Sunday, September 4, 2022

The Ham Radio Series 34 – One Antenna To Rule Them All – Part 2

I was just going to call this “Multiband Antennas,” because that’s what it’s about, after all, but you’ve got to admit this is a catchier title.  

We started out talking about the EFHW – End Fed Half Wave (on its lowest frequencies) – because of their wide popularity these days.  I referred to these as being decades old, and they really are about as old as radio itself.  “I’m so old…” that I don’t think of these as EFHWs, but as Zepp antennas, named after the long wires hung out of the back of a Zeppelin in the early days of both air flight and radio.  Feeding an antenna at the end instead of in the center, like a dipole, is simply more convenient; especially when the far end is flapping in the breeze like it was for the Zeppelins!  Sure beats putting up three really tall poles for a dipole – two for the end and one for the feed point.  For those original Zepp antennas, the set of reasons for wanting a 50 ohm antenna like we do today just weren’t there.  There were no solid state transmitters, tube transmitters could be tuned for different output impedances than we design for today – 50 ohm coaxial cable hadn’t even been standardized on -  and more.

The drawback of high impedances like the Zepp/EFHW is that for a constant amount of power, there’s a couple of important relationships that point to trouble.  

Relationship 1:  Power is Voltage times Current, P = V*I  (current is usually abbreviated as I)  If you’re used to the way ham and technician study guides write that, they tend to use E for voltage – short for electromotive force – that conveniently turns that into P=I*E, or P=IE and who can forget pie?   

Relationship 2:  Voltage is Current times Resistance,  or V=I*R and substituting back into the first equation, we get P=V*I = (I*R)*I or P=I2*R.  If P is constant, and R goes up, I2 (current) has to go down.  Since I has gone down, and the power has remained the same, the voltage V has to go up to keep the product I*E constant.    

So what?  The higher the voltage, the more likely that arcing over becomes.  Another phenomena is called corona, a type of discharge different from arcing but that can sometimes be seen (at night) as a glowing area around the feed point, or highest impedance points.  

Let’s put some numbers here.  A typical HF ham rig puts out 100W.  In a 50 ohm system, the current would be SQRT(100/50) or 1.4Amps, and the voltage is 70.9V (SQRT is "square root of").  In a 5000 ohm system, the current drops to 141 milliamps while the voltage goes up 709V.  At the end of a coax run, due to the loss in the coaxial cable, both voltage and current values will be smaller.  In commercial and NASA specs, they consider voltages above 250V to be the start of extra precautions due to corona.  Because of reduced air pressure at higher altitudes, corona forms at lower voltages than at sea level.  Those of you in mountain states will have more ideas of how much this affects you than I will.  It’s similar to how the air pressure at high altitudes reduces water’s boiling point and affects cooking times.

To borrow a line from an old commercial, so what’s a mother to do?  Use something lower in impedance.  Voltage comes back down, current goes back up, chances of arc or corona go down.   It’s one reason that resonant dipoles and monopoles became popular.  Recall that a dipole in free space has a feed point impedance closer to 75 ohms than 50 and a vertical (monopole) is ½ of that or closer to 36 ohms.  (If you read around enough, you’ll find stories from hams who have put down more radials (quarter wave ground wires) to make a more perfect ground and found that their antenna’s impedance match has gotten worse because they’ve driven the impedance closer to 36 ohms.)   

Still, a dipole or monopole cut for one band is only good for that band and the subject here is multiband antennas so this sounds like a dead end.  To be completely accurate, a dipole or monopole is also good on it’s odd harmonics.  That is, a 7MHz dipole or vertical will work on 21 MHz (3rd harmonic) or 35 MHz (5th harmonic; there’s no ham band but just to show the example).  An 80m antenna can give you 30m on its third harmonic, if tuned for the very bottom of the band.  It turns out that with those two exceptions, this is an almost useless thing to keep in mind.  

This is a subject that has been studied for the history of radio, so let’s cut to it.  There are two main ways to make multiband antennas: nesting together several independent antennas at the feed point or using circuits to make them multiband.  These are called traps.

A trapped dipole uses parallel resonant circuits to isolate sections of the antennas from each other.  Parallel resonant circuits present high impedance - an open circuit - which isolates parts of the antenna. An example from the ‘net. 

