The mobile industry seems to have just finished the World Mobile Communications Conference (MWC) Carnival in Barcelona, ​​Spain. Suppliers, system OEMs and mobile operators in the technology industry are immediately facing a series of 5G development obstacles that need to be resolved. In fact, these questions are just a starting point.

The technical issues of 5G development are multifaceted. Among them, smart antennas and radio frequency (RF) front ends for 5G millimeter wave (mmWave)-which are expected to be implemented at 28GHz, 39GHz or 60GHz frequencies-may seriously affect the performance of 5G mmWave phones that have not yet appeared.

Claire Troadec, head of the RF electronics business of Yole Développement, accepted an interview with EETimes after MWC. She said: "Although many companies like Qualcomm, Intel, MediaTrk and Samsung use mobile phones as their 5G mmWave display platforms, we believe that mobile phones will not be the first to implement 5G mmWave applications.” In contrast, 5G mmWave is more likely to become a stationary computer on a desk or desktop, allowing consumers to download or stream a large number of broadband applications.

NXP wireless network solution introduction

Claire Troadec
why? This is actually because the 5G mmWave band has problems such as high propagation loss, directivity, and sensitivity to obstacles. Therefore, it is not an easy task to design a 5G mobile phone that can always operate without losing the signal. Imagine that consumers may therefore be forced to stay on a certain page and must constantly look for other signals.

Another challenge in deploying 5G mmWave wireless signals in mobile phones is battery life and power consumption. During the 2018 Winter Olympics (Winter Olympics 2018) held in Pyeongchang this year, Samsung reportedly demonstrated its own 5G tablet. Although the operation of this device is very smooth, but the evaluation on MWC is quite surprising: the battery runs out after 30 minutes.

In response to this rumor, Troadec believes that “the 5G mmWave signal transmission of mobile phones will have excessive power consumption.” She guessed that “most leading manufacturers should be paying attention to this field.” But she added that she found these Technology vendors have not proposed many remedies for this obvious system-level power consumption problem in 5G New Radio (5G New Radio; 5G NR) applications. She said that no one wants to discuss this issue further at the exhibition.

The interference that 5G mmWave RF modules will bring to the emerging 5G market is not limited to technological changes. Also deeply affected are the entire "industrial supply chain" that currently supplies 3G and 4G RF components and modules.

Yole explained that because 5G mmWave allows suppliers to use CMOS or SOI technology to design RF front-end modules in SoCs, this field will open to the RF market for “advanced CMOS designers and manufacturers” currently in the mobile phone ecosystem architecture Door. In addition to Intel and Qualcomm, players in this field include Samsung, Huawei and MediaTek.

The more 5G frequency bands, the more RF front-end modules
As technology suppliers begin to work on building complex RF front-end (RFFE) modules that can handle increasing frequency bands, the mobile industry has made great progress. According to Troadec, with the development of cellular standards from 3G to 4G, the number of frequency bands that the RF front-end must respond to has increased significantly from 4 to 30.

The number of frequency bands supported in smartphones is increasing, which only increases the complexity of the RF front-end.

As 5G technology and applications are gradually put in place, the situation will become more complicated. Although 5G is a standard in theory, it has three important components: 5G for the Internet of Things (IoT), 5G using the sub-6 GHz frequency band, and 5G using mmWave. In terms of RF technology, Troadec observed that "this means that components with different performance must be integrated."

This means that 5G will follow the direction of "parallel development of different 5G versions at different implementation stages." In other words, there will not be a unified 5G RFFE, but "5G IoT, 5G sub-6 GHz, and 5G mmWave will develop in accordance with their respective paths, and will build parallel ecosystems with their RF system-in-package (SiP) progress. ."

So, what RFFE path will each 5G technology take? Troadec believes that 5G mmWave technology will bring the most disruptive innovation. She also anticipates that the design will need to be redesigned and new materials used.

Fortunately, 5G mmWave can end complex front-end module applications based on SiP technology currently used in 2G, 3G and 4G RF front-end systems. Troadec explained: "You can design every building module based on advanced CMOS or SOI technology-including power amplifiers (PA), low noise amplifiers (LNA), filters, switches and passive components." This will It brings opportunities to develop SoC front-end modules for many digital chip suppliers who previously lacked RF expertise.

At the same time, for the 5G technology below the 6GHz frequency band (sub-6 GHz), Troadec believes it will be built on the basis of incremental innovation. She explained that in this frequency band, it is expected that only the current RF packaging architecture needs to be changed with the smallest bill of materials (BoM).

Since 5G IoT will use frequencies below 1GHz, Troadec believes that in this frequency band, 5G RFFE semiconductor packaging "requires little or almost no innovation." Despite this, the 5G IoT specifications and protocols that have caused data transmission problems for a large number of IoT devices have not yet been defined and standardized.

Parallel development of various versions of 5G technology (Source: Yole Développement)

Who are the "big guys" in the RF supply chain?
Before delving into the details of 5G RF solutions, let us first take a look at the current major RF component and module suppliers.

