Thursday, June 23, 2016

Switching Matrix Test Bed

Many of the antennas I've described on this blog include a simple or an elaborate switching systems. These are useful for direction switching in low band multi-element antennas, selection of receive antennas and for stacked yagis (e.g. up, bottom, both). The more elaborate switching systems consist of many relays whose operation must be orchestrated so that every array element is correctly configured and the antenna behaviour can be switched with low-voltage cable from the operating position that may be 100 meters or more distance from the antenna.

Switching matrices for antennas are straight-forward and use common components in a repetitive pattern. Even so it is worthwhile to pay attention to details. With the low-band directional receive and transmit antennas I aspire to build in my next station this is a topic that will become a priority. Some experimentation with these devices is time well spent.

The transmission line cannot be easily used for this purpose since there may be many configuration states and there is only one DC path (assumes a bias-T to allow coax to carry both RF and DC). Instead we use a multi-conductor cable with one common wire and other wires assigned to each configuration state. For example, in the 3-element vertical yagi for 80 meters I earlier described the cable would require 7 wires: 1 common; 4 for each antenna direction (unpowered it is in omni-directional mode); 1 to select CW or SSB band segments; and 1 to select 160 meters.

Since a picture is reputed to be worth 1,000 words I'll reproduce the schematic of a switching matrix taken from ON4UN's Low-band DXing book. I chose this one for its clarity and no other reason. Also, I was reminded of the book when I had a short QSO with John on 20 CW while writing this article.

In this matrix the common line is negative; it can be the positive line if you prefer and you do not share a common return path with the coax outer conductor. One of the horizontal wires is energized by the operator activating a switch. Each relay coil has a vertical conductor to complete the matrix. A switching diode is placed at the junction of each horizontal and vertical line where for that operator selection the relay is to be energized. Depending on the application the relays and what they switch can be quite different, however the circuit topology remains the same.

I think any ham looking at this schematic would quickly understand it despite its size. The topology is straight-forward and the circuit elements are basic. Perhaps there's nothing to learn here, just a standardized approach for those planning switching arrays for receiving antennas or even wire yagis. There are other topologies possible, such as those requiring fewer wires but requiring binary coders and decoders. Since a long run of small gauge wire is inexpensive it is rarely worth the added complexity.

Despite the simplicity of the approach there are benefits to having a test bed for evaluating components and circuitry before diving into building switching matrices and their attendant interfaces and supporting hardware (and software). To this end I dropped by a local electronic parts retailer and filled a bag with some odds and ends to build a combination test bed/prototyping tool. It was very inexpensive. Here it is:

As you can see it doesn't need to be pretty! The components of the test bed include a punch-down prototyping board and a bundle of switching diodes. This board is large enough to prototype a switching system with 5 or 6 relays rated for 10 A. The 3 A PCB-mount relay looks tiny compared to the board. It was purchased only for the purpose of testing the test bed.

Wires, power supply and multi-meter were already in my workshop. I grabbed wires from my junk box that were already suitable for the task rather than make new ones. That contributes to making it look messy. Colour-coded wires cut to size are recommended when building a prototype.

The multi-meter in the photo is measuring DC voltage across the relay coil. At other times it was in resistance mode to monitor contact open and closure.

Now that the test bed is built let's review the ways in which it can be used.

Component testing

There was a time when semiconductors, relays and other components that were available retail had a DOA rate of several percent. Even though this has improved there is some comfort to be had by testing relays and diodes before soldering them to the PCB. Some use sockets for relays, but that can be expensive and introduces another source of future trouble: contact corrosion.

Not all relays have their terminals labelled. The one I purchased doesn't. The test bed can be used to identify the terminals. You can see that I drew a diagram to record what I found. Make sure that all the relay contacts -- NO for normally open and NC for normally closed -- show the expected zero or infinite resistance in each state.

It can even be useful to test diodes since the cathode marking may be unclear unless you are familiar with the component and its package style. I mounted two diodes in opposite directions to confirm the markings are as expected.

Circuit topology

The prototyping board has a similar layout to off-the-shelf pre-drilled PCBs. There are just fewer bus lines, thus requiring more jumpers. After the full prototype is built take a picture of it and transfer the components to the PCB, retaining the same positions. Solder and test in stages and you're done.

