Antennas for Communications (AFC) Logo

Radome Transmission Loss
and Antenna Pattern Degradation

grn_ball.gif - 969 BytesIntroduction

HIT_radome&dish1.jpg - 5616 BytesRadomes are composed of panels, which when assembled form a truncated spherical shell to protect the enclosed antenna from the environment. Each radome panel is surrounded by a flange perimeter enabling adjacent panel assembly. After assembly, the panel flange perimeter members form a framework characteristic of the panel shapes. Both the radome shell wall and panel framework are in the path of the shielded antenna. Radome transmission loss is the sum of the ordinary insertion loss of the antenna (radar) signal passing through the radome wall plus the scattering loss off the radome panel framework blocking (shadowing) the antenna aperture. From an antenna point of view, just as the antenna feed assembly, feed support struts or cassegrain subreflector block the antenna aperture, in the same fashion so does the radome framework that shadows the antenna aperture. While radome wall insertion loss is typically less than 0.1 dB, surprisingly, the scattering loss off the framework is 4 to 100 times larger than the wall insertion loss. Therefore, one must take into account the radome panel framework scattering loss in order to understand radome performance for protected antennas or radar.

In general, a larger scattering loss is encountered for a longer length framework shadowing the antenna aperture. For example, if the framework shadowing length is twice another radome, so is the radome scattering loss (transmission loss). What happens to this scattered energy is that it spreads out into the antenna sidelobes. The antenna pattern is corrupted by the scattering loss. Since sidelobes have small energy, the larger the radome scattering loss, the larger the effect on antenna sidelobes. The precise details of where this energy goes is dependent on the shadow length and radome transmission loss, the electronic signature of the framework and the positions and orientations of the framework members shadowing the reflector. For the most part, the scattered energy is directed into the far out sidelobes where the antenna sidelobe pattern is very weak. Here the antenna sidelobes may be affected from framework scattering by 6 to 12 dB. In contrast, where the antenna pattern is strong, the first several sidelobes are affected by only 1 dB.

This result should not surprise you. Antenna engineers have long since learned to orient feed support struts at 45-degree angles relative to the principle horizontal, vertical planes. Then when azimuth and elevation antennas patterns are measured, these measurement pattern cuts do not show the effects of feed support strut scattering. Nevertheless, the antenna pattern is still degraded by scattering from the feed support struts, but mostly in the not measured 45-degree pattern directions.

red_ball.gif - 950 BytesRadome Geometry and Framework Shadowing

Radome diameters 6.7m (22ft).
ornagepeel-geometry.gif - 1617 Bytes quasi-geometry.gif - 1889 Bytes
Figure 1. Orange peel
radome Geometry.
Figure 2. Quasi-random
radome geometry.

Consider a perfect radome (truncated sphere) without any panels. If we cut the radome vertically in half, we now have 2 panels and one framework seam member. As the radome protected antenna scans in azimuth, the framework shadows the antenna aperture starting from one edge, continuing over the antenna and then passing out the other side. The framework blockage length depends on the antenna scan angle and may vary from 0 up to the antenna diameter length when the framework member just bisects the reflector diameter.

If we now vertically cut this radome in half once again, we now have 4 orange peel shaped radome panels. Again as the antenna scans in azimuth, the framework shadows the antenna aperture. As the antenna diameter gets larger, it is now possible to have 2 framework members pass in front of the antenna aperture ---- one framework member on one edge of the dish reflector and the other framework member on the opposite reflector edge.

With actual radomes, framework shadowing is more complicated and depends on the radome geometry. Radome geometry is a term used to describe how the truncated sphere is separated into panel shapes. The mathematical process is known as tessellating the sphere. From a practical point of view, the radome must be packaged for shipment anywhere worldwide. What this means is that the maximum panel size must conform to standard shipping sizes. Taking into account shipping size constraints, two common radome geometry types are the symmetric orange peel and quasi-random geometry shown in Figures 1 and 2 respectively. Quasi-random geometry radomes may have triangular, hexagonal or pentagonal panel shapes. A geodesic radome using triangular panels is an alternate implementation of the quasi-random radome geometry.
Orange PeelQuasi-random
orangepeel-proj.gif - 1809 Bytesquasirandom-proj.gif - 3097 Bytes
Figure 3. Radome framework shadow pictures for 6.7m
(22ft) diameter radome with orange peel shadow (left) and
quasi-random shadow (right). Antenna diameter is 4.3m (14ft).

