Traditionally, electronics have been designed with static form factors to serve designated purposes. This approach has been an optimal direction for maintaining the overall device performance and reliability for targeted applications. As the world enters the age of ubiquitous computing, the need for reconfigurable hardware operating close to the fundamental limits of energy consumption becomes increasingly pressing. Militaries are demanding technologies that could enable reconfigurable multi-function systems able to switch quickly between radar, electronic warfare (EW), signals intelligence (SIGINT), and communications. This kind of equipment has the potential to adapt to different conditions and operating modes to change rapidly depending on pressing military mission needs and threats.
Reconfigurable electronics might use reprogrammable RF and microwave components and signal-processing components such as field-programmable gate arrays that can change their modes and capabilities quickly. Other technologies could involve new software algorithms able to learn how to collect the most valuable information autonomously. Another technique is to use new materials with built in reconfigurality.
Reconfigurability refers to specific features that can be changed (to include dynamic and partial reconfigurability), whereas adaptiveness addresses the behavioral constructs that inform how reconfigurable features might best be exploited in live use conditions. They can have synergistic effects, as in a system that might be able to morph in response to changing conditions (such as battery saving, reduced bandwidth, noisy communications channels) by altering a few of its own ‘‘knobs’’ or, more radically, by reshaping and replacing complex circuits in real time, says IEEE.
Adaptable electronics is generating significant interest in the scientific community because of the many applications. Imagine for a moment that one single microchip was capable of accomplishing the tasks of several different circuits. For example, a circuit assigned to process sound information could, when not being used for this purpose, be reassigned to process images. This would allow us to miniaturise our electronic devices.
At the same time, it would become possible to develop resilient circuits. Whenever a microchip is damaged, it could theoretically reconfigure itself so that it could still function using the components that remain intact. “An effective way to keep faulty devices working when they are in hard-to-reach places, like space,” says Leo McGilly, the article’s lead author.
Adaptive reconfiguration can also be used as a technique to expand the operational envelope of analog electronics for extreme environments (EE). Conventional circuit design exploits device characteristics within a certain temperature/radiation range; when that is exceeded, the circuit function deviates from its set-point. On a reconfigurable device, although component parameters change in EE, a new circuit design suitable for new parameter values may be mapped into the reconfigurable structure to recover the initial circuit function. Partly degraded resources can still be used, while completely damaged resources may be bypassed.
Laboratory demonstrations of this hardening-by-reconfiguration technique were performed by JPL in several independent experiments in which bulk CMOS reconfigurable devices were exposed to, and functionally altered by, low temperatures (~ −196°C), high temperatures (~280°C) or radiation (300kRad TID), and then recovered by adaptive reconfiguration using evolutionary search algorithms.
If electronic circuits could automatically reconfigure their internal conductive pathways as required, microchips could function as many different circuits on the one device. If many of these devices were then incorporated into larger pieces of equipment, such as robots, it is possible that self-sufficient, self-sustaining machines could change to suit their environment or even reconfigure broken or damaged pathways to repair themselves.
Highly Configurable RF Platform Supports Increased Capacity, Functional Modes
In 2020, Analog Devices is introducing what it calls a mixed-signal front-end (MxFE) RF data-converter platform. It combines high-performance analog processing with digital signal for 4G LTE and E-band 5G mmWave radios, broadband systems, and even phased-array radar, electronic warfare, and test-and-measurement systems.
The MxFE platform embeds digital-signal-processing (DSP) functions on-chip so that the user can configure the programmable filters and digital up- and down-conversion blocks to meet specific radio-signal bandwidth requirements. The high level of integration lowers chip count and results in a 60% reduction in PCB area compared to alternative devices. As a major added benefit, the configuration yields a 10X reduction in power dissipation, when contrasted to architectures that use the system FPGA for RF conversion and filtering. Commensurate benefits are due to newly available FPGA resources, or the option to use a smaller, less expensive FPGA.
“Cell towers are nearing saturation based on the number of antennas they must support, and our customers want lighter-weight, multiband radios that fit into today’s radio form factor,” says Kimo Tam, general manager, High-Speed Mixed-Signal group, Analog Devices. “They are also asking for software-defined RF platforms with the configurability and scalability to enable one platform to be used across multiple geographies and use cases.”
