Wednesday, May 29, 2013

Choking the End-fed Antenna

In an earlier article I discussed the ramifications of common-mode currents on the coax feeding a nested delta loop antenna. I used EZNEC for that demonstration and also referenced other material. Since my guest room is free once more I decided to deal with a persistent problem I had when I reassembled my temporary station (which took all of 10 minutes).

Unbalanced antennas are very prone to currents on the outside of the coax. This can be used to good effect when making the coax part of the antenna, while at other times it is a pattern-altering, RFI/EMI-inducing nuisance.

Loading the aluminum eaves trough with a tuner is very much an unbalanced antenna system, and there are unwanted currents on the coax. This does not compromise the antenna pattern much, sadly deficient as it is. But since the KX3 is less than 3 meters of coax away from the tuner (antenna feed point) that current does wreak some havoc.

The reaction of the KX3 to the current is quite stark: it aggressively folds back the power. With only 10 to 12 watts available (8 on the highest bands) the power reduction can be a problem. I am ok with QRP but QRPp with this poor excuse for an antenna is more painful than I like. I could only go up to 5 watts on 15 meters and 3 watts on 10 and 6 meters.

I grabbed a 1 meter long RG-58 patch cord from my box of cables and added it in line (with a T-connector, which was the best I had) to add the slack I needed in the transmission line. I then made a coax choke from that slack, which worked out to a 6 turns coil with a diameter of 12 cm. I proceeded to tune and test the altered antenna system on all bands from 20 through 6 meters.

Even with this suboptimal coax choke the results were remarkable. The RFI incursion into the KX3 became negligible. When matched with the tuner, the rig now puts out full power on all bands. Not only that, the KX3 exhibits no audio artifacts on QSK and the SWR behaviour is smooth, not abrupt, when adjusting the tuner. Antenna current now appears to be staying where I want it, on the wire counterpoise and antenna proper.

The lesson here is that perfection is a laudable objective, but don't let that stop you from trying something of lesser quality. Experiments like this are cheap and can have a large benefit. Coax chokes are not ideal since they are reactive and therefore cannot cover all the HF bands. However they can work well. If one doesn't it's easy to add or subtract a few turns. There really is no excuse not to have a choke on your coax transmission lines.

In the interaction article I pointed to a comprehensive discussion on the choking of common-mode currents on transmission lines by K9YC. There are examples in there that are quite good (high resistance), but that require ferrite cores. I will now point you to some air core alternatives, with test data, made by G3TXQ.

Monday, May 27, 2013

Scoring a Second-hand Tower

In an earlier article I described my antenna siting options. Three of those options (and two that I plan to use) require a small tower: one bracketed to the south side of the house (B) and one guyed in the yard (C). The missing piece in my expansive junk box was the tower itself.

It seems that even buying a new tower is not the easiest of propositions. Finding local dealers and arranging transport are neither easy nor inexpensive, yet my requirements are modest. A tower of 30' or so suits my antenna objectives. Additional height would come from a mast to support the apex of wire antennas.

The industry has changed in the 20 years I've been out of the hobby. There were 2 major domestic manufacturers of residential-grade antenna towers: Delhi and Trylon. While Delhi is no longer in existence, its tower business now belongs to Wade Communications. Trylon is still Trylon. Common US tower brands such as Rohn are not often seen in Canada since towers do not travel well. By this I mean that transport is expensive (due to size and weight) and not always easy to arrange, and domestic dealers are few.

The Delhi towers are the DMX series that are ubiquitous in the backyards of Canadian hams and the lighter-duty Golden Nugget series that are typically used to support TV antennas. Both lines appear unchanged and are on offer from Wade. After a brief and fruitless search Wade pointed me to a local dealer. But then I went online to first look for used towers.

Finding used TV towers was becoming difficult around 1990 since the migration to cable was largely complete. Except in rural areas there were few used TV towers for the taking. The market has since reversed course due to the resurgence of OTA (over-the-air) TV viewing by so-called "cord cutters". Used Golden Nugget towers are showing up everywhere.

I now have a 30' Golden Nugget 18-gauge tower (actually about 28.5' when built) sitting on my deck, including a 10' length of 1.5"-OD mast. It was free for the taking, provided of course the purchaser took care of taking it down. So I dusted off (and tested!) my climbing gear and quickly relieved the homeowner of it. I left him the TV antenna and rotator so that he could sell it, and so come out ahead as well.