Keep in mind that the orientation of the diodes (direction the cathode faces) depends on whether the circuit uses the negative or positive DC bus as the common point for all the relay coils.

Component ratings

Relays and diodes are simple components yet they do have ratings you should be aware of. These can be tested.

First, you should check two voltages of relay operation: turn-on and turn-off. The first is the minimum voltage to positive relay closure. The other is the voltage below which an energized relay full opens. These can be read from the spec sheet if it is available (typical values are 70% and 20%, respectively). But that may be impossible for old stock or parts pulled out of a junk box. The more forgiving the relay operating voltages the more flexibility you have in choosing long runs of small gauge wire. By turning off the power supply and watching the voltage drop I estimated at most 3 volts when the relay opened (< 25%). I didn't measure the turn on voltage.

Second, you need to account for the diodes. Notice in the picture the voltage reading is 13.05 VDC even though the power supply is 13.8 VDC. A silicon junction has a forward voltage drop of 0.7 volts (±0.2 depending on current and temperature). The multi-meter probe is connected to the relay side of the switching diode so that the voltage drop can be seen.

When multiple relays are energized by one diode the diode current and power ratings can become important. The 1N4148 diodes I am using have a current rating of 200 to 300 mA. The coil of the test relay has a DC resistance of  700 Ω (measured as 680 Ω). For a supply voltage of 12 VDC (e.g. 13.8 minus wire loss), leaving 11.3 VDC across the coil, the current through the circuit is 18 mA, with the diode dissipating 13 mW due to its 0.7 volt drop.

This isn't much. However this is a small relay. For a typical 10 A rated SPDT relay for this RF application (e.g. Omron G2RL: 12 VDC, coil resistance 360 Ω) switching systems the current is 31 mA and the diode dissipates 22 mW. Multiply these amounts by the number of relays simultaneously energized by one diode in the matrix.

Go beyond a diode load of 6 relays and you might run into trouble, or even with fewer relays on a hot day when the temperature soars inside the enclosure. A higher capacity switching diode may be desirable in those cases. The 3-element vertical yagi for 80 meters comes close because the switching matrix powers up to 2 relays (float and reflector/director) at each of the 4 parasitic elements.

Summertime notes

I now enter the summer season. Expect the pace of articles on the blog to slow during the next 2 or 3 months. It's not just summer but also that some changes are afoot, about which I expect to say more later as my plans firm up.

I also have a trip back home to VE4 coming up very soon. I may come back with a few pictures related to amateur radio, something I neglected to do on my previous trip west.

For those operating in ARRL Field Day this weekend, have fun out there. From past experiences I've decided to avoid operating out of doors since it doesn't agree with me. I have many stories of past FD misadventures that only seemed funny to recount well after the fact. I enjoy outdoor adventures and amateur radio, just not at the same time!

Wednesday, June 15, 2016

Informal Station Comparison on 6 Meters

With sporadic-E (Es) season underway I decided to be active even though, unlike last year, I chose to not bother with a 6 meter antenna. In fact I've made no changes to my station this year, which is a subject for another time. For the present I am selecting the antenna that seems to work best on 6 meters, on the basis of receiving signal level, which is my Hy-gain Explorer 14 tri-band yagi up 15 meters.

The antenna works surprisingly well once the mismatch has been tamed by the transceiver's internal tuner. In the past several days I added 7 DXCC countries (now 18 on this band) and made close to 100 QSOs in the ARRL June VHF contest while operating part time.

But how to tell how well it works? Perhaps the best way is to compare performance with another station while quantifying as far as possible the differences. The opportunity arose when I recently added 6 meters to a friend's K4KIO Hex Beam, a kit he omitted when the antenna was bought a few years ago.

I only had the idea when I found that we were calling the same DX stations. Although we are some distance apart we can hear each other on ground wave. We are enjoying the friendly competition. Since we are both experienced DXers our techniques are comparable. This matters since a timid, overly aggressive or inexperienced operator will often lose out when the opening is marginal and the competition fierce.

I'll step through the elements of the station comparison then discuss how we fared against each other.

Height and terrain

While I was on his tower I noted that his antenna is at the very same height as mine -- 15 meters -- with a similar flat topography in the direction of Europe. There are more houses around me while he has more trees in his semi-rural location. A more exact comparison would be extremely difficult. Let's call this one a draw.

