It isn't quite as easy as described above. Consider the following points:
- The loop is close to the ground. That interaction will be quite different over the range of 7 to 28 MHz. That is, the pattern and match could be very different than what it is on the fundamental frequency.
- While the loop is vertically polarized on 40 meters that is not necessarily so on its harmonics.
- A loop that is larger than 1λ, as with a dipole that is longer than λ/2, has minor lobes in its far-field pattern, and deep nulls between those lobes. The number of lobes increases with each harmonic.
- The loop has an impedance over 100Ω, which the ¼-wave transformer converts to 50Ω. The transformer is cut to be λ/4 only on the design (fundamental) frequency. It behaves as we want on odd harmonics but not even harmonics. For this antenna the transformer works on 21 MHz (¾-wave). On 20 or 10 meters the RG-11/U adds a reactance to the load that worsens the match to 50Ω coax.
The first thing I did was measure the SWR on 20, 15 and 10 meters. As expected the SWR was over 3 on 20 and 10, most likely due to the RG-11/U section acting nothing like a ¼-wave transformer. On 15 meters the antenna resonates a little below the band, at about 20.950 MHz. At 21 MHz the SWR is 1.6.
But as the title of this blog declares, it is not sufficient to have a match. We also need to look at the pattern. The pattern is a mess since each leg of the delta loop is now λ rather than λ/3, and the feed point is not well positioned.
The affect is obvious on the adjacent EZNEC plot of antenna current. Notice also that the tower strongly interacts with the loop on 15 meters which does not occur on 40 meters. This is due to the tower height being more compatible with the shorter 15 meters wavelength.
The tower model that I built in EZNEC only approximates the reality, especially when we consider the fact that the tower is (obviously) ground mounted. The SWR predicted by EZNEC looks nothing like what is measured, calculated at over 3 across the band.
Those ground interactions should also affect the pattern. The degree of interaction is hard to model due to that long and low horizontal leg of the delta loop plus the mutual inductance with the tower. Even so it is worth looking at how EZNEC predicts the antenna pattern.
The pattern is quite poor, as the plot demonstrates. Most of the radiation, though vertically polarized, is lost to high angles. At low angles, which are key to DX, the gain is atrocious. It gets worse in other azimuth headings (not shown).
Of course the real test is putting the antenna on the air. Tuning in to any DX station on 15 meters and switching from the inverted vee or dipole to the loop drops the signal by a significant amount. For long paths it is worse. Switching to the loop while listening to K9W (Wake I. DXpedition) caused the S5 signal to vanish into the noise
On 20 and 10 meters, in addition to the poor match, the antenna is horizontally polarized. A horizontally-polarized antenna at such a low height is certainly a poor DX antenna. I won't even bother to show those patterns.
It is always possible to match an antenna and (usually) get a reasonably radiation resistance on frequencies higher than its fundamental frequency. However that says little about performance, and especially DX performance. Beware antenna advertising that promises multi-band performance by means of a sophisticated matching network! It'll work better than a dummy load though perhaps not by much.
The pattern looks not too bad, considering that the antenna apex is up less than λ/4 at 3.5 MHz. The ground losses are high, though not too different than at 7 MHz, and the polarization is vertical.
There is however a serious problem with this antenna. That is the radiation resistance. Across the 80 meters band the antenna's radiation resistance is less than 0.2Ω. It doesn't rise to anything near reasonable below 6 MHz.
A matching network to bring the feed impedance to 50Ω is difficult, and very difficult to do so without excessive loss. It will also have a high Q and therefore a narrow bandwidth. This is not a challenge I am willing to tackle.
An alternative is to use a relay to break the loop on 80 meters. The antenna would be very unbalanced but with care could be made to behave as an end-fed antenna. Again, I doubt this is worth the effort since the performance would remain poor and it adds a lot of complexity in design and construction. The modelling I've done for this configuration has so far been inconclusive. I prefer to just not operate on 80 meters for the next while.