These towers are relatively easy to put up and take down if you have experience with the bigger, amateur-grade towers. I easily handled it on my own, with the homeowner as ground crew. I did make a sight on the drive across town with the tower dangling out the back of my sports car.

You do get some surprises when you see how some of these TV towers are installed. It's a bit scary what passes for a residential tower installation. I'm sure the reason the majority of these towers don't come crashing down is because the bottom 2/3 is sheltered from the wind by the house it's bracketed onto.

There are always going to be trade offs when you go the used route rather than buying new equipment. Towers are no different. A compromised tower is not one that should be repaired; it should be thrown out!

Rust is common. Galvanizing is not a miracle cure, it only delays the inevitable. Cheaper tower uses the cheaper galvanizing process of electroplating rather than the better hot dipping. This tower appears to fit into the former category. Apart from cosmetic rust there is more serious deterioration at the bottom of the bottom section and the base plate since the previous owner buried the bottom with the intention of making the installation more sturdy. That is unnecessary when properly mounted. Burial greatly reduces the life of any tower, especially those with tubular legs since water will pool inside the leg.

You can get an idea of the rust in the close-up view I've attached. Although the base plate looks quite bad it is in fact perfectly usable; most of the "scale" you can see is debris that can be cleaned off. The base metal is mostly intact though little of the zinc coating remains. The base plate is heavier gauge steel so that it can withstand ground contact for the life of the tower.

Of more concern is the bottom of the tower. Golden Nugget is 12"-face, 18-gauge steel (I weighed a 10'-section a 12.1 kg, or 26.5 lb) and is vulnerable to weakening due to rust penetration. The tower also comes in 16-gauge steel, but you don't usually get to choose when you buy used. I doubt that most OTA users are aware of the difference and simply steer toward the lower-price option.

On first inspection the tower bottom looks good enough for reuse. The rust scale will have to be removed and then painted over with a suitable product. Even so I am reluctant to use this tower at site B, bracketed to the house. It may be best suited at site C where it can support a fibreglass mast and wires for the low bands, in particular a delta loop for 40 meters.

The other thing I've collected recently is stackable 4'-sections of 1.75"-OD surplus fibreglass mast from Maple Leaf Communications. The price was right and it looks as if it can do the job I require of it (without actual specs there is some risk which I will mitigate as well as I can). As I pointed out in the article on the nested 20-15-10 delta loop antenna, the mast must be non-conductive and would have to vertically span 7 meters. I bought enough of the fibreglass for both this antenna and for the top part of the supporting mast of the 40 meters loop.

I'll have more later when I am ready to erect antennas. I still have a little more work to do to prepare the basement shack, work that is harder for me to get motivated to complete as the season grows warmer. For now I am QRT since I ran into the foreseeable problem of setting up in the guest bedroom: guests arrived.

Saturday, May 18, 2013

QRP Antenna Tuners

In my junk box I have a few antenna tuners, small and large. There is also a selection of components that can be cobbled together to make more should I be so inclined. One of those commercial products was put to work in my current station to coerce my eaves trough into an impedance that my KX3 will (mostly) tolerate.

The one I chose is a small L-network tuner by MFJ that is suitable to end-fed wires. Other network topologies would also work, but I chose this one because it is small enough to comfortably (and unobtrusively) sit on the narrow window sill just under the wire coming in from its attachment point to the eaves trough.

The thing is, aesthetics aside, was this the proper choice for my QRP station? Consider the two (very dusty) tuners in the picture below, both of which I dug up from my basement horde of ham paraphernalia. If you had to choose, which do you think is the best antenna tuner for QRP? (For the moment please ignore the fact that these are very different tuner circuits, one being an L-network and the other a general-purpose transmatch.)

Have you picked one? The thing is both are correct, under the right conditions. It all has to do with why one uses QRP and one's personal circumstances.

