IC Substrate - Test platform for all-digital phased array radar

Radar PCB revolution
In the past 15 years, ARRC has participated in the U.S. National Multifunctional Phased Array Radar (MPAR) project, followed by the U.S. National Surveillance Radar Spectrum Efficiency (SENSER) project, which was originally developed by the Federal Aviation Administration (FAA). ), the Department of Defense (DoD), the Department of Homeland Security (DHS) and the National Oceanic and Atmospheric Administration (NOAA). Therefore, ARRC is working on a scalable S-band fully digital polarized phased array to meet the needs of weather and long-range aircraft scanning. The array will also support other important operations, including MIMO and conventional communications.

Flexible beam control and multi-function implementation make the phased array the best candidate for multi-task radar systems because it provides an efficient and cost-effective solution. Advances in GaAs, SiGe, CMOS and PCB technologies have provided reliable, highly integrated RF devices that make phased array antennas the core of modern remote sensing and communication technologies. Highly integrated and efficient devices allow the phased array antenna architecture to contain multiple transceivers. Compared with the previous generation phased array antennas that only use analog beamformers, these devices can reduce the cost and reduce the cost of phased arrays. For size and weight, and to optimize system functions and improve system performance, 5G will of course use such phased array technology. Arrays that use analog beamforming are naturally limited to beamforming schemes that can be achieved through the precise setting of front-end beamforming electronics.

At present, the use of digital beamforming (DBF) at the sub-array level is a common method to improve the flexibility of phased array radars. The 76-panel Advanced Technology Demonstrator (ATD) operated by the NOAA National Severe Storm Laboratory (NSSL) and Massachusetts State University (UMass) Raytheon low-power radar (ie Skyler) can prove this. However, the shift to a cell-level DBF architecture will enable unprecedented functions. For example, Australia’s CEA-FAR naval radar, US Navy’s FlexDAR radar 2, Israel’s Elta’s MF-STAR, AFRL’s BEEMER (the baseband digitization of MIMO experimental radar antenna units), and space fences. In addition, the digitization of each antenna unit allows people to accurately control the polarization, and can control pure H polarization or V polarization, or control both H and V polarization at 45 degrees, as well as LHC, RHC or any one of them. Polarization state.

Digital array technology is a nascent research direction. An important contribution of the Combat Capability Development Command Army Research Laboratory (CCDC ARL) is the development of powerful techniques for array calibration. The operation of phased array radar in a crowded environment depends to a large extent on the measures to protect the radar and the continuation of the calibration work in a dynamic environment. Factory calibration is not enough for digital arrays, so powerful on-site calibration techniques are needed, which also have advantages in computational efficiency. OU and CCDC ARL are developing calibration technology based on mutual coupling to solve the dynamic calibration problem. CCDC ARL is conducting a proof-of-concept experiment, using a unit-level digital array laboratory test system to quantify the performance of the initial algorithm. Looking to the future, CCDC ARL will optimize these technologies to achieve wider bandwidth performance, and will focus on the scalability of large arrays and their applicability to operating environments other than laboratory test platforms.

Complete digital architecture

Although it has proven challenging to achieve dual polarization on PAR, recent radar technology exchange seminars sponsored by the National Science Foundation (NSF) have made significant progress5, such as the S-band control panel of MIT Lincoln Lab in ATD6, BCI/LMCO's S-band prototype, NCAR's C-band airborne phased array radar system, UMass's X-band radar and OU's S-band cylindrical polarized phased array radar (CPPAR) demonstrator 7. In order to improve the time resolution of the "spotlight" operation, ARRC produced a single-polarization X-band atmospheric imaging radar (AIR) a few years ago, as shown in Figure 1. The AIR works in "flooding" mode, launching a 20-degree vertical fan beam, and using 36 receiving arrays for large-scale digital beamforming. In other words, the range height indicator (RHI) measured by the radar can be formed at the same time, similar to taking a photo with an electromagnetic camera. This architecture combined with 20 degrees/second azimuth mechanical scanning will allow the existing AIR to collect information in the range of 180*20 degrees in about 9 seconds. Therefore, this is also the world’s highest resolution for tornado cause observation. 8. Another similar system with flood resolution is Osaka University’s X-band PAR.

These advanced imaging surveillance operating modes require multiple digitized sub-array channels. The improvement of digitization level will also make adaptive digital beamforming (ADBF), spatiotemporal adaptive processing (STAP) and even MIMO operation modes possible. The ideal phased array architecture will have digital functions, and the transmit and receive signals of each antenna unit are controllable, so it also has wide bandwidth coverage. Since the unit-level processing and subsequent beamforming are both digital, they can be reconstructed and optimized for different application scenarios. Unit-level digitization opens the door to new beam processing and beamforming solutions, and provides maximum flexibility through unprecedented dynamic adjustable range in large systems. For example, given M antenna elements and the noise between the elements is uncorrelated, the signal-to-noise ratio of the system is improved by 10 log(M). However, this is accompanied by inherent technical risks and practical challenges, such as the amount of data that needs to be processed and the use of uncomplicated transceivers.

Figure 3 shows three examples of all-digital PAR systems. The leftmost image in Figure 3 depicts several typical high-sensitivity beams and several low-priority beams, which are necessary to collect important information in an area. The picture in the middle of Figure 3 shows an example of spatio-temporal multiplexing, through which multiple sets of independent samples can be collected from the monitoring area; in this way, fewer samples can be used to collect data. Since adaptive spatial filtering can be achieved by a phased array4, this greatly proves the correctness of using a phased array on a typical parabolic dish antenna. Finally, the rightmost image of Figure 3 describes how the mobile demonstrator will use the team’s imaging expertise to achieve fast batch scans8.

