In November 1984 I wrote an article on the Curtain Quad for QST [1,2,3]. The Curtain Quad is a broadside array of 7 to more than 200 elements using a very powerful but seldom used method of power distribution. Instead of feeding each element of the broadside array in the center and using power splitters with a corporate feed to distribute the power to each element, the antenna itself is used to distribute the power from a feedpoint in the center of the antenna. The resulting structure has high aperture efficiency and negligible ohmic losses. The array can consist of as few as 7 elements (with about 15dBi gain) or more than 200 elements (with about 30dBi gain).
Figure 1 shows a reproduction of Fig.2 in my original paper on the Curtain Quad. The basic building block consists of a 3 wavelength loop. The arrows represent a half wave section of the loop, with the ends of the arrows representing low current/high voltage points. Additional loops are connected at high voltage points. The resulting structures have the currents in the horizontal elements all in phase (arrows are pointing in the same direction), and the currents in the vertical wires are all out of phase. The resulting radiation is horizontally polarized. The number of elements can be considered to be the number of horizontal wires. The vertical wires are used for power distribution. The horizontal wires are the radiators as well as used for power distribution.
This structure works ok for a small number of elements, but the power distribution results in less current being delivered to the outer elements than to the inner elements, reducing the aperture efficiency of large arrays.
The current distribution can be adjusted by varying the lengths of both the horizontal and vertical elements. The easiest way to determine the lengths for best power distribution is to not worry about the power distribution at all, but to submit the antenna to an antenna simulation program optimizer to determine what lengths give the best gain. I use the genetic evolver in 4nec2 .
21 element Curtain Quad
Figure 2. 21 elements. http://home.comcast.net/~ross_anderson/CurtainQuadFigure2.htm
Figure 2 shows a 21 element version of this antenna at 1296MHz using #12 copper wire. In order to reduce the number of variables (and make the antenna easier to construct), all vertical elements are fixed to the same length and all wires intersect at 90 degrees. So all vertically stacked horizontal elements are the same length. Also shown in the link to Figure 2 are the V and H plane patterns, and the 4nec2 program which contains all the dimension information for the antenna.
The gain of an ideal aperture antenna is Gi = 4*pi*A, where A is the area of the antenna in square wavelengths. To compare the gain of an antenna to an ideal antenna with the same aperture, the formula is G = H*Gi, where H is the aperture efficiency. To convert to dBi, Gi(dBi) = 10*logGi. For a bi-directional antenna add 3.01 to Gi(dBi). The antenna is designed at first as a bi-directional antenna. Later I will discuss reflectors to convert to uni-directional. The area under the 21 element antenna is 7.58 square wavelengths which results in an aperture efficiency of 109%. This is not a bogus number; the electrical aperture is somewhat larger than the physical aperture. The ohmic loss is 0.04dB. The high aperture efficiency shows that power is being distributed efficiently to all 21 elements. Note that the length of the vertical wires are about 1 1/3 times the length of the horizontal wires instead of being equal as in the prototype antennas of Figure 1, but the circumference of the loops are still about 3 wavelengths.
45 to 221 element Curtain Quads
Simulations for other versions of the antenna can be found in the following links as follows:
Figure 3. 45 elements. http://home.comcast.net/~ross_anderson/CurtainQuadFigure3.htm
Figure 4. 77 elements. http://home.comcast.net/~ross_anderson/CurtainQuadFigure4.htm
Figure 5. 117 elements. http://home.comcast.net/~ross_anderson/CurtainQuadFigure5.htm
Figure 6. 165 elements. http://home.comcast.net/~ross_anderson/CurtainQuadFigure6.htm
Figure 7. 221 elements. http://home.comcast.net/~ross_anderson/CurtainQuadFigure7.htm
The Table tabulates some of the important information from these simulations. Notice that the smaller versions have the best aperture efficiencies and lowest losses. But even the 165 element version has an aperture efficiency of 70%, a bi-directional gain of 25.7dBi, and ohmic loss of only 0.11dB.
Notice that the shape of all these antennas is diamond shape. I found that this shape gives the best aperture efficiency. Actually the elements can be arranged in any shape such as rectangular, square, circular, elliptical and still work well.
Of course, a bi-directional antenna is usually not desirable. It needs to have a reflector to direct the energy in one direction. I have considered three types of reflectors, illustrated in Figures 8, 9, and 10 for the 21 element array.
1. An identical array to the driven array is placed behind the driven array. This works surprisingly well, giving nearly 3dB more gain; but the f/b ratio is not good. See the link to Figure 8.