The AD9081 integrates eight RF data converters (six for the similar AD9802), enabling designers to upgrade to multiple wireless radios from single-band units in the same footprint (see figure). This results in an increase in call capacity of up to 3X for 4G LTE stations. Both devices offer what’s claimed to be industry’s widest instantaneous-signal bandwidth (up to 2.4 GHz), which simplifies hardware design by reducing the number of frequency-translation stages while also relaxing filter requirements.
The AD9081 has flexible interpolation/decimation configurations that enable direct-to-RF multiband radio applications. It supports four transmitter channels and four receiver channels with a 4D4A configuration via a 16-bit, 12-Gsample/s maximum sample-rate RF-DAC core and a 12-bit, 4-Gsample/s rate RF-ADC core. In addition, the device supports a complex transmit input data rate up to 6 Gsamples/s and a receive output data rate (in single-channel mode) up to 4 Gsamples/s. Maximum radio band spacing supported in multichannel modes is 1.2 GHz. It also features a 16-lane, 24.75-Gbit/s JESD204C or 15.5-Gbit/s JESD204B data-transceiver port, an on-chip clock multiplier, and DSP capability targeted at single- and dual-band direct-to-RF radio applications.
Target dissipation for these ICs, housed in 324-lead thermally enhanced BGA packages, is between 6 and 7 W. Samples of the AD9081 (quad 12-bit 4-Gsample/s ADCs; quad 16-bit 12-Gsample/s DACs) are expected in September 2019 with full production in March 2020. Samples of the AD9082 (dual 12-bit 6-Gsample/s ADCs; quad 16-bit 12-Gsample/s DACs) are also expected in September, but full production is slated for a month earlier. Both devices will be priced in the $1,500 range (1,000-piece lots).
Reconfigurable electronics based on memresitors
Emerging nanoelectronics technologies bring forth new prospects yet a significant rethink of electronics design is required for realising their full potential. The rich landscape of modern electronics design became even more diverse with the steady introduction of memristive devices into the family of standard electronic components. The ability of memristors to act as thresholded electrically tuneable, multi-level, non-volatile resistive loads, combined with their inherently scaling-friendly, low power and back-end integrable fabrication processes has rendered them a highly promising candidate for use in future electronics applications. These properties promote memristors as ideal candidates for achieving in silico reconfigurability in a post-Moore context, i.e., without relying on front-end integration density for performance and operating on the principle of separate, dedicated memory and processing elements.
In 2018, the University of Southampton has been awarded a multi-million pound programme to lead the development of innovative nanotechnology that could open the door to a new generation of electronics. Professor Themis Prodromakis from the University of Southampton is the principal investigator of the predominantly Engineering and Physical Sciences Research Council (EPSRC) funded programme, which, along with industrial contributions, exceeds £11million.
Working with Imperial College London and the University of Manchester, as well as industrial partners, the project will centre on memristors and their ability to enable electronics systems to be configured with increased capability, as opposed to transistors. Prof Themis Prodromakis said: “Memristor technologies bring great prospects for next-generation chips, which need to be highly reconfigurable yet affordable, scalable and energy-efficient, not to mention secure. Traditionally, the processing of data in electronics has relied on integrated circuits (chips) featuring vast numbers of transistors – microscopic switches that control the flow of electrical current by turning it on or off. The size of transistors has reduced to meet the increasing demands of technology, but are now reaching their physical limit, with – for example – the processing chips that power smartphones containing an average of five billion transistors.
Memristors could hold the key to a new era in electronics, being both smaller and simpler in form than transistors, low-energy, and with the ability to retain data by ‘remembering’ the amount of charge that has passed through them – potentially resulting in computers that switch on and off instantly and never forget. The University of Southampton has previously demonstrated a new memristor technology that can store up to 128 discernible memory states per switch, almost four times more than previously reported.
Professor Themis Prodromakis added: “For decades we have followed the pattern that computers should have separate processor and memory units, but these are now struggling to cope with the masses of data in the public domain. Soon the span of functionality in future Internet of Things (IoT) systems will be much wider than what we know from today’s smartphones, tablets or smart watches.
