You are viewing compilerbitch

f_default

Making stuff: A 20m-15m-10m multiband HF sloping wire dipole antenna

Image:Sloper-array.png
The rationale behind this antenna design was the following:
  1. Simplicity/low cost
  2. Field deployable with little effort by 1 or 2 people
  3. Able to work effectively without a tower or rotator
  4. Directional, ideally electrically steerable
  5. Multiband capability, ideally without a tuner or at worst with a limited range built-in autotuner
  6. Able to survive extreme weather conditions
  7. Should work well with a relatively short mast

Though slopers (sloping dipoles) are most commonly used for 40m and 80m as an adjunct to a large tower with a multiband beam antenna, the basic design scales fine to higher frequencies. My first prototype was a monoband 20m Sloping Dipole that proved itself very effective, so I put a bit more effort into extending the design to cover multiple bands.



 

Theory

The design presented here is an open-sleeve dipole. Sleeve dipoles are a classic approach to creating wideband antennas:
Image:Sleeve dipole.png
Here, a coaxial 'sleeve' surrounds the dipole. It is capacitively coupled to the driven dipole element, though not otherwise electrically connected to it.
Such an antenna has two significant dips in its SWR frequency response: one at the resonant frequency of the dipole, and a second at the resonant frequency of the sleeve. Usefully, though the SWR does not remain as low as it does at the two resonant frequencies, it nevertheless tends to stay fairly low (e.g., of the order of around 10:1 or so, well within the range of most antenna couplers) between those frequencies.
Since coaxial sleeves are difficult to construct, an alternative approach is often recommended in the literature:
Image:Dp sleeve dipole.png
Here, the sleeve is replaced by a pair of conductive elements either side of the main dipole. Again, as with a coaxial sleeve, these elements (usually called an open sleeve in the literature) are not electrically connected to the feed point and couple only capacitively.
My first attempt at modeling such an antenna involved trying to place resonances at 20m and 10m in order to keep the gap well-behaved enough to work well with a tuner, but NEC2 results weren't impressive -- seemingly, a whole octave was a bit too much. My second attempt modeled resonances at 20m and 15m, by extending the sleeve elements a couple of feet and tweaking their length to get the resonance in the place I wanted it. This time, I got good results across the 20 and 15 bands, and fairly decent SWR across the 17m band which would require a tuner, but only a relatively slight adjustment to get a perfect match. The 10m response sucked, however, so I kept working.
Image:Ul sleeve dipole.png
This time, I tried two unequal length sleeves, one tuned for resonance at 15m, the other at 10m. This retained the nice wide SWR dips at 20 and 15m, and gained another in the middle of the 10m band -- according to NEC2, such an antenna would not need a tuner for 20 or 15 and for most of 10, and with a fairly narrow range tuner would be good for all of 20, 17, 15 and 10, and with a slightly wider range tuner also good for 12m. This seemed good enough, so I went ahead with construction.

Element Spacing

From modeling results, element spacing (i.e., the distance between the driven dipole element and the parasitic elements) wasn't too critical, but it was apparent that getting them too close resulted in reduced feed point impedance, and getting them too far apart seemingly had them not coupling well enough to get good, low-SWR dips at the other resonant frequencies. After a bit of experimentation and genetic optimization using 4nec2, it became apparent that 1.49" element to element spacing was the preferred distance, so I rounded that to 1.5" for ease of construction and went on to design the antenna.

Sloping Dipoles

Sloping dipoles, also called slopers or full-slopers, are an interesting compromise between the classic flat-top dipole and a vertical configuration. In this approach, one end of the dipole is suspended (via an insulator) from a tower, and the other is staked to the ground, again via an insulator, such that the lowest part of the dipole is a few feet above the ground. Some of the literature recommends a 45 degree angle, though a slightly shallower angle actually gives better front-to-back.
Sloping dipoles are directional antennas. The reason for this is demonstrated by the following diagram:
Image:Sloping dipole.png
Here, we show a cross-section through the classic doughnut-shaped polar response of a typical dipole. Mounted vertically, dipoles are essentially omnidirectional. Mounted horizontally, typically a figure-8 pattern is seen. Tilting that structure through 45 degrees, as shown above, results in one side of the pattern pointing up in the air and forwards, and the other side of the pattern pointing at the ground. This results in an antenna response that is within a couple of db anywhere from the sides to straight ahead, with a heart-shaped null at the rear which is at its maximum pointing directly behind the antenna. The pattern resembles that of a 2-element Yagi or Moxon, though with a wider front lobe and a little less gain.
Interestingly, usual experience about soil quality is turned on its head with this kind of antenna, particularly with regard to front-to-back performance. In this case, you actively want the ground to eat the back lobe rather than reflect it, so the poorer the ground, the better -- exactly the opposite of the usual approach.

