The Art of the Fat Dipole
I'm on a quest to find an antenna for Jove and radio astronomy in general good enough to keep me interested all winter. Since I've started sweeping the 30 - 60 MHz range I want something that works up there too, not just at 20.15. In my experience broadband antennas of any sort are a compromise. The fact that this only has to work for receiving, not transmitting, makes this a little easier. This came to mind after the fan dipole and T2FD.
A fat dipole, at least my interpretation of one, is just a dipole with elements much bigger in diameter than they need to be for a single frequency. I had in mind to build this without spending any money on it, otherwise some aluminum downspout would probably be my choice for materials. If I can pull off the logistics I intend to make this out of galvanized steel electric fence wire, strands from used network wire, and dead CDs for spacers.
I made a fat dipole once back around 1992 out of used alumimum storm window frames from tall church windows and it worked great as a scanner antenna. I mounted the aluminum pieces on bits of hardwood slabwood with my coax connected to the center and stood it on end. It was about 15 feet tall and worked very well on low VHF (30 - 50 MHz), not so well on high VHF (148-174 MHz) and not terribly well at all on UHF. But any handy pieces of metal junk that are about the right size can work. I didn't spend money on that either.
This looks good at least in modelling it. The SWR isn't very important for receiving, but maybe I should use a balun and aim for somewhere around the middle of the impedance range predicted, maybe 300 ohms. I tried using 4 parallel wires, but 8 works better.
This was modeled using NEC2 with output screens by Xnecview under OpenBSD. A halfwavelength at 20.15 MHz is 7.4390188 meters, but I used +/- 3.72 from the center here since this is broadband anyway. That's 24 feet 4 7/8 inches overall for those of us that are metric challenged or without metric tape measures. There's a copy of the NEC2 input file here. The antenna height above ground in these is 2 meters, or a handy height to work on for most of us. Putting it higher will pick up more manmade trash, lower changes the impedance. There's nothing critical at all about the dimensions, this is just what I happened to model so the more you deviate the more unpredictable the results will be. They might be better. The antenna's worst modelled performance is at what should be its best frequency.
The set of 8 wires in parallel approximates a single fat conductor. I planned on CDs because I've got a stack of "coasters" I saved knowing I'd find a use for them someday. If I can make holes in them about 3/8 inch from the edges that will be close to the +/- 50 mm I used here. If you've ever tried to do anything with CDs you've probably noticed that drilling holes in them tends to split or shatter them. Melting holes with a woodburning kit or a hot nail, or an old soldering iron tip might work better. Don't do this with a soldering iron tip you plan to actually solder with again since it will be about impossible to get the melted plastic back off. And do the melting outdoors with good ventilation.
Some far field plots at different frequencies. The vgain number is the estimated gain in the direction toward the viewer, which changes as you tilt and rotate the antenna on the screen. I wouldn't put a lot of faith in the maxgain figure either, but that's the maximum strength of the strongest lobe, whichever way it happens to point. It's possible that to some degree instead of acting as one big dipole it works as 8 little ones in phase.
An important consideration for me is that at most frequencies it doesn't pick up much off the ends, so I can point the end at the house and not see noise from TVs, computers, touch lamps, vacuum cleaners, blenders, etc. And that just happens to orient it east-west which is right for Jove stuff. So put this east or west of your house and line the axis of it up with the house.
If for some crazy reason you're going to use CDs too and you want a drilling
template, I uploaded the one I made. It isn't perfect, but it works well
enough. This is a PDF, print it out and use it to drill 1 CD as a master,
then use that CD as a template to drill the rest. Drill where the lines
cross the inner circle. Finally, a use for those AOL CDs, old
Windows copies, every CD or DVD you screwed up on. The fact that there's
some conductive stuff inside that makes them interesting to microwave
doesn't matter here at all.
Drilling went much better than I expected. I taped a disk to the back of the template with 6 or 8 pieces of tape, then turned it face up on some scrap wood. I have numerical drills so I started with a #38 (0.101 inches) and drilled the master where all the marks on the template were. Then I untaped it and stacked all my CDs on an old 100-pack spindle so the center holes stayed lined up with my master on top. I used a battery powered hand drill at full speed, about 1500 RPM. Once I got the first hole drilled I stuck the next size smaller drill bit in the hole to keep the disks lined up then did a hole on the opposite side, stuck in another drill bit, then drilled the other 6 holes. I see a little delamination of the disks around the holes but no splitting. These drill bits aren't ground for brass/plastic either. It was a mixture of CDs, DVDs and the blank clear disks that come at the ends of 100-packs. One of the clear disks cracked and I had to throw it out.
I played around a little with stacking a few of these. It works, but because they're multifrequency antennas which cover more than an octave of frequency the optimum stacking spacing changes with frequency. I modelled them at 25 feet spacing, the same as a pair of Jove dipoles. With 3 of them the maxgain hits 16.51 DBI at 36 MHz and the 300 ohm SWR at 20 MHz is onscale at about 8.2:1. With 2 it's about 9.4 and maxgain is 14.28 DBI at 36 MHz.
I assume the optimum spacing changes with frequency for reasons not unlike how an interferometer works, and it could be corrected by switching phasing chunks of coax in and out. Here are NEC2 files for the double and the triple versions. Both have a usable pattern from 20 to 44 MHz without fudging the phasing.