By way of example, the innermost dipole would be cut for the highest frequency band; let’s say 10 m.  The trap would be designed to be parallel resonant just below 28 MHz, so it wouldn’t allow the 10m signals to “see” (be affected by) the next sections of wire.   The next section of wire between the traps (plus some residual reactance from the trap, typically inductance which makes the antenna shorter) lets the antenna resonate on the next band down that’s desired; perhaps 12m, 15 or 20.  The second trap would cut off below the bottom of the second band so neither 10m or the next band “see” that next section but allow lower frequencies past it.  Finally, the last section of wire would allow the lowest band.   Designs that resemble this drawing for 10, 15 and 20m are extremely common, and are widely sold as “triband beam antennas.”

There are free pieces of software that help you design and build traps.  Some use coaxial cable, some have you create air wound inductors and use lumped capacitors. A good source of information and ideas about software is at W8JI.

The other main type of antenna is multiple single-band antennas fed at a common point, and often called a fan dipole.  

They don’t have to look like this; in fact, they probably shouldn’t.  This website presents information from the Stanford Research Institute that imposes some physical distances that seem to be necessary.

“The SRI found that the wires at the center feed point had to be separated by at least 5 1/2 inches vertically and the ends separated by 38 inches in the 2 to 18 MHz range. As in any fan dipole construction, all of the dipoles are connected in parallel but in the SRI method, the separation between them at the feed point must be maintained.

By this simple change they found that you could accurately cut the antenna element lengths for given frequencies and eliminate the need for pruning.

In the drawing above, the lowest frequency antenna is on top and is cut 4% short of the standard 1/2 wave length. (Length in feet= 0.96 times 468 divided by the operating frequency in MHz).

The middle frequency antenna (lower in frequency), is cut for an exact 1/2 wave length. (length in feet= 468 divided by the frequency in MHz)

The highest frequency antenna is at the bottom and cut for 1% longer than the 1/2 wavelength (length in feet= 1.01 times 468 divided by the frequency in MHz)

Compared to the construction effort of a standard multiband dipole the only difference is the fabrication of a feed block or center insulator that is about 12 inches vertically by 3 inches wide, so make sure this is made of a good insulating material such as Lucite, Bakelite, fiberglass, or PVC.” 

I’ve run a fan dipole before, although it has been a while.  The SRI information wasn’t available back then, so I trimmed it manually with an SWR meter.  I think the wires were cut for 40, 30, 20, and 10; the antenna also worked on 15m because of the third harmonic of the 40m wires.  There was some interaction with the 30m element and I don't remember how I got around that.  Probably an antenna tuner.  The spreaders were cut from some six inch long, inch wide, 1/16” thick pieces of plexiglass, drilled with four holes on every piece and enough on each side to keep the wires parallel to each other and not touching.  I worked many countries on that antenna, with it mounted like an inverted Vee sloping to ground on both ends and folded in the middle so that the two sides of the antenna were around 90 degrees to each other.

Trapped and fan antennas are widely available and (of course) can be homemade.  Don’t forget that these can be monopoles or dipoles; and the monopoles are very widely sold as multiband or trapped verticals.  Pro tip: if you see a vertical sold as not needing radials, it might well be a vertically mounted dipole that is actually fed well above ground.  A good example is antennas sold by a Florida company, GAP Antenna Products, like their Titan DX antennas.  There are many multiband verticals on the market.  

The major difference between a fan dipole, or different, full-sized antennas and trapped antennas is traps will have some loss associated with them while full-length wires don’t have that loss.

Finally, you’ll note there’s a type of multiband antenna I haven’t mentioned at all, the log periodic dipole array, also called LPDA or just LP.  I’ve talked about these antennas in the past. (Here for example)  They have their place, but are pretty advanced as a home project

A Cushcraft (brand) multiband vertical called the R8 – for the number of bands it covers. 

Something that I need to mention is that with both the trapped or fan dipoles, it might not be possible to get every band in the HF spectrum.  Before the bands that we were given in World Administrative Radio Conferences, the amateur bands were separate enough that wires didn't interact.  Now, with new bands like the slivers called 60 and 30meters, and the slightly wider slices at 17 and 12m, it's harder to deal with the interactions.  This guy describes how he addressed it:  two antennas: one trapped dipole for the 10/15/20/30/40/80 bands and another antenna for 17 and 12m.  Not everyone is interested in operating on all bands though, and antennas for the older, larger ham bands might be all some people care about.



14 comments:

  1. Those are sort of approximate fan dipoles. A fan dipole is based on a self-similar antenna, often a skeletonized version of a biconical dipole. Here is an example of a commercial fan dipole (opens a pdf): https://www.antenna.be/tci-613.pdf
    Note the VSWR spec is 2.5:1 max, 2.0:1 typical. The max is typically found at the low frequency (where the antenna is getting small). The high end isn't a problem if you sweat the small stuff. The big version covers a 15:1 frequency range. A 3 to 30 MHz version is only 10:1 and covers all of HF. The feed is in the middle, note the pole with a box on it. It's a balun with one of E.M.T.Jone's wonderful balun designs on the top. I can't quite recall but I think it's a 200 ohm balun, available for up to 20 kW. 4NEC2 is the antenna nerd's friend....

