Regionalization Semiconductor Technology for Critical Infrastructure – Analog Devices
John Cowles - Senior Director of Technology
It has been recognized for years that advanced semiconductor manufacturing and packaging capabilities mostly shifted to east Asia over the last several decades. Competitive costs, continuous reinvestment in next generation technologies and excellent customer service all justified this regional consolidation. The near-fabless model in the West allowed for a flexible, as-needed outsourcing model and obviated the need for huge capital and operational investment in fabs and packaging facilities. In just a couple of years, the geopolitical strains around the world and the supply chain misalignment to demand during Covid have raised alarms across the West around the availability, security and integrity of critical technologies that underpin its infrastructure. This goes far beyond Defense applications; it encompasses all critical segments like Telecom, Energy, Automotive and Industrial, all of which are benefiting from increases electrification and semiconductor content.
- Semiconductors will increasingly be viewed as a key element of national and regional security. The increased dependence of nations, governments, industries and citizens on the electronics behind the infrastructure will make access to these technologies national priorities, either through strategic alliances or on-shoring critical capabilities. Government investments and policies will pair with industry to ensure the supremacy in semiconductor research and manufacturing. Both the Chips for America Act and the European Chip Act will inject critical funding in the US and Europe to incentivize the private, commercial sector to invest further in on-shoring critical semiconductor manufacturing.
- Investments across multiple regions will skyrocket to provide more resilience against supply chain disruption, whether by natural or geopolitical events. Such investments are also likely to drive local policies to prioritize domestic sources over foreign ones. The question of scale within smaller geographies will challenge their profitability. The end-result is likely higher costs of semiconductors across the board.
- Different models will exist for different technologies. Legacy nodes on 8 inch wafers like 0.18um will require a different approach than 65/28nm nodes on 12 inch wafers which will differ from the most advanced FinFET-based nodes at 16nm and below. Then there are specialized technologies like BCD-CMOS, SiGe, SOI, GaAs/GaN and MEMS that may be lower volume but are nevertheless critical in certain applications. The scale and costs for each of these families will dictate whether internal, dedicated manufacturing investments are needed or shared, foundry entities need to be established.
- Besides the availability of the components, there is growing concern with the intrinsic security of the underlying IP and tampering of the content. There are creative ways being proposed to address these issues with everything from mixed manufacturing where the IP is applied late in the process at a secure facility to clever identification of unwanted changes to the IP. Even today, FPGAs do not host IP until they are programmed close to the end application. As long as the hardware has not been tampered and malicious software has not been embedded before shipping, the risks are much lower than a digital ASIC. A systematic process of assessing these risks will become part of the decision tree around component selection and possibly drive the system architecture around security risk, rather than just cost, schedule and performance.
- While the focus of investments has been primarily around semiconductor manufacturing, packaging is also of critical importance. Lead times attributed to packaging such as lead-frames and laminates, are a significant contributor to the overall delays that have reached an astounding 12 months or longer. The dynamics around packaging are different from semiconductor manufacturing. While a fab requires huge capex investments in advanced equipment and installations, a packaging facility is more heavily influenced by opex around labor. Local investments in packaging to accompany those in semiconductor manufacturing are lower but also need to be made.
Multi-Function Radar and Electronic Warfare (EW)
Ben Annino - Systems and Applications Director
Multi-Function Radar and Electronic Warfare (EW) applications put a high value on wideband, high dynamic range, agile spectral monitoring. Digitization has been progressively inching closer to the antenna enabling wider instantaneous BW, mission reconfigurability and smaller SWAP-C, all while maintaining or even improving dynamic range. The coming year will see increasingly higher sample rate data converters allowing architecture changes to the radio front end that shrinks SWAP-C, maintains performance, and evolves toward software programmability. We predict that the wideband digital radio is on the verge of a complete architectural overhaul.
Spectral Sensing In Olden Days
Not too long ago, data converters could only cover a few hundred MHz instantaneous bandwidth (iBW). They’d be sampled up to a few GSPS, and the bandwidth might be centered around DC (Zero-IF aka ZIF) or centered around an IF offset. ZIF requires IQ modulators and demodulators as well as quadrature error correction (QEC), which make it unattractive for many ADEF applications requiring very wide iBW and spectral coverage. Bandwidth-hungry Radar and EW prefer the latter IF sampling approach, direct sampling wide iBW in the 1st and 2nd Nyquist zones. To cover spectrum outside the Nyquist zones, an RF tuner uses a swept LO mixer to frequency translate a sliding block of IBW into the fixed IF that matches up with the data converter direct sample zone. Due to the low IF sampling, old systems use dual mixer stages in order to get adequate image rejection. These are large and expensive.
Spectral Sensing Today (MXFE)
Today’s wideband spectral sensing approach has improved because present day ADC sample rates in the several-GHz range are high enough that you can direct sample the intermediate IF following that first mixer above. The RF tuner often doesn’t need 2 mixer stages anymore. The 2nd Nyquist IF direct sampling is high enough in frequency to allow adequate frequency spacing of the desired input RF band and image band so that an attainable RF filter can do the job. RF filters are still complex, large and expensive, but big savings is had eliminating an entire frequency translation stage.