Apart from the obvious size difference there are others.
  • Price - Price tends to rise in proportion to size and features. For example, a little coil is much cheaper than a big coil, a built-in SWR bridge costs extra, and increased size and weight mean higher costs of transport and inventory for the retail supply chain. Buying used equipment can change the equation, as will the price sensitivity of the buyer.
  • Power - When properly designed bigger components means a higher maximum power and mismatch tolerance. While this is not relevant to QRP, the operator may have aspirations for QRO in the future and so may wish to plan ahead.
  • Performance - As a general rule, the greater the mismatch the greater the losses within the tuner. Even if you find that perfect 50+j0 match to satisfy your finicky transmitter it may be that 9 or more of your precious 10 watts are spent heating the shack (and the tuner). This is not an exaggeration. It often occurs with small tuners driving non-resonant antennas.
For the QRP operator who does it under the principle that "small is beautiful" the choice is clear: pick the smaller, cheaper alternative, even if it entails reduced performance. If that is not your style you will need to review your options.

My choice was not based on price or principle. I simplify opened boxes until I found some tuners and proceeded to pick one that met my immediate needs. That was one suited to an end-fed antenna (the eaves trough) and would be easy to get close to the feed point. There was just the one that fit on the window sill. That it is the "best" tuner configuration for the antenna was a bonus, not mandatory.

The price I pay is performance. In practice it is a high price on several bands, particularly on 160, 80, 10 and especially on 15 and 6 meters. This is unsurprising since tuners, and most especially the small ones, get terribly inefficient at their range extremes, in regards to both frequency and mismatch. If you happen to read QST you should pay attention to these measurements in their tuner reviews so that you'll see how true this is.

I did not bother to open either unit to take pictures of their insides. However I'm sure you can guess the difference in coil wire gauge and capacitor plate area. Unlike with a kilowatt, even if every one of my 10 watts is dissipated by the tuner there will be no meltdown. Nevertheless the losses are there just the same. If the tuner has a 3 db loss (500 watts out of a kilowatt) it will also be 3 db with QRP (5 out of 10 watts).

The losses are due to resistances in the coils and capacitors, and core heating of any toroidal transformers. Ohm's Law tell us what we need to know:

P = I²R

For variable coils the resistance increases as the length of wire in the active section of the coil increases and as the wire gauge decreases. For variable capacitors the resistance increases as the plate overlap decreases or the plates are smaller to begin with (less surface area). We typically use more of the coil as the frequency decreases (A moves to the left in the diagram) and less of the capacitor plate area as frequency increases. That helps explain where much of the losses occur on the lowest and highest bands, respectively.

Current compounds the losses. Inside a matching network -- whether in the transmitter, tuner or antenna -- the currents circulating through the components are often higher (sometimes much higher) than the current in a perfectly matched antenna system:

I² = P / 50

This is why the best (and most expensive) components have large surface areas, and those surfaces are silver-plated. For a fixed R the loss increases in a component by 6 db for a doubling of the current. The square of the current quickly rises as the mismatch between the transmitter and antenna system grows.

I would be better served by the larger antenna tuner in my present situation even though it can be tricky getting it to match an end-fed antenna. There is no QRP "aesthetic" standing in my way. It's just that it too big to place near the antenna and it takes longer to change bands since the roller inductor takes more time to adjust than a tapped inductor. Speed and annoyance matter.

For an outrageous example of how inefficient the little tuner can be I tried to use it on 6 meters. To be fair this is well outside its design range. I just had to make the attempt when I noticed a sporadic-E opening one evening. I heard many stations but not one of them heard my 3 watts -- the most power I could coax out of the KX3 into a poor match. When I bypassed the tuner the band noise and received signal strength rose at least 10 db. The SWR was a high though usable 2:1. This time I did log a few stations.

What the actual loss of the tuner was I don't know since I can't easily measure it. I expect that the results on the other problematic bands (mentioned above) are of lesser degree though still significant. It's a handicap to add to my other handicaps of QRP and poor antenna, though a necessary one.

I do enjoy a challenge but this is ridiculous. Should I require a tuner when I get some real antennas erected it will be a big one.

Sunday, May 12, 2013


Antennas interact with their environment. In one sense that is obvious since that is the essence of what an antenna is all about: electromagnetic coupling to free space. That is a "good" interaction. There are also "bad" interactions. By this I mean interactions that cause unwanted effects (EMI, coupling, etc.) that degrade antenna performance from the objectives set out in its design.