    ReplyDelete
  2. Hey, SiG. You know the reason for the terms in the equation; it's using the physical entities, rather than their measurement units. Thus, V=AΩ, but we write E=IR, though some mix that together as V=IR. But I do like pie!

    I've been playing with antennas off/on for years. Not that I'm Maxwell reincarnated - I like to think I've learned a lot, but there's always more. I did build a Zepp (not a Zepp variant) and the feedpoint* impedance at the end of the transmission line is 14Ω. Fun stuff. One of my "light bulb" moments was when I realized that a J-pole is a Zepp.

    Biggest challenge for new hams is contending with all the "old knowledge" floating around, that keeps getting repeated, and is misleading and/or wrong. My personal pet-peeve is the treatment of resonance as some sort of magical talisman of antenna construction

    I've only recently begun to really appreciate the utility of Smith charts, thanks to some great videos from W2AEW. It took a few tries to get it to sink in, but the Smith chart, and the physics behind it, got me to better understand what was going on in my latest build - the "end-fed" dipole from AD5X in the May QST.

    * Yes, I know, the "feedpoint" should refer to the electrical feedpoint of the radiating portion, not where it is you connect your feedline to the antenna assembly.

    ReplyDelete
    Replies
    1. There have been so many times I've wanted to go into Smith Charts. So much of antennas, impedance matching, and more is easier to see if you grasp Smith charts that I really need to do that. I keep avoiding it because I'm assuming everyone who reads this is going to need a very deep introduction and that's a week worth's of writing and reading.

      Have you started using SimSmith by AE6TY? It has gone beyond just being a Smith Chart calculator to doing network analysis and now he's embedding NEC into it. Very useful software for free.
      http://www.ae6ty.com/Smith_Charts.html
      There's a ton of videos on learning the software from W0QE
      https://www.youtube.com/channel/UCKSyLSu4fm_1RHoO3Jvk4YQ

      Delete
    2. I think the basics of the Smith chart aren't too difficult to explain. Again, I think it was W2AEW who did a great job showing how an X/Y graph gets curved into the chart. And once I saw him demonstrate calculating a matching network by following arcs up/down from/to the center line, the configuration of an L network suddenly made much more sense. I suppose the approachability of it varies a lot, depending on what level of foundation you already have. When I first got into the hobby, I certainly would've glazed over.

      I did look at Sim Smith. Don't recall, ATM, whether it runs on Linux, but since it's Java, I should be able to run it, though I haven't set up a JRE here. That's a drawback for me, and with NEC as well. There are NEC programs for Linux, but nowhere as well done and easy to use as 4NEC2 for Windows. For now, I'm using a NanoVNA app on my phone. Saving the Touchstone file and using a web app to plot it helps a lot.

      I'd be happy using printed Smith charts, if my eyeballs could read that tiny little type.

      Delete
    3. Huzzah! SimSmith runs fine, at least at first look. I guess I'll need to RTFM. Then try to remember which Touchstone file is that latest incarnation of the most recent project. IIRC, the first thing I want to do is rotate the plot for adding 1/8 wavelength of feedline at 14.175. It should still be inductive.

      Delete
    4. Once you get used to the interface adding a transmission line is easy in SimSmith. There's an icon for a transmission line and the program includes a bunch of transmission lines by brand and model number. You specify length in options for the line as either physical or electrical. Very quick to see what it does.

      Delete
    5. jed, I find that 4NEC2 runs great with WINE on Linux. I have an occasional hiccup but no great problems; and the results are the same as on Win 7. My ham shack system runs Ubuntu 20.04.x and the latest release (32 bit -- there are reasons) of WINE from winehq.org (or at least a recent release -- later than what Ubuntu puts out).

      Delete
    6. After watching a few of W0QE's videos, I played with it some, and found that once I had the L/C network figured out, the added transmission line wasn't needed. My reason for it was to get total line length ~ 1/2 wavelength, which translates to better values if you want to measure the antenna itself - effectively cancelling out the feedline, except for line loss. What I did was load up my Touchstone file from the NanoVNA into the file field of the load. Didn't know if that'd work, but it plotted right there on the chart in the same place as the one in the VNA. Now I know what values I need for an L/C match to put all of 20M inside the 1.5 SWR circle. Next up, I guess is making my own 690nF inductor.