Considering the radome geometry of Figures 1 and 2, what we see is that the framework shadow on the reflector surface is a complicated geometry problem dependent on the radome diameter, panel size, the antenna diameter and antenna scan angle. Figure 3 shows the radome framework shadows for both an orange peel and quasi-random 6.7m (22ft) diameter radome geometry protecting a 4.3m (14ft) diameter reflector antenna. To keep scattering loss low, normal practice is to use large radome panels. Therefore as the radome diameter gets smaller, the radome is manufactured from fewer panels.

jointlength.gif - 5885 Bytes
Figure 4. Radome framework shadow length comparison
for the orange peel and quasi-random 6.7m (22ft) diameter
radome shown in Figure 1,2 and 3 above.
So what does one mean by a radome with quasi-random radome panels. From a geometry point of view, one refers to a radome with several different panel shapes. From an electromagnetic point of view, the framework shadow must be quasi-random. Clearly if the quasi-random radome panels are large relative to the antenna diameter, only parts of the panels shadow the reflector at any one time. Therefore for small radomes (6.7m and smaller) with large radome panels, there is no such thing as a quasi-random radome shadow and the concept has no meaning.

On the other hand, we can purposely make the panels smaller where the framework shadowing would mimic the quasi-random nature of the radome (compare Figure 2 and 3). Such a quasi-random radome with smaller panels would have a larger transmission loss due to the longer length of framework shadow members. Figure 4 graphically shows, as a function of scan angle, the extra shadow length for the quasi-random shadow compared to the orange peel shadow. The quasi-random shadow contributes 44 percent more blockage than the symmetric orange peel geometry radome.

We wish to point out to the reader that the extra shadow length problem associated with the quasi-random geometry is a property of smaller diameter radomes where the radome panel size, antenna diameter and radome diameter have similar dimensions. For larger radomes, 8m (26ft) diameter and above, panel sizes are small relative to the antenna diameter. Here, the shadow length geometry differences becomes insignificant. We will see later that the for large radomes, the quasi-random geometry is fundamental to achieving enhanced RF performance.

grn_ball.gif - 969 BytesTransmission Loss and Framework Electronic Signature

W.V.T Rusch, A.F. Kay and other investigators have shown that the framework scattering loss is approximately:

tl.gif - 1471 Bytes

where TL is the radome transmission loss (scattering loss in dB units), l is the total length of the framework shadow, w is the framework width, A is the antenna aperture area and IFR is the framework Induced Field Ratio. The w*IFR product is known as the framework electronic signature. The framework electronic signature may be measured in an anechoic chamber or calculated using Method of Moments electromagnetic simulation techniques. w*IFR measurements yield two complex numbers (amplitude and phase); one where the electric field polarization is oriented parallel to the framework; and the second when the electric field polarization is perpendicular. Electronic signature w*IFR product amplitude is a measure of the effective electronic shadow width. Its value depends on:
      • the physical dimensions of the flange framework.
      • the dielectric or inductive framework properties (dielectric or metal framework material).
      • the details and peculiarities of the framework design shape.
      • the antenna or radar frequency.
      • the orientation of the framework member relative to the antenna aperture.
      • the antenna polarization.
Note that the ratio w*l/A is the percentage framework shadow area blocking the antenna aperture area. We also wish to point out that the framework width w has structural properties consistent with the maximum rated radome wind speed specification. To enhance RF performance, a balancing act takes place between a stronger structure with heavy-duty sized members (making w larger) and RF performance. This RF optimization process often determines that radomes are designed with minimal structural safety factors. It is for this reason that AFC has defined radome structural safety by a criterion based on the catastrophic failure wind speed. We refer the reader to the AFC’s
Structural Analysis web page.

When structural requirements force w to exceed RF electromagnetic specification limits, two approaches are available to reduce transmission loss and electromagnetic degradation. The first method appeals to the design shape of the framework member. The second method appeals to a process known as impedance matching where the framework electronic signature w*IFR is reduced in value. Both methods are described below.

red_ball.gif - 950 BytesRadome Framework Design Shape

There are 2 flange framework forms common to the radome industry. The first type is called a perpendicular joint. The second type is called a parallel lap joint. Figures 5 and 6 picture both flange framework forms.

perpendicularjoint.gif - 1375 Bytes lapjoint.gif - 1542 Bytes
Figure 5. Perpendicular joint's framework
fastener hardware is internal to the radome.
Figure 6. Parallel lap joint framework has
fastener hardware external to the radome.

