Wednesday, July 20, 2011

Curtain Antennas

Curtain Antennas

Curtain antenna at Radio New Zealand

Three towers support two 2-bay wide 4-bay high 5 band curtains at Radio New Zealand International.

curtain antenna HRS Sterba or Bruce array high gain towers

These two 300+ foot tall towers will support my new distributed feed curtain array for 80 and 40 meters. From all indications this will be the highest gain array ever installed and used exclusively for amateur radio service on 80 and 40 meters. Read how the curtain evolves in the article below.

Reference for gain, the DIPOLE

In order to understand gain, we have to understand the signal from a good basic antenna like the dipole. A dipole is the normal reference for comparing antennas.

The dipole is also a basic building block of many antennas. Let's dispel a common gain misconception about dipoles and isotropic radiators.

A dipole does NOT have 2.2dB gain over an isotropic radiator when the dipole is placed over earth. At optimum heights, a common 1/2 wave dipole actually has about 8.5 dB gain over an isotropic radiator! Always remember that when you see antenna models over earth that tell you an antenna's gain in dBi.

If a model over earth shows a "gain" of about 8.5 dBi, the model effectively has the same gain as a dipole at optimum height over typical earth! We cannot add 2.15 dB to the isotropic gain to get the dBi gain unless ALL of the antennas are in free-space! The instant the earth is involved in a model or measurement the 2.15 dB rule flies out the window.

The plots below are for a 145-foot high copper wire dipole modeled with high-accuracy ground over medium real earth on EZNEC:

reference dipole pattern for curtain antennasreference dipole properties

You can see the gain is 8.5 dBi and it is just a simple dipole just over 1/2 wave high. Any antenna we model should be compared to a standard like a dipole over real earth (unless we intend to install the antenna in outer space)!

What is a Curtain Antenna?

Properly constructed curtain arrays or curtain antennas are a form of broadside radiators. Their name derives from the similarity to a wire "curtain", or a wide flat sheet hanging vertically. The curtain's flat wide form occurs because broadside radiators generally use elements aligned in a flat plane at right angles to maximum radiation. Properly designed curtains require a minimum number of elements to obtain optimum gain for a given physical area. Other antennas may be less complicated to build, but a properly designed curtain has the highest gain for a given volume of space of any large array. Rhombics are no match for curtains, and neither are Yagis, unless perhaps the curtain is a curtain of made of small Yagis!

A few advantages of properly designed curtain arrays are:

  • Bandwidth of usable gain at least one octave
  • Pattern largely insensitive to frequency changes within a 2.5:1 frequency ratio
  • Maximum possible gain in a given physical area
  • Ability to electrically skew or steer the pattern over at least 20 degrees of azimuth and elevation

Curtains can assume several common forms. These forms include vertically polarized arrays like the Bobtail, Bruce arrays, and H arrays. They also more commonly include horizontally polarized arrays like bedspring arrays, Sterba curtains, lazy-H antenna arrays, or distributed feed curtains like the USIA arrays or HRS arrays used for shortwave broadcast.

Two common installers of large broadcast curtain arrays were TCI and Weldon and Carr. While E J Sterba (E. J. Sterba, "Theoretical and practical aspects of directional. transmitting systems," PROC. IRE, vol. 19,. pp. 1184-1215) pioneered some early closed-loop curtains, they were mostly single frequency designs.

This article covers the design of several common curtain arrays.

Lazy H Antenna

The Lazy H is the most elementary form of a curtain antenna. It consists of two horizontally polarized doublet or dipole elements that are fed in-phase. It might or might not have collinear gain from the elements being extended beyond 1/2 wave but less than 1.3 wavelengths, but it always has broadside gain from parallel in-phase elements.

There are several Internet pages and publications showing incorrect or substandard feed methods. These substandard systems include feed methods that are critical for electrical length, needlessly sacrifice gain, or restrict bandwidth!

This layout appears many places, including an often referenced internet article:

Lazy-H antenna typical but bad feed method

The article originally containing this drawing claims the elements are being fed out-of-phase. It further states the array will work multiple bands with the feed system shown on the left. The reasoning behind that statement is understandable. After all, the feedline is transposed. At a glance we might assume the feedline transposition causes the elements to be out-of-phase.

In reality, all broadside curtain arrays feed elements in-phase, not out-of-phase. In-phase elements radiate broadside to the element plane, or broadside to the plane curtain. If elements are fed out-of-phase there would be very little radiation broadside to the array. If the elements were out of phase, there would be a deep null broadside to the array. Most of the radiation would be at very high angles, or skewed toward the ends.

Any curtain array must feed the elements in-phase, not out-of-phase. In the simple array pictured here, this is accomplished by having a feedline electrically 1/2 wave long and a 180 degree flip. A 180 degree feedline plus a 180 degree flip results in 360 degrees of shift, which is the same point on a circle as zero. The total phase shift is zero, so the elements are in-phase.


We have a feedline that behaves like it is 250 degrees long. We transpose the feedline 180-degrees.

antenna phase wheel

Inverting or transposing the feedline moves the phase around the wheel 180 degrees. This changes the 250-degree lag to a 70-degree lag.