      I've tried Wine a few times, and while it will run things, the way it handles font mapping and rendering, at least on my system, is atrocious. I've had some issues with keystroke handling too. YMMV, I guess.

      Antenna in question is the "end-fed" 20M dipole made from RG58 in the May QST. That one is a head-scratcher, until you understand inside vs. outside currents on coax shield.

      Delete
    7. The first time I used a file in SimSmith it was written by my AIM4170 antenna analyzer, and that was in CSV notation. If you had to, you could take something that allows you to read R +jX but doesn't write files, write them down, and type up a CSV file with it. Although with the price of NanoVNAs, it's almost a no-brainer to get one even if you use it rarely.

      690 nH at 14 MHz? Sounds like a job for a -6 toroid material, although with transmit powers, you might prefer something air wound with big wires. Whatever number of turns you calculate and whichever way you make it, measure its value with the NanoVNA.

      Don't know if I'm preaching to the choir here, but those A sub L values in the toroid data sheets have a tolerance associated with them. I never wound a coil to the number of turns it said to use and had it come out on value.

      Delete
    8. Another ham has offered my choice of toroids from his pile, plus the loan of an LCR meter. But using the NanoVNA might be fun. Or, I could use the MFJ-259C. I tend to forget that it can measure capacitors and coils. Seems best to measure at the target frequency as well, from what I've read.

      But I was thinking more along the lines of an air-core, probably 16ga, because that's what I have, other than some really small gauge stuff.

      Delete
  3. The SRI information is very interesting. Though I have a few acres, it is heavily wooded and am limited to a 40M dipole without a lot of tree trimming. I have modeled a 40/20/15/10M fan dipole of three dipoles of 12 gauge wire in 4NEC2. I like 4NEC2 because it allows embedded equations and optimization while running the program. Back to the fan dipole. I modeled it with 6" offset between the individual dipoles at 1' out and a common feed point as in the diagram above. It was not very pretty impedance/SWR wise. I then move the spacing to 12" and it cleaned up a lot. But with the 10M dipole on the bottom there was interaction with the 20M dipole. I moved the 10M to the top and things became better. I did do optimization during this modeling process.

    I added a bunch of equations to the model be able to make an Inverted-V out of the antenna with variable angles and the ability to optimize. With the initial values from the previous version, things got wonky. 40M and 20M were not too far out, but 10M was off the chart with about a 5:1 SWR at best. I optimized for each band starting with 40. The 10M dipole shortened about 8 inches from the previous version to get a reasonable match in band but with only abut 300KHz under 3:1 SWR. I don't know what to build.

    Another thing, trap dipoles get such limited bandwidth at the low end due to the multiple reactances introduced by the traps. When you get down to the 75/80M band on an 80/20/15/10M trap dipole, there is generally less than 100KHz that has less than a 3:1SWR around the tuned frequency.

    ReplyDelete
    Replies
    1. Another thing, trap dipoles get such limited bandwidth at the low end due to the multiple reactances introduced by the traps. When you get down to the 75/80M band on an 80/20/15/10M trap dipole, there is generally less than 100KHz that has less than a 3:1SWR around the tuned frequency.

      How long are the sides? A good rule of thumb to remember is that it's tough to get much BW with a side that's under 1/8 wave long. 1/8 wave on 80 is 1/4 on 40, so about 31 feet at mid band (3.75 MHz). If the overall length per side is much less than that, it might be worth taking a look at.

      Linear loading - coils in series in the dipole wires to shorten them - tends to work, so something else may be at play. I've never spent time trying to model that, though.

      The other side of that is I've bought two different verticals, that worked on 80 and neither was over 100 kHz wide. Both were under 1/8 wave tall.


      Delete
    2. I didn't even think about the overall length factoring into the bandwidth on a trap antenna. I was basing my assumption on what I had seen of trap verticals. I did a little look around the web on trap dipoles. A number are longer than the on one side than the verticals you mainly see. W8IJ though points out that coaxial traps are not really great; wire and capacitor traps are much better.

      Delete
  4. Linear loading is the use of transmission lines as reactances, not coils. A section of transmission line less than a quarter wave, shorted at the end, as an inductor is very common.

    I found the learning curve for SimSmith to be steep. As an intro to Smith Charts, I like JJ Smith: http://tonnesoftware.com/jjsmith.html. It's simpler.

    I learned Smith charts from the ARRL antenna book, circa 1975. I even figured out how to do shunt components by inverting through the center. That trick lead to my first real job in antenna engineering a few years later!

    ReplyDelete