Electronic coupling is the transfer of energy from one circuit or medium to another. Sometimes it is intentional and sometimes not (crosstalk). I hope that this