In the lazy H with transposed feedline, a 180-degree long feedline would move to the opposite side of the wheel, and the net phase shift would be zero degrees.

If we moved frequency so the 180-degree line was now 90-degrees long, the effective phase would be opposite the 90 degree point of the wheel. The actual phase would be a 270-degree lag, which is the same as a 90 degree phase lead.

This is why a transposed feed is a narrow bandwidth feed for a broadside array.

There are three major problems in this first design example:

  1. 1.) The feedline is too long. Open wire line has a velocity factor of less than unity, so spacing must be less than 1/2 wave.
  2. 2.) This is a single band antenna design, because phase is only zero degrees (in phase) on frequencies where the feedline is odd multiples of 1/2 wave long.
  3. 3.) Current is not evenly distributed in the array. Current distribution is only proper on frequencies where the elements are exact multiples of 1/2 wave, and when the feedline is nearly lossless and is an exact multiple of 1/2 wave.

This antenna's pattern, with the bottom element 20-feet above ground, is shown in the following models:

lazy H crossed feed pattern ten meterslazy H azimuth crossed transposed feed on ten meters

Remember a dipole over earth at optimum height has about 8.5 dBi gain. To convert dBi to dBd we should subtract 8.5 dB from the dBi value given in the model.

11.92 dBi - 8.5 dB dipole over earth = 3.42 dBd. The above antenna has about 3.4 dBd gain on 28 MHz, which is slightly more gain than an Extended Double Zepp. Now let's see if it is really a multi-band antenna.

Looking at 21 MHz:

transposed feed lazy H antenna pattern 21 Mhz21 MHz 15 meter lazy H azimuth patterns

The transposed feed Lazy-H antenna produces less than one dBd gain on 21 MHz. This is essentially the same gain as a dipole. Because of feed phase errors, this feed system tries to force a high angle lobe.

Going to 14 MHz we have:

lazy H 14 MHz pattern of cross feedLazy H base frequency 20 meters crossed feed azimuth

On 20 meters, this antenna shows negative gain over a dipole. The antenna is about -3 dBd on 14 MHz, or half the radiated power of a regular dipole!

The article also suggested using wider spacing. Let's expand the spacing and look at 28 MHz again:

wide spaced crossed feed lazy H pattern 5/8th wave spcing lazy H antenna gain crossed transposed feedline

Wider spacing without an element length change results in 3.6 dBd gain, about .2 dB gain increase over 1/2 wave spacing. We now have higher angle lobes, a result of incorrect phasing, but we did gain something on 28 MHz with wider spacing.

If we change the same expanded spacing antenna to the EDZ (extended double Zepp) element version suggested in the article we have:

Lazy-H transposed open wire feed gain Lazy-H antenna system pattern and gain

The antenna actually lost gain!!! It is now about 1 dBd gain, not the additional 3 dB proposed in the article.

Why did it lose gain? The phase and current distribution is wrong because of the feed method.

Upper element left current maximum and phase
.65 A 13.4 deg

Upper element center (feedpoint) current and phase
.436 A 27 deg

Upper right element current and phase
.65 A 13.4 deg

Lower element left current maximum and phase
.41 A -3.14 deg

Lower element center (feedpoint) current and phase
.312 A -15.4 deg

Lower element right current maxima and phase
.41 A -3.14 deg

For maximum gain and cleanest pattern, the current maximums must have the same relative phase and equal currents. The lower element has 20 log (.41/.65) = 4 dB less power than the upper element!

Phase error at element current maximums is about 16.5 degrees, but this is for a minimum possible length high velocity factor feedline. A real antenna would likely be even worse.

The commonly-suggested transposed feed arrangement works best with half-wave element spacing or slightly less, there is next to no improvement in performance by making spacing wider. With dimensions shown above, the Lazy-H will only work optimally when the velocity factor of the open wire line is nearly the same as freespace and only then on one band!

This is not a good feed method. It is a feed system to avoid unless we are planning on single band operation and closer element spacings. To ensure the elements are fed in-phase, this feed method requires the open wire line be exactly 1/2 wavelength electrical length and be properly transposed.

What causes the elements to be in-phase, when at first glance they appear to be out-of-phase?

This is probably worth going through again in more detail, because it shows that the simplest feed systems are not often the best feed system.

The upper element is fed through 180 degrees of transmission line and the line is transposed 180 degrees at the upper element. 180 degrees electrical length rotated another 180 degrees by transposing the connection at one end is 0 degrees! The elements are not out-of-phase, they are in-phase. The problem is they are in-phase only on one frequency. It is a simple feed system, but it is a seriously flawed feed system unless we want a single band antenna with less than 1/2 wave element spacing!

