Multiphysics simulations for 5G RFICs and SoCs

The transition to 5G is exciting but no small task given the degree of complexity at various points in the system. Multiphysics simulations simultaneously solve power, thermal, variability, timing, electromagnetics and reliability challenges across the spectrum of chip, package and system to promote first-time silicon and system success.

Laurent Ntibarikure S ystem-on-chips (SoCs) and radio frequency integrated circuits (RFICs) for 5G smartphones and networks need to manage huge amounts of antenna data and offer significantly high processing capabilities in thermally and power-constrained environments. The growing interdependence of various multiphysics effects like timing, power, electromagnetics, thermal and reliability in sub-16nm designs poses significant challenges for design closure. Traditional margindriven, silo-based design approaches to the chip, package and board have limited simulation coverage and fail to unravel potential design weaknesses, causing field failures.

Multiphysics simulations simultaneously solve power, thermal, variability, timing, electromagnetics and reliability challenges across the spectrum of chip, package and system. Early analysis is key, but this comes with some requirements for SoC and intellectual property (IP) designs. The most important ones pertain to power efficiency, power integrity, reliability, advanced packaging and electromagnetic crosstalk.

Power The shift from 4G to 5G is expected to deliver a spike in cell edge data rates from 10 Mb/s to more than 1 Gb/s, plus a 50 percent gain in energy efficiency. For evolving generations of 5G implementations, the main focus is on predicting power profiling early in the chip design phases. Specific attention has to be paid to spectrumrelated issues, traffic characteristics, radio interference and interoperability and network access-related issues.

Power efficiency is a key design consideration for 5G devices. Average power, peak power, peak change in power and sustained worst-case average power are all important for thermal robustness, power integrity and cost of system operation. Early feedback is critical to achieving 5G power targets.

For 5G SoCs, power grid signoff through traditional approaches isn’t feasible. This is due to severe routing constraints that can potentially cause timing convergence issues down

The need to model electromagnetic effects from DC up to mm-wave calls for special handling of layouts.

stream. For advanced FinFET technology processes, the power grid’s node count is very high and any reduction in node count will affect accuracy. With very small design margins, power signoff solutions leave little margin for error. The slightest inaccuracy can result in product failure. It’s important, therefore, to analyze the entire power grid flat rather than partitioning the design with a “divide and conquer” approach.

Reliability Design for reliability is another key consideration for advanced SoCs used in 5G communication systems. These SoCs, for example, will be instrumental in enabling future mission-critical applications like self-driving cars. Reliability issues can be challenging at advanced Fin

FET nodes. FinFET designs have a high dynamic power density, and power directly impacts the chip’s thermal signature. Accurately modeling the temperature distribution onchip by considering the chip in the context of the system is critical for ensuring its reliable operation.

Electrostatic discharge (ESD) design and verification are also becoming extremely challenging with the prominent use of IPs and high-speed interfaces in SoCs. ESD checks are now one of the key signoff metrics. Almost 55 percent of the failures are interconnect-related and can be avoided by performing systematic ESD checks during the design phase. But ESD protection that works at the IP level may not work at the SoC level due to poor connectivity to other IPs and circuits in the SoC. Therefore, it’s important to analyze the ESD protection schemes at the SoC level, across multiple voltage domains, to make sure they provide the intended low-resistance path for discharging a potential ESD event without stressing the functional devices.

Packaging Advanced packaging technologies will be the key driver of heterogeneous integrations in next-generation edge compute data centers and 5G electronics systems to achieve extreme performance, high system

bandwidth, low power and low cost. The Internet of Everything – enabled by 5G infrastructure – will generate huge amounts of data to be processed and stored. The ability to handle such large volumes of data will be threatened by limited system bandwidth between the traditionally packaged processor and the memory integrated into the system.

Hence, advanced 2.5D/3D IC packaging technology will become a popular choice for 5G system designs. Short interconnection paths enabled by through-silicon vias between stacked chips lead to higher performance because of increased I/O speed. They also consume lower power because of their reduced capacitance and their smaller form factor due to the stacking of multiple dies. This is indeed a very promising technology, although it’s fraught with many challenges owing to its complexity.

