In July of this year, the Global Semiconductor Industry Association reported that the semiconductor industry has continued to see new venture capital (VC) investment diminishing. For the last three years, the amount VCs have invested in semiconductor start-ups has steadily decreased by 5.0 percent, 13.1 percent, and – most recently – 17.2 percent, a sign of maturity and consolidation.
This means that an aging semiconductor industry – that makes the digital life nearly everyone in the world enjoys – must continue to innovate and integrate without easy access to venture funding. Nowhere is the need to innovate and integrate more apparent than in the hotly contested market for chips that populate today's popular consumer devices: smartphones, tablets, and the emerging intelligent devices popping up in energy, medical, and automotive monitoring applications.
Compounding the funding problem is consumer expectation of continual price reduction and new product introduction cycles on a yearly basis. This short cycle provides barely enough time to develop innovative breakthroughs; in turn, this provides an opportunity for new start-ups to provide innovation. If the next generation of a product has only evolutionary improvement, such as a new industrial design, the consumer expects a price reduction. If the OEM hasn't boosted its volume with the price cut, the results is revenue decline with no hope of maintaining margin: a going out of business plan as seen by the fall from grace of yesteryear's leaders.
It is very difficult to maintain leadership in the electronics industry; to do so companies have to execute to a roadmap that includes both innovation and integration. Initially, OEMs provided innovation, such as IBM with the IBM PC. Semiconductor companies provided integration. As the PC evolved over time, the semiconductor company began driving both innovation and integration.
With the advent of the smartphone and tablet computers, the OEM once again is driving innovation and dictating integration. This discussion will examine the circumstances surrounding the back and forth role of the OEM and semiconductor company as system innovator.
The tasks of marketing and engineering
One innovation marked the dawn of a new category of consumer electronics product: the smart phone. That innovation was the multi-touch technology user interface. The development involved a small amount of hardware – a capacitive sense function – and a large amount of software.
What made this function revolutionary was that it changed the user's entire experience with the phone. It's not going too far to say that this change was on par with the introduction of the graphical user interfaces (GUIs) that replaced the command-line interpreter interfaces on the PC nearly 25 years earlier. Both represented a watershed event in the evolution of the PC and phone, and both opened the functionality of the PC and phone to a much larger base of users. In the process, both created previously non-existent ecosystems for their respective products: software developers for the PC and apps developers for the smart phone.
The early PCs allowed users to evolve from computer science sophisticates to applied users who ran shrink-wrapped software, though they still had to deal with an arcane command line interpreter and they still had to understand the vagaries of directories and file structures. The system company, IBM specifically, determined the hardware architecture, the software, and the future roadmap for both hardware and software on this new computing machine.
The dawn of the Graphical User Interface
The advent of the Windows GUI removed the need to interact with the command-line interpreter and facilitated a dramatic expansion of the number of applications beyond the rudimentary productivity software of the day. The PC now became a productivity tool for nearly every occupation: middle managers planning new product rollouts, investment bankers running numbers on a merger or acquisition, a lawyer performing searches of legal data bases, and countless others.
With Windows also came more structure to the architecture of the PC. This enabled the establishment of an orderly, evolving product roadmap that more easily lent itself to semiconductor integration. The system and applications software that Microsoft and its development community were creating determined the software roadmap of the PC. The Intel x86 CPU and supporting chipsets determined the hardware roadmap due to the enormous base of software written in Intel binary code. Intel determined new functions to integrate into silicon – math coprocessor, graphics processing, Ethernet connectivity, etc. – driven by demand from the software that would run on the PC.
The advent of the Internet in 1989, the first Web Browser in 1993, and the first Search Engine in 1994 set the stage for the next evolutionary step of the PC: from a compute engine into a personal communication and collaboration device accessing an expanding amount of on-line data and enabling remote interaction. This still on-going evolution has forever changed the distribution and consumption model of all published media: print, audio, images, video, and gaming. This ubiquitous information network, combined with affordable computing, set the stage for the next major evolution: the smartphone.
Until the advent of the smartphone in 2007, semiconductors drove mobile phone architecture and roadmaps by providing integrated solutions to handset OEMs: baseband processors, applications processors, and RF radio components. With the smartphone, the handset OEM took control of the roadmap and architecture of this new category of product. It was a handset OEM that determined the user interface, which now came with icon-based interface along with zoom, pan, and stretch capacitive-sense input capability.
The citizen's personal digital concierge
The new handset and the tablet added functionality that vastly expanded the definition of the mobile phone. It was no longer a device only for for talk, text, contact manager, and e-mail. The smartphone brought the multi-touch graphical user interfaces (replacing earlier dysfunctional interfaces), and added the concept of pay-per use media, a larger display screen, a usable web browser, and the concept of third-party applications. With the addition of GPS, compass, accelerometer, gyroscope – coupled with a proliferating number of apps written to use this on-board hardware – the smartphone began to morph into a personal digital concierge. The subsequent rise of social networking enlisted the smartphone for always-on ubiquitous personal interactions.
Just as Windows redefined the use model of the PC, the smartphone and its on-board functionality redefined the use model of the phone. In the process, the use model is changing the nature of information processing. Everyone in the world with a Twitter account has become a media provider publishing 140-character messages as well as video clips and still images by the millions: 750 million photos were uploaded to Facebook over the 2012 New Year's weekend. The innovation in the smartphone and emerging social networks symbiotically created a cultural revolution that has altered how everyone interacts.