Many hams are often unaware of or outright ignore interactions. Once the first QSO is made with a new antenna all thoughts turn to operating and away from the antenna itself. This isn't surprising. While we like to promote our hobby as one promoting technology and technological expertise we know that we are in it for the operating. The technology, antennas included, are a part of the journey, not the destination.

As you might guess since I am writing this article is that there is good reason to stop for a moment and consider the antenna's interactions during its design and siting. It can prevent many problems and achieve better performance.

What I will do here is talk about some of those interactions in regard to the nested 20, 15 and 10 meters delta loops antenna I described recently. In the following model the antenna is mounted on a mast at the apex of my roof, which is site 'A' described in the site planning article written earlier. Although, as I said in that article, I rejected that location I am being lazy and using an interaction model I had already created for that site.

The antenna view at right includes a model of a coaxial transmission line, consisting of wires 10 through 12. It is approximately to scale but I made only a modest attempt to be precise about direction and length. My purpose was to get a sense of the interactions rather than to make the coax part of the formal model. You'll have to imagine an outline of my house to see that wire 10 angles down to the edge of the roof (tied to the eaves trough that is my no-antenna antenna). The same goes for the remaining wires, with 11 dropping down to the back edge of the lower roof, and 12 takes us down to near ground level, close to where it will enter the basement.

Two more things about the model: there is only one coax drawn but there will in fact be two, one for the 20 and 15 loops; and, the top of wire 10 ends between the two feed point (V1 and V2) and is connected to nothing. This is obviously simplified though sufficient for our purposes here. By not connecting the coax to the antenna I am modelling the presence of a high-resistance common-mode (current) choke on the coax. This ensures the coupling to the coax is by mutual conductance only, and not direct conductance.

Using a high-resistance choke is important to avoid turning the coax into part of the antenna, with a negative consequence on pattern and match, among other difficulties. Mutual inductance is bad enough, so don't make it worse. I won't get into chokes here, so instead I will refer you to a truly excellent document on the topic by K9YC (PDF). If you have May 2013 QST you'll notice this work referenced in the article on making a common mode choke.

Notice in the antenna view I am showing the plot of currents when fed at 21.1 MHz. This is the worst case scenario for induced currents on the model coax. The currents are much lower on 10 and 20 meters, with lesser impact on the antenna performance.

The currents, especially on wires 10 and 12, are enough to cause a significant change to the far field plots I presented in the antenna article, where I modelled none of the environment. The coupling occurs even though I kept wire 10 nearly perpendicular to the antenna plane. I did not model the aluminum eaves troughing or the house wiring in the 2nd-floor ceiling, all of which are in play when it comes to interactions.

Compare the far field plots shown here for 21.1 MHz to those in the earlier article. It is no surprise that the pattern is no longer symmetrical. The major changes are in the azimuth omnidirectionality and the increase in the amount of horizontally-polarized radiation. The SWR is not shown since it isn't much different, and is besides easily remedied during tuning when the elements are adjusted after installation on the mast.

Are these interactions bad? That depends on your objectives. The increase in azimuth asymmetry (~5 db rather than 3 db) will be noticable though probably not by much in actual use. The elevation pattern, while lopsided, is not too far off the ideal model.

An antenna like this is bound to have unwanted interactions of this sort since it is dimensionally large and close to the house, including all of its many embedded conductors. The difficulties are less for a yagi atop a tower where the coax (and tower) are orthogonal to the antenna plane. Less, but still there. Chokes will at the very least protect loss of depth in the front-to-back and front-to-side nulls.

These are not the only reasons to assess whether the interactions are acceptable. There are others that give me pause and that is why I am still thinking about whether I should build an antenna that is otherwise suitable for my purposes.

Coax currents are reciprocal as they are in any antenna: they act the same on receive as on transmit. With the antenna so close to the house (and neighbouring houses) there are interactions that could be more serious than on the far field pattern or the match.

First, there is the obvious matter of EMI. Using QRP as I am at present the risk is low. That will change should I increase power to 100 watts. With a low antenna where the main field intersects houses that is already a problem. Add in radiation from the coax and the probability of EMI increases. In the above model there are radiating surfaces all the way down to ground level, adjacent to inhabited areas. While a tower and yagi might appear more problematic to neighbours, the EMI risk is often greater from low antennas and all transmission lines.