A phase problem can occur. The article suggests spacing can be increased to 5/8th wave for additional gain. If the open wire line is made long enough to reach between the elements the upper antenna element will be 45 degrees longer than the ideal 1/2 wave. Both elements will not have equal power unless the open wire line is exactly 1/2 wave long. If the builder uses normal ladder line with normal sag as the open wire, the error becomes even more severe, perhaps 50 degrees electrical length error to the upper element. This is because the velocity factor of the line is closer to .90 than to 1.0, so the ladder line we commonly use line is electrically longer than the physical length. As a matter of fact, even real open wire line with bare copper conductors and ceramic insulators is electrically longer than the physical length! When we count feedline sag we are lucky we can run the elements 17 feet apart using the feed system above.

Let's go through a few curtain arrays to see how they evolve. Let's learn how to avoid feed system problems so we can use different spacings and use the antenna on multiple bands.

Good Feed System Lazy-H

The feed system below allows any element spacing and the use of any velocity factor balanced line. It also allows multi-band use.

Lazy H antenna best design feed method distributed feed

If the array is expanded, additional cells like this are built. These cells are connected in groups of cells, and eventually in layers of branches containing cells, with branch or splitting points fed through equal length feeders from common matching points.

This feed method is the most stable in weather and feeds the elements in phase regardless of line length, velocity factor, frequency, or impedance so long as the two equal length lines are the same type.

This array can also be oriented so the elements are vertical. Years ago I had just such a vertically polarized array for 20, 15 and 10 meters. It was one of my better working curtains. It was much better than my bobtail curtains, and performed better than a Bruce array in on-the-air comparisons. It also worked three bands, not just one!

Lazy H Antenna and Distributed-feed Curtain Array

The Lazy H is actually two stacked dipoles fed in-phase. The gain varies with dipole length and spacing, but through proper feed techniques and antenna size the array can be made to operate with gain and good pattern over nearly a 2.5:1 frequency range.

Let's look at a 3.5 MHz to 7 MHz Lazy H. (This could of course be scaled to 28 MHz, so we can consider 7 MHz as 28 MHz and 3.5 as 14 MHz with proper size scaling.)

First we would set the element length and spacing for the highest frequency. We do this by making each element an extended double Zepp, or a 1.3 WL long center fed doublet.

For 7.1 MHz we would have:

984/7.1 = 138.6 feet for one wavelength

138.6 * 1.3 wavelengths = 180.2 feet doublet length

This length is the approximate maximum collinear length. The actual optimum length will be slightly shorter. Now we know the absolute maximum length elements can be.

To find optimum spacing we can look at the graph on my Broadside and Collinear gain page.

broadside element gain graph

Optimum gain for two broadside elements occurs near a spacing of .65 to .7 wavelengths. We want to use the maximum broadside spacing possible at the highest planned frequency before gain falls off to ensure the best low frequency performance. This will also be the minimum height of the lowest element above ground if we want near-optimum gain on the highest band. .7 wavelength * 138.6 feet per wavelength on 40 meters = 97 feet minimum height for lowest element on 40 meters.

This means we need two "dipole" antennas up to 180 feet long stacked up to 97 feet apart with the bottom element 97 feet above earth for a maximum frequency of 40 meters. (This is, through no accident, as close as I could get to that height and spacing for my stacked 40 meter beams.)

proper feed lazy h picture drawing

1 is the upper "dipole" element. 2 is the lower element. Each are 180 feet long and fed with 450 ohm lines in phase at point 3, the feedpoint. Feedline lengths are not critical because this is a distributed feed from one central common point.

Modeling the Lazy H with an element length of 180 feet and element heights of 97 and 194 feet, we have the following patterns:

Lazy H antenna elevation pattern distributed feedLazy H antenna best feed method pattern

The antenna above produces an overall gain of 14.6 dBi, or 14.6 - 8.5 = 6.1 dBd This antenna has 6.1 dB over a dipole at optimum height. This compares to only 3.4 dBd gain for the twisted feed method lazy-H shown at the start of this article. With only a feed method change and a slight increase in spacing, gain has almost doubled!

At 5.4 MHz we have:

lazy H antenna gain lower frequency bandsLazy H antenna for 5.4 MHz band

13.44 - 8.5 = 4.94 dBd almost 5 dB over a dipole at optimum height. We now have improved gain-bandwidth significantly!!

Now let's change the frequency to 3.5 MHz, with no changes in the antenna:

Lazy H antenna on base band lazyH antenna pattern 80 meter band small antenna

We have 9.21 dBi, or 0.71 dB gain over a dipole at optimum height. This isn't much gain, but at least it is not significant negative gain like the twisted feed antenna produces at half-frequency!!

By using two equal-length open wire feedlines to a common junction, and feeding that junction with another open wire line to a tuner or matching system, we can have a gain antenna over a 2:1 frequency range (and beyond). This is the basic principle of distributed feed broadside arrays. You cannot do this with a transposed feedline or a series feed system. It just will not work if you want gain or good performance on multiple bands. You cannot do this with a Sterba curtain either, since a Sterba is a narrow bandwidth antenna when we consider gain and pattern.

