Most silicon (Si) used in defense applications is designed by domestic chipmakers but fabricated in overseas foundries, which introduces vulnerabilities to the microelectronics supply chain—from weather catastrophes to geo-political disruption. To reduce these risks, technology organizations with expertise building products for aerospace and defense are expanding onshore microelectronics facilities to enhance their design and manufacturing capabilities.
One of the early initiatives in the industry is the Mercury Systems microelectronics manufacturing facility in Phoenix. The company made a $15 million investment in 2019 to upgrade the facility and enhance its 2.5D and 3D custom microelectronics capabilities, followed by another facility expansion in 2022. Mercury’s manufacturing facility is recognized as a trusted U.S. defense microelectronics supplier, DMEA accredited. Having a trusted onshore supplier enables military customers, including the Department of Defense (DOD), to access the leading commercial Si technologies, and using the trusted open system architecture addresses the DOD requirement for onshore manufacturing of critical microelectronics.
As a proof point of this capability, Mercury Systems has developed an RF system in package (RF SiP) that integrates high speed data converters, digital processing, memory and power management in a single package. The RFS1140 is the first to combine data converters operating at 64 GSPS with powerful AI core FPGA processing and power management.
Heterogeneous integration of these circuit functions reduces system cost, size and complexity. Perhaps more important, it reduces latency because the SiP can be placed close to an antenna, enabling new sensor processing applications for defense platforms and programs (see Figure 1). Countering the latest radar and electronic warfare threats requires extremely low latency responses driven by an intelligent, adaptive strategy. Addressing this rapidly evolving need requires innovation from chip to system, which defined the powerful functionality integrated in the RFS1140.
The RSF1140 uses the AMD Xilinx Versal AI Edge adaptive compute acceleration platform (ACAP), a heterogeneous processor incorporating three types of compute engines and fabricated in 7 nm CMOS technology. Complemented with high speed direct digitization up to 32 GHz and four channels each of analog-to-digital and digital-to-analog data conversion at 64 GSPS per channel, the RFS1140 provides ultra-low latency processing. Integrating on-chip memory and power management in the SiP completes the processing system and contributes to overall system longevity (see Figure 2).
Versal devices are the industry’s first ACAP, combining adaptable processing and acceleration engines with programmable logic and configurable connectivity. The customized heterogeneous hardware enables applications such as data center, automotive, 5G and wired telecom, as well as defense. With transformational technologies like intelligent engines, adaptable engines and scalar engines with an integrated Si host interconnect shell, the Versal device provides superior performance per watt compared to conventional FPGAs, CPUs and GPUs.
To track potential targets, including those moving at hypersonic speeds, the RF SiP uses high speed data converters to provide the massive processing required. Two 10-bit Jariet Electra-MA transceivers in each RFS1140 directly digitize waveforms, from 40 to 64 GSPS per channel and analog frequencies as high as 36 GHz, with instantaneous bandwidth exceeding 4 GHz. Designed and fabricated in the U.S., the converters are fabricated on a 14 nm CMOS process, which provides high DC power efficiency.
Each transceiver channel in the RF SiP uses interleaved analog-to-digital and digital-to-analog converters followed by programmable digital up- and down-conversion; linear equalization, decimation and interpolation; and a 16-bit SerDes baseband data interface.
TRUSTED AND SECURE
The RFS1140 enables edge processing by maximizing performance through a highly customizable architecture developed, designed and manufactured from a source trusted for DOD programs. The RF SiP offers direct digitization and massive processing, minimizing SWaP-C by eliminating multiple boards required in a traditional system.
By using this SiP architecture, future generations of semiconductors created by continuous commercial R&D investments can be used to rapidly upgrade military systems. Systems using the RF SiP will be able to add new capabilities in the same physical form factor, saving development cost and speeding system upgrades, while decreasing the complexity of upgrade cycles.
More than ever, critical military systems such as EW, phased array radar and C5ISR need trusted onshore microelectronics from a company with expertise in secure, heterogeneous packaging. The advanced SiP technology from Mercury Systems’ onshore design and manufacturing capability brings commercial technology to mission-critical applications, enabling sensor processing at the edge.