Electromagnetics Integrating a high-power beamforming module with sensitive analog and RF circuitry can lead to substrate noise propagation between the two, which can impact the overall performance. For accurate power noise analysis, it’s important for a designer to model the propagation of substrate noise in a dynamic voltage drop analysis. Integrating the digital beamforming module with sensitive analog and RF circuitry can cause switching noise to propagate through the substrate if insufficient isolation is guaranteed.

For reliable operation of the millimeter-wave RF module, it’s critical to have a methodology that allows for modeling of the substrate noise generated in a digital beamforming module using digital noise injection. The methodology also needs to perform analyses to determine the frequency and time-domain response of the analog/RF blocks.

For 5G, RF front-end circuits, high-performance reference oscillators and associated interconnects must be designed properly to ensure reliable operation at 6 GHz up to mm-wave frequencies. On-chip mixed-signal components are affected by electromagnetic effects and their design considerations should include self- and cross-coupling among various sensitive mixed-signal circuit blocks. Careful examination of the layout, parasitic inductance and capacitance, substrate modeling and trace resistance is critical for reliability. The need to model electromagnetic effects from DC up to mm-wave calls for special handling of layouts.

Laurent Ntibarikure is an application engineer at Ansys.

Bram Nauta is a professor of IC design at the University of Twente.

Smoke signals

Along time ago, we already had wireless communication. Like we know from the western movies: people were making a small fire on a hilltop and with a piece of cloth, they shaped the stream of smoke in a sort of on-off-keying modulation.

Since we’re among nerds here anyway, let’s have a look at how power efficient this type of communication was. First, the harvesting of the energy we need: a decent fire has – say – 5 logs of wood, which burn well for half an hour. Extrapolated to one year, 24/7 communication requires 90,000 logs per year. With 500 logs per cubic meter of wood, this is 180 m³ per year. In a typical forest, wood grows with 7 m³ per 10,000 m² per year, so we need 250,000 m² of forest to keep this single fire burning.

Now the bitrate: my guess is that with a few smoke symbols and a bit of practice, one can send about 2 bits per second. This is 64 megabit/year.

My mobile subscription today gives me 20 gigabyte/month, which is 1920 gigabit/year. This is equivalent to 30,000 fires in parallel, requiring 7,500 km² of forest to keep my communication going. And since communication to my ‘base station’ is two-way, we can safely say I need 15,000 km² of forest.

The earth has about 200,000,000 km² of land, so in theory, there’s room for a maximum of only 13,000 people to communicate like me on this planet. So, I fully understand that the people making these kinds of calculations back then freaked out and searched for other ways of wireless communication.

We invented electrical energy, batteries, radio communication and cheap microchips and now billions of people can communicate wirelessly gigabits per day with simple handheld devices at very low cost. What progress! Yes, we want more! We want the Internet of Things, 5G, 6G, more bandwidth! We want to increase our bandwidth with at least a factor 100 in the next decade.

A conservative estimation says that all wireless communication today consumes about 1 percent of the total energy used on this planet. And given that power dissipation in wireless communication scales linearly

The 100 times 1 percent becomes a bit of a problem

with bandwidth, without additional measures, the 100 times 1 percent becomes a bit of a problem.

Communication roadmaps predict the migration to “free frequency space” at tens of gigahertz frequencies because that’s where the free bandwidth is. But at those frequencies, communication will be limited to line of sight and relatively short distances. Beamforming techniques to aim those radio beams require a lot of transmitters and receivers per terminal. The 5G terminals being developed at 28 GHz already have a serious cooling problem. Without cooling, they may catch fire! (And we don’t want to go back to smoke signals.)

But we actually may use a modern version of smoke signals: for certain IoT applications, for example, we may actually go back in time and ‘see’ the bits again. If we have a short distance and line of sight anyway, why not make an ultra-low-power monochrome display – like an e-book – tag, on which a type of QR code is visible representing the data. One central optical camera can then see a lot of these tags and receive the information in a massively parallel way at very low power.

Finally, we should re-think radio communication completely. We’re still sending too much radio power, too long, to places where it’s not being used, and that even harms other users. I’m sure we can come up with other ideas if we think energycentric. We might go back to our oldschool IR remote control technology or even throw USB sticks to each other. But anything is better than smoke signals.