LOOK UP IN THE SKY! IT'S A BIRD! IT'S A PLANE! IT'S A YAGI!!
We were discussing yagi and log periodic antennas and their related brethren, and the fact that those antennas we refer to as "yagis," often are something else, entirely.
Both these types are frequently stacked vertically and/or horizontally, to make up dual and quad arrays. An array may be used to increase the gain of the antenna even further, or to further tailor the directional pattern of the basic antenna. When antennas are stacked, the physical space between the units is critical, as is the length of the harness cables used between the antennas and RF splitters/combiners.
The quad array pretty much represents the maximum practical gain available from a given antenna. Theoretically, in order to realize a 3 dB gain over a quad array, we would need to go to eight antennas. In the real world, the additional cable harness, connector and combiner losses eat pretty badly into even that gain figure. So we've reached the point of diminishing returns.
It's also common when using these antennas for transmitting purposes to use arrays of "skewed yagis" (antennas pointing in different directions) to produce many weird and wonderful directional patterns. In this case, the power dividers used can also be arranged to produce unequal power divisions, to even further enhance the number of choices available.
Most of the time, these antennas are oriented for horizontal polarization, although they may be used for vertical polarization as well. An additional consideration when mounting them for vertical polarization is that the antenna should then be located several wavelengths above ground in order to function as specified. Otherwise ground effects can affect the impedance as well as the directional pattern and front to back ratio of a vertical yagi or log periodic antenna.
Yagi's can be used in creative ways to eliminate co-channel interference. One technique is to mount two yagis, both oriented in the same direction, but staggered such that the incident desired signal arrives at the first antenna one-quarter wavelength before it reaches the second antenna. A special harness is constructed such that an additional quarter-wavelength delay is encountered by the feed from the first antenna before it's combined with the output of the second antenna. Net result: incident signals on the main lobe of the antennas are delayed equal amounts, and sum normally. Signals coming in from the back of the antennas end up opposite in phase and cancel out at the summing point. The front to back ratio of the antennas is increased significantly (at one frequency of interest).
A more general technique to reduce co-channel interference from a specific known direction involves using trigonometry and the known velocity of wave travel to calculate the phase delay between two antenna positions as seen from the source of interference. This distance is adjusted until the undesired signal arrives at the two antennas 0.5, 1.5 or 2.5 wavelengths apart. Summing the antenna outputs causes phase cancellation of the undesired signal, in effect placing a deep asymmetrical null in the array's directional pattern at that frequency -- in the direction of the interference. Reception of the desired signal is relatively unaffected.
It is even possible to make up a circularly polarized array by coupling a horizontal and vertical antenna through a phase delay harness. The real benefits of circularly polarized FM transmissions have never been fully realized: a CP FM receive antenna has a powerful mechanism to reject multipath reflections (in addition to directivity, that is), since reflected circularly polarized signals "spin" in the opposite direction as the incident signal. This would be an advantage for receiving fringe CP FM transmissions at a rebroadcast site or cablevision headend, for instance.
The sheer physical size of such a contraption would pretty much prevent its acceptance on FM frequencies by consumers, however.
Coming up next, we'll stir the pot a bit in a discussion of circular and horizontal polarization for FM broadcast stations.