sop-vs-qr.gif - 4722 Bytes
Figure 7. Transmission loss comparison for the
6.7m (22ft) diameter symmetric orange peel and
quasi-random geometry radome.
Geodesic dielectric or metal space frame radomes typically use the perpendicular joint shape. Note from Figure 5 that the hardware for the perpendicular joint is internal to the radome. In contrast from Figure 6, the parallel lap joint hardware punctures the radome surface and is both external and internal to the radome. Such parallel lap joint hardware external protrusions collect dirt, grime and fungus and allows corrosion to attack metals exposed to the outside environment. From a framework shadow width point of view, the perpendicular joint has a very narrow cross section. This contrasts significantly with the large lap joint cross section shadowing the dish reflector. Due to its smaller width, the perpendicular joint has a scattering width 8 times smaller than its parallel lap joint counterpart.

A radome transmission loss (scattering loss) comparison for the symmetric orange peel and quasi-random radome is shown in Figure 7. To help with the contrast, the perpendicular joint is used with the orange peel geometry radome and the parallel lap joint with the quasi random geometry radome. Figure 7 shows that the transmission loss and framework electronic signatures w*IFR increase from 0 at low frequencies to some maximum value dependant on the detailed design and dielectric joint properties. Clearly, the framework electronic signature w*IFR has fine structure that contributes to the complex nature of the transmission loss curves. In summary, the superior transmission loss performance of the symmetric orange peel random over the quasi-random unit is both a property of the shorter shadow length and smaller electronic signature.

grn_ball.gif - 969 BytesRadome Framework Impedance Matching

tl-im.gif - 5072 Bytes
Figure 8. Reducing radome transmission
loss by impedance matching radome
framework tuned for C-band.
One method of reducing radome transmission loss (scattering loss) is to reduce the framework electronic signature w*IFR. Clearly from the transmission loss equation above, if the magnitude of the electronic signature decreases by a factor of 2, so does the radome transmission loss. To exploit this electronic signature property, AFC engineers appeal to a process known as "impedance matching." For a particular design frequency, by adding a properly designed inductive circuit to a dielectric framework, an impedance match results tuning out the capacitive reactance of the joint dielectric framework. With an impedance match, the framework no longer scatters energy. In effect, the framework disappears (becomes stealthy), reducing transmission loss and thereby removing the scattered energy degradation from the antenna sidelobe pattern. Concentrating on the symmetric orange peel radome, Figure 8 shows the radome transmission loss with a framework impedance matching network tuned for C-band. Note that impedance matching decreases C-band transmission loss 5-fold from 0.5 dB to 0.1 dB.

Clearly for a metal space frame radome, impedance matching is impossible. With a metal space frame radome, there is no approach for improving the framework electronic signature.

red_ball.gif - 950 BytesRadome Noise Temperature and Antenna System dG/T Degradation

Radome noise temperature is the sum of the power absorption in the radome wall, noise reflection off the radome wall and noise scattering off the radome panel framework. In a similar process to radome transmission loss, noise from the warm earth scatters off the radome framework into the antenna feed system with a magnitude several times greater than the other two mechanisms. Noise temperature NT and dG/T is therefore proportional to radome transmission loss or electronic signature w*IFR as shown in the equation below, where Tsys is the antenna system noise temperature before the radome is installed.

Radome Noise Temperature Equation

G-T-degradation.gif - 2336 Bytes

From the noise temperature equation, impedance matching the framework electronic signature w*IFR not only decreases radome transmission loss, but also simultaneously reduces noise temperature and dG/T degradation. This feature of impedance matched "tuned" framework radomes provides more than a 3-fold performance enhancement for receiving systems as shown in Figure 9 and 10. Figure 9 compares the impedance matched framework radome to the standard radome. Figure 10 shows the significant improvement in signal-to-noise ratio (dG/T degradation) by impedance matching the framework.

nt.gif - 4872 Bytesgt.gif - 4996 Bytes
Figure 9. Radome noise temperature comparison between standard and impedance matched framework radome.Figure 10. Antenna system dG/T degradation for impedance matched and standard radome.

grn_ball.gif - 969 BytesAntenna Pattern Degradation

How the framework shadow members disturb the antenna pattern is a complex function of panel shape (framework orientation), their length and electromagnetic properties. Referring to Figure 3, the framework shadow forms line segment members. Labeling a member length li, the total shadow length l (from the transmission loss equation above) is the sum of the member lengths l = S li. Framework members each have an element pattern that scatter into the sidelobes in preferred directions according to their length, orientation and electronic signature. By keeping track of all the framework member element patterns shadowing the antenna, one has a linear array antenna. Each element of the scattering array has an element gain given by w*IFR. This linear array scattering antenna is characterized by its electric field, Es with scatter pattern proportional to |Es|2. Adding the scattering array pattern to the antenna pattern electric field, Ea, we arrive at the radome-enclosed antenna pattern, |Et|2, where Et is given by:

Et = Ea + Es

Es.gif - 1556 Bytes

where N is the number of framework members shadowing the antenna aperture A, ai is the antenna aperture field pattern on the ith framework member, g(q,f)i is the ith framework member scattering element pattern and
k*ri = k(q,f)*ri
is phase distance from the antenna aperture to the ith framework member in the field sampling direction q,f.

Antenna engineers view the scattering array pattern |Es|2 as degradation and therefore seek radome solutions where the scattering array pattern and radome transmission loss are very small. Over the main beam and first several antenna sidelobes, where the antenna power is strong, |Ea| > |Es| and radome scattering has little affect on the total pattern. In contrast for the far out sidelobes, where the antenna pattern is very weak, |Es| > |Ea| and the total radome-enclosed sidelobe level pattern is limited by the scatter pattern |Et|2 @ |Es|2. In between values, where that antenna sidelobe level and scatter pattern have equal amplitude, |Ea| = |Es| and the total radome-enclosed pattern |Et|2 is modified by +/- 6 dB. Figure 11 compares the framework scattering array pattern for both the smaller symmetric orange peel radome and quasi-random radome for the shadow blockage appearing in Figure 3.

Orange PeelQuasi-random
scatter-pattern.gif - 9045 Bytes
Figure 11. Scatter array pattern comparison for orange peel
and quasi-random radome geometry radome.

The scatter pattern figures are the standard elevation and azimuth cuts common to the antenna industry. Comparison of the scattered field over the +/- 4 degree range shows that the quasi-random scattered field has a 12 dB higher amplitude than the orange peel design. What this means is that the quasi-random radome will start degrading antenna sidelobes at the -22 dB level compared to the orange peel radome at -35 dB level. Both radomes types have higher azimuth scattered energy than elevation cuts. We will see in the next section that this focusing of scattered energy is a useful property to be exploited in radome geometry panel design. For example with a radar antenna designed to minimize ground clutter returns, the orange peel scatter pattern shows (compared to the quasi-random radome) significantly smaller effect on antenna sidelobes pointed toward the ground (elevation pattern).

In contrast to small radomes, for large radomes, where numerous framework members shadow the antenna aperture, quasi-random shadow blockage provides superior antenna performance over the symmetric radome counterpart. When the radome geometry shadow pattern is symmetric, as with large orange peel radomes, the scatter pattern equation, above, may simplified by bringing the framework element pattern, framework electronic signature and framework member length outside the summation sign:

Es-simplified.gif - 1415 Bytes

Now for the symmetric radome, the scatter pattern, Es, resembles the standard linear array pattern from antenna theory. When the spacing between elements is greater than l/2, as is the case for radome panel framework scattering elements, the appearance of secondary beam peaks are introduced into the scatter pattern. These secondary beam peaks are called grating lobes. Arrays with element spacing greater than l always have grating lobes (multiple main beams), but the scattering element pattern or antenna aperture field pattern may reduce the grating lobes amplitude. The next set of figures (Figure 12) shows the grating lobes for a symmetric 10.7m (35ft) diameter radome in comparison to the quasi-random radome. To the detriment of satcom and radar applications, these symmetric radome grating lobes significantly degrade antenna sidelobe performance to that of the radome.

Quasi-random Symmetric Orange Peel
quasi-radome-p1.gif - 14789 Bytes sym-radome-p2.gif - 14217 Bytes
quasi-p1.gif - 5217 Bytes sym-p2.gif - 5017 Bytes
quasi-p3.gif - 1975 Bytes
Figure 12. Scattering pattern for 10.7m (35ft) diameter radome with quasi-random and symmetric orange peel geometry. Note the grating lobes for the symmetric radome. Antenna is the ASR-9 radar operating in S-band.

red_ball.gif - 950 BytesCompliance of Satellite Earth Station Antenna Patterns

Stringent sidelobe level (SLL) envelope requirements apply generally in SatCom applications that employ satellites in geosynchronous orbital slots. The paramount operational requirement is that the earth station antenna (ESA) neither illuminates via sidelobes other satellites in the geosynchronous plane nor receives signals, via sidelobes, from such satellites. Compliance with the SLL envelope requirements assures that such crosstalk is at an acceptable level. For satellite earth station transmit antenna patterns, the key issue is SLL compliance with FCC Part 25.209 regulations (as well as similar requirements for Intelsat IESS-207 and IESS-601, Eutelsat EESS 500, Asiasat and PanAmSat) known as:

