The evaluation kit for the Linx Technologies units (which work at 908-928 MHz) only comes with two antennas. There is no T/R switch on the evaluation boards, so some additional antennas will be necessary to try a full duplex (or, at least, half duplex) link. After casting about for ideas on what sort of antenna to build (other than just a 1/4 wave ground plane...), I decided to build a pair of helical antennas, one for each circular sense. Helical antennas have broad bandwidth, loose construction tolerances, and for 900 MHz, wouldn't be too big. They also have a bit of gain (10-14 dBi), so they'd be a nice choice for talking to the robot at a distance (10 dB would be about a factor of 3 range increase).
The basic design rules for helicals (in Kraus' book, or the ARRL antenna handbook) say that the circumference of the winding should be on the order of a wavelength (33 cm in this case). This works out to 10.5 cm, or about 4", a convenient size to wind. I found a 3" PVC coupling that is 103 mm OD, which served as a nice form. I had some AWG 10 (or is it 6?) bare copper wire that I wound on the form. It springs back a bit after winding, so the resulting coil is 110mm in diameter, which is certainly in the right range. A typical winding pitch is a quarter of a wavelength per turn, or, in this case, about 8 cm/turn (3 1/4 inches). I have 11 turns of the copper wire total, so if I split it into 5.5 turns per antenna, I can have a left and right pair. The approximate axial ratio is 2n/(2n+1), so I should get an axial ratio around 0.9. To make the opposite sense, I will have to rewind the coil though (slightly...)
Actually, I foolishly thought the winding pitch was a quarter of the diameter at first (25 mm, 1 inch), and ran a bunch of NEC models at that pitch, and the gain is still reasonably good (9dBi), but the axial ratio wasn't that hot.
The ground plane should be on the order of a wavelength in diameter (i.e. a pizza pan 14" in diameter should work nicely). I picked up some inexpensive steel pans at the super market ($4 each). They're plated with some sort of shiny metal and not real thick. Copper sheet or some strips of copper tape might be necessary to get the ground plane resistance down, but without a calibrated antenna range, it will be hard to tell.
Supporting the helix is a bit tricky, depending on how rugged I want it to be. Finding a suitable piece of pipe and just gluing it to the ground plane might be one approach. This has the advantage of being quite rugged. Another approach is to use struts, which makes it a bit more fragile, but a lot lighter weight. Making a tube from thin plastic sheet rolled up (like a plastic placemat, for instance) is another possibility.
For now, I've chosen to cut some lengths of 3" PVC Sch 40 pipe. (3.5 inches OD, 3.068 ID) 45 cm long ( a bit less than 18") and epoxy them to the steel pizza pans. A TNC connector is used to feed through and connect to the helix (it's a bit more rugged than an SMA, and they also match what I have for old cell phone antennas). The pair of ground planes are bolted with 10-32 machine screws (4 places) to a piece of 1"x2"x1/8" aluminum channel that is 2 feet long. It looks like some sort of science fiction death ray, and will probably get the "homeland security" folks after me. Upon an improvised stress test (I dropped it), the epoxy failed between the PVC and the steel (probably not enough epoxy in the right place, and not cured long enough). I'll re-epoxy it, and put some small screws into the PVC. [Note from 10 Nov, reglued, added some #4 wood screws, and it's pretty rugged, although with the heavy PVC pipe on it, the ground planes are now the "weakest link".]
The actual helix is wound from 1" copper foil tape. I marked the tubing at the appropriate spacing (8 cm) on two roughly opposite sides. I then wound the tape (with its paper backing) to hit the marks, adjusting the pitch and windings so that the tape lays flat. The paper doesn't stretch, so when you're done, it is a truly even spaced helix. A small piece of tape at each end holds it in place. Then, I marked along the edge of the foil tape with a marker to give me a guide for when I stuck the tape down for real. Peel the backing, stick the tape, and it's done.
To tune it and adjust the feed impedance, a standard technique is to bend the first turn a bit so that it is closer to the ground plane, and forms a transformer to change the nominal 140 ohm input to 75 or 50 ohms, as needed. With the copper foil tape, I just set the helix so it ended right before the connector, and then trimmed it back a bit. Some more on matching calculations.
It's a bit heavy (3 feet of 3" Sch 40 PVC pipe weighs a lot!). With no adjustment, I checked out the match using a fancy network analyzer. The return loss on both was better than -10 dB, and more like -14 to -15 dB. Part of the reason the match is so poor is that it was tested sitting on a bench next to me, a shelf, and the rack with the network analyzer, as you can tell from the overall waviness in the return loss curve (particularly look at the Left Hand helix, which was closer to me). Even aggravating this is that the test was done in a shielded room, so the antenna was pointing at the metal ceiling about 1 meter away (although, given that the antenna is CP, the first reflection shouldn't make any difference).
Here are thumbnails of the return loss plots (5 dB/div) (click them for the full size picture in a new window)
10 dB return loss = 1.95:1 VSWR, 14 dB return loss = 1.5:1 VSWR
The following are a series of thumbnail photos of the completed antenna. Clicking on the links gives you the full res JPEG image in a new browser window. The scale in the photo is a 12" long machinist's scale. The full size images are about 100K.
This is a NEC model (for a modified version of NEC2 using P3)
CM 918 MHz Helix CM 110 mm in diameter, 5.5 turns, 2.54 mm (AWG 10) conductors CM pitch is 82.5 mm/turn CM groundplane 36 cm (14") in diameter CM Pattern at 918 CE Evaluated at 08 through 928 MHz SZ 1800,0 GW, 200,36, 0,0,0, 0.18,0,0,.00127 GM, 1, 17, 0,0,20, 0,0,0, 200 GM,0,0, 0,0,0, .18,0,0,200 GW, 400,36, 0,0,0, 0.18,0,0,.00127 GM, 1, 17, 0,0,20, 0,0,0, 400 GM,0,0, 0,0,0, -.18,0,0,400 GM, 0,0, 0,0,0, 0,0,-.005 GH, 300, 201,.0825, -.45375, .055,0, .055,0, 0.00127 GM, 0,0, 0,0,0, 0.18,0,0, 300 GH, 100, 201,.0825, .45375, .055,0,.055,0,0.00127 GM, 0,0, 0,0,0, -0.18,0,0, 100 GW, 999, 3, .235,0,-.005, .235,0,0,0.00127 GW, 900, 3, .125,0,-.005, .125,0,0,0.00127 GE EK GN,-1 PT,-1 EX,0,999,3,1,1,0 FR,0,1,0,0,918 RP,0,91,1,1000,0,0,1,3 EN