2. Individual reflectors placed behind each radiating element. This works very well, giving about 3.5dB more gain than the bi-directional antenna. See the link to Figure 9.
3. A screen reflector placed closely behind the driven array gives more than 4dB more gain and a f/b ratio of greater than 30dB. The exact performance depends on how much larger the reflector is than the driven array. Figure 10 shows a design with an aperture efficiency of about 70% as calculated from the area of the screen. A smaller screen will give better aperture efficiency at the expense of f/b ratio and gain.
Why Curtain Quad?
The attributes of the Curtain Quad are:
1. High aperture efficiency. This means that this antenna gets nearly the maximum gain expected for its size.
2. Very low ohmic losses. There is nearly no loss in the power distribution over the antenna.
3. Simple feed. Just feed the center of the antenna.
4. Low side and back lobes.
5. Thin and flat.
6. Easy to model.
Building a Curtain Quad
I leave it up to you antenna builders to come up with the best construction techniques. Obviously the construction techniques will differ greatly depending on whether you are building a high gain antenna for eme at VHF, a fixed antenna for a specific path such as for tropo ducting or a long distance link, or a high gain antenna for UHF or microwave. Before ending this paper, I do want to discuss one construction technique, ending with a description of an antenna for 1296MHz that was built and tested.
Building this antenna would be a lot easier if we could find a ready made supply of wire mesh with wires crossing at 90 degrees and already welded. It turns out that such a supply exists in any hardware store. It’s called galvanized welded wire fence. The fact that it is galvanized is not a problem. Zinc has about 1/3 the conductivity of copper, but that is more than adequate for this antenna. Of course you won’t find any fencing material with the element lengths calculated in your antenna simulator, but that also is not a problem. If you want to use 2 inch by 1 inch galvanized welded wire fence for the 21 element array in Figure 2, set h=6, l0=1, l1=l2=l3=4 (See the 4nec2 programs for the length definitions). Also use 0.039 for the wire radius and 16600000 for the wire conductivity. This will yield an antenna with G = 16.28dBi and Z = 87 + j61, nearly as good as the optimized version, CQ21. If you want to use 3 inch by 2 inch fence, set h=6, l0=l1=2, l2=l3=4. This will give G = 15.46dBi and Z = 180 – j208. Of course the gain will go up 3+dB and impedance will change when a reflector is added.
13 element Curtain Quad for 1296MHz described
A 13 element version of this antenna (with a different form factor than the diamond shaped antennas so far described) was built and tested and will now be described. Refer to the Figure11 link. The material list is as follows:
The antenna was modeled at 1332MHz because that was experimentally found to give the measured performance at 1296MHz. The reason for the discrepancy is probably mostly due to the fact that the dielectric constant of the foam is not 1 as is assumed in the model. The antenna was matched to 50 ohms using a short piece of two conductor speaker wire as the matching transformer. The estimated impedance of the speaker wire is 126 ohms and velocity factor is 0.74. The simulation program optimized the driven array, reflector, and matching transformer together. I will now give step-by-step instructions for the building of this antenna (refer to the Figure 11 link).
1. Cut the antenna out of the fencing wire to the dimensions shown in Figure 11a.
2. Remove about 3/8 inch from the center of the center wire. Squeeze the vertical wires near the center wire until the gap is about 1/8 inch.
3. Solder a 2+ inch piece of speaker wire to the N connector.
4. Mount the N connector onto the piece of aluminum sheeting.
5. Cut out a 21 by 28 inch section of the foam sheet and glue aluminum foil to the backside.
6. Cut a small hole through the center of the foam sheet.
7. Mount the antenna to the other side of the foam sheet.
8. Pass the transformer through the foam and solder to the center wire. The calculated final length of transformer is 1.93 inches (2.61 times 0.74).
9. Secure the mounted connector to the back of the antenna.
This antenna was tested on an ANA, and showed nearly unity SWR at 1296MHz. Antenna patterns were taken on W6VSV’s backyard test range and agreed closely with the modeled patterns.
For an 802.11 version of this antenna for 2.4GHz, go to the following link.
Ross Anderson W1HBQ firstname.lastname@example.org March 13, 2007 June 10, 2007
 Anderson, “Meet the Curtain Quad Antenna”, QST, Nov 1984.
 This antenna was invented by Kraus. See Kraus, Antennas, 2nd edition, 1988, pp 491-495.
 Haviland, The Quad Antenna, 1993, pp 135-138.
My homepage “Ross’s Antennas”, with links to my other pages, is http://home.comcast.net/~ross_anderson
Key words: planar array, panel antenna, Kraus grid array