it is important to have an analyzer with the requisite performance and to use it properly. The RigExpert AA-54 is quite good considering the price. There are better products on the market for a higher price.
The analyzer should be able to get the R and X values correct within a few ohms when the SWR is high, which most of the cheapest units cannot do. Not only is accuracy important but so is repeatability. That is, the values should be the same between readings, including readings made after the unit is turned off for a while and reconnected to the antenna under test.
Make sure your body is not influencing the measurement by moving around. Do the same for the analyzer itself, including when in your hand or supported elsewhere. Keep the test leads short or compensate for the length; better units can calibrate on the transmission line length, or use software such as TLW to convert the measured impedance to that present at the load.
Doing it right with a finicky antenna like this one can save you a lot of grief. Buy the best analyzer you can afford if you like experimenting with antennas. A 2-port VNA is favoured by those with a need to design and test more complex antennas or to analyze impedance transformation networks. A handheld unit like the RigExpert is perfect for me.
On the air
This article is being written after 3 evenings with the antenna in its hard wired direction to the northeast (Europe, North Africa and Indian Ocean). Not only has activity on 80 meters has perked up after the summertime lull there are several current DXpeditions roughly in that direction, including 9X0T and TO6OK. As always there are many Europeans active overnight and their sunrise.
When I first lit it up I was briefly disappointed. Switching from the inverted vee (now with an apex of 19 meters) to the vertical yagi the band noise dropped precipitously. It seemed that the antenna efficiency was low, so that the analyzer measurement might be hiding a deeper problem. Or, perhaps, the transmission line loss is higher than calculated. My worries dissipated when I tuned in a signal from Europe.
All European signals were stronger versus the inverted vee just as they were when the antenna was configured as an omni-directional vertical. But now the difference is substantially greater. I did extensive comparisons to ensure that Faraday rotation and other causes of QSB were accounted for by comparing through signal strength peaks and valleys. The improvement was consistent. However there is variation with different stations and path lengths. The variation -- elevation angle, polarization and skew -- demonstrates the value of having more than one antenna per band.
Signal differences were as little as 1 S-unit and as much as 4 to 5 S-units. That is promising. But what happened to the band noise? One hint was listening to W8 and W5 stations off the back of the yagi. The F/B is very good. Signals that were S-9 on the inverted vee would drop to S-3 or lower; as a vertical the difference was typically 2 S-units in favour of the inverted vee. What these differences are in decibels is hazardous to estimate since an S-meter is not a reliable instrument.
What appears to be happening is the attenuation of early fall atmospheric QRN from warmer locales south and southwest of VE3. The northern path to Europe is of course colder with less of the weather that contributes to noise. That is, the antenna has good directivity. I confirmed this by comparing reception against my 175 meter long northeast Beverage. The Beverage is still the better receive antenna, though by far less than with an omni-directional antenna.
Keep in mind that a well tuned 4-square can achieve similar or better F/B over a wider bandwidth than this yagi. I knew this before proceeding with this antenna. For me the ability to experiment and make a less expensive directive array of similar performance was a deciding factor.
I can see myself living with the yagi on 80 meters and reserving the Beverage and other planned receive antennas primarily for use on 160 meters. I have no plans for a directive transmit array for 160 meters, although that could change.
Up next
The switching units must now be assembled, installed and adjusted. The 4 parasitic element units must be identical to ensure identical performance, which is important when switching direction. This applies to the coils, relays, lengths and routing of internal wires. The unit at the driven element will contain the switching matrix, signal distribution and relays for the matching networks.
 |
| QRP friends visiting the 80 meter array (Credit: VA3RKM) |
The driven element stinger may need to be replaced, if not this fall then next year. It is proving to be mechanically marginal when tension is put on the catenaries. I would like to take up more of the slack. Replacement would take a few hours and should only affect the L-network due to a change in the electrical length of the driven element.
I don't anticipate completion before November. There are too many other projects to be done this fall and this job can be done during the cold weather. I may even lay more radials before the snow flies.
With the rough wired yagi performing so well I am really looking forward to completing this project.
No comments:
Post a Comment
All comments are moderated, and should appear within one day of submission.