On receive there is also risk of EMI to us. Our homes are filled with computers and computer-driven phones and other appliances, switching supplies in wall warts and light fixtures, and more. Their radiation will make its way into our sensitive HF receivers via the exterior of the coax. With an eaves trough for an antenna I currently have this problem in spades. In my house I have at least 3 Ethernet-based devices within a few meters of the eaves trough and there are more further away, including my next-door neighbours. If the coax radiates, even if its current is a tenth of what is on the antenna proper, the noise becomes a problem on at least parts of the HF bands. It is easier to install a choke one run of coax than to choke ten appliances.

Yet even with the common-mode choke at the antenna feed point there can still be radiation on the coax, as I've already demonstrated. If it is necessary to go further to solve problems we will have to install one or two more chokes along the length of coax, ensuring that the length of coax between chokes is non-resonant on all bands of interest. If you concurrently operate on more than one radio or band this will include radiation from antennas for those other bands since they will happily couple to coax runs of other antennas.

Many hams do not realize the true source of EMI problem, often blaming it on high power or just bad luck. Yet it can often be cured or at least reduced with more and better common-mode chokes on their runs of coax. Ignorance is not bliss. Just because one is unaware of the root cause of EMI, or poor antenna performance, it will not go away.

Interactions between antennas and all conductors in the vicinity is an unavoidable consequence of the laws of physics. We can ignore those interactions but they will not ignore us.

Sunday, May 5, 2013

Nested Delta Loops for 20, 15 and 10

I am finally getting around to this post. Because of my antenna siting criteria there was, and is, some doubt that I'll put up the antenna described in this article. Regardless of what happens I would like to document the design. Later I will come back to my evolving antenna options, and where this design stands. There are a few interesting ideas in here that may be of value to others in their own antenna designs.

As should be obvious from my previous writing I do have some love for delta loops. Although they are big and (to some) ugly they do have certain advantages:
  • Inexpensive - lots of performance per dollar
  • Multiband - on harmonics
  • Easily configurable for any linear polarization - vertical, horizontal or in between
  • Can be matched to typical transmission lines with a quarter-wave transformer - doing this does unfortunately remove the multiband capability
  • Omnidirectional when configured for vertical polarization
  • Like any full vertical (dipole or loop) it is a decent DX performer even at low heights
  • Requires only one support mast, although it should be non-conducting if vertically-polarized
  • Broadband - low SWR across the band when matched to 1.0 SWR at resonance
  • Loops are quieter in rain, snow and low humidity since there is no corona effect off the ends
There are of course disadvantages, including:
  • If visible to non-ham neighbours it is often perceived as an eyesore
  • There are ropes going everywhere, both to hold the shape of the loop and to guy it and its mast
  • Heavily dependent on ground quality when vertically-polarized (as for any vertical) for far-field performance
  • Can be fragile if not well designed and constructed
For this article let's put all this aside and just design the thing. No one has to use it after all, and that includes me.

From the adjacent EZNEC antenna view it is a somewhat unusual looking design. There are nested loops cut for 20, 15 and 10 meters.

You'll see that there are two feed points: one each on the 15 and 20 loops. They are fed with 70Ω λ/4 transformers (RG-6/59/11) to match the high loop impedance to 50Ω. There is no feedline to the 10 meters loop since it will be parasitically fed. I'll come to that in a moment.

If you're familiar with quad beams you'll likely know that they are fed at a common point, where the loops for each of the driven elements are brought together. This work since the impedance of a quad beam is typically very close to a 50Ω match. That doesn't work in the present case since this is not a beam and matching is required. I did try various techniques but none worked well. Since I am pretty much swimming in coax I decided to go with more than one transmission line.

I mentioned how 20 and 15 are matched, but 10 is special. The 20 loop resonates harmonically on 10 but not where we want it. It resonates well below the band. Also, that λ/4 transformer for 20 is now a λ/2 on 10. What this does is leave the impedance unchanged; it is equivalent to making a full circle on a Smith chart. That isn't what we want.

So what I did was, in effect, make this a 2-element parasitic array on 10 meters, tuning the inner loop such that the array resonates where I want it. This has several affects: the impedance drops to near 50Ω at resonance (which the 20 meters transformer leaves alone); the pattern gets more complex, but does not become a unidirectional beam; and, the bandwidth is reduced due to the increased Q.