29-25*log(Q )

We wish to point out that the enclosure of the antenna in a fixed panelized radome introduces a critical and novel distinction between operational system performance and performance validation by certification tests. The reason is that when the antenna is operational, both the antenna and radome are stationary with the antenna pointed at the satellite. In contrast, when the antenna is undergoing certification tests, the radome and satellite are at rest with the antenna moving to construct the sidelobe pattern. Under such circumstance, the antenna aperture blockage afforded by the radome framework continuously changes as the antenna sweeps over the sky. One therefore must make sure that both the operational antenna/radome sidelobe pattern and the antenna/radome under test both meet the intended specifications.

We also note that in general there is a reduced FCC operational requirement to limit SLL outside the geosynchronous plane 32-25*log(Q ). Such a reduced requirement is measured by an elevation antenna pattern cut. This comment is of practical significance for center fed antennas. In FCC/INTELSAT applications, 4 struts that run diagonally to the outer portion of the dish typically support the feed/subreflector. The scattering by these struts, or by any long scatterer illuminated by the antenna, is in the plane transverse to the axis of the strut. The strut scattering degrades the pattern in diagonal planes. In consequence, such FCC/INTELSAT qualified antennas often do not generally comply with the SLL envelope in diagonal planes. It is of tremendous practical import that the strut scattering does not degrade the pattern in either principal plane. Such antennas comply with SLL requirements in both azimuth and elevation cuts. This allows elevation cuts to be used as validation pattern measurements in transmission using a cooperating satellite. With some limitations, the like measurement cannot completely be performed on azimuth cuts as it involves sweeping the antenna main beam along the geosynchronous arc with unacceptable illumination of nearby geosynchronous satellites. Under such circumstances, antenna symmetry may be assumed for the equivalence of both elevation and azimuth pattern cuts.

Here is where the complication begins. Introducing the radome into the antenna pattern reduces the antenna symmetry to that of the radome framework symmetry. Elevation and azimuth pattern cuts for the antenna/radome combination are now independent. Indeed, it is often the objective of the radome designer to put the scattered energy into the elevation cuts so as to minimize sidelobe error for the azimuth pattern. Even though scattered energy is very small, it has to go somewhere. The problem for the radome designer is that the antenna sidelobe energy is very small as well. Fortunately, the radome designer has two tools at his disposal. The first tool is impedance matching to reduce the framework scattering loss. The second tool is radome geometry to orient the framework members to scatter energy in appropriate directions.

Figures 13 through 16 illustrate the distinction between the measured pattern of the radome-enclosed antenna under test and the actual pattern of the radome/antenna system (the pattern on the sky). The figures are for AFC's 10.4m (31ft) diameter radome protecting a Viasat 4.5m earth station antenna. Operation is at C-band for linear polarized transponders. Figure 13 is the actual azimuth cut pattern, the illumination along the geosynchronous arc when the ESA is boresighted on the primary satellite. Figure 15 is the actual elevation cut pattern. In contradistinction, the measurement of the radome/antenna is made by rotating the antenna inside the radome and observing the pattern at a fixed far field point. Each incremental rotation of the antenna with respect to the radome modifies the set of illuminated framework joints (framework shadow). The cumulative effect is significant. Figures 14 and 16 show the measured pattern. Both measured and sky patterns sets clearly comply with the SLL envelope. Compared to the sky SLL patterns, the measured azimuth pattern has somewhat higher and less distinct sidelobes beyond -9o.

azcut1.gif - 28435 Bytes
Figure 13. The azimuth cut pattern on the sky as radiated by the radome/antenna configuration. The pattern is fully compliant with the SLL envelope.

azcut1.gif - 28435 Bytes
Figure 14. The apparent azimuth cut pattern measured by rotating the antenna in the radome. The measured pattern is compliant with the SLL envelope.

azcut1.gif - 28435 Bytes
Figure 15. The elevation cut pattern on the sky as radiated by the radome/antenna configuration. The pattern is compliant with the SLL envelope.

azcut1.gif - 28435 Bytes
Figure 16. The apparent elevation cut pattern measured by rotating the antenna in the radome. The measured pattern is compliant with the SLL envelope.

red_ball.gif - 950 BytesRadome Design Considerations for Satellite Earth Stations

regpattern.gif - 5199 Bytes
Figure 17. Regulatory antenna pattern requirements
along with radome scatter pattern level as
a function of radome transmission loss.
We must point out that having both elevation and azimuth pattern SLL compliance is usually not the case for radome-enclosed earth station antennas. For a practical radome geometry, the panel configuration, which is optimum for azimuth cuts, is essentially the worst case for elevation cuts; and in most circumstances, one must give up elevation cut compliance to get SLL compliance in the more critical azimuth cut. This is the final consideration particular to radome-enclosed earth station antennas. And in general for a radome-enclosed ESA, one cannot validate azimuth cut performance by elevation cut measurements.

