元器件 > 高级半导体博客



Tales from the Compound Semiconductor Road

by Eric Higham | 3月 30, 2016

I spent much of March away from my office. I was at home, in transit or at some far-flung location and that got me thinking about broadband connectivity and convergence. For a business traveler, nirvana is ubiquitous, seamless, inexpensive and simple connectivity. We aren’t quite there just yet, particularly the inexpensive part, but broadband connectivity is becoming an expected cost of doing business. Airports and hotels were quick to adopt Wi-Fi and airlines are starting to embrace satellite broadband. Most of these operators still differentiate their services by offering faster tiers for a price or to reward loyal users, but the availability and convenience of broadband connectivity is undeniable.

However, availability is not a foregone conclusion. The website Internet Live Stats has a counter showing a running total of the number of internet users. As I am writing this blog, the number is approaching 3.338 billion users (and counting). According to them, this means that nearly 54% of the world’s population does not have access to the Internet. The other 46% of the population with access has a different problem. They have the fear of waking up some day and finding that they have a “connectivity issue”! I am being a bit facetious, but we all know how life changes during an internet outage. The cloud that enables so much of our connectivity and productivity suddenly gets very stormy and we all hope for a fast resolution to the problem, lest we are forced to embrace a bygone era of communications!

During my travels this month, I had the pleasure of attending the CS International and OFC  conferences. I got a chance to hear about some of the developments that aim to ensure broadband network connectivity continues to expand in reach and performance. The CS International Conference was in Brussels in the beginning of March, so before I share some of my thoughts about what caught my eye at these conferences, let me take a moment to pass along my condolences and best wishes for all those people affected by the senseless violence that took place in Brussels.

Here are some of the interesting nuggets of information that I heard at both conferences:

I had the distinction of presenting a market overview to set the stage for a 5G session at CS International and there is a groundswell of 5G component development activity:

  • Qorvo discussed some different architectures for beamforming solutions required by MIMO antennas. The ideas all rely on higher performance compound semiconductor and digital devices.
  • Skyworks looked at the front-end architecture for mobile phones and pointed out that 3GPP Release 13 will bring the number of band combinations for carrier aggregation to just under 200! They are developing devices to enable these and higher frequency architectures using SoI (switches & amplifiers), GaAs (pHEMT & HBT), GaN and InP (for amplifiers).
  • SEDI touted their dominance in GaN amplifiers for base station applications. They indicated their commitment to GaN for future base station and wireless backhaul applications, even as frequencies increase. They showed some good performance results for PAs in the 3.5 GHz range, but defaulted to GaAs technology for a 30 GHz amplifier.
  • Peregrine Semiconductor displayed the performance advantages of their CMOS-based “Intelligent Integration” concept that allows them to integrate the switching/phase shifting/amplitude adjust networks with digital shift registers for faster programming, beam steering and nulling of multi-element antennas.

From here, it was on to Anaheim and the Optical Fiber Conference (OFC). As the name implies, this conference focuses on fiber in the network transport layer, but it’s really just a different solution to the same problem of handling and enabling data traffic. The growth in wireless data traffic gets a lot of attention, but wireless represents less than 6% of the total data a network carries. The OFC Conference addresses the trends and components that form the enterprise, data center and transport networks that carry the other 94%.

Some of the interesting trends that surfaced at this conference:

  • Data centers remain a hot topic of interest because of the amount of data traffic they must handle. Even though data rates are increasing quickly, there is a surprising amount of copper interconnects at the “top of the rack”. Copper’s biggest advantage is cost.
  • With broadband speeds increasing quickly, 100 Gbps networks have become the “table stakes” or the lowest network speed that is economically feasible. There is development on long-haul networks beyond 1 Tbps, so capacity is ramping quickly.
  • A big discussion surrounds the appropriate number of lanes to achieve the required data rates. The industry workhorse remains 10 Gbps, but manufacturers do not view this as an efficient architecture for the future, so there is development and debate around the best evolutionary path for the lane, as well as the transport network data rates.
  • This is an interesting time in this industry. There are several standard package types, modulation schemes, lane rates and network rates for the equipment. Couple this with many market participants and no dominant supplier, along with volumes that are not that high and this industry is ripe for consolidation, as some will struggle to meet financial objectives.

The common theme from both conferences is the entire electronics industry is working feverishly to develop business models, equipment and devices that enable the seemingly endless growth in data traffic. Not every technology and certainly not every company will survive, so stay tuned in the coming months as I dig into some of these topics in more detail. Most of all, pray that you don’t wake up tomorrow morning with no broadband access!

  • Eric
Previous Post: Will 5G Benefit the Compound Semiconductor Industry?...Yes! | Next Post: In a Galaxy Far, Far Away; GaAs and GaN-based Communications
Leave a comment