ICO S-BAND ANTENNAS
Peter A. Ilott, Ph.D.; Robert Hladek; Charles Liu, Ph.D.; Bradford Arnold
Hughes Space & Communications, El Segundo, CA
The four antenna subsystems on each of the twelve ICO satellites, includes two eight foot diameter S-Band active arrays, driven by a digital signal processor (DSP). These phased arrays, each consisting of a triangular lattice arrangement of 127 radiating elements, must be tested for functionality and workmanship, before being integrated onto the spacecraft. With a two-month center to center delivery requirement, standard fabrication and test procedures had to be modified and automated in order to meet schedule without compromising the traditional conservative approach for performance verification. This discussion of the ICO S-Band test program includes descriptions of the nearfield testing, Field Aperture Probe tests, and other tests related to EMI problems (such as transmit to receive isolation and PIM) on the spacecraft, as well as a brief description of the PC-BFN, a rack of special test equipment designed to allow testing of the passive array without the satellite DSP. Emphasis is given to the design of tests compatible with a mass production environment.
Keywords: Nearfield, Field Aperture Probe, PIM, ICO
The current trend in the satellite industry towards lower cost and ever-shorter schedules has challenged the
Figure 1-1: ICO Satellite
fabrication and test approaches of the past. The ICO satellites, currently under construction and test at Hughes Space and Communications (shown in figure 1-1), are the first of a kind for the commercial mobile communications market. Twelve will be placed in two 45° inclined, six hour orbits (5 operational, 1 spare in each), providing world wide mobile coverage for ground users. The payloads will be digitally controlled. The Hughes digital signal processors (DSP) on board will perform all channelization, as well as serving as the beam forming engine for the two large S-band phased arrays which provide the mobile link to hand held units. Careful attention was paid to test designs which would ensure the performance of the antenna payload, yet allow the quick delivery of twelve of each subsystem. These considerations were important to every aspect of the spacecraft, including the antenna subsystems of which there are four; the C-band downlink arrays, the telemetry and command (T&C) antennas, a calibration antenna for use by the S-band system, and the mobile link S-band phased arrays. This discussion covers only the S-band phased array antennas.
Prior to and immediately after the successful proposal campaign, the Hughes and ICO technical teams worked out a series of tests for every subsystem on the spacecraft. The S-band arrays, being the direct link to the user, were of special consideration.The test strategy was to conclusively prove the design on the first few antennas. Element level tests will allow both performance and workmanship verification on later flight models The initial tests would also validate the prediction software models, on which was based the performance claims for the antennas. It was agreed that radiation patterns of the first two flights, as well as the engineering models, would be measured by standard near-field testing. To provide a signature test, or measure of quality, for the remainder, return loss tests and field aperture probes (FAP) of every element in the arrays would be performed. Together they would permit an assessment of the integrity of the units through environmental testing, by measurement of the S parameters. The FAP process also provides a method for electrical alignment of the arrays at spacecraft level. Since the DSP was not scheduled to be available when the arrays were ready to test, a test rack, coined the PC-BFN (for PC controlled beam forming network), was designed and put into service providing the beam scanning excitations for the S-band arrays. It proved one of the most useful pieces of special test equipment (STE) of the program, and was used in several tests to verify and quantify the errors of system concepts, as well as for array near-field tests.
After a brief system summary the tests will be discussed, and results presented.
2.0 S-Band System Summary
The system allows for the available channels to be divided among 163 beams (see figure 2-1), which simultaneously cover the visible earth in 163 cells (see table 2-1 for cell parameters). A four-cell frequency
Figure 2-1: ICO Beam Coverage
reuse scheme is used to maximize capacity. This reuse, of course, puts constraints on the array performance. Power efficiency considerations of the
Cell Size, degs
Cell Size (km for nadir beam at earth surface)
Cell Reuse scheme
Reuse Cell area, degs2
Center to center spacing (reuse cells)
Reuse cell side lobe spacing
Table 2-1: Cell Parameters
solid state amplifiers for the transmit array coupled with the need to maximize EIRP limited the depth of the transmit (Tx) array taper. The same was not true for the receive (Rx). Fortunately the link requirements did not dictate the same level of reuse cell isolation, and therefore, pattern side lobe suppression, for the Tx array as for the Rx. The arrays, shown in figure 2-2, have 127 (RHCP) radiating elements in a triangular 1.3 l center to cener layout. The hexagonal structure generates a six-
Figure 2-2: ICO S-Band Arrays on satellite
spoked side lobe pattern (see figure 2-3). The sun-nadir steering maneuver of the spacecraft, designed to
Figure 2-3: Tx Array Side lobe Pattern
maximize power and minimize heat build up, requires that the beams are continually deyawed, or steered, to compensate for spacecraft rotation. This did not allow the antenna team to take advantage of the side lobe structure to minimize reuse cell interference. The Rx taper both minimized the side lobe levels, and made the first side lobes much more symmetric about the main beam. Insertion phase and amplitude errors in the 127 active chains that feed the array would degrade the performance of the arrays. These errors were modeled extensively to quantify their magnitude and effect. The antenna team designed its first near-field tests to allow for performance evaluation due to chain to chain errors, based on the system error budgets.
The DSP gives several advantages over a passive beam former, such as the ability to scan beams to any angle, and thus allows for service reallocation as the mobile market becomes more well defined. A practical design advantage is the ability to compensate across the frequency bands for unequal line lengths within the spacecraft. (See  for greater system details).
The size of the arrays required changes in the standard 601 bus, with the arrays becoming part of the bus structure. The payload design team took a modular approach, with five major components on five panels, two of which are the S-band arrays. Each radiating element in the arrays has a band pass filter. After testing at the passive level, 127 SSPA’s and LNA’s are installed and the arrays tested before spacecraft integration.
3.0 S-Band Antenna Tests
The test program can be categorized in 4 groups; radiation performance verification (near-field), Tx to Rx isolation verification (coupling and PIM tests), environmental signature tests (return loss and signature FAP tests), and array alignment (FAP). The signature tests were used at various times during assembly and environmental test of the arrays. The return loss and FAP tests provided a signature of the r/f scattering parameters of the element filter pairs, which could be verified whenever the array was subjected to such mechanical tests as acoustic, vibration, and thermal cycling. The return loss was measured using a network analyzer, controlled by a PC which logged the data, and allowed for immediate evaluation of all the chains. The significant feature of this test was the ability to review the large amount of data in real time, and make immediate decisions about whether to proceed to the next test in the program.
To obtain a measure of the through characteristics of the array elements, a FAP test was devised. A field aperture probe is inserted into each of the elements in the array at a standard depth. The probe is designed with a collapsible front sleeve to prevent flight hardware damage. Two rotation angles are used to extract the principle polarization component of the element (and filter out the cross pol component most of which is produced by the probe insertion into the element). Several pairs of angles provide an average of the S21 transfer function of the element filter pair, an average that compensates for the disturbance of the fields in the element by the introduction of the probe. A robot scanner (actually an industry standard nearfield scanner) is used to position the probe and insert it into the elements. A power divider feeds the entire array. The measurement of each element filter pair yields a signature of the r/f performance. The signature tests would be performed several times on every payload, and therefore had to be fast, easy to perform, and allow for quick evaluation. The entire FAP signature test is automated and takes less than two hours to complete. Once again the automation process allows for quick testing and real time evaluation of the repeatability of the data compared to the initial measurement.
To perform radiation pattern evaluation the passive array was excited using the PC-BFN. This test tool consists of a power splitter feeding a bank of phase shifters and attenuators, controlled using National Instrument I/O boards. The boards are controlled using a Pentium class computer using the Lab Windows software package. With phase and amplitude resolutions of 1.5° and .06 dB, any scanned beam of the ICO set can be simulated. The PC-BFN provided a method to generate beams for near-field tests, as well as for studies of the Tx to Rx coupling and coupling tests involving the arrays and the calibration antenna. Figure 3-1 shows the PC-BFN connected to one of the S-band arrays during a FAP test.
Figure 3-1: S-Band Array in FAP Test
A typical near-field test begins with a calibration of the PC-BFN. A desired taper and phase distribution (usually uniform phase for the normal (nadir) beam), is input to
Table 3-1: Multiple Angle FAP measured data averaged to find amplitude and phase. Note improvement in standard deviation in RHCP extracted data.
the PC-BFN. The S21 of each channel is measured, from PC-BFN input to end of cable output, and the PC-BFN compensates for the internal differences in the power slit and line lengths. At this point the desired amplitude and phase is met at the cable ends. The cables are then connected to the array. Since each element filter pair will have some amount of variation in S21, there is a certain amount of error across the array. At this point the FAP process is used to electrically align the array. Each element is FAP’ed at the desired frequencies, and the data processes as described above. The safety feature of a collapsible FAP sleeve has the disadvantage of introducing more error into the process. This error can be greatly compensated for by averaging the results of multiple angles of FAP measurement. Table 3-1 shows an example of averaging the information from 8 different angles of rotation of the FAP.Figure 3-2 shows the measured errors for the phase of half of the array (the amplitude errors are insignificant in this case due to the small variability from element to element). This information is fed back to the PC-BFN to align the apertures of the elements, and the FAP repeated. The result is shown in figure 3-2. The standard deviation has improved from 5.1° to 0.75° , and the error range has reduced to ± 1.75° compared to ± 10° before alignment. It is easy to assume that the FAP has performed well, however it may be just correcting for the errors inherent in the FAP process. To verify that we are in fact aligning the array excitations near-field scans were performed with and without the alignment corrections. The results (to be reviewed in the next section) confirmed that the FAP process does align the array. Part of the test process early on was to verify that the prediction that array analysis
Figure 3-2: Typical Pre and Post Alignment FAP Data for 1/2 Array
gave was accurate. It was seen that agreement was much better after the array was aligned by the FAP method. As a test of the sensitivity of the radiation patterns to excitation errors expected during normal operations, and our ability to predict them, we deliberately introduced errors using the PC-BFN. The radiation pattern for a given set of beams was measured with a chosen set of errors introduced, while the same errors were input to the analysis. The agreement was excellent, and further gave the Hughes and ICO teams confidence in the process. The FAP process is also used at spacecraft level to align the excitations of the complete payload. One great advantage of using the FAP is that there is no need for removal of the elements to do hardline alignments. It is even possible to set the absolute gains of the channels by calibrating out the transfer function of the filter/element/FAP probe. An average over several element filter pairs mounted on a test stand provides the calibration data.
The PC-BFN was used to quantify the coupling between the Tx and Rx arrays. The two arrays were placed in
Figure 3-3: Scematic of Isolation test Setup
their flight configuration and the PC-BFN connected to the Tx array (figure 3-3). The Rx array was only partially assembled, with the filters not in place. This was necessary, as the 80 dB or so of rejection at Tx frequencies would have made the test impossible. The big problem with this test was that coupling could be
Figure 3-4: Isolation between Tx beam and Rx array as a function of Rx beam angles
expected to be different for different beam combinations, both Tx and Rx. With only one PC-BFN we had to rely on much post processing. The PC-BFN was used to generate several (>20) Tx beams, and S21 measured at each of the Rx elements (with appropriate calibration) for several frequencies. Using the measured data and MathCad analysis simulating the beam forming process of an Rx BFN, we were able to quantify the range of beam to beam coupling for these antennas. Figure 3-4 shows an example Tx beam which couples most strongly for an Rx beam around q = -22° . The study found a small number of similar Tx beams that added coherently across the Rx array, evident in coherent phase fronts (figure 3-5). Fortunately the numbers of beams which phase up to create these significant coupling was small, and we demonstrated that the coupling
Figure 3-5: Phase coherency Across Rx Array for Tx beam in figure 3-4
requirements of the system were met with margin. This work would have been impossible without the PC-BFN.
Figure 3-6: Phase Excitation Error Across Rx Array due to 0.4" positional error (in 2-d) in Calibration antenna. Normal to fitted plane gives 1st order beam pointing error.
Another study was conducted with the PC-BFN to verify predictions about the performance of the calibration antenna. This antenna is deployed from the front of the spacecraft and measures the relative phase and amplitudes of the array elements, providing data on the excitation errors in the chains. Early analysis quantified system errors that could be expected in the event of calibration antenna deployment position errors. The PC-BFN was connected to a Tx array in an anechoic chamber, and a calibration antenna placed in the flight configuration. Several studies were carried out, one of which was to move the calibration antenna slightly from its nominal position, and verify the predictions. It is expected that the final calibration antenna position could be off by up to ± 0.4 ". To first order this results in a beam pointing error if the calibration data is used to correct the array excitations. A beam deviation factor was calculated based on analysis, which gave the team an estimate of expected beam pointing error. Part of this test would confirm that factor. Figure 3-6 shows an example of 0.4" movement in two axes (up and to the left). A plane is fitted to the data from which the pointing error is deduced. The measurements confirm the calculated beam deviation factor, which predicts an error of about 0.1° in azimuth and elevation.
Other tests using the PC-BFN included EMI tests, where, for example, the array is excited and leakage behind the array is measured. PIM (passive intermodulation) was also of great concern for this system. Standard approaches to thermal PIM tests applied to the ICO arrays would have required a large dedicated capital equipment investment, and would make PIM detection and location extremely difficult. A further complication was that the R/F environment varies as different beams are created with differing amounts of traffic. Given the large number of antennas to test, an innovative PIM test was designed which minimizes high power R/F equipment, and which allowed for quick location of any PIM source on the arrays. It provides sufficient overtest of the arrays so that the customer is not concerned about the different beam effects. A thermal chamber large enough for the ICO arrays was built at Hughes. The efficiency of the test allows Hughes to run a three thermal cycle PIM test in less than three days. Currently more than 4 flight arrays have been sucessfully tested.
Figure 4-1: Pattern Cut comparsion shows how FAP data can improve the prediction model.
Figure 4-2: Pattern comparsion after FAP data is used to realign the array excitations, and improve pattern performance.
The PIM test approach is currently being applied to other programs.
4.0 Pattern Measurement Results
Figure 4-1 shows a pattern cut comparsion example (for the nadir beam), between measured and predicted , where the FAP excitation data has been used to correct the prediction model. Note the side lobe asymmetry. In figure 4-2 the FAP data has been used to realign the array, and is compared to nominal (no errors) predictions. Agreement is excellent in both cases. Note the improvement in first side lobes when the array is realigned. The prediction models were shown to be accurate, both for patterns and absolute gain.
The ICO S-Band test program has been a great success, technically, and from a programmatic standpoint. Several innovative tests were designed and executed with high schedule and resource efficiency all with the goal of faster, better, cheaper. Lessons learned in the ICO program are already being applied to the next generation of communications satellites under construction at Hughes ensuring its future competitiveness.
(The authors wish to acknowledge the team members including; Stan Gillespie Jr., Gordon Howard, Gina Schmidt, Fernando Vega, as well as the mechanical team who helped us make it happen).
 P.A. Ilott PhD; R. Hladek; "Active Antenna System for the ICO Satellite", to be published in Proc. 1998 Microwave & R/F Meeting, London, 1998.
 Mailloux, Phased Array Antenna Handbook, Artech House, 1994.