Bi-Square Array

The bisquare array is actually a single support lazy-H antenna. This makes the bisquare antenna very easy to install, requiring only a single center support. Instead of having the elements perfectly horizontal like most collinear/broadside antennas, the bisquare has V shaped elements. While a series feed is often shown, the bi square antenna can actually be fed either with the superior broadband distributed feed system, or the easier but very narrow gain-bandwidth series feed system. The standard narrow bandwidth series feed does not use a vertical balanced transmission line like the lazy-H antenna. The more popular narrow bandwidth series-feed system uses the elements themselves as the transmission line. While this eliminates some open wire feedline, it does not result in maximum performance.

bi-square antenna open loop W8Ji feed distributed feedThe antenna to the left is a distributed feed or W8JI bi-square. This antenna will have a broadside pattern with equal or better-than-dipole performance on frequencies where each one of four sides of the diamond making up the antenna are between 1/4 and 5/8 wave long.

This antenna does not quite have the gain claimed in many books and articles. Those articles assume Lazy-H caliber performance, but the bisquare doesn't quite match performance of a Lazy H. The bisquare is a folded in Lazy-H. Folding the lazy H in allows much easier mechanical construction, but the folding narrows array width (collinear spacing) and element stacking spacing (broadside spacing) substantially.

For example, on the band where each side is 1/2 wave long a lazy-H antenna would have 1/2 wave stacking distance and 1/2 wave collinear distance between current maximums. When we fold the open ends of the lazy-H together to form a bi-square the stacking distance is between current maximums is only 127 degrees electrical and the collinear distance between current maxima is 127 degrees. This is the electrical equivalent of a lazy H antenna with 0.7 wave long dipoles for each element (0.35 wave collinear separation between current maximums) and only 0.35 wave stacking distance between the dipoles on the primary band. Shorter stacking spacing of current maximums causes the bisquare antenna to lose most of the stacking gain available in the lazy H, and much of the collinear gain developed in the lazy-H.

This is why the bi-square, even with optimum feed, has about half the gain of a lazy H antenna. This doesn't mean it is a bad antenna, it will work quite well on at least two bands with the distributed feed system. The mechanical simplicity of the bi square's single support often has more value than maximum gain.

conventional bi-square poor standard feed method

The antenna to the left is a more common bi-square antenna feed method. This feed system results in a series-fed antenna. The upper inverted-V element obtains its power from the connection to the outer-end of the lower V element. This antenna only works well when each side (or face) of the diamond's four faces is 1/2 wave long. This results in a total of two wavelengths of antenna wire.

This feed method suffers the same electrical spacing issues at the current maximums as the distributed fed bi-square, except outside of the design band the phase and location of the current maximums is wrong. This current location and phase shift error prevents the antenna from having a good pattern over wide frequency ranges.

If we find a way to electrically short the upper insulated break in the conductors on half-frequency, this bi-square will act like a conventional full wave loop on half-frequency. One way to do this would be to connect an open stub at the top that is 1/2 wave long on the band where the face lengths are each 1/2 wave long. In this case when frequency is moved to half frequency and each side is 1/4 wave long, the upper inverted V element is shorted. This makes the antenna behave like a full wave loop on the half-frequency, and behave as a bi-square on the primary design frequency. The voltages across the open ends can be very high, many kilovolts at high power levels, so this system requires a good relay or stub even at low or medium power levels.

The patterns below compare conventional feed bi squares with distributed feed bi-square antennas.

Bi-square Patterns

Distributed feed 7 MHz bi-square patterns for a 138-foot high distributed feed W8JI style bisquare. This antenna is a "square" 68 feet on each side, for a total of 272 feet of wire. The upper two wires apex is at 138 feet:

bisquare antennabi-square antenna pattern gain

The bi-square above has 10.34 -8.5 = 1.84 dBd gain. Increasing height only has a minimal effect on gain on 40 meters (the higher band). For example, at 200 feet apex height the 40 meter gain only increases 3/4 dB over performance at 130 feet.

The 80 Meter pattern of 40 meter distributed feed bi-square is shown below:

distributed feed bisquare half frequency W8JI feed methodbi-square w8ji feed

The 80 meter gain is about 2 db negative from an optimum height dipole, but increased height will help 80 meters. Let's raise the height to 200 feet apex height......

Distributed feed bi-square on 80 meters, apex 200 feet::

bi-square pattern good W8JI distributed feedbisquare distributed W8JI method of feed

With the antenna at 200 feet apex height, we now have about the same gain and pattern as a dipole at 150 feet

Conventional feed pattern on 80:

bi-square traditional poor conventional feedBi square pattern traditional feed system

Conventional feed distorts the pattern and produces about -4 dBd gain on 80, making this a lossy antenna on 80.

With increased height 40 meter gain and pattern of distributed feed has changed to this:

Bi-square pattern W8JI feed methodBisquare antenna

We now have about 2.6 dBd gain on 40 meters, up about .75 db from the 138 foot high bi square.

Looking at the same antenna with conventional feed we have:

Bisquare quad antennaBisquare antenna patterns

We have identical 40 meter gain, about 2.6 dBd gain on 40 meters with conventional feed.

On 30 meters with distributed feed we have the following pattern:

Bisquare quad antennaBisquare system operated above cutoff frequency

Here we have 9.5-8.5 = 1 dBd gain on 30 meters with distributed feed.