Spectral Sensing in the Near Future
The coming year will see high sample rate RF sampling digital converters that gets us to true wideband software defined radio. These digital converters direct samples at an even higher IF, separating the desired and image band far enough that lower-Q tunable MMIC filters are adequate. You will finally eliminate a great number of planar hi-Q ceramic filters which is a big SWAP-C savings. Even better, the filters will go from fixed (every use case has a custom set of filters) to tunable. This means a single wideband hardware configuration will be software programmed to optimize the right performance trade for many customer frequency schemes across many use-cases.
Integrated frequency translation ICs employing sub-octave RF filtering and gain control were hard to nail down in the past because everybody’s use case, frequency plan, and resulting RF/IF filtering was different and high performance. Things are about to change drastically. The single IC integrated tuner will be natively wideband with built-in RF filtering capability to software-define operating bands and rejection bands. So, the integrated tuner becomes more one-size-fits-many across customer use-cases and frequency plans.
In the coming year we will see radio performance differentiation accelerate its shift from high performance RF to high performance RF direct sampling, with an emphasis on software defined, adaptive RF signal conditioning. Flexibility and multi-function configurability becomes the focus.
2022 Predictions Phased Array Radar and Development Platforms
Jerome Patoux, Marketing Director - Aerospace & Defense
Jon Bentley Product Marketing Director – Aerospace & Defense
Phased Array technology continues to evolve driven by a need for increasing performance, flexibility and capability. Hybrid architectures employing distributed mixed signal converter nodes feeding RF beamforming subarrays are commonplace today. In the coming year we will see a continuing trend to more mixed signal nodes feeding smaller RF sub-arrays as RF sampling pushes toward the individual elements. Challenges such as digital processing bandwidth at acceptable cost and power will continue to prevent wideband every-element digital beamforming in the near term.
Large investments are pushing rapid advancement in mixed signal data converter bandwidth and power efficiency, every-element wideband digital beamforming is becoming more practical. This trend will continue into 2022 and gather pace as investments in semiconductor technology (i.e. Silicon, Gallium Nitride and Gallium Arsenide processes), integration and digital processing capabilities, make full elemental digital beamforming at higher frequencies more realizable, efficient, and scalable across large arrays.
At higher frequencies, challenges with data throughput are linked to higher processing requirements for the baseband processors and power consumption will increase. To cope with this problem, some compromises are made with regards to the converter performance (lower resolution and power). This leads to usually undesirable trade-offs in signal degradation and overall performance and flexibility at the system level.
With more converter channels and with those converters located closer to the array antenna elements, array gain improves SNR but front end adaptive RF signal conditioning is required to preserve dynamic range in blocker environments. Digital beamforming better supports adjustments to the mission or multiple missions and this can be all configurable via software. The multi-mission capability of digital beamforming systems allows for Size and Weight optimization in space constrained radar end equipment, such as airborne systems.
New solutions and platforms alleviate these challenges by offering lower power, ultra-high performance mixed converter front ends, standalone or implemented as part of sub-system solutions or development platforms. These sub-systems or platforms reduce engineering efforts and time to market for radar designers who seek industry leading performance while optimizing size, weight, power and cost (SWAP-C) with higher reliability even in severe environments.
Incumbent hybrid architectures offering digital simplicity and trade-offs in SWaP-C will remain. An X-Band hybrid beamforming bits-to-beam radar development platform that offers a complete 32 channel transmit and 32 channel receive antenna-to-bits solution is an example of a platform that demonstrates full array performance from antennas to bits and helps customers design-in more quickly so they can focus on higher system level issues. This platform integrates ADC and DAC solutions with RF microwave up and down converter circuitry and integrated analog beamforming ICs to evaluate a full hybrid beamforming signal chain solution.
Another example is the 16 channel transmit and receive Quad MxFE phased array direct sampling system solution, supporting L, S and C bands that can assist in the design of a full digital beamforming system. This architecture showcases multiple-chip synchronization and system level calibrations. The system plugs into a COTS FPGA board with reference HDL code and MATLAB software. A separate calibration board is used to facilitate calibrations and verification of system level phase noise, spurious and dynamic range performances.
Digital beamforming offers several benefits particularly the flexibility to program multiple antenna beams simultaneously in many directions but has its challenges with synchronization, SWAP-C trade-offs and processing the high volume of digital data. A system alternative that we believe will remain for some time is to use a subarray architecture with a mix of both analog and digital beamforming where a full digital implementation is considered not practical or constrained by system cost.
Development platforms that showcase capability, help reduce time to market and solve complex engineering challenges faced during the design cycle, start to gather momentum. We predict an uptake of these platforms to assist in current and future R&D efforts.
Go to the next page for Renesas predictions.