Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
ANTENNA PATTERN The powered elements of the antenna are all driven by transmitters iden- tical in design. Taper of illumination across the aperture is achieved by powering only a fraction of the elements in the outer part of the array. The thinning ratio is l:2, hence the presence of the 885 dummy radiators. Given that each element of the antenna is excited with a signal of known phase and amplitude, the radiation pattern of the antenna can be calculated with great accuracy. The availability of simple approxi- mate models allows for independent checks of the correctness of compu- tations. The functional demands of the PAVE PAWS mission require that the antenna, as it actually operates and not merely in some ideal state of perfect excitation, meet stringent conditions. The design then requires that tolerance limits be set on the amplitude and phase of each radiating element for each operating condition. It must be verified that when each element is within tolerance, the antenna still meets its requirements. Finally, controls must exist to hold the antenna elements within toler- ances, and to detect departures from satisfactory operation. Given the tolerance limits for, or statistical descriptions of, the amplitude and phase of the excitation at each radiating element, straight- forward calculations can be made with great accuracy of the limits of departures of the antenna pattern from the ideal. As a necessary part of the design process, the radar contractor, Raytheon, has made such analyses. The information below is based on the contractor's analyses. The panel did not, of course, repeat the many calculations involved. The panel's confidence in the results described here is based on the consistency of the quoted results (a) with simple models, (b) with the performance requirements of the PAVE PAWS system, and (c) with other systems, including the PAR and the FPS-85, with which some of the panel members have been directly involved. The PAVE PAWS antenna radiates like any antenna of comparable aper- ture (in wavelengths) and taper. The main beam is nominally 2° wide at its half-power points. The first sidelobes are 20dB or more below the main lobe in power gain and are contained within a cone around the main beam of about 4° half angle (second nulls at about 4° off the main beam). l2
l3 Secondary sidelobes are at least 30dB below the main lobe in power gain; they are distributed in a roughly random manner across the angular field around the main lobe, tapering in density but not in peak gain at angles remote from the main beam. This is the kind of pattern that results from a design that minimizes the maximum secondary lobes. The ground level in the neighborhood of the PAVE PAWS site slopes away from the radar at l° or more below the horizontal. Consequently, it is principally the secondary sidelobes that intersect the ground during normal operation. Much of the rest of this report deals with just two issues: the power in the secondary sidelobes and the possibility that, because of some departure from design conditions, a first-order side- lobe or the main beam may illuminate areas of public access. Each of the l,792 nominally identical transmitters in the radar is an amplifier provided with a switchable phase shifter. Steering is by what is known as "row and column" orders. The elements lie naturally in rows and columns on the antenna face. Although many row-column inter- sections do not contain active elements, each active element of the array is identified by the unique combination of row and column in which it appears. To steer the beam in a given direction a desired phase-shift is determined for each row. These progress in uniform steps from row to row. Similarly, a desired phase-shift is generated for each column pro- gressing uniformly from column to column. A particular transmitting module simply adds the phase-shifts given by its row and column instructions to determine the phase-shift it then introduces into the signal applied to its input. Were this process carried out with perfect precision and were the individual radiators themselves emitting spherically symmetrical waves, the resulting field would constitute a discrete approximation to a plane wave leaving the antenna face in a direction determined by the phase-shifts. The approximation replaces by a discrete set the plane continuum of spherical sourcelets that, by Huygens' Principle, generates a plane wave. In the actual antenna further departures from the discrete approxi- mation to Huygens' Principle result or can result from several sources: o The individual radiating elements are not isotrophic radiators. o Each phase shifter is quantized to steps of 22.5°. o There can be departures in each phase shifter from the desired 22.50 steps. o The transmitting modules are excited from a common source by way of cables. The delays along these separate distribution paths are held to uniformity only within some limit of tolerance. All of these effects are controlled in one or more ways in the design, manufacture, and test of the radar. All of them can be analyzed and are accounted for in the contractor's analyses. Some basic features of the analysis or control are discussed below. Element directivity: Each face of the antenna scans at most 60° off boresight in azimuth and elevation. The radiating pattern of an individual element is broad. The effect of the pattern of the element is to modify the whole antenna pattern as a function of angle off bore- sight; the modification is not large in comparison to the inevitable change in pattern with beam position that results from the fact that
