Dipoles are in theory the simplest antennas. They are 1/2 wavelength long consisting of two 1/4 wavelength elements insulated from each other. The typical impedance of a dipole is 75ohms, so you will often see variations on the dipole theme in order to get a better match to 50ohm feedline. Some of these variations include flange matching, gamma matching and slot matching. Flange matching is the simplest and there are many variations for this in itself. The circuit board dipoles you will find on other pages here use this method. I use a gamma match on my VHF dipoles. The dipole is a single 1/2 wavelength element and the coax shield is fed to the center of the element and the core is feed to about 1/3rd of the element distance with a variable tuning capacitor. I don't recommend this sort of feed at 2.5GHz because the capacitor itself would become parasitic and would most likely detune the antenna.
I use a slot match on my 1.3Ghz dish. It works very well and has a wide relative bandwidth. This sort of feed is complex and takes quite some time to construct. I made mine out of heliax cable that was carefully hacksawed. I later made a second feed out of copper waterpipe and used a blowtorch to solder the pieces together.
The radiation pattern of a dipole is omni-directional around the axis of the element. The gain is 2.2dbi
Another variation of the dipole is a folded dipole. It's characteristic impedance is 300 ohms so you will need a balun transformer of some sort to match the dipole to 50ohm coax. There is also a double folded dipole, it's characteristic impedance is 600ohms. There is no reason why you could not use folded dipoles. Their wide bandwidth compared to standard dipoles makes them a good choice, but they can be tricky to construct.
Yagi antennas are a variation on the dipole theme. Adding a reflector increases the front to back ratio and adding directors increases gain. The limit occurs at about 32 elements before you no longer get any practical gain. Yagis tend to be narrow bandwidth antennas. You can improve the bandwidth by making the elements thicker or having a variation in the director lengths. The variation on this is commonly known as a log-periodic antenna. Of course you never get something for nothing and the tradeoff for bandwidth is gain. Other type of yagi antenna is a hybrid yagi-slot type antenna known as a skeleton slot antenna. This has the best of both worlds, good gain with wide bandwidth. At 2.5GHz they will be easy to construct.
Phased arrays are just dipoles stacked to give more gain. There are even more variations on phased arrays of antennas than I can care to imagine.
Loop yagi are a spin off of yagi antennas. They are very easy to construct and are tolerant of even the worst home constructor. They provide wide bandwidth, and with a good reflector and a number of elements, very high gains and excellent front to back ratios. They are a very good compromise for dish antennas since they are small and have very similar gains.
Quad Antennas are another wide bandwidth antenna. They are usually used on HF, frequencies below 50MHz, but they also work well up into UHF/SHF. Quads can be phased and stacked into arrays or the can be configured like a yagi antenna (Quagi). They can be center fed, corner fed or even gamma matched. There are also Delta Quads (Triangular) and Swiss Quads, and Double Diamonds.
Helical Antennas are a very good option since they are very easy to build with a good former.
They have high gains and good front to back ratio's with a good wire reflector. The only problem
with these antennas is Polarisation. Helical antennas produce "rotational polarisation". This is
still workable with linear polarisation, but there is a 3db loss incurred. (half the signal is lost.)
The other problem is that there are two types of rotational polarisation.
LHCP & RHCP left hand and right hand circular polarisation. The isolation between the two is about 30db (1000 times) and all it takes to make the mistake in building the antenna by winding the wrong direction by accident. To add to the confusion. If you are reflecting the signal off anything then the polarisation is reversed. There is another form of antenna called a 'dielectric antenna' that have about 5dbi gain circular polarisation. They are a type of flat panel antenna. Turnstile antennas also produce circular polarisation, they are mainly used for LEO satellites but a small version could be used to feed a dish.
Dish antennas are the ultimate. High gain and high directivity is ideal for end users, but for those of you who want to run an Access Point, high directivity is not a desirable quality in an antennas. They are also big, difficult to mount, have high wind loadings and are generally expensive. Corner reflectors or parabolic mesh reflectors (galaxy MDS antennas) are the next best thing. Again, they have very directional radiation patterns, so they are really only good for peer to peer use or peer to AP paths.
I've noticed several web sites explaining how to build "cantennas". This is a simple form of waveguide antenna is easy to build but has some rather annoying characteristics. The gain of a cantenna is about 6dbi, but the antenna itself is rather inductive and the phase distortions that occur between the E & H planes causes unpredictable radiation patterns. The advantage they do have is that they are easy to mount to a dish.
A better option is to use rectangular waveguide. One side of the waveguide should be 1/2 wavelength and the other 1/4 wavelength. The waveguide will then operate in single mode and phase problems will not occur. You can then add a horn antenna to increase the gain of your waveguide antenna.
Horn antennas and waveguide are very easy to test with simple equipment. All you need is a small speaker and a 2700Hz signal source. Sound waves have the same wavelengths as microwaves, so a good horn antenna will also be a good speaker and have the same audio patterns as it does RF radiation patterns. All you have to do is walk around the antenna and listen to where the signal is strong and weak.
Slot antennas. There are a few different types of slot antenna. Slots cut across the E-plane of the waveguide work fine, they can also be cut along the waveguide, but this is more difficult to do. Skeleton Slots are more like a yagi than a slot antenna, but the principles are the same for the driven element. Alford Slot antennas are very easy to build and they produce very good omni radiation patterns on horizontal polarisation.
Dipoles make simple omni antennas when vertically polarised. A Quarter wave vertical with a ground plane is the next simplest design. It has an radiation pattern of 45 degrees. A Half wave vertical with a ground plane behaves the same as a quarter wave except the radiation pattern is lower, with an angle of about 30 degrees.
A 5/8 wavelength vertical on a ground plane gives an even lower radiation angle, somewhere about 15 degrees. The lower the radiation angle the more apparent gain the antenna has at the horizon. Co-linear antennas are basically these three types of elements stacked one on top of the other with some sort of phasing arrangement between them. The phasing elements can be capacitors, inductors, tuned circuits, delay lines or phasing collars. There are plenty of combinations and most of them work with any major hassles.
You will often see gain in dB. When you see this unit you will need to consider what it is your seeing carefully, for it is not really a unit of measurement but a ratio of units. A good example is dbi and dbd. One is in reference to isotropic and the other in relation to a dipole. A lot of people seem to have trouble thinking about isotropic radiators, so a lot of ham radio operators just compare their antennas to a dipole for reference. Therefor, if their antenna is twice as good as a dipole it has a gain of 3dbd. A dipole is 2.2dbi itself so the antenna in question really just has a 5.2dbi gain.
An isotropic radiator isn't such a hard thing to comprehend. A good example is the sun. It emits radiation equally in every Direction. It has 0dbi gain. But if I take a very big mirror and place it behind the sun so that half of the radiation is reflected back, then the sun will appear twice as bright on this side. It now has a gain of 3dbi. If I again place another mirror below it as well as behind it is now reflecting twice as much power again.. four times which is 6dbi gain. If I had another mirror so that the sun sits in a 3d corner reflector, the gain is now 8 times or 9dbi. And so on till you have an infinite gain, an example of which would be a perfect laser. Lasers have such high gains in relation to isotropic radiators, this is why a 5mW laser can blind you where a 40W light bulb can barely light the room.
Quad Phased Array