The azimuth pattern is fairly clean, but starting to show a cloverleaf pattern from the elements being slightly too long.

With conventional feed we have the following 30 meter pattern:

Bisquare single feedline bi square traditional style above cutoff

We now have about .3 dBd gain on 30 meters with conventional feed, a loss of .7dB peak gain but the gain it has is mostly at a very high wave angle.

The bi-square antenna is obviously a compromise from the Lazy-H array. The peak gain of a bi-square with optimum feed is only about 2.5 dB over a dipole at optimum height, while a Lazy-H would be about 3 dB higher. With distributed feed it can be used on three bands, but it isn't a high gain antenna on any band. While the bi-square has about the same gain as the conventional transposed-feed lazy-H antenna, it is about 3 dB down from a distributed-feed lazy H antenna with optimum spacing.

Collinear or Stacked BiSquare or "BiQuad" Antennas

Bi-square antennas, like Lazy-H antennas, can be expanded into a larger array by using proper expansion methods. We must be mindful that series feed, while simpler in construction and less expensive narrows bandwidth significantly, degrades pattern, and decreases system efficiency.

Base antenna, 40 meter bi-square 70 feet on a side with 140 foot center height:

Bisquare 40 meters

40 meter bi-square

40 meter bi-square elevation plot

Basic monoband antenna has 11.04 dBi gain, or about 2.6 dBd gain.

By adding a second element collinear, we have:

ual collinear bisquade biquad

40 meter biquad

40 meter dual bisquare

We now have 13.1 dBi, up from 11.04 dBi gain.

This is about 2 dB additional gain, or about 4.6 dBd (dB over dipole) gain.

In similar fashion, one antenna can be stacked above the other vertically. This does not quite produce the same gain increase. As a general rule we are better off using the multiple supports to build an H antenna.

Feed Methods Dual-bisquare

There are numerous methods of feeding dual bi-squares or double quad antennas, some of which result in vertical polarization and some of which result in horizontal polarization.

The widest bandwidth systems are distributed branched feed systems, while the most simple would be a series feed near the center.

Lazy H Antennas Transposed Feed and Distributed Feed

The next logical step in the Lazy H, other than adding a reflector, is to expand either the stack height or the stack width. Let's look at width first, but let's also keep in mind a reflector at this point would add over 3 dB gain.

Making the Lazy H two bays wide we have:

Lazy H double array distributed feedpattern VOA style antenna distributed feed curtain HRS, not Sterba

Pattern width becomes narrower, and we have 9.35 dBd gain on 40 meters. This would be a bit more than 3 db higher if we made it unidirectional, or over 12 dBd gain. This would be MORE gain than a pair of 4-element Yagi's stacked at optimum height!

On 80 meters we have:

curtain antenna 80 meter pair lazy H2-bay lazy H antenna 80 meter plot pair broadside

Which is 4 dBd gain on 80 meters. We have the approximate gain of a two-element full size Yagi, except it covers the whole band and well beyond with increasing gain as we increase frequency! Again if we added reflectors, the gain would exceed 7 dBd or be about the same as an optimum 4 element Yagi. One notable difference is the curtain would cover ALL of 80 meters, not a small portion of the band.

The next step could be to narrow elevation pattern by adding an upper element. This is a three high stack two bays wide or an HR 32 array (HRS if steerable).

curtain three high two widecurtain antenna distributed feed 3 high 2 wide

The 40 meter gain is now 10.8 dBd. This is more than a stack of full size three-element Yagi antennas. On 80 we have:

curtain 40 meter 3 high two wide 40 meterscurtain array 40 meter gain 3 high two wide HR 32 curtain

We now have 5.8 dBd on 80 meters, better than nearly all 80 meter Yagi antennas. Again if we add a reflector system, the system would have around 9 dBd gain. This is better than any stack of 80 meter yagi antennas in existence.

The final step was to add a reflector screen 35 feet behind the array; although this screen could have been added anytime in the process to pick up at least 3 dB. This "screen" only needs to contain horizontal wires. Resonant reflector wires, one 35 feet behind each dipole element, would also work. We now have:

curtain 80 meter plot antennaW8JI curtain array azimuth directivity HR32 style

The 40 meter gain is now 13.9 dBd, which is virtually unheard of gain in an amateur station at HF. With 1500 watts, this is like running 37 kilowatts into a high perfect dipole.

On 80 meters we now have:

curtain antenna array elevation plot gain Curtain array azimuth plot

On 80 meters we now have almost 8.5 dBd at the bottom of the band, climbing to 9 dBd at 3.8 MHz. This is about equal to a stack of full size three-element Yagi antennas, except it will work on any frequency from 3 to 7.5 MHz with increasing gain as frequency is increased. This is a properly working distributed feed (USIA/VOA style or HRS style) curtain.

While the operating SWR bandwidth is not as great as with Rhombic Antennas, the curtain antenna can easily have significantly more gain than the very best Rhombic designs. It does this while occupying a tiny fraction of the Rhombic's physical space. Note, this is NOT a Sterba curtain!! VOA did not use Sterba curtains. Look at pictures, and you will see VOA actually used distributed feed arrays. Let's look at why they did NOT use Sterba curtains in the text below.