l4 the effective aperture of the antenna is already reduced by the cosine of the angle off boresight. All of these essentially geometrical effects are incorporated directly into the analyses. Errors: The other effects listed above result from departures from ideal gain and phase in each radiator; they can properly be called errors. Indeed, consider the antenna pattern as being generated by two sets of radiators: (a) the ideal radiators that all produce fields of exactly the same amplitude, having respective phases exactly equal to the desired values, (b) parasitic radiators, located coincidently with each ideal radiator, that emit error signals. The error signal at each radiator is simply that signal which when added to the ideal produces the actual signal. One aim of the design is to keep the error signals small. There are other considerations in the design process that also bear on the character of the error signals. One critical requirement of the radar is that the true direction of the main lobe be known--i.e., be exactly what is ordered by the computer. Another requirement is that echoes returned from the ground or sea because of sidelobes that lie below the horizon not interfere with the detection of targets. Errors in the phase-shifts that are systematic across the face of the antenna--or are systematic among a considerable fraction of the antenna elements- - can deflect or broaden the beam or increase the sidelobe returns. Ac- cordingly, errors must be kept small and must avoid systematic patterns. Control is exercised in several ways: o The design itself undertakes to randomize across the face of the antenna unavoidable departures of the actual phase-shift from that ideally ordered. o Manufacturing tolerances are set on the cables and phase shifters as well as on amplifier gains^ o During operation, the antenna is systematically tested to verify that performance in gain and phase remains within specified tolerances. The intent of the elements of control in combination is to make it possible to think of the antenna pattern as composed of the ideal pattern plus a sum of error signals from the various radiators that are not only individually small but also have no coherent or systematic pattern across the antenna aperture. Therefore, these error signals cannot focus into an undesired beam or sidelobe that is of high intensity. Thus it happens that in order to meet the primary criteria of antenna performance, the design process is such that it also limits the extent to which errors in individual elements can disturb the "ideal" pattern of sidelobes. As to error control: row and column steering orders (phase-shifts) are presented to the individual transmitting element with a precision of six binary places (one part in 64). These result from calculations that start with the desired position of the beam expressed as sines of angles off boresight to 60 binary places. At four stages in the cal- culation, results are rounded to fewer places; at one of these roundings, a randomization of the least significant figure is introduced to avoid bias. At the antenna element, the row and column orders are added together and the result rounded to four binary places; the least sig- nificant, figure then corresponds to 22.5° of phase-shift.
l5 Phase-shifts in the feed cables to the antenna elements are measured to eight binary places (approximately l°). Cables are cut to a length specified to six binary places plus an increment (of maximum value about 4° in phase) that is deliberately controlled so as to appear spa- tially random across the face of the array. Manufacturing tolerances are set on gain and phase of the indi- vidual transmitting modules and phase shifter with the intent that the root mean square (RMS) amplitude variation across the population of l,792 elements be held to 0.ll4dB and the RMS phase variation (including rounding) to l6.4°. The panel is of the opinion that the processes as described can be expected to attain this level of sta- tistical control over pointing errors. Given error control to the degree described, the effects on the sidelobe patterns of the antenna can be given a precise statistical description. For example, the likelihood of a variation in a parti- cular sidelobe from the nominal pattern by as much as 2dB is less than 5 percent. Quite apart from the design calculations, measurements have been made to validate the design, and further measurements will be made during operation of the radar. In one part of the contractor's test program the transmitting antenna pattern is probed by tracking satellites: with the receiver pattern fixed so that the satellite appears on the receiving main beam, two transmissions are made, one with the satellite on the main transmitting beam to calibrate the roundtrip loss, and a second with the transmitting