Antenna gain

The hex beam is a two-element yagi on all bands from 20 through 6 meters, with separate wire elements for each band. The elements, and especially the driven element, have numerous bends. The bends reduce gain a small amount below the 7 dbi that theoretically possible for a 2-element yagi. I would estimate the gain on 6 meters to be 6 dbi. The Moxon-style coupling between elements affects F/B and match but not gain.

Much to my surprise the Explorer 14 has a F/B on 6 meters. I expected it to behave more like a long dipole. While it does exhibit the additional lobes (6 total) of a long dipole there is about a 2 S-unit different in ground wave received signal strength between the forward and backward (broadside) directions. I measured this with the VE3WCC beacon on 50.009 MHz in FN15. I receive the beacon at S5 strength when pointing its direction.

With no way to measure the actual gain I can only go on what I can measure and theory. For theory I followed the principle of subtractive arrays that when signal is removed in one direction it appears as gain in another direction. However it takes a lot of F/B to add noticable gain with the yagi elements loosely coupled on 6 meters due to element separation. My guess is 3 dbi gain, or about 1 db better than a simple dipole.

Transmission line loss

My transmission line length is about half his: 30 meters vs. ~60 meters. Mine is a single length of RG-213 in very good condition; his is approximately half "well-aged" RG-213 and almost new LMR400. Loss can be accurately measured though not easily or conveniently enough for this casual comparison. Instead I plugged some numbers into TLW to estimate the transmission line loss.

I assumed that the RG-213 in my station has a loss comparable to new cable. In his I mentally balanced the higher loss of his weathered RG-213 and the better quality LMR400 and judged it to be comparable to a single run of new RG-213. The input side SWR at his station is 1.5 and 2.7 at mine.

TLW does the rest once I find a combination of resistance and reactance to give the measured SWR. It is sufficient to set the reactance to zero and only adjust the resistance in this exercise since both components rotate around the Smith chart many times along the coax which is long compared to wavelength. I could have measured these values at my station easily enough but there was no point.

Tuner loss

I use a tuner; he doesn't. Compact tuners for such a wide frequency range -- 1.8 to 54 MHz -- can be quite lossy at the extremes of the range. Loss further increases with the ratio of the impedance mismatch. I'll be optimistic and guess -1 db of tuner loss for the input side SWR of 2.7, which equates to 4.8 at the load due to transmission line loss.

It is possible to achieve greater accuracy by comparing power output to my Bird dummy load. I have done so in a casual fashion, enough to tell that the loss is probably not as great as I estimated. I am being conservative in my estimate to compensate for my avoidance of rigourous testing.

Adding it all up

With all the numbers in hand I proceeded to plug them into a spreadsheet and do the sums.

The difference is ~3 db. Although not large it is significant. Who wouldn't want 3 db of performance, on any band? Now we come to the on-the-air comparison. Does his 3 db beat me out?

To my surprise the answer is a firm "no". A few DX stations would respond with "VE3?" when we called simultaneously, yet it is a toss-up for who got through first. Repeated several times with different DX stations, and the vagaries of different fading patterns at our separated locations, we seem to do equally well. Perhaps equally poorly is more correct.

In a couple of cases neither of us made the QSO, never getting beyond copy of our commonprefix at the other end. Alternating our calls rather than calling atop one another made no appreciable difference.


One has to question whether the comparison is valid or whether 3 db is less of an improvement than we might wish to believe. To that I have no good answer. What it ought to make you wonder is how far you will or should go and how much you will spend to gain a decibel or three. Some will go quite far. Others refuse to go there at all.

There is no magic balance point that represents an optimum of performance or finances. It's a personal choice. Then there are the exogenous factors of terrain and propagation that are quite a lot more difficult to deal with.

A comparison like the one described serves only to demonstrate how elusive performance gains can be in practice.

Wednesday, June 8, 2016

Adjust a Yagi by Pointing It Up

The bigger the antenna the more difficult it is to tune. When done near the ground you are at risk of seeing its performance be worse when raised to full height due to the change in ground interaction. When done in the air the tuning process is difficult and potentially dangerous. Modelling can help to compensate for the height differential, but it is no panacea since software ground is rarely similar enough to real ground. Also, there are other environmental influences at low heights.