For any multi-task radar in the future, the simultaneous realization of multiple functions is the only way to meet mission requirements on a given time axis. Therefore, it is essential to realize the flexibility of advanced beamforming through digitalization. In addition, this method can implement other tasks in the entire life cycle of the digital PAR through software upgrades rather than expensive hardware transformations, thereby saving a lot of operation and maintenance costs. The next part will give an overview of the development of the S-band dual-polarization PAR that is being designed and manufactured at ARRC. The S-band dual-polarization PAR will achieve these goals. This system is called Horus. It has a digital transceiver for each polarization and each antenna element. It will become a valuable research tool to evaluate the advantages and challenges of this method.

Horus radar design concept

ARRC is currently developing a mobile S-band dual-polarized phased array system. The system has an all-digital architecture, composed of 1024 dual-polarized antenna units, divided into 25 8*8 panels (16 of which are equipped with electronic devices), as shown in Figure 4. Each panel is equipped with eight "OctoBlade", almost all of the radar electronics are located in it. Each OctoBlade is carefully designed to excite the 8-element column of the high-performance antenna array in the panel and achieve an almost ideal polarization state on the main plane. The main plane consists of a metal cooling plate (heat transfer tube) with a PCB on each side to accommodate a total of 16 GaN-based front ends (each unit, each polarization> 10W), eight of which are dual analog devices Channel digital transceiver, four front-end FPGAs for processing and two FPGAs for control. The assembly of the antenna subsystem and its related electronic components can be classified into one of the following three main architectures: conformal patch assembly, panel assembly (with a slide-out OctoBlade) or independent structure separated by cables (Figure 4) . A design with a slide-out OctoBlade requires the lowest maintenance costs because these electronic components are easily hot-swappable. This convenient function is perfect for foundation systems that require decades of service life.

Generally, the performance of a large array depends on the digital interconnect structure behind the array. The traditional hierarchical topology is currently being used, but some of their characteristics, such as scalability, flexibility, and bandwidth, are restricted. For example, some arrays use a mesh topology. When using a mesh topology, the burden on the central channel is heavy. This usually leads to congestion in the central area of the network. The solution to this situation is to add routers to the mesh network or use a ring topology. This ring topology has symmetry on the routers on the opposite side, which can reduce unnecessary congestion with a small increase in resources. . But there are still many unresolved issues. We believe that the three main issues are: the data transmission mechanism (ie RapidIO, Gigabit Ethernet, etc.), the degree of partial beamforming, and the data path topology (ie, hierarchical structure, etc.). A good balance of these issues will allow the array size to be easily expanded to meet various tasks.

Horus' ordinary radar is fed into the back of the panel through the RapidIO network to complete digital beamforming. This will enable the beam-bandwidth product of a conceptually multifunctional PAR system (such as a 200MHz beam in an appropriate dynamic range). The hierarchical beamformer reduces the number of data streams at each level of the hierarchical structure, and performs partial weighting and aggregation in the process. Pulse beamformer is also similar, but instead of summarizing data at a given stage, it sends data along a series of nodes or units. Part of the beam data is summarized in this process to be used in subsequent processing stages. Output. Almost every medium-sized digital array known to the author is using some form of layering/pulse processing to implement a digital front end. Importantly, unlike analog arrays, the use of layered/pulsed beamforming can balance the number of beams with the signal bandwidth in the digital domain, so that the fixed overall beam-bandwidth product remains approximately constant at each point of the front-end processing chain .

For a multi-level structure, the interconnection cost is proportional to the logarithm of the number of units M, and the data and front-end processing are roughly linearly proportional to M. Both are proportional to the bandwidth of the entire system. These types of considerations have guiding significance for the design of any front-end DBF architecture in the trade-offs of calibration, beamforming, and adaptation. In the end, RapidIO can support any network architecture, such as a folded ring can reduce latency and improve reliability, which will be explored in the future.

Figure 5 shows the laboratory measurement results of the mobile demonstrator9. This fully digital active dual-polarized phased array antenna is designed to completely control the transmit and receive signals of each antenna unit. Compared with the WSR-88D parabolic antenna, the antenna design of the ARRC project focuses on achieving the same function or improving performance. In view of the fact that meteorological missions have higher requirements for polarization than aircraft surveillance missions in terms of target recognition, these design specifications are crucial. Dual-polarization radar requires both a low cross-polarization level (less than -40dB) and a well-matched pattern (less than 0.1dB) to successfully determine the polarization variable of the scanned atmosphere.

Generally, as the cross-polarization level of an antenna increases, all deviations in the polarization variable increase. During the PCB design of the 8*8 array, many elements in the antenna unit were studied. These elements include: edge diffraction suppression; bandwidth with a center frequency of 2.8GHz over 10%; isolation between ports is about -50dB; azimuth angle ±60º In the scanning range, the cross-polarization level is lower than -45dB and the co-polarization mismatch is lower than 0.1dB within the scanning range of the pitch angle ±10º. After careful calibration, the active reflection coefficient of at least -10dB can be obtained for the scanning range of azimuth angle ±60º and the scanning range of elevation angle ±10º. Therefore, this designed a new type of stacked cross microstrip patch radiator with electromagnetic coupling for Horus. The leftmost picture in Figure 5 is an 8*8 panel of these radiators. The radiator and the feed network are divided into two different parts to prevent them from bending after manufacturing. The radiator consists of two conductive layers and a radome bonded by RT/Duroid 5880LZ and RO4450F.