Generally, the RF front-end module is composed of RF components such as RF switches, PA/LNA, RF filters, and antenna components (tuners and switches).

In this crowded RF supply chain, major RF vendors include Sony, Murata (Murata; acquired Peregrin Semiconductor at the end of 2014), Skyworks, Qorvo, Infineon, Broadcom/Avago, Cavendish Kinetics, TDK EPCOS Wait.

Each company has its own proprietary RF components, usually using different substrates and process technologies. These technology choices include everything from RF-SOI and BiCMOS to bulk CMOS, gallium nitride (GaN) and RF MEMS.

Because different types of RF components use different process technologies, the current approach for integrating RF modules is usually SiP instead of SoC.

At present, for 2G, 3G, 4G, and 5G frequency bands below 6 GHz, Troadec confirmed that "the only way to meet the strict wireless performance requirements of smart phones is the SiP approach."

Currently, no RF component supplier has the ability to have every best technology. Troadec explained that in the integration of RF front-end, "Each building module requires very specialized technology: the best PA using gallium arsenide (GaAs) technology, the best switch using SOI technology, and the use of surface acoustic wave (SAW) The best filter for bulk acoustic wave (BAW) and the best LNA using silicon germanium (SiGe) technology."

Troadec also said: "Broadcom, Murata, Qorvo, Skyworks and TDK/Qualcomm are currently vendors that can provide SiP process technology for RF front-end modules."

She explained that each product has its own characteristic requirements, such as high-frequency modules, intermediate-frequency modules, low-frequency modules, and diversified receiving modules, each using the "PA Module with Integrated Multiplexer" (PAMiD) Or the "Front End Module Integrated Multiplexer" (FEMiD) form. PAMiD is a highly integrated custom module, performance-oriented but high in cost, and is limited to several companies such as Apple, Samsung and Huawei; while FEMID offers better performance and cost compromise, and is more popular than LG and mobile phone companies. Favored by Tier 2 smartphone manufacturers.

She concluded: "We do see only a few companies that can play a role in this highly technologically mixed environment."

5G sub-GHz: Continue to use SiP...

As the cellular industry develops towards 5G, it is expected that the same principle-SiP integration will continue to be used for 5G sub-GHz RF front-end modules.

However, according to Yole, there will be some changes in the future for more integration within SiP and packaging. Troadec explained that these new measures include integrating LNAs and switches on the same chip on SOI-based platforms, and adopting more wafer-level packaging (WLP) approaches for filters to save chip space (for example, the current Only Broadcom uses this method, and Qorvo is developing this method). In addition, the wafer-level packaging approach is also suitable for packaging PA (wire bonding is still used so far).

5G mmWave: From SiP to SoC

Undoubtedly, 5G mmWave RF front-end modules will completely change the most complex RF component/module supply chain. Due to the use of various process technologies, a large number of complex RF components are manufactured. In the future, it is possible to introduce mmWave front-end modules into SoCs based on advanced CMOS or SOI technologies.

There are many reasons why 5G mmWave can design RF modules in SoC.

First, Troadec explained that 5G mmWave means that it is shifting to the spectrum area where bandwidth is available. "Therefore, we don't need many frequency bands to send information, which further simplifies its wireless architecture."

Therefore, this also reduces the restrictions on filter technology. "There is no need to perform high-order filtering in the module," she reminded, however, "we still have to switch high-order switches (high isolation, linearity) between different wireless technologies (4G or 5G sub-6 GHz and 5G mmWave). .

She also pointed out that for 4G technology, “we use carrier aggregation (CA) with a bandwidth of 20 MHz per frequency band, and also use multiple frequency bands. Therefore, it requires high-order filter technology to distinguish between each frequency band. Each signal, but currently only BAW components (MEMS technology) are available.

Another important factor is that 5G mmWave will use beam-forming technology to form beams and transmit information to multiple users at the same time. "This will reduce the limits and requirements of PA power emission. On the other hand, it also means that CMOS technology can play a role." She added: "At mmWave frequencies, the inductance becomes smaller; therefore, it is possible to use CMOS/ SOI technology integrates passive components."

However, Troadec once again emphasized that for 5G mmWave RF modules, one of the limiting factors seems to be the power consumption of the entire system. "Why do we have to clarify this problem? Moreover, no one has technically explained why and what must be done to solve this problem."

New entrants entering the RF field
Once the industry switches to CMOS or SOI technology and designs 5G mmWave RF front-end modules in SoC, the current RF ecosystem will change from a seemingly harmonious RF front-end module supplier club (such as Broadcom, Murata, Qorvo, Skyworks and TDK/Qualcomm) began to change.

Troadec pointed out that Intel and Qualcomm are ready to enter the mobile phone modem and transceiver business, and they are very hopeful that they can master the wireless RF field and provide end-to-end solutions. The goal of these companies is to "bring a complete top-down in-house design to the RF industry chain."