For yagis one technique many hams have had success with is to point the yagi straight up from a low base height. The idea is to get the antenna (reflector element) just high enough off the ground that its performance is comparable to that in free space (and therefore also at its intended height), and laterally offset far enough from the tower to avoid mutual coupling that will skew its behaviour. But close enough that the feed point is within easy reach. I've used this technique several times. It can work very well indeed.

Hy-gain 155BA prepared for tuning at
VE3CRG's station circa 1985
An example of the technique is shown in the adjacent photo, which I scanned from an old 35 mm print. The yagi isn't quite in tuning position for this photo op, with one side of the antenna too close to the tower. The rope attached to the reflector was used to stabilize and orient the yagi for tuning.

A casual internet search however shows an awful lot of unsubstantiated opinion and misinformation mixed with the good advice. When you do it right the technique works, so that's what we'll explore in this article.

Yagis are not dipoles -- even though they are comprised of dipole elements. Yagi behaviour does not go through the wild swings as height is varied, unlike what we see with single element antennas. Yagis are more immune to ground effects, becoming stable at surprisingly low heights (in terms of wavelength). Why that is true is interesting and worth exploration.

What the technique can and cannot do

The only performance parameter that can be conveniently adjust by this technique is the match; that is, optimizing the SWR across the band or band segment of interest. Gain and F/B optimization require other techniques, such as accurate modelling, including stepped diameter correction (SDC), field strength test range or testing with a cooperative ham within ground wave range.

Apart from that, like most hams we will have to trust the manufacturers' published dimensions for each band segment and for gain and F/B performance. If you don't trust the manufacturer (this can be an issue) design your own yagi.

With today's accurate software models and antenna analyzers it is still difficult to test and tune large yagis at the tops of towers, even if only for purposing of achieving a good 50 Ω match. For yagis with other than 3 elements the feed point is often not even within reach from the top of the tower.

Statement of objectives

We want the best possible match to minimize transmission line loss and maximize flexibility in the shack by having antennas that work well with transmitters and amplifiers that require a low SWR. The vertical pointing method is useful for conveniently adjusting the SWR of large yagis.

Or is it? Does it work, and how well does it work? Computer modelling helps us to answer these questions. The less we have to adjust our antennas at height the less chance for an accident. The best tower safety practice is the one that doesn't require a trip up the tower at all.

Test antenna model: 3-element 20 meter yagi

Rather than explain what's going on and then verify with models or measurements I will start with the model to see what it can tell us. For this exercise I will use a 3-element 20 meter yagi on a 0.35λ boom, a model I've used before as a reference for designing and evaluating tri-band yagis. This antenna is similar in scale to a Hy-gain TH6DXX but without the performance impact due to trap loss and element shortening. The model includes a hairpin (beta) match to achieve a 50 Ω match.

The following chart is from the referenced article, reproduced here for convenience. In particular notice the variation in F/B with frequency: its importance will become clear later in this article.

First we need to model the reference antenna in free space as our basis of comparison. Next, the antenna is rotated upward and placed over real ground (EZNEC medium ground). Its height is then varied and its SWR and gain calculated and compared to the reference. For what ought to be obvious reasons the F/B cannot be compared (see the adjacent elevation pattern).
Sample elevation pattern of a vertically pointing yagi;
the pattern can be quite odd in this configuration

When the impedance across the band is comparable to the antenna in free space we can usually trust that the gain and F/B are also in accord with the free space model. So if we can achieve this with the yagi in its vertical test position it should do fine when raised to the top of the tower.

Proper test setup

The yagi must be carefully positioned on the tower for this tuning method to be reliable. First, by height I mean the height of the centre of the reflector (bottom) element. I am ignoring element droop since its effect is only noticable at very low heights and for light duty elements with significant droop. Coupling is greater for the element tips and these will be closer to ground than the rest of the element.

Yagi model pointing up
The model does not include the antenna boom, tower and guys. The boom and element clamps do affect element tuning but are not relevant to our model experiment since we are looking for a change in behaviour not the absolute performance data. I further assume that guy wires are non-conductive or well broken up so as to have near zero mutual impedance with the antenna, and that the tower is "invisible" when the boom is outboard at least 1 meter from the tower and the elements are orthogonal to a line between tower centre and antenna boom.