“This unique programme of activities will allow us to develop reconfigurable electronic systems that are at the forefront of innovation through being embedded almost everywhere in our physical world; within vehicles and infrastructure or even within the human body. “We are thrilled that our vision is shared with world-leading industry and together we look forward in bringing the change in modern electronics.”
Electronic circuits with reconfigurable pathways closer to reality
At EPFL, the researchers demonstrated that it was possible to control the formation of walls on a film of ferroelectric material, and thus to create pathways where they wanted at given sites. The trick lies in producing a sandwich-like structure with platinum components on the outside and a ferroelectric material on the inside. Underlying this promising technology are so-called ‘ferroelectric’ materials in which it is possible to create flexible conductive pathways. These pathways are generated by applying an electric field to the material. More specifically, when the electric current is applied, certain atoms moves either “up” or “down,” which is known as polarisation. In recent years, the academic world has observed that conductive pathways several atoms wide – called ‘walls’ – form between these polarized zones. The only problem is that, until now, it was impossible to control how these pathways form.
“By applying electric fields locally on the metal part, we were able to create pathways at different sites and move them, and also to destroy them with a reverse electric field,” says Mc Gilly. Low conductive electrodes were used to surround the ferroelectric material. This means that the charge spreads very slowly in the structure, making it possible to control exactly where it is applied. “When we use highly conductive materials, the charge spreads rapidly and walls form randomly in the material.” Specifically, lead zirconium titanate (Pb(Zr,Ti)O3) – a ceramic perovskite material that shows a marked piezoelectric effect – was used by the researchers, in which they created conductive pathways parallel to an applied electromagnetic field. Known as “walls,” these conductive paths form between polarized zones of atoms. However, until the EPFL research, previous work had shown how exceptionally difficult it was to control the way these pathways formed.
At this point, the researchers have tested their research on isolated materials. The next step consists in developing a prototype of a reconfigurable circuit. Leo McGilly would go even further. “The fact that we can generate pathways wherever we want could allow us to imitate in the future phenomena that take place inside the brain, with the regular creation of new synapses. This could prove useful in reproducing the phenomenon of learning in an artificial brain.”
Reconfigurable Electronics: Disappearing Carbon Circuits on Graphene
Using carbon atoms deposited on graphene with a focused electron beam process, Fedorov and collaborators have demonstrated a technique for creating dynamic patterns on graphene surfaces. The patterns could be used to make reconfigurable electronic circuits, which evolve over a period of hours before ultimately disappearing into a new electronic state of the graphene.
“We will now be able to draw electronic circuits that evolve over time,” said Andrei Fedorov, a professor in the George W. Woodruff School of Mechanical Engineering at Georgia Tech. “You could design a circuit that operates one way now, but after waiting a day for the carbon to diffuse over the graphene surface, you would no longer have an electronic device. Today the device would do one thing; tomorrow it would do something entirely different.” The change usually occurs over tens of hours, and ultimately converts positively-charged (p-doped) surface regions to surfaces with a uniformly negative charge (n-doped) while forming an intermediate p-n junction domain in the cours“
There are multiple ways to modulate the dynamic state, through changing the temperature because that controls the diffusion rate of carbon, by directing the atomic flow, or by changing the carbon phase,” Fedorov said. “The carbon deposited through the focused electron beam induced deposition (FEBID) process is linked to graphene very loosely through van der Waals interactions, so it is mobile.”e of this evolution. “The electronic structures continuously change over time,” Fedorov explained. “That gives you a reconfigurable device, especially since our carbon deposition is done not using bulk films, but rather an electron beam that is used to draw where you want a negatively-doped domain to exist.”
Beyond allowing fabrication of disappearing circuits, the technology could be used as a form of timed release in which the dissipation of the carbon patterns could control other processes, such as the release of biomolecules. Fedorov and his collaborators have so far shown only the ability to create simple patterns of charged domains in the graphene. Their next step will be to use their p-n junctions to create devices that would operate for specific periods of time. Reported in the journal Nanoscale, the research was primarily supported by the U.S. Department of Energy Office of Science, and involved collaboration with researchers from the Air Force Research Laboratory (AFRL), supported by the Air Force Office of Scientific Research.