Sterba Curtain

The Sterba curtain antenna is sometimes misspelled sturba curtain. The Sterba curtain is in the same antenna family as Bruce arrays. They are series fed antennas, where outer elements obtain power from a long series path through all conductors closer to the feedpoint.

(You can read about distributed feed curtains in Jasik's "Antenna Engineering Handbook" as well as here.)

Please note: There is a large difference between distributed feed or branched feed curtain arrays like USIA arrays and Lazy H antennas and narrow or single band curtains like Sterba, Bobtail, and Bruce arrays. The Bobtail isn't even a true curtain array in the class of high gain curtains. The Bobtail is a simple three-element vertical broadside-array with a unique feed system that produces binomial current distribution, but it is still commonly referred to as a curtain.

Many people incorrectly call USIA or distribution / branch-fed curtains "Sterba curtains". Factually there are HUGE differences in performance and construction of broadcast curtain arrays like the USIA arrays used at VOA and Sterba curtain arrays.

Because the Bobtail isn't a true curtain (and doesn't have the potential for significant gain), I won't include them in this discussion. We are actually installing a large USIA style or distributed feed Curtain here at W8JI. You can see the new tower that will eventually support one end of the curtain at this link.

The planned curtain at W8JI will be aimed at Europe, it will have a reflector, and will cover both 80 and 40 meters. The upper element will be 300 feet high! This antenna will have about 18dBi gain on 80 meters, and almost 23dBi gain on 40 meters.

Sterba Curtains

Sterba curtains are modest-gain single-band antennas. They are named after EJ Sterba, who developed a simple curtain for Bell Labs in the 1930's. There are multiple feed arrangements for the Sterba. They provide a very limited gain-bandwidth product and are critical to construct. You can find details of Sterba curtains in William Orr's Radio Handbook.

Let's look at an actual Sterba array so we can understand why Sterbas have narrow bandwidth and limited gain:

Sterba curtain antenna types construction

Let's walk through the current distribution of Ant 1.

  1. The feedline connects to the middle of a lower 1/2 wl "dipole" section. As with any dipole, high voltage appears at the ends.
  2. Vertical sections D are 1/2 wl transmission lines. One terminal is excited by the voltage at the end of the lower dipole.
  3. Section A top is excited by the high voltage on the wire of transmission line D that connects to the dipole. Maximum current in section A is at the bent end where it transitions to vertical sections C.
  4. Vertical Section C has highest current (maximum radiation) at the bends. The current gradually transitions to a high voltage at the large black "dot" in the middle of section C.
  5. The bottom of vertical section C again has maximum current at the bend to lower horizontal section A.
  6. The inner area of lower section A has high voltage, that excites the second conductor of vertical transmission line D.
  7. The upper end of the second conductor of vertical transmission line D voltage feeds the outer ends of the upper middle 1/2 wl dipole.

This poor method of excitation produces three very undesirable circumstances:

  • The overall path through conductors that supplies current to the center upper current maxima is through2-1/2 wavelengths of wire on each side! On very high frequencies the physical length and the series resistance of that wire length might not be significant. On lower frequencies the long physical length means appreciable series resistance is added to the current path. On 20 meters, for example, the upper center half-wave is excited through 160 feet of wire, the entire current path being over 300 feet long. On 80 meters the current path would be through more than 650 feet of conductor length to the current maximum, with a total current loop distance of 1/4 mile! That can produce significant resistive losses in the antenna.

  • The phase of current in the upper 1/2 wl section depends very heavily on the accuracy of wire length in terms of wavelength. There are 2-1/2 360-degree long segments, or 900 degrees conductor distance, in series with the feed to the upper current maxima. Even if we ignore other effects and consider the phase solely dependent on conductor length, an error of 5% in electrical dimensions will result in a phase error of 45 degrees! That means if we could somehow build a perfect small Sterba that is the equivalent of two half waves stacked 1/2 wl over two more half-waves (~5dB maximum gain) on 7.2 MHz, the system would start to lose gain with a move to the CW band of 40 meters! We are fortunate when a Sterba covers a single band. (A six-section Sterba I used was only a few dB over a dipole at the design frequency, and fell equal to the dipole at the extremes of the band.)

  • Collinear and stacking distances are limited by the necessity of maintaining 1/2 wl long transmission lines. Gain is further limited by the requirement of zero distance end-to-end element spacings. This means the antenna sacrifices potential gain by using spacings less than optimum.

The amateur radio fascination with Sterba and Bruce arrays probably stems from confusion. I've noticed most amateurs incorrectly call large distributed-feed or branched-feed curtains used by short wave stations "Sterba" curtains, but they are definitely not!! The Sterba and (if ground conductivity is high enough) Bruce arrays can be good antennas, but they are frequency sensitive and require careful construction.

I can't recall ever seeing an actual Sterba curtain used at any commercial SW BC (shortwave broadcast) station. Distributed or Branch Feed curtains are used, not sterba arrays. These distributed feed curtains are also called HR arrays, with the H and R standing for Height and Rows. When they are steerable, they are sometimes called HRS arrays, the S representing "steerable". An HR 43 would be an antenna 4 elements high and 3 elements wide. If it was an HRS 43, it would be a steerable array of the same element configuration.