beam directed in another direction . Comparing the returned signals from the two transmissions provides a sample of the sidelobe pattern in a particular steering configuration. Over the course of time, a map of sidelobe performance is thereby established. The contractor reports that, as measured to date, near sidelobes are at least 22dB below the main lobe. This observation is confirmed by plots of data shown to the panel. An independent series of measurements, sampling the radiation field at points near ground level, is being conducted by the Air Force. Measurements are further discussed in Section 6. In addition to errors of the kind just discussed, including small departures from nominal conditions, gross malfunctions must be considered such as failure of a module to transmit at all. When a module fails to transmit, the effect on the antenna pattern is to add coherently to the unperturbed pattern an error signal from the failed module equal in amplitude but opposite in phase to the ideal signal from that module. Since one module contributes only .06 percent of the power of the antenna, such an error signal can have but a trivial effect on the antenna pattern. Even if many modules fail, the effect on the main beam, where the power density is high, is small. The effect on sidelobes will also be small unless the failed modules lie in some systematic pattern across the face of the antenna, so that their error signals constitute a well-formed beam. For a given number of failed elements, a worst-case configuration (not the only one) is that of uniform spacing along a line. Such a
l6 line of radiators is an antenna having a conical main beam, (one concentrated near a central surface that is a cone).. The axis of the cone is the line along which the failed radiators lie. The half-angle of the cone is the angle between this axis and a line, which is then a generator of the cone, that is parallel to the direction of the main beam of the antenna. If a column of radiators were to fail, the axis of the cone points upward, and no generator of the cone lies at lower elevation than the main beamâhence the main lobe of the error pattern cannot strike the ground. If a row of modules falls, the axis of the cone is horizontal and the coneâi.e., the main lobe of the error patternâ always intersects the ground. This is an illustrative severe case. The longest row in the PAVE PAWS antenna contains 32 elements. There are several such rows, none of them in fact contain as many as 32 powered elements.. The maximum possible flux density at a given point from 32 radiators is l,024 (= 32 x 32) times the flux density at that point created by one radiator. In the case of PAVE PAWS, at a point on the ground where such a maximum occurs,_ if that point is not illuminated by significant radiation from a secondary sidelobe of the unperturbed antenna, the flux density from the error pattern is 5dB below that of an unperturbed secondary sidelobe--i.e., 35dB below the main lobe-. If the error lobe falls exactly at a point already illuminated by the maximum of a secondary sidelobe and if the error signal adds constructively to the unperturbed sidelobe at that point (constructive addition cannot be ruled out), the effect is to create a perturbed sidelobe about 4dB higher than the unperturbed one--i.e., a sidelobe 26dB below the main lobe. Were a row of elements to transmit in reversed phase, the effective error signal would have twice the amplitude of that used in the calcula- tion above. The main lobe of an error pattern from such a malfunction would be slightly more intense than a secondary sidelobe of the unper- turbed antenna, and could,, in the worst case, cause a perturbed secon- dary sidelobe 24.5dB below the main lobe. Failure of five adjacent rows to transmit would create an error lobe that is about 2ldB below the main beam. This when added constructively to a secondary sidelobe of the unperturbed pattern could create a perturbed sidelobe l8.5dB below the main lobe. These estimates are worst-case bounds, since it is unlikely that the main lobe of the error pattern where it intersects the ground will coincide both in position and in .phase with the peak of an unperturbed secondary sidelobe. The estimates show, however, that the simultaneous failure of several rows of antenna elements or of their driving modules can have an effect on the secondary sidelobes of the PAVE PAWS antenna that is quantitatively much more severe than the effects of random small errors that were examined earlier. Written into the specifications and designed into the PAVE PAWS system is a testing feature that continuously and automatically veri- fies the operation of the transmitting modules during operation of the radar and monitors the ability of the antenna to form proper beams. Some distance in the front of each antenna face mounted on a pole well