Some graphics for the triple version:
One caveat here is that the ground in the models is probably better than a realistic one, unless you stretch hardware cloth or chicken wire on the ground under the antenna. Do that on a lawn and the grass should grow up through it in a few months so it isn't visible if you get it flat enough. But unless it's stainless steel it will also rust out in a few years. I don't plan to bother.
It took me a few days, but I really did build it and hang it. The left picture below mainly shows my wire connection method. I looped the steel wire around a #2 Phillips screwdriver to make a 1/4 inch loop, then put the small wires through that and soldered them. Each wire goes through its hole in the CD twice so it has an overhand knot of sorts.
The right picture shows how I held the other end in place. I made the same sort of loop in the steel wire, but through it I put 2 short pieces of steel wire, then stuck them back through the hole in the last CD and bent them at right angles. The short pieces are soldered to the main wire so they don't make noise when the wind moves everything. The small wires are tied off with some variation of two half hitches in the last CD.
Here it is up in the same spot I had my fan dipole hanging, maybe about 8 feet above ground. The gray PVC conduit at the center houses a homebrew 4:1 balun to match my 75 ohm downlead.
How does it work? I don't really know. I'm new enough at this Jove stuff that I haven't actually heard anything that I can confirm as either Jupiter/Io or a solar burst, but I keep improving things hoping some day I will. It's not really the best season for Jupiter according to Xephem.
This antenna sounds more lively, except for 6 meters. Running the spectrograph program the lines for the 6 meter repeaters aren't as strong as they are on the T2FD. But then this is more directional. One repeater is line of sight 20 miles off the end of this: the T2FD would pick it up better.
I went out with my MFJ-269 antenna analyzer and 3 sheets of paper, sat in the same lawn chair as I did when reading my fan dipole and took readings. I tried to duplicate what I did in testing that, which was hanging in the same position and I was connected through the same 6 foot piece of RG6. Adapters are a kludge: the MFJ has an N fitting, which is adapted to an SO-239 "UHF" fitting, then a BNC, then an F fitting. Same for both antennas though.
This is the same test from the fan dipole:
Of course SWR isn't really appropriate for a receive antenna, but it does indicate a relative quality of match. The fat dipole uses a 4:1 balun, the fan dipole is directly connected to RG-59. I think the match is better over most of the 30-60 MHz range I'm interested in with the fat dipole, except for 6 meters (50-54 MHz).
I read somewhere that any antenna looks good at the end of a long enough piece of coax, so mostly on a whim I disconnected my receiver and connected the antenna analyzer to the coax from in the house, through about 200 feet of RG-6. I got this:
It probably doesn't show much that's useful, except notice that the real low point in the SWR is about the same frequency as in the more directly connected plot at the top. There were many strange narrow resonances that didn't come through on this plot because they were less than 500 KHz wide.
|My raw data and Gnuplot control files.||
And an end view of it hung:
I measured the center (steel) wire and made it the same length as a regular Jove dipole. I didn't make any attempt to measure the little wires because I wasn't sure what effect tying knots in them after measuring would have. I calculated that about 1 foot of each wire went into the knots, about an inch at each disk. I also didn't worry about the velocity factor involved: The little wires are insulated but the steel wire isn't. The spacing of the CDs is about every 2 feet but it's not at all critical, more CDs would be better.
OK, I get it: the sun sounds like bacon frying. Not so much the constant sizzling, but the little spits and spatters that will burn you with hot fat if you're standing too close to the pan. There's a random assortment of little sounds, among the longest something a little like a cough or at times like somebody blowing out a candle. My strip chart recordings couldn't have told me that, I had to spend several hours listening after I knew Jupiter had set. All that quiets down after sunset. I added triggered audio capturing to my stripchart program.
I also hear something that's definitely 60 Hz modulated that reminds me of arc welding, like you've just struck an arc but it goes out again a fraction of a second later. That I worry about but it's so intermittent I don't know how to track it down.
6 meters came back.
This may change slightly before about 9/26/2012 and I'll replace the image but here's the fat dipole compared to an 18 foot T2FD. I've got an RF survey program I wrote which listens every 5 KHz for 1 second and logs the integrated audio level into a database then moves on. It covers the HF range on my IC-7000 from 35 KHz to 60 MHz, then starts back at 35 KHz and averages each run into the database. One pass takes about 3 1/2 hours. I have data from a 4 day, 27 pass run I did on the T2FD and I'm comparing the fat dipole against it. So far the fat dipole is on pass 13. The T2FD passes were all on weekdays so I want to get some weekdays into the fat dipole data too.
I lump all the results for each 1 MHz range into a bin by subracting the T2FD values from each bin and adding the fat dipole values so the T2FD is negative. There are actually 200 frequencies and 400 readings in each bin. I'd heard about the T2FD for years so I finally built one but I'm not terribly impressed by it. The fat dipole is just sitting out there stuffing signals down 200 feet of RG6 and makes a pretty good general purpose receiving antenna. I set my computer's clock from the 3.33 MHz CHU with a modified CHU program the other night, first time I've seen a good enough signal to noise ratio for it to work in a few months. Neither antenna has a preamp, but they both have homebrew baluns. If I had to live with one receiving antenna this might be it. Of course a single antenna won't work for interferometry.
Copyright 2012 ab1jx