Sterba Curtain Model

The following is a model of a center-fed Sterba. This antenna is identical to the one on HamUniverse (link to models of this antenna), except this antenna is properly scaled for 40 meters.

Height above ground of bottom wires 16,12, 3, 7 = 67 feet height at bottom. 134 feet high at top.

Length of 1/2 wave wires 3, 4, 6, 8, 9, 12, 13, 15, 17, 18 = 67 feet

Length of 1/4 wave wires 2, 5, 7, 10, 11, 14, 16, 19 = 33.5 feet

Fed in center

model EZnec sterba curtain array

Pattern of correctly dimensioned 40-meter Sterba curtain antenna using REAL open wire line for vertical sections (not window ladder line).

sterba curtain gain elevation plotsterba curtain antenna patterns

Source impedance on 7.1 MHz is 69.92 - J 8.59 ohms
SWR (50 ohm system) = 1.440 (75 ohm system) = 1.148

Gain is 12 dBi or 3.6 dBd. This large complex antenna has 3.6 dB gain over a dipole on the design band at optimum height. This is actually about the gain of a simple double extended Zepp antenna.

Now let's see what happens to gain and pattern on half frequency, which in this example is 3.5 MHz:

Sterba curtain gain elevation base frequencySterba curtain array picture base frequency

This antenna now has 6.8 dBi gain, or about -1.72 dBd gain. It has almost 2 dB negative gain over a dipole. This is quite different than a properly fed Lazy-H or other distributed feed curtains!

Feed impedance on 80 meters is 8.166 - J 35.29 ohms

SWR (50 ohm system) = 9.229 (75 ohm system) = 11.238

On double the design frequency, in this example 14 MHz, it has the following pattern:

Sterba curtain pattern 2nd harmonic

This is a very poor multiband antenna. It has about the same gain as an extended double Zepp on the design frequency, the gain of a dipole on the second harmonic, and about -1.7 dB negative gain over a dipole on half-frequency. While this antenna can be loaded on multiple bands it is NOT a multi-band curtain.

Sterba Gain vs. Ground System

Claims have been made a ground system will increase gain 3-6 dB with a Sterba curtain. This is not true.

The only possible change a ground can make to a balanced antenna, such as a dipole or array of doublets making up a curtain, is to reduce induced or coupled losses in earth by "shielding" the earth from the intense fields of the antenna. It is true that a very low dipole (low means a small fraction of a wavelength) can benefit greatly by the addition of a grid of wires parallel with the antenna below the length of the antenna. For example a dipole at .05 wavelengths above ground could show several dB increase in field strength when a ground screen or system of wires are laid parallel with the antenna element. The effect is caused by the reduction of current in the lossy soil. The overall efficiency of the antenna can increase so the pattern has increased intensity at all angles and directions.

This effect is significant only for very low antennas. It does not apply to a curtain at any reasonable height. If the antenna is so close to earth as to benefit significantly from a ground screen, then it would be definition be a very poor array to start with. Let's look at the curtain first:

Sterba curtain ground gainSterba Curtain gain from counterpoise

Average soil pattern of curtain Perfect zero loss ground pattern of same antenna

Net gain difference 0.53 dB by adding a perfect zero loss ground in all directions for infinite distance. That isn't much change!

sterba curtain gain dipole ground reflection reference dipoledipole gain low dipole

Dipole at 1/10th wave high over medium dirt Dipole at 1/10th wave high over perfect lossless groundplane

Low dipoles can have a significant change, in some cases over 6 dB, with the addition of a large reflective ground screen below the antenna. There is almost no change in antenna pattern, but the antenna efficiency increases a great deal.

Bruce Array

The Bruce array is a vertically-polarized single-band curtain antenna. The primary advantage of the Bruce array is the low height requirement. The array elements being only 1/4 wavelength tall, 3/8 wave tall supports will often provide enough overall height. The longer the array, the more narrow beamwidth and bandwidth becomes, and as typical in series-fed arrays system efficiency decreases with longer antenna lengths. As with all vertical antenna near earth, the Bruce array is seriously impacted by soil conductivity in the Fresnel region (area out a few wavelengths from the array where pattern is formed).

Some general guidelines for Bruce array performance are:

Ground conductivity is critical for gain. Like all vertically polarized arrays close to earth, Fresnel region losses have a major effect on pattern and gain

While the array can be fed in multiple places, Bruce array construction is less critical when the feedpoint is closest to the array center

This is a series-fed array. As such, conductor losses and earth losses below the array reduce current available at points distance from the feedpoint

As with all series fed arrays, bandwidth is progressively more restricted as the antenna is made longer and pattern is more distorted from errors in element length

Bruce array over perfect groundBruce antenna array elevation over perfect ground

layout 4 element bruce array

In this array, wires 1 and 9 are 1/8th wave long counting end-effect. This slightly shortens wires 1 and 9.