l7 above ground is a monitoring receiver that measures the amplitude and phase of the signal received and reports the measurement to the control computer. Slightly more often than once per second this receiver is used to test features both of the transmitting and of the receiving operation. Signals are distributed to the l,792 powered elements of the PAVE PAWS antenna in such a way that these elements divide into 56 groups of 32 elements each. (Figure 4 shows how these groups lie on the antenna face.) Each group is simply a subarray forming a small phased- array itself. Each subarray is subject to the steering orders that are given to the whole antenna. The modules of the subarray share a common driver amplifier. Any subarray can be turned on or off inde- pendently of the others. Once every 972 milliseconds (during one 54 millisecond period out of every l8) subarrays are steered one by one to the test antenna and their gain and phase measured. Each subarray is tested once every 30 seconds. Since a single transmitting module contributes 3 percent of the power of a subarray, a malfunction of even a few modules within a subarray is readily detected. Measurements are reported to the con- trol computer. A major fault is reported if the gain of any subarray is not within IdB of the design value. Individual transmitting modules can be tested by this same procedure, but that process is not done during the normal cycle of operation. Figure 4 shows that the specific severe malfunction examined earlier, the simultaneous failure of one to several rows of modules, is a most unlikely event unless it results from the simultaneous turning off of about l2 subarrays. Whether or not subarrays as a whole fail, the malfunction of one to several rows of modules would appear as a malfunction in enough subarrays to be detected within a few seconds. In fact, almost any distribution of 30 to l00 malfunctioning modules that could create a reasonably well formed beam from error signals is one that would put enough malfunctions into some one subarray that the disturbance would be detected within 30 seconds. If a subarray is turned off the resulting error signal is one generated by 32 elements. The arithmetic is therefore the same as that used above in estimating the effect of a malfunctioning row. However, a subarray is a compact antenna having not a conical beam but a broad main lobe oval in cross section and steered in the same direction as the main lobe of the unperturbed antenna. It is fringes of this broad lobe, or sidelobes from the subarray, that intersect the ground. Hence, the effect of turning off one to five subarrays is less severe than the bounds estimated earlier for the respective effect of one to five malfunctioning rows. In fact, measurements of sidelobes on the ground have been made during periods when several subarrays were not functioning. Within the capability of the measurement process, there was no difference detected from the sidelobes of the full antenna. If a subarray were to be steered independently and its main lobe directed toward the ground, the effect could be the same as that esti- mated earlier for a malfunctioning row. The subarray alone illuminates the ground with 5dB less intensity than a secondary sidelobe of the
l8 Figure 4 Subarray Positions Courtesy of the United States Air Force
19 main antenna. Added constructively to a secondary sidelobe, this signal could create a perturbed sidelobe 4dB above the unperturbed one. To cause such faulty steering many malfunctions would have to occur, each in a very specific way. The event is most unlikely but in any case could not pass undetected for longer than 30 seconds. An accumulation of water, ice, or snow on the antenna face can alter the gain and the phase of antenna elements across the antenna aperture both in a random and in a systematic way. Therefore, the effects of accumulated precipitation are amenable to analysis in the manner already illustrated. Analyses and measurements of these effects were made in connection with the design and installation of the PAR in North Dakota. These have been supplemented by measurements specific to PAVE PAWS. Rainwater cannot accumulate deeply, and it takes a l0 inch accu- mulation of snow to have an electrical effect comparable to that of l inch of ice (or water). Therefore, in the climate of Cape Cod, an accumulation of ice is the condition most likely to be severe. Antenna elements are provided with heaters to prevent the accumulation of glaze in ordinary storms and to remove glaze after a severe storm. Past weather records indicated that even without ice removal, there will be only a few brief periods per decade in which enough ice can accumulate to affect the operation or result in shutdown. Degradation of performance by the presence of ice will be detected by the regular .process of moni- toring the subarrays. Even if- ice accumulation should occur, a uniform accumulation will not affect the sidelobe pattern. A systematically wedge-shaped accumu- lation that tapers by as much as l inch across the antenna face (an extreme condition) would not affect the sidelobe structure, but would deflect the.whole pattern by not more than l/20 degree. Such a de- flection has no effect on the estimates of radiation intensity from the sidelobes that are discussed here. A sufficiently severe random component in the accumulation of ice could increase sidelobes. A varia- tion in thickness by one-half inch distributed across the aperture in such a way as to have a severe effect could increase sidelobes by 2dB. Ah accumulation having random variations as severe as this is unlikely.
