Stacking for 50MHz
Zaba, OH1ZAA/NN0Y
Issue 65, May 2000

Reprinted from Cheesebits, February 1996

With additional contributions of WA1YHO, WD4ECK, NOØY, NU6S, WB2KMY and WD8ISK after the KP4XS, W9IP, W7LZP, PA3BFM, WA9JML and OH1ZAA opinions, I think that we have a delightful collection of practical experience and backgrounds, that will be useful for future stacked antenna systems, and not just for 50MHz. It is easy (but lots of hard work) to expand the whole thing toward a book, but this time I would rather like to conclude with some remarks that hopefully give still more precision to the thoughts. We found that there was no controversy as such, but the subject has been ‘illuminated’ from a number of personal viewpoints with an emphasis on particular features.

If you use an antenna program like Yagi Optimizer with the horizontal (E-plane) plot you really don’t know what happens to the top/bottom lobes, since the 3 D-pattern is always at minimum toward the end of the dipoles. Therefore a smooth E-plane plot can show good F/B ratio, but it can hide big lobes in the vertical plane. Therefore I recommend to observe the H plane, and always optimise with that option activated. I only have the old 1988 version of YO but I have seen the newer one, and it is really a marvellous tool. I have also the Antenna Optimizer 6.35 (1994). In clear environments it may be worthwhile for best F/B ratio to slightly adjust antenna dimensions when running a stack instead of a single yagi. On a cluttered tower it’s hard to predict all the de-tuning unless it is completely in the model.

If you know the H-plot, you automatically know the E-plot, since it is the H-plot multiplied with the dipole pattern. Small H-plane sidelobes give still smaller sidelobes in any other 3-D direction. In total integrated noise power this means that the bulk of the galactic noise is trying to creep in from over your head, unless your preventive measures are rigid.

Look at the plot of two in-phase dipoles (in almost any antenna book) with about 0.6 wavelength spacing (like the two driven elements of stacked yagis). The pattern is compressed enormously to the sides, and gain increases to 4.9 dB over a single dipole. However, the 0 dB F/B ratio does not change.

With stacked yagis at 0.6 wavelength a poor F/B ratio is not forgiven, but anything to the top/bottom sides (and skew) is tremendously attenuated. H-plane sidelobes in wide cones around the Z-axis are virtually eliminated. However, the more complex mutual coupling due to parasitic elements limits stacking gain to lower values than with dipoles, but with very short yagis the 3 dB mark can even be slightly exceeded.

There are two simultaneous benefits: Overall parasitic noise power is greatly reduced, and nearly 3 dB additional gain is produced with the stack. The net S/N ratio gain with uniform noise from all directions is at least 5 dB, but with the cosmic hot spots much bigger contrasts occur. Remember also that the earth turns with a speed of one degree in every four minutes, so certain azimuths are not ‘jammed’ all the time. The sharper your main lobe, the shorter the suffering. We need someone with a good ‘electronic noise map’ (a sky noise data map on a PC) to calculate the advantage for a specific stack compared to a single yagi.

Stacking at 0.6 wavelength always assures the noise power cancellation, but if booms get long (say longer than 8 m/ 27 ft), the stack starts to look as a single yagi gain-wise. The stacking gain may fall well under 2 dB (which is still a lot), but if the single yagi pattern is relatively clean it is better to go to for 1.2 wave stacking for moderate length yagis and 1.4 wavelength stacking for the long ones. This is just due to the mutual impedance: a 0.8 wave separation is always poor for gain and sidelobes (see the consistent gain dips in W2PV’s yagi book).

With four antennas it is superior to use non-uniform stacking: first stack 0.6 wave with each pair, and then allow the two bays to cancel the residual sidelobes of the separate bays (then there will be hardly any noise power from unwanted directions). For maximum gain shortcuts can be made, but due to the S/N issue on 50 MHz, I would never neglect the 0.6 wave basic cell as the building block for optimal listening. This is also why VE7BQH’s collinears on 144 MHz work so well. Close spacing and careful current balancing eliminates all parasitic noise leaks.

It is a fairly wide-spread misunderstanding that you can simply connect a low and high antenna to one cable and that the coverage is then the same as with each one separately. Remember that the pattern is a summation of field vectors and that the combination generally forms a zero field at at least one angle of takeoff where each antenna would launch a considerable field, when used alone. Also the result is 3 dB worse at an angle where one of the antennas has a pattern null due to ground reflection (since the power has been shared; half lost).

