Wireless Hacks. 1917 IndustrialStrength Tips and Tools [Electronic resources] نسخه متنی

Hack 77 Slotted Waveguides

Make a high-gain, horizontally polarized omni
or unidirectional antenna. And it looks cool too!

[Hack #75], slotted waveguides are
resonant
antennas, and have a relatively narrow operating frequency range. The
designs described in this Hack have an adequate bandwidth for any
WLAN, but they have been carefully designed and must be equally
carefully constructed.

The major attraction of a slotted waveguide design is its simplicity.
Once you have built the first one, it is very simple to build many
more. The gain varies little across the 802.11b spectrum, dropping a
little bit at the extreme ends. A finished 8-element directional is
shown in Figure 5-17.

How a Waveguide Antenna Works

A waveguide is a very low-loss
transmission line. It allows us to propagate signals to a number of
smaller antennas (slots). The signal is coupled into the waveguide
with a simple coaxial probe; as it travels along the guide it
traverses the slots. Each of these slots allows a little of the
energy to radiate. The slots are in a linear array pattern, and the
total of all the radiated signals adds up to a very significant power
gain over a small range of angles close to the horizon. In other
words, the waveguide antenna transmits almost all of its energy at
the horizon, usually exactly where we want it to go. Its exceptional
directivity in the elevation plane gives it quite a high-power gain.
Additionally, unlike vertical colinear antennas, the slotted
waveguide transmits its energy using horizontal
polarization, which is the best type for distance transmission.

Unidirectional Waveguide Antennas

I am going to describe 2
unidirectional designs. The first has 8 slots and is about 30 inches
long. The second has 16 slots and is about 5 feet long. Simple to
construct, the 8-slot has been provided as a good starting point for
an antenna novice. I built my 8-slot prototype using only hand tools.

The 16-slot design has been made to radiate over a wider beamwidth by
the addition of "wings" to each
side of the guide, flush with the front (slotted) surface. It is, of
necessity, higher Q, and the higher gain is obtained over a narrower
bandwidth. They can be expanded aluminum or sheet, and should extend
9.6 inches beyond the sides of the guide. They act as a ground plane
for the slots. Do not change this dimension; it is two electrical
wavelengths.

Omnidirectional Slotted Waveguide Antennas

The
slotted waveguide has achieved most of its success when used in an
omnidirectional role. It is the simplest way to get a real 15-dBi
gain over 360 degrees of beamwidth.

Horizontal polarization[Hack #84] in a wide area network
can often double the number of users that can interconnect without
interference. When using horizontally polarized BiQuads or Patch
antennas (provided that they have been tested for good
cross-polarization performance) at the client site, these omnis will
be 20 dB stronger than the signal from a similar vertical collinear.
Conversely, vertically polarized receiver antennas will prefer the
vertically polarized colinear over the slotted waveguide by a similar
amount. Transmission on an immediately adjacent channel (say,
channels 5 or 7), normally not permissible because of interference,
is now possible. So a judicious intermingling of horizontally
polarized clients can talk with a horizontal central station on the
same or adjacent channels that other clients are using with vertical
polarization.

To make the unidrectional antenna radiate over the entire 360 degrees
of azimuth, a second set of slots is cut in the back face of the
waveguide. When looking straight at the face of the waveguide, you
will be able to see clearly through both slots.

Unfortunately, unless a lot of slots are used, the antenna becomes
more like a bidirectional radiator, rather than an omnidirectional.
This antenna was invented in the 1940s, and as our simulation and
measurement technologies have become more accurate, it is apparent
that the slotted waveguide designs we used in the past are far from
optimum. The most common defect is a
"tilt" in the radiation pattern at
the extreme ends of the frequency range. This occurs when the
wavelength of the signal travelling down the guide differs from the
slot spacing.

My current favorite uses 32 slots to
get 15 dBi of gain, radiated in a uniformly omnidirectional manner.
The large number of slots makes it easier to dissipate the energy
from the waveguide. As with the 16-slot unidirectional, two sets of
"wings" (one set at each slot
surface) are required to get equal radiation of energy over a full
360 degrees. Note that a higher Q is necessary to get all the slots
illuminated evenly.

Note that the gain versus frequency curve peaks at 2,440, and it
radiates well over all 14 channels.

Highly Directional Slotted Waveguide Antennas

Sometimes
it is useful to have a highly
directional antenna. For example, when installing a point-to-point
link between two buildings, it is not desirable to have a wide angle
of coverage. Any interference from other 802.11b devices (or
microwave ovens) that are in the radiation zone will affect your link
integrity.

The ideal antenna for such a situation is a dish, such as
Primestar's. When using my BiQuad feed, it is
possible to reject interference outside the dish's
primary 5-degree cone by 30 dB or more.

But if a 16-slot waveguide antenna is turned to a horizontal
position, parallel with the ground, it radiates vertical
polarization. Its directivity in this plane is extremely good. So, if
you don't have a dish handy, consider the
possibility of using a pair of these slotted waveguides, parallel to
the ground. They will work very well.

Construction Details for the 8-Slot Unidirectional Antenna

The base extrusion for all of
my slotted waveguides is 4" x 2" O.D. rectangular aluminum
tubing with approximately 1/8-inch thick walls. Inside dimensions are
3.756 x 1.756 inches (95.4 mm x 44.6 mm). These
inside dimensions are critical, and must be within +/- 0.040 inches
or +/-1mm if the antenna center frequency is to be +/- 1 channel. I
cut the end inserts from a 5/16" x 1 3/4" flat aluminum
bar extrusion.