As usual I cut my antennas for CW. I also have no interest in 10 meters above about 28.6 MHz. You can adjust the antenna to your operating preferences.

The 20 and 15 loops are fed 1/4λ from the apex (~25% of leg length from the bottom) to make the antennas vertically-polarized and omnidirectional (within 3 db). Although this is covered in an earlier article, the elevation pattern for 20 meters is posted here. Although ground loss is ~3.5 db the gain at 10° is an impressive 1.5 dbi. That isn't possible for a dipole at the same height over suburban ground.

The pattern for 15 is similar. Ground loss drops to 2.8 db and the 10° gain rises to 3.1 dbi. This is quite good.

The actual maximum gain for both 20 and 15 are in higher angle lobes. This makes the antenna usual for domestic contacts, which in my case would only matter during contests if at all. Unlike a low dipole there is a deep null at high angles. This is not too interesting on the high HF bands but it may help remove some domestic QRM while chasing DX.

For purposes of SWR and pattern the ground is modelled on the poor side of medium, typical of suburban areas. The antenna (bottom of the 20 loop) is set at 10 meters. This would put the apex at 16.3 meters. If I build it I will drop it a bit to keep it under the 15 meter height threshold that requires undesirable city (Ottawa) policy conformance.

The pattern on 10 meters is not ideal due to being both parasitic and having one element 2λ long. Of course there is additional gain in some directions, but also less in others. The antenna is fixed so I would just have to live with it as is. I am not too concerned since 10 will soon (2014) begin to decline in importance as the current solar cycle declines. It's a compromise. I could always fill the "gaps" with another antenna.

The surprising thing about the elevation pattern on 10 is that it is horizontal off the ends (antenna plane). That's the effect of driving it from the 20 loop. Even so it does well at low angles, equalling the gain on 15. The high angle radiation is less useful for DX but will come in handy for aurora and sporadic-E openings. Ground loss is a little over 1 db, which is negligible.

The antenna is not omnidirectional on 10. Unlike 20 and 15, at most elevation angles the gain is higher off the ends than broadside. The pattern shown here is for 10°, which is most pertinent for DX. The differential between the gain off the ends and broadside is larger at higher angles. Although this isn't perfectly aligned with my design objectives the compromise is acceptable.

Some construction details follow.

The loops do not have to be perfectly concentric but should not snuggle up to each other. I confirmed this to some extent through modelling, but it is a lot of work to be comprehensive and certain. Therefore it is better to err on the side of making the loops close to concentric. As shown here the separation between the 20 and 15 loops is 141 and 70 cm at the apex and bottom, respectively. Between the 15 and 10 loops the separation is 61 and 40 cm at the apex and bottom, respectively.

The antenna is cut for CW so the SWR climbs towards the high end of the bands, particularly on 10. The model wire is 12 AWG THHN copper. Increasing the gauge to 10 is a good idea for a 40 meters loop, but 12 should be ok here. The bottom of the outer loop in the model is wire, however I would build it using aluminum tubing of about 25 mm diameter (1-inch). I have this already so the cost of aluminum tubing is not relevant. The tube will lower the total 20 meters element length a small amount.

Element lengths are as follows, with 1/3 of the total in each of the 3 sides (equilateral triangle):
  • 20 meters: 21.78 m
  • 15 meters: 14.47 m
  • 10 meters: 10.98 m
Use of a bottom tube avoids the need to guy the antenna to support its shape. The inner loops can be tied down to the same tube. The 20 loop can be tuned by changing the attachment points of the vertical legs on the bottom tube. Remember to tune the 20 meters loop first before tuning it on 10. Apart from this you'll find that the antennas don't interact much (confirmed in the model) so tuning interaction shouldn't be a problem.

Although the interior angles don't have to be exactly 60°, staying close (equal side lengths) will keep the performance in line with the design. Just ensure there is tension on the wires so they don't sag, even under temperature changes (metal shrinks or expands in accord with the temperature) and ice and snow loads.

The apex of each loop is tied to the mast, which must be non-conducting. Fibreglass is a good choice since the antenna stands fairly tall and must be tough to survive the weather, and it can be guyed. When mounted at some height a metal mast or tower can be used under the fibreglass mast provided it doesn't get close to the antenna bottom centre, since it is a high-impedance point. Keep the separation at least 1/2 meter.