As successively smaller CMOS process generations reach manufacturing volume, semiconductor companies' marketing and engineering teams are being pressured to integrate the functionality handset OEMs' roadmap are dictating. The recent supply crunch at 28nm shows a jump in demand from semiconductor manufacturers, all of whom are vying to achieve the highest levels of integration ahead of their competitors.
The mobile phone camera is an example of the integration that has occurred on this platform. A modern camera's mixed signal CMOS sensor contains an all-digital preprocessing unit that performs data enhancement before the image is compressed. This image signal processor (ISP) compensates for any non-linearity, lens effects, etc. in the sensor. The result is a mixed-signal chip containing the analog sensor and a digital ISP.
Adding full motion video capture to cameras that previously processed only still images came about when the increasing number of available gates on-chip at 65nm, 40nm, or lower process nodes enabled the integration of this capability – a function of Moore's Law. Semiconductor companies' marketing and engineering teams are constantly assessing at which process node new features become practical and cost little or nothing to add.
Concurrently, to drive their need to expand market share, OEMs are pushing chip suppliers to add still more functionality on-chip. One example is the RF section of the phone. Consider an early smartphone OEM that made a strategic decision to choose one radio frequency (RF) radio chip to implement on its first phone. As the handset OEM sought to expand his market share, it demanded chip vendors to produce an RF radio to handle multiple frequencies.
A recent example of such integration came from Qualcomm Inc. In January this year, the company introduced the Gobi modem MDM8225, MDM9225, and MDM9625 chipsets. The chipsets integrate seven different radio access modes on a single baseband chip: cdma2000 (1X, DO), GSM/EDGE, UMTS (WCDMA, TD-SCDMA), and LTE (both LTE-FDD and LTE-TDD).
The benefit to the mobile phone OEM is a large increase in handset volume with no increase in bill of materials. Whereas before, the handset manufacturer was limited to individual service providers such as AT&T or Verizon, he is able to supply both along with variants of GSM and CDMA to support Sprint in the U.S. as well as service providers elsewhere in the world.
The electronics industry is experiencing a major change in the products being produced and consumed. Semiconductor companies that long influenced what system OEMs were building are now finding their role serving the OEMs. In the process, they are being driven to integrate increasing amounts of functionality on-chip. Thanks to the relentless semiconductor process technology advance, that integration continues unhampered.
If history is any guide to the future, the role of system OEM and semiconductor supplier will eventually alternate as it did with the PC years ago. It may take time, and if the PC is an object lesson, it will demand that chip suppliers understand the end products better than the system OEMs they are supplying. This will happen when the market matures.
And one last thing…
Looking to replace the aging serial EEPROM
Since its development in the late 1970s, the serial EEPROM remains with us to this day, performing the yeoman duty of supplying a small amount of boot code and configuration settings to initialize a processor after power on. Since its invention, it has been primarily integrated in systems and packages, but not in silicon.
Today, however, technology and economics may finally be catching up with this 20th Century throwback. Integrating this function provides area savings, improved quality, and operational cost savings for OEMs. For chip manufacturers, integration provides margin dollars previously earned by the EEPROM chip suppliers. However, integrating the EEPROM function in advanced process geometries is not practical using floating-gate technologies.
One technology to achieve this integration cost-effectively is the antifuse non-volatile memory IP developed by Kilopass Technology Inc. This IP affords the storage capacity of EEPROM with the benefit of being manufacturable on standard logic processes, including the latest advanced process nodes. Antifuse, which does not require special processing, can be used to emulate EEPROM functionality. This is achieved by over-provisioning the amount of embedded antifuse OTP NVM to provide multiple-time programming capability.
The OEM buying the SoC benefits by reducing his bill-of-materials by one serial EEPROM (which runs around $0.15 per chip based on a 100mm2 SoC). The SoC designer eliminates four pad ring pins at a savings of $0.08 (which subtracts the licensing fee for the NVM IP plus royalty expense over several million chips from the $0.15 cost of the four pad ring pins.) Eliminating the external serial EEPROM will save the SoC system customer logistics and inventory cost, and this could result in savings of around $0.30.
Thus, the 20th century artifact of the serial EEPROM may finally give way to the 21st century antifuse NVM IP.
About the author
Andre Hassan is Field Marketing and Applications Director at Kilopass. Andre is an industry veteran with over 20 years of semiconductors and systems experience. He brings a broad business experience in marketing, sales and operations, as well as depth in multiple engineering disciplines.
Prior to Kilopass, Andre held senior management and engineering positions at Sigmatel, Monolithic System, S3, Sun Microsystems and Digital Equipment
If you found this article to be of interest, visit Programmable Logic Designline
where – in addition to my Max's Cool Beans
blogs – you will find the latest and greatest design, technology, product, and news articles with regard to programmable logic devices of every flavor and size (FPGAs, CPLDs, CSSPs, PSoCs...).
Also, you can obtain a highlights update delivered directly to your inbox by signing up for my weekly newsletter – just Click Here
to request this newsletter using the Manage Newsletters tab (if you aren't already a member you'll be asked to register, but it's free and painless so don't let that stop you [grin]).