The FCC, Intelsat and other regulatory agencies have yet to come to grips with a set of standards suitable for a radome-enclosed ESA. While radome design criteria directs scattered energy into the elevation plane where SLL interference is a non-issue, compliance regulations preclude such an obvious solution. Another issue relates to ESA site compliance measurements as performed for a regulatory agency and radome scattering into the far out sidelobes. For example with respect to AFC’s 10.4m (31ft) diameter radome protecting Viasat’s 4.5m ESA described in Figures 13 through 16 above, Figure 17 constructs the regulatory 29-25*log(Q ) antenna pattern envelope superimposed over the scatter pattern level. The scatter pattern levels (dBi units) are shown as a function of radome transmission loss. While ESA site compliance measurements are typically limited to scan angles +/- 10-degrees or less, Figure 17 demonstrates that the radome far out scattering pattern level actually causes non-compliance issues at pointing directions not measured during certification tests; namely 12-degrees for 0.5 dB transmission loss and 18-degrees for 0.25 dB transmission loss. On the other hand, due to the limited scan range with cooperating satellite measurements, certification tests would be found compliant. It is only when the radome transmission loss is less than 0.1 dB that strict compliance is recorded for elevation and azimuth pattern measurements.

Over standard communications bandwidth, the dilemma for radome electromagnetic design is that radome technology, through impedance matching and radome panel geometry, yielding near zero (< 0.1 dB) transmission loss is beyond the present state-of-the-art. As pointed out earlier, antenna engineers have the luxury of only 4 feed support struts oriented at 45-degrees angles. Their antenna pattern non-compliance is masked by the specific elevation and azimuth antenna pattern cut directions. In contrast, radome engineers have no such comfort. Numerous radome shadow members, oriented at various angles, preclude such an obvious deception.

afc41ft-vertex8-1f1a.jpg - 10520 Bytes
Figure 18. Construction of AFC's 41-ft diameter radome
shielding a Vertex 8.1m satellite earth station.
As more system engineers and planners use radomes to protect vital ESA communication systems from wind, weather extremes and corrosion, these regulatory agencies will need to develop radome-enclosed SLL standards based on satellite interference principles and azimuth SLL patterns alone. In any case, some means of validating operational system performance needs to be established for radome-enclosed ESA systems.

The following four Adobe Acrobat PDF files illustrate radome-enclosed earth station SLL performance. The azimuth and elevation pattern measurements are for a Vertex 8.1m antenna protected by AFC's 12.6m (41ft) diameter radome as shown in Figure 18. These measurements were conducted to verify Intelsat certification for C-band satellites on circular polarized transponders.

The four pattern measurements are:

The above cooperating Intelsat regulatory pattern measurements show "compliance" for azimuth patterns and "non-compliant" for elevation patterns. For the elevation pattern cut, SLL compliance issues begin at the measured 8-degree scan angle where the SLL pattern is approximately –46 dB down.

AFC manufactures, markets and sells worldwide satellite dish antennas, conical horn antennas, radomes, antenna feeds, microwave and waveguide components, ultra low loss waveguide transmission line Tallguide ®, and shelters. Our customers serve the broadcast, communications, radar, weather and cable industry, defense, government, and government agencies worldwide. AFC's quality control manufacturing standards are certified under ISO 9001 : 2008.

A complete Internet WWW AFC site index may be found in Antennas for Communications (AFC) Home Page Document Summary List. Additional radome information is contained in AFC's Radome Capability.

Top of Page   Return to AFC Profile   Return to Radome Network Home Page Return to Radome Network Home Page.

Antennas for Communications
2499 SW 60th Ave, Ocala, FL 34474
Telephone (352) 687-4121 Fax(352) 687-1203 E-mail sales@afcsat.com


Tallguide is a Registered Trademark of Antennas for Communications
Copyright © 2002 - 2012 Antennas for Communications