All other wires are 1/4 wave long, assuming bare wire.

This model includes copper losses.

Gain over a reference dipole is about 3 dB, although wave angle is significantly lower. This would be representative of a Bruce array over saltwater.

Bruce Array over real medium earth

Bruce array over real earth

Bruce array gain over real earth elevation plot

In this case the Bruce array has about -3 dBd gain. This is why I had a difficult time justifying a 75 meter Bruce array I installed back in the early 1970's. My Inverted Vee dipole at 130 feet was about even with the Bruce array into Europe night after night on 75 and 80 meters. If you have high ground conductivity can install the antenna very high above earth, this antenna shows some merit.

With a good counterpoise system below the antenna, it is possible to pick up a few db regardless of soil type. A good counterpoise system can increase efficiency from 25 percent up to about 40%.

Summary, USIA or Distributed/Branched feed-system Curtains

Distributed feed curtains use a series of common points, each fed from equal length low loss transmission lines, to distribute power. Conductor loss is less, phase error is significantly reduced, and all elements receive equal currents.

This feed method places conductor resistances in parallel, and makes array patterns stable over very wide frequency excursions. In addition to having more gain, a distributed feed curtain (such as USIA arrays used at VOA sites) can be used over a 2:1 or broader frequency range with minimal gain and pattern change. It is very easy to make a distributed feed curtain operate on 80 and 40 meters with full gain and no pattern distortion.

A distributed or branched feed curtain also allows designers to use optimum element spacing, both in collinear and broadside (stacking) distances. This means a 4-element branch fed curtain can provide the highest gain per acre of any antenna design.

  • The stacking height compresses the signal in elevation. This is the vertical count of layers.
  • The number of columns, bays, or array width compresses the signal in azimuth.

Despite occupying a tiny fraction of the space required for their rhombics, the USIA style curtains are the highest gain arrays used at VOA's International Broadcast and the largely defunct VOA relay sites.

Note: The Lazy H, when center fed, is a distributed-feed or branched-feed curtain.

The antennas below are similar to my planned 80/40 meter antenna. Since I am not running 100 KW AM transmitters, I'm not worried about element voltages or open wire line arcing. I'll use single wire elements instead of wire cages and operate low loss open wire lines with standing waves. This simplification is acceptable for amateur use because voltages and currents are much lower with 1500 watts CW or SSB rather than 100 kW or more carrier power (400 kW PEP on AM) used by VOA.

My curtain will have the upper element at 300 feet, and the lowest element at 100 feet. It will be three layers high and three sections or "rows" wide. Another name for the array, in common SWBC descriptions, would be an HRS 3/3/1 array. This is a Height Row Steerable 3/3/1 array. My variation models just over 18 dBi on 80 and over 23 dBi on 40 meters, with ability to steer the beam in azimuth and elevation.

I've had similar scaled down distributed feed curtains for 20 meters and up in the past, and they worked quite well. Unfortunately I don't like 20 meters and up, so I only used them as a brief experiment lasting a few months.

By the way, curtains like this have considerably more gain than Rhombic antennas. The curtain would also occupy a tiny fraction of the space required by a Rhombic producing significantly less gain. For example, my planned 80/40 curtain is only 350 feet wide and 300 feet high yet has almost 15 dB gain over a dipole on 40 meters. A 1500 watt transmitter into the curtain will produce the equivalent of 37 kW to an optimum height dipole.

Of course this all hinges on getting new 300-foot Rohn 55G and 320-foot Rohn 65G towers installed, which is actually more than 50% complete now.

My Curtain Array

Large curtain antenna array

This is an HRS 331 array. There are three bays high, three bays wide, and the system is steerable in elevation and azimuth. The upper element is at 300 feet, and the array is about 350 feet long with 30 foot spacing to the reflectors, allowing simple 30 foot long booms from Rohn 25G to support the elements at the proper spacing.

You might wonder how I can get away with fewer wires in the elements, and reflectors instead of large complex screens. The multiple conductor elements in a SWBC array limit impedance changes as frequency is varied, and the also keep voltages low. This is very necessary when transmitters run ten's of kilowatts on AM.

  • PEP without negative peak limiting is typically four times carrier power on AM. A good 25 kW AM transmitter has at least 100 kW PEP output. The large cages are used to reduce corona and arcing in the antenna because they reduce the intensity of the electric field around the conductors.

  • The complex screen reflector is because multiple very close spaced bands are used, rendering tuned reflectors impossible. The only solution is a multiple conductor grid or screen.

For Amateur use, neither of the above is problematic. Power levels are low and the system operates on two or three fairly wide spaced bands. This allows the use of resonant reflectors.

Here are a few pages my planned system is derived from:

Basic curtain with reflector. In amateur applications it is possible to use a system like this over a 2.5:1 or wider frequency range! The system can also be scaled down to use fewer elements, and the reflector can be omitted. You can see the tower being installed that will support one end of this curtain on my Rohn 65G page.

High frequency arrays curtain antenna

Curatin array pictures

Beam slewing HRS array

curtain array schematic

typical SW BC curtain pattern NOT a Sterba curtain

No comments: