First, find a parabolic reflector designed for Ku band direct-to-home satellite service. These are now inexpensive and abundant - the one pictured above was discarded by a local resident. After washing off the dirt, spiders, and millipedes, I found the dish to be in excellent condition. Similar reflectors can be found in thrift stores, flea markets, and other habitats of electronic scroungers. Seek large reflectors (at least 76cm or 30" diameter), since these will provide greater gain and directivity.

Consider how to feed the antenna. I have seen web pages featuring tin can feedhorns, biquads, patch antennas, and helical radiators. A helical feed seemed suitable for this project as it was simple to build, inexpensive, and provided good gain and directivity. Being circularly polarized, a helical eliminates sensitivity to antenna orientation, and resists attenuation in rain. Though there is a 3 dB loss working linearly polarized stations, designing for more turns brings the gain to parity with the tin-cans and biquads. To Calculate dimensions for your antenna, use this excellent calculator, plugging in a frequency of 2450 MHz.

PARTS REQUIRED:

• one square piece of single sided copper clad PC board for a ground plane
• one PVC kitchen drain tailpiece (3.8 cm / 1.5"diameter) to hold the helical windings
• six 1/8" plastic cable ties
• a length of copper circuit tape (adhesive backed, width 3mm or 1/8") or #14 copper wire
• one suitable chassis connector (I used a reverse sma type matching the connector on my adaptor)
• one 90 degree angle bracket with screws and bolts to fit

CONSTRUCTION:

1) Center the tailpiece on the PC board, copper side, and mark the circumference in ink.

2) Mark four locations on the circumference, spaced 90 degrees, where the cable ties will hold down the PVC tube.

3) Mark one location on the circumference, exactly between two 90 degree markings, where the coaxial connector will be mounted.

At this point you should have a PC board with a circle in the center, four tick marks on the circle at 90 deg intervals, and one tick mark exactly between two others.

4) Drill 1/8" holes on the inside and outside of the circumference at the cable tie locations.

5) Drill a hole directly on the circumference suitable for the chassis connector. Carefully measure and drill other holes for this connector if necessary.

6) Drill four holes, spaced 90 deg apart near the bottom end of the PVC tailpiece.

7) Drill holes to accomodate a small 90 degree corner bracket.

8) Drill holes on opposite side of board to accomodate USB wireless adapter that will be affixed with cable ties.

9) Tin the copper around the connector mounting hole, then mount the connector. Clip the center pin to keep it only long enough for connection to the helix windings.

10) Cut out a notch to accomodate the connector; it should clear center conductor, but avoud cutting out excess PVC material.

11) Feed cable ties through from the back side of the board, through holes in the tube, and back through the board. Tighten the cable ties, making sure the tube is firmly held to the copper ground plane.

12) Use a ruler and the edge of a sheet of paper to create a template for positioning the windings on the PVC tube. Distance zero represents the ground plane, then add the feedpoint distance, then ticks matching the turns spacing. Use the template to mark your tube on both the feedpoint side and the opposite side. The objective here is to have a guide while precisely winding the helix...

Here is a table used for my prototype helix and its connector. Note that turn 1 starts at 0.8 cm (height above ground plane of feedpoint). Turns Spacing is 2.5 cm, and the diameter is 3.9 cm (close enough for 1.5" PVC tailpiece)

13) Carefully wind the helix, using copper circuit tape or wire, then solder to center conductor of chassis connector. Double check against the turns template. Polarization will be right-handed if the turns spiral clockwise (looking outward from feedpoint).

15) Attach the angle bracket and wireless adapter, making sure all parts are secure and ready for service.

The antenna should resemble the prototype, pictured below with the angle bracket removed.

At this point, the antenna is ready for its smoke test...plug in the cables and look for some signals! Theoretical gain of the prototype was about 18 dB over an isotropic radiator; it beat my biquad by about 7 to 13 RSSI units, and indeed seemed less sensitive to polarization and rainfall. Signals still seem to fluctuate much from second to second. If your antenna is functioning satisfactorily at this point, I suggest spray painting two layers of clearcoat onto the windings and groundplane for stability and preventing corrosion.

Currently, I use a short version of this antenna, left hand polarized, to feed a parabolic reflector and note signals about 22 dB stronger than on a simple dipole. Mounting the offset fed dish was tricky - it required inverting the mounting hardware and angling the dish to about 75 degrees to look for signals on the horizon. It rests atop a vertical length of 2" sch 40 PVC pipe. After not initially seeing much strength on a station until finding the right elevation, the signal strength suddenly pegged full scale when I aimed it dead-on! Aiming properly in both azimuth and elevation is important with these larger parabolics - you can aim to reject interference as well as maximizing weak stations. Mounting the helix was simply a matter of removing the LNB and (using the angle bracket) bolting the helix into place at the focal point. It was necessary to bend the bracket just a bit to aim the helix at the center of the dish.

Note that a biquad feed would also work well with a parabolic reflector, and would have a slight advantage over a helical - but would show more sensitivity to polarization and rain

Source:http://www.geocities.com/ab9il_worldwide/wifi3.html