This can be accomplished with a length of pipe or angle iron projecting outward from the tower with a pulley at its extremity and secured with ropes so that the antenna doesn't rotate in the wind, as shown in the photograph earlier in this article. It does make a difference, as I have discovered in practice when a breeze comes up and the SWR starts climbing.

If the boom is especially long it can be difficult to avoid guy wires when lifting the yagi into position for tuning. If necessary remove those elements, lift the antenna and reattach the elements.

As with all antennas a common mode choke should be used to prevent feed line radiation from disturbing the performance. The coax should be dropped straight down, either off the bottom end of the boom or along the tower vertex nearest to the boom.

Performing the comparison

I start with the antenna at a height of 2 meters (centre of reflector element to ground). The antenna is then raised in increments and its SWR and gain recalculated. The SWR and gain are calculated every 50 kHz from 14.000 to 14.350 MHz. This is sufficient granularity to derive an insight into what is going on, as we'll see; we don't need to know the R and X values of the feed point impedance, although that can be interesting and important in other instances.

At just 1 meter height (0.05λ) the SWR is already barely distinguishable from the antenna's SWR in free space. Going higher changes little. Are you surprised? Let's do it again, this time with the antenna in its normal horizontal orientation, stepping through the same heights.

Now we see that low heights have a large impact. Only when at a height of 5 meters (0.25λ) does the SWR become close to that in free space. At 10 meters height (0.5λ) the SWR is nearly indistinguishable from what it is in free space.

This confirms the technique where many hams with big towers mount their large yagis for tuning on a short tower before placing them at their final height. But now we also know that the "yagi tuning tower" should be no less than 0.5λ, at least for this antenna. Keep this in mind should you find it easier at your station to tune yagis horizontally rather than in the vertical orientation we are focussing on.

Let's move on to the gain comparison with the antenna rotated back to its vertical orientation. The results may be less surprising now that you've seen how the SWR responded. Yet the behaviour of the gain is somewhat unusual.

The yagi's gain has a complex relationship with height; there is a cyclic component due to minor lobe ground interactions. Influences include: mutual impedance with ground; ground reflections; and, directions and magnitudes of rearward nulls and minor lobes. While not shown the patterns develops odd lobes with small changes in height. The gain variation is within about 1 db, which is quite modest. If it were bigger we'd have likely seen greater fanning out of the SWR curves.

What is clear is that frequency has a stronger influence on the gain differential. SWR variation shows a similar though less pronounced trend (weak but visible correlation). This is where we need to step back and review how yagis work. Therein lies the explanation.

Why it works: mutual impedance

A yagi is a subtractive array. Strong mutual coupling among close-spaced elements causes field cancellation in some directions and field reinforcement in other directions. The relationships can be quite complex, which is why yagi optimization was so difficult in the years before software tools such as NEC were available.

Field cancellation reduces the radiation resistance which in turn increases element currents. When those currents are approximately in phase there is gain in that direction, accentuated by the higher current. Recall that the current (and antenna aperture) determine field strength. In contrast, additive arrays -- ones where the elements are far enough apart to have low mutual impedance -- work on superposition alone since the near field effects are relatively small.

Yagi elements also interact with the ground. Ground is in essence a flat, non-resonant medium that is lossy. That loss and ground's dielectric properties in combination with the induced current cause the phase and magnitude of the re-radiated field to vary with angle and polarization. It's messy but (mostly) able to be modelled. NEC4 perhaps does it best. EZNEC does pretty well with NEC2 and its real ground ground models.

Off the back end of a yagi that coupling is reduced since the cancellation of fields leaves less to interact with the ground. As long as the magnitude of minor lobes in the back 180° is small relative to the forward lobe the effects can be quite small. We can say that the strong mutual coupling between the yagi elements dominates the mutual coupling between the yagi and ground.

The gain differential with height for the yagi pointing upward correlates with the F/B. That's why I suggested keeping the plot of F/B in mind for this discussion. Notice that the free space F/B is highest at 14 MHz and falls off as the frequency increases. There is enough rearward field at the top of the band to more strongly interact with ground and thus influence the gain. Happily the effect is modest so SWR tuning is little effected when the reflector is quite close to the ground.

Yagis with poor F/B fare less well. This is especially true of 2-element yagis which have poor F/B (except for the Moxon with its critical coupling). These antennas should be raised higher, whether oriented vertically or horizontally, for the tuning to be reliable. Optimized yagis with 4 or more elements typically have excellent F/B across the band and respond especially well to the vertical tuning technique.