Waveguide antennas are fairly critical in their constructional
dimensions, and are easiest to make with a CNC milling machine. I
have computed these designs so that they would be easy to replicate;
if you are plus or minus 1 mm, the design will work finebut
you must be careful. I used a jig, a hand operated DeWalt heavy duty
cut-out tool, a 1/4-inch router bit, and lots of water to machine the
slots. This worked fine (even if it was a little tedious). Really,
folksplus or minus 1 mm will not kill your antenna!

Coupling the Signal into the Waveguide

As I said previously, we are
propagating the WLAN signal down a waveguide and then using it to
excite a number of elemental radiators, or slots. The first task is
to get the signal into the waveguide with a feed probe. First,
obtain a suitable N connector. Take a piece of 20 mm x 40
mm copper or brass shim, and form it into the shape of a cone. Use
Figure 5-18 as a template.

Solder it to the inner conductor of your Type N connector. Its length
should be 20 mm, and its largest diameter should be about 15 mm. When
soldered to the N connector, it should protrude exactly into the
center of the waveguide and no further. The finished feed cone is
illustrated in Figure 5-19.

Both ends of the waveguide need to be terminated for RF. The easiest
way I found to do this was to cut 3.75-inch pieces of 5/16"
x 1.75" aluminum bar stock. I do not recommend that you
make the end plugs sloppily, but good electrical contact is not
required.

Remember not to have any screws protruding into the waveguide for
more than 1/8 inch, especially the screws holding down the N
connector. They will affect performance.

8+8 Slot Omnidirectional Antenna

The total length of air inside the
8+8 slot omnidirectional waveguide,
from end to end, is 765 mm. Mount the N connector in the center of
the widest side: 27.5 mm from one end (the
"base") of the airspace in the
waveguide, and offset 10 mm from the centre line of the face, in the
direction as the offset of the first slot. The wavelength of the
radiation passing down the waveguide is longer than a wavelength in
free air (it is 161 mm in this design).

The first slot is centered 1.0 wavelength from the base, at a maximum
of the H field in the waveguide. This length is 161 mm from the base
of the airspace. It is the H component of the field that induces the
energy into the slots and makes them radiate. Each slot is 59 mm
long, and extends outwards from the centerline for a width of 17 mm.
The waveguide excites each edge of the slot, depending on its
position across the wide surface of the guide. If it straddled the
exact center, each edge of the slot would be excited in antiphase
(the waves cancel each other out), and there would be no radiation.
So as we offset the edges of the slots, the more the offset, the
greater the energy that is dissipated into each slot. The electrical
length of each slot should be 59 mm. Do not allow too much kerf at
the ends (it should be 2 mm radius max). I recommend finishing the
cut with a 1/8-inch router bit (or a file). Or you might use the 1/8
bit in a CNC machine to cut the entire rectangular outline. Remember,
even though these slots are arranged vertically,
they radiate horizontal polarization.

For the 8+8 slot omnidirectional, slots 2-8 are centered at distances
of 241, 322, 403, 483, 564, 644, and 724 mm from the base of the
airspace, staggered across the centerline. It
doesn't matter which direction the first one is cut,
but they must alternate. The end plate should create a 765-mm
airspace. Looking straight on at the front of the guide, you can see
right through both the front and back slots.

8-Slot Unidirectional Antenna

The total length of air inside
the 8-slot unidirectional, from end to end, is 760 mm. Mount the N
connector in the center of the widest side, 25 mm from the base of
the airspace in the waveguide. The wavelength of the radiation
passing down the waveguide is 160 mm in this design. The first slot
is centered 1.0 wavelength from the base, at a maximum of the H field
in the waveguide. This length is 160 mm from the base of the
airspace. Each slot is 58 mm long, and extends outwards from the
centerline for a width of 20 mm. The waveguide excites each edge of
the slot depending on its position across the wide surface of the
guide. If it straddled the exact center, each edge of the slot would
be excited in antiphase and there would be no radiation. So, as we
offset the edges of the slots, the more the offset, the greater the
energy that is dissipated into each slot. The electrical length of
each slot should be 59 mm. Do not allow too much kerf at the ends.
Remember, even though these slots are arranged vertically, they
radiate horizontal polarization.

Slots 2-8 are centered at distances of 240, 320, 400, 480, 560, 640,
and 720 mm from the base of the airspace, staggered across the
centerline. It doesn't matter which direction the
first one is cut, but they must alternate. The end plate should be to
create a 760 mm airspace.

Constructional Details for 16- and 16+16-Slot Design

The correct wavelength for these
designs is 161 mm. The gain for the 16-slot unidirectional is 15
dBi-17 dBi, verified on my test range, across the whole band. On the
range, the 16 slotter gives slightly higher gain than my Hyperlink
Technologies model 2419G Mesh Parabolic, which is
"rated" at 19.1 dBi gain.

The slot width for the 16 slotter is 15 mm, and it is 12 mm for the
32 slotter; otherwise, the key dimensions are the same.

The original article upon which this Hack is based is available
online at http://www.trevormarshall.com/waveguides.

Trevor Marshall