Building the bicircle

I also tried different heights off of the reflector. I varied the distance from 1/16 to 1/4, with the impedance being better at 1/8th. I had eliptical shapes and some very odd ones, but they all worked as well. The shape of the loop, and where it gets feed, does change the impedance. Practically, these antennas worked equally as well, though given the readings I have now made, some of these shapes have a horrible impedances.

We hadn't originally taken into account the velocity factor of the wire in the antenna. i.e. the speed of light in the wire is less than in air. Having a slower speed in the wire, means that the frequency shifts downward (frequency = velocity/wavelength), so the wire needs to be shorter to resonate at the same frequency. Measuring a 2.442GHz antenna, 20 year on, shows it was tuned better to 2.35GHz, so was too long.

The velocity factor is going to vary with different wire thicknesses. In our case the overall wire length should be reduced by 4% to 5%. Plastic close to the wire will change the velocity factor too, further reducing the speed of the radio wave. This happened to us, when we put the antenna inside a plastic case. Copper oxide on the wire will have this effect too, so an aging antenna is going to slowly detune to a lower frequency.

The velocity factor represents the percentage slowdown of the incoming signal in air, compared to the signal in the copper wire. This is the same affect you see with light going through water or glass (light being a radio wave). The light induces a wave in the electrons in the glass, but the speed of that induced wave is less than the speed of light in air. If you add both the original light's radio wave, and the slower induced wave in the material, you get a resultant wave that we see. When the light leaves the glass it speeds up again. The light isn't magically gaining energy. It has just stopped interacting with the electrons in the material. The material doesn't need to be conductive either, just a dialetric like plastic. In a dialectric, the radio wave alternates the polarization (alignment of charge) in the material, but no current flows.
See also BiQuad Calculator

Frequency	b/g Chnl 1 b/g Chnl 2 b/g Chnl 3 b/g Chnl 4 b/g Chnl 5 b/g Chnl 6 b/g Chnl 7 b/g Chnl 8 b/g Chnl 9 b/g Chnl 10 b/g Chnl 11 b/g Chnl 12 b/g Chnl 13 b/g Chnl 14 -NZ Indoor- a Chnl 34 a Chnl 36 a Chnl 38 a Chnl 40 a Chnl 42 (mid-range) a Chnl 44 a Chnl 46 a Chnl 48 -NZ outdoor low- a Chnl 50 a Chnl 52 a Chnl 56 a Chnl 57 (mid-range) a Chnl 58 a Chnl 60 a Chnl 64 -NZ Outdoor High- a Chnl 149 a Chnl 152 a Chnl 153 a Chnl 157 (mid-range) a Chnl 160 a Chnl 161 a Chnl 165	GHz
Free Space Wavelength		mm
Velocity Factor		%
Loop Wavelength		mm
Offset from reflector		mm
Reflector height		mm (Minimum)
Reflector Width		mm (Minimum)
The size of the reflector and its distance from the active element affect the feed impedance. The ARRL handbook says there should be a 1/4 wavelength edge beyond the active element.

Misc Notes

• Use 1.5-2mm copper. I originally used 3.2mm (1/8") Mang-Bronze welding rods. These are really hard to bend. I built 2 with bits of broken toy railway track, and also with 2mm copper wire (strands from a discarded offcut of 16mm power cable). All worked equally as well, so my suggestion would be to use the copper wire, as it is by far the easiest to work with.
• The N-Socket is riveted to the back plate (cut from an old copper water tank). I initially used nylon bolts, which can be seen in some of the construction pictures.
• The antenna is held off of the back plate (1/8 wavelength) by 2mm diameter copper wire (the original plan had coax, but I see no difference in performance). From experimentation, the wires should about 1mm apart. Moving closer to the backplate increased the impedance too, so is an effective tuning method.
• You will notice the use of nylon mounting nuts, so as not to have metal bolts above the backing plate surface. We now just use 3mm rivets, and file the bulk of the top off, after it is in place. Again, it hasn't really affected the antenna performance. It might well be doing odd things to the impedence.

These have long been retired, but I got a nanoVNA recently, and decided to go back and see how well tuned these antenna were. They all had been working 500m to 2.5km, so from a practical point of view, they were great antenna.

• They are all tuned low, between 2.35 and 2.36GHz, so the wire needed to be shorter.
• At 2.35GHz, the SWR on many is really good at 1:1.2
• And most had an impedance about 46ohms (some were as low as 40ohms, but squeezing the feed lines together bought them closer to 50ohms). The best was 48ohms, and was 1mm closer to the back plate than the others.

  From the front (vertically polarized).	From the Back	Looking down on it when it is aligned for horizontal polarization (i.e. the circles are one above the other) This one is a test version at 3/16 of a wavelength from the reflector.	Variant with parabolic backing plate. Something we tried for long links, which worked well (toy Rail used for the circles)