Also stacking of two antennas does not lower the takeoff angle, but it will correspond to the average height of the two (when fed in phase). Therefore the maximum field of the higher antenna will peak at a lower angle, but the higher gain of the two makes that the combination will still produce a stronger field at this same lower angle, though the peak strength of this lobe is a bit higher.

Varying the phase between the two yagis gives new orders of freedom, but it always weakens the field in an initial maximum (any new field maximum will be lower in strength and offset in elevation).

A relatively flat environment (or sea/lake) will produce a fairly neat analysis assuming that things are fairly homogeneous over the penetration depth of the wave. This is a completely neglected issue, since a thin layer of wet clay over rocky ground may not suffice as a real substitute on lower frequencies.

Most hilly landscapes tend to act like optical lenses, but it is hard to get into a clean focus. The problem about a focus is that when that makes a signal terribly strong in one place, it must be weak somewhere else. This is why a Fresnel zone is always an area and has to be considered in full for even a two-dimensional far-field analysis.

A nice example of the preceding is the story of VK3MO, who used to drive around in Australia (probably on his motorcycle), while listening to the BBC transmissions. When he finally found an area where signals peaked awfully high, he bought his new QTH at the hot spot. I remember working him on 14 MHz, while turning my power down to about 20 milliwatts, and still got a 57 report (loud and clear). My sea reflection was a big help too (though only 20 m asl at best, near perfect on 14/50 MHz horizontal -polarisation).

Most so called one-way propagation I tend to explain with local S/N ratio and the fact that signals from small antennas tend to drown in the noise, but big signals keep at least head and shoulders dry. A pile-up is the most brilliant example of that. Why is one calling for eight hours without results, and the other makes it with one shout, while both receive the DX at reasonable strength? It is not one-way propagation at the DXer’s end, and even less at yours! On receive, think about the pile-up as the cosmic noise and the weak DX as the new DXCC you need. Your stack pushes the noise floor down to get one more, and otherwise you would witness a dead band.

No one can understand the difference of a few dB unless he has worked a couple of years with a piece of wire, and is then allowed to work with a six-element monoband yagi for a day or two. The contrast will be clear when returning to his own setup - this is why newcomers should always play first with a simple antenna, albeit just for a while. Due to the inflation of S-meter scales, most reported differences of a few S-units are not more than those few dB, but in S/N ratio those values may often approach the truth.

With microwave equipment, Magic T’s and the like are used for one-way propagation (isolators etc.). These mostly require magnetic fields and coupling of different modes. In geophysics we may expect one-way propagation when the geomagnetic field is ready for the game. That is when the magnetic field is perpendicular to the movement of free electrons induced by the impinging radio wave. Therefore there are ordinary and extraordinary waves in ionospheric propagation, and the MUF is about 1.4 MHz (the gyro frequency) higher for the highest extraordinary mode. This may explain an effect like WD8ISK’s vertical beating a six-element yagi, but it could as well happen that the elevation of the yagi was such as to make a null for the vertical angle of arrival. A polarisation rotation is also possible, but will gradually continue to the other polarisation in time.

One-way propagation may occur when the extraordinary wave is excited to one path direction and inhibited to the other. The cause may be the initial state of polarisation. Since waves tend to gradually change polarisation in ionospheric layers (Faraday rotation effect like in EME) it is not that simple, and we may again play with the idea of crossing marginal S/N levels when it seems to appear. This means that the process is there but it is not completely on/off. The stack will again provide a rescue close to ‘off’ state.

For poor ground the Brewster angle is high and this makes patterns of horizontally and vertically polarised yagis or groups very similar for the DX angles. Mainly at the ocean shore (‘liquid copper’), the maxima are interlaced to cover practically all angles, while switching between the two polarisations. Therefore, just switching the polarisation is not sufficient to cover all angles inland. Antennas are required at least at two different heights, and should also have provision to operate separately. The vertical reflection coefficient from poor ground is worst at the Brewster angle (with a 90 degree phase shift) and is always weaker than the horizontal coefficient.

With sliced patterns due to high antennas and good ground reflections, there will be a multitude of common slices in the cross section. The total common volume and the scatter angle determines average levels (height is a bonus, even on one side).

I remember a Boulder Foothills Field Day in Colorado with the W0DK group in 1985. It was a steep down-slope so ground reflection was not useful. The city covered most of the below horizon angles, and most noise was spiky man-made. Since the main lobe of a single yagi is fairly wide, tilting did not help much. Any 0.6 wave stack would have been the answer (and partly hiding behind the edge, if the top had been wide enough with a proper flat tilt for ground reflection gain).

To return to the archive page click here