Myth and lore

When I first heard about vertical tuning of HF yagis many many years ago I had no understanding of why it worked. All I knew was that reputable names with super-stations used the technique and recommended it. I tried it and it worked. So I kept using it.

A lot of ham radio is like that: we do things that we have heard about but do not understand. Unfortunately this is also a good way to propagate nonsense, of which there is a lot out there.

By gaining a deeper understanding we learn to separate the wheat from the chaff, rejecting the "myth and lore" and adopting that which is science and evidence based. My hope is that this article contributes to the latter even though it only touches on the subject with a few examples. You can run your own models for the antennas you have or are designing to confirm how best to tune them.

Wednesday, June 1, 2016

7¼ Radials

Last week a few labourers were working in my yard with shovels. With barely visible radials of my 80 meter tower vertical lurking in the grass and weeds the inevitable happened. I was lucky that only one radial was severed. I watched from some distance away while talking with the supervisor and several others with no opportunity to politely intervene.

I knew it had happened when the young man with the shovel looked down with a puzzled look on his face. Perhaps I should have rolled up the radials after all. No matter, the deed was done and I was left to deal with the consequences.

But were there consequences? In fact I was not particularly bothered by the radial damage even though CQ WPX CW was days away. Neither was I concerned about replacement cost since wire is cheap and the radials are only 8 meters long. I went ahead and used the antenna to make a dozen 80 meter contacts in the contest without a qualm. If its performance changed I couldn't tell.

It is worth taking a few moments to delve a little into what is going on here, and to thus explain my lack of concern. Radials are often misunderstood by many hams. Let's start with the most easily measured data: impedance and SWR.

The model view of the disfigured antenna is at right. Wire 24 was chosen to represent the radial damage since it is parallel to an axis, making it easy to change its length, and orthogonal to the 40 meter inverted vee, just like the damaged radial. Modelled SWR curves show the effects.

Resonance drifted upward by about 25 kHz (~0.5%). I see greater variation than this when it is raining or snowing, and the ground is thawed or frozen. That is, the change is negligible. The increased ground loss is similarly small, being well under -0.1 db in the model.

With my antenna analyzer I swept the SWR from 3.5 to 3.8 MHz and confirmed that there is no discernible change. I have no means to measure ground loss, other than noting that performance seemed no worse than before; the antenna performance remains poor though servicable.

Returning to the model I calculated the radial currents. As expected the current in the short radial is so small as to be equivalent to the radial being absent. I could just as well have titled this article 7 Radials. Current in the two adjacent radials was reduced less than 10%. Current in the other 5 radials was equal and of the expected magnitude; that is, they are unaffected by the cut radial.

What can we take away from the radial-cutting incident and my analysis? The points to be made are modest but I believe important in better understanding radial behaviour.
  • With 8 short radials adding or subtracting 1 radial has a small but measurable effect on antenna resonance. That is, the radials are still acting in part as resonant counterpoises. The effect diminishes with increasing radial counts. With a large number of radials the radial system is more ground plane than counterpoise.
  • The loss of one radial increases ground loss only a small amount. Again, the effect declines with increasing radial count. One aspect I did not attempt to isolate was the degree to which the increased ground loss, which is in series with the radiation resistance, affected the frequency of minimum SWR versus frequency of X = 0 Ω. The two frequencies are typically not the same.
  • With 8 radials the loss of symmetry -- equal length and spacing -- due to one deleted radial is small, even with the short 0.1λ radials I am using. A similar loss with 4 radials is more dire.
Adding or subtracting one radial is no big deal for other than the most minimalist radial system, whether on the ground or above ground. It is not surprising that studies of radials and actual antenna construction typically discuss performance changes when the radial count is doubled or halved. Large changes in radial count are what make a measurable difference.

For example, if you want to improve the efficiency of your vertical double the radial count. Don't settle for lesser increments. Doing it this way also happens to be the most convenient since you are laying one radial centred between every two existing ones. Nothing needs to be moved. But you'll want to stock up on low-cost wire since doubling radials requires a lot of wire.

[Note: All the links in this article are to other articles in this blog. The referenced articles contain links to outside sources that you may find of interest.]