Baluns

Slopers are inherently somewhat unbalanced as a consequence of their ends being in rather different electrical environments. Consequentially, in order to avoid RF feedback on the screen of the feed line back to the shack, it is necessary to use a balun. My preference is to use a 1:1 voltage balun mounted right at the feed point. Though it's probably also feasible to use a choke balun in line with the feed line, e.g. mounted at the mast, radiation from the feedline between the choke and the feed point would most likely mess up the antenna pattern. Since such baluns are pretty cheap even for full legal limit power handling capability, I'd recommend just using one.

Electrically Steered Arrays

Image:Sloper-array.png
3 slopers attached to a single mast, 120 degrees apart
Since slopers have a wide, flat polar response curve, they are well suited to the construction of electrically steered arrays. Two antennas back-to-back on a single mast would allow, by selecting between the antennas with an antenna switch, the null to be flipped in the forward/backward axis. Adding a third antenna and spacing all three 120 degrees apart makes it possible to rotate the beam (and, actually, more usefully the rear facing null) a bit more accurately. Going to four or more elements is probably unnecessary, given the relative lack of directivity of the pattern.
Nothing more complicated than a 3-way antenna switch is necessary. A remote switch mounted at or near the mast is ideal, though it's also quite valid to simply run a separate feed line from each antenna back to the shack (as is the case with my own setup).
Note that this is not a phase-steered array -- only a single element is driven at a time.

Design Methodology

My design approach initially involved creating an antenna model in 4nec2, which was parameterised for the length of each element and for the element spacing. I used the genetic optimization features in 4nec2 to evolve a first-cut design, which was then optimized further by hand-tweaking until I got something that did what I wanted it to. This is an oversimplification, somewhat, because there were a few false starts. The tweaking overall took a few evenings work. Eventually, this resulted in a successful model that had the behaviour I wanted. I then exported the geometry by cutting and pasting from the NEC2 output file to a Mathematica workbook, which was then used for detailed calculation of the wire lengths. Since NEC2 gives only the coordinates of the ends of the wires, it was useful to be able to calculate the wire lengths in Mathematica. I also used Mathematica to create the visualisation of the 3-element antenna shown above, and also to help figure out how many of each kind of spacers I needed to make.

4nec2 Model

This is actually a second attempt at a 4nec2 model -- the original ended up very messy after many hours of hacking around trying various ideas. This is the result of redoing the model so that it could be included in a relatively understandable form in the slides for the talk I gave at the SBARA (South Bay Amateur Radio Association) meeting on January 14th 2011.
Final dimensions for this version of the antenna were:
  • 10.17858m 20m Dipole
  • 6.899761m 15m Sleeve
  • 5.047893 10m Sleeve

Model Results

Image:Full-sweep-swr.png
SWR across full design spectrum


Image:Full-sweep-gain.png

Gain across full design spectrum


Image:Full-sweep-directivity.png

Directivity across the full design spectrum


From the above analysis, it's pretty clear that things get very weird between about 23MHz and 25MHz. My suspicion is this is the frequency range within which the antenna switches from working predominantly as a half-wave dipole to working as a full wave dipole with respect to the driven element. That said, I have still managed to work the 12m WARC band, which sits in the zone of weirdness, just fine with a tuner. The 17m band, being in the 'nicer' part of the curve between the 20m and 15m resonance points works fine with very little for the tuner to do. Nevertheless, the main point of this antenna is to support working without a tuner in the 20m, 15m and 10m bands, so here are some more detailed SWR curves:

Image:20m SWR.png
SWR across the 20m band


Image:15m SWR.png

SWR across the 15m band


Image:10m SWR.png

SWR across the 10m band
As you go up in frequency, the antenna gets more narrow band, which is as expected. It's possible to put the SWR dip somewhere else in the bands -- here the dip on 10m is at the upper end of the voice band, but you could put it elsewhere without difficulty just by tweaking the length of the 10m element. In practice, the modeled peaks will be somewhat off because an actual installation will inevitably have different electrical characteristics to the model, so you always need to tweak the wire length empirically. The model gives a good starting point, however -- from experience, it generally suggests element lengths that are fractionally too long, which is generally preferable because it's easier to trim an element than it is to extend one.
One interesting characteristic of these antennas is that their directivity improves as the ground gets worse, because poor ground tends to eat the back lobe more effectively.
Image:Good ground.png
Pattern at 14.1MHz over Good Ground


Image:Average ground.png

Pattern at 14.1MHz over Average Ground


Image:Poor ground.png

Pattern at 14.1MHz over Poor Ground


Both gain and directivity actually improve on the higher bands:

Image:15m directivity poor ground.png
Pattern at 21.2MHz over Poor Ground


Image:10m directivity poor ground.png

Pattern at 29MHz over Poor Ground
For reasons that are not entirely clear, the antenna actually has a deeper rear null at 10m, which probably indicates that something other than the usual mechanism of the ground eating the back lobe is contributing to the null:
Image:10m directivity good ground.png
Pattern at 29MHz over Good Ground

Construction Details

Wire element lengths for the initial prototype:
  • Driven Element (dipole): 33ft 6.2"
  • 15m sleeve: 22ft 7.8"
  • 10m sleeve: 16ft 9.1"
Note that these lengths are a starting point -- actual results are rather dependent on the antenna's environment, particularly if it is put up in a sloper configuration. Therefore, it is recommended that you first cut the antenna to the lengths specified above (or slightly longer if you prefer), then trim them based on in-situ tests with an antenna analyzer. The lengths given in the 4nec2 model above are slightly different, because this is the result of a second modeling attempt that used slightly different optimization techniques.
I used 14 gauge stranded insulated wire (approx $55 for a 500 foot reel from Home Depot). You could substitute for-real antenna wire, but it's unlikely to work any better than the cheap stuff.

Wire standoff/insulator construction

Image:AntennaParts-sloper-array-web.png
The only critical dimensions above are the 1.5" wire to wire spacing. I cut out all of the parts from a single sheet of acrylic with an Epilog laser cutter, purely for speed, but there is no reason why a drill press and hand tools could not be substituted.

Tuning procedure

  1. Measure minimum SWR (or better still X=0 point if your analyzer can do it) for the 20m band. Call this frequency Fa.
  2. Assuming that you didn't get lucky and that the desired frequency is different, call the desired frequency Fd (I'd recommend 14.15 MHz as a good compromise).
  3. By the formula Final length = Initial length x (Fa / Fd), calculate how much you need to trim from (or extend) the driven element.
  4. Trim the antenna, and re-measure all of the resonances with an antenna analyzer.
  5. Repeat the same procedure for 15m and 10m in turn. Be aware that all of these changes depend on each other to some extent, so it may be wise when trimming to chop only about half the length the formula suggests -- I over-trimmed my 15m element on my first attempt as a consequence of trusting the formula too much.
All this said, if you have a decent antenna tuner, you can probably just construct the antenna with the dimensions above and have it work completely fine, but you may get slightly better performance by tweaking the resonances.

Conclusions

Though this antenna is by no means a large Yagi on a 100ft tower, it certainly works, and has been pretty effective for DX. Gain, as expected, is a bit less than a large flat-top dipole, though comparable with a 1/4 wave vertical -- this comes out of the modeling, but has also been supported by practical A/B tests. The response appears better behaved than the multi-element fan dipole design that it superficially resembles. My modeling has been based on running the antenna in a sloper configuration, but there is no reason at all why it would not also work well as a flat-top or inverted-V, though the exact wire lengths would change somewhat -- the method given for tuning will work fine, just allow a bit more length to start with because getting an antenna higher usually results in shifting its resonance(s) higher, so the default dimensions may be a bit short.
Construction is pretty straightforward -- make the spacers, cut the wire, and thread everything together carefully, starting from the balun outwards. I'm fond of using $42 Home Depot 'Longarm' painters poles, 23' long, rather than a pushup mast. If you decide to make one, I hope you have as much fun doing so as I've had in designing it!
-- Sarah Thompson/NQ6K

References

Comments

We were just going to solder a connector to coax wire for Christine's birthday to make an antenna for her short -medium wave Sangean radio for her birthday on Wednesday.

Somehow our baby step seems really small, now.

Christine says, "If she's in the area..." and then trailed off. I just remember calculating fields from antennas was not my favorite part of EM.

Excalibur

I went to the local electronics shop, bought a spool of wire, connected it to my new WinRadio Excalibur. Hard to do much better living in a 7th floor apartment.
Yay for invention documentation!

Ppst - lets plan April UK *nudge*
Interesting!

I've just got a fan array of dipoles up at the moment, but they are going to be replaced soon (they have lasted nearly two years for a design lifetime of about 3 months)

May have to clone your design for some portable work in the summer.

(Anonymous)

I always use the FIT2700 antenna swr analyzer (http://www.lbagroup.com/technology/fat2710-antenna-swr-analyzer.php) over the MFJ series.

Hmm...

Looks like the FAT 2700 has a frequency range from 30MHz to 2700MHz. Since this is an HF antenna designed for 14Mhz - 30Mhz specifically covering the 20m, 15m and 10m ham bands, it wouldn't seem to actually be useful? Or is the spec wrong on the link you included?

I'm no huge fan of the MFJ, don't get me wrong, but I've got one and it works, so... :-)
f_default

April 2013

S M T W T F S
 123456
78910111213
14151617181920
21222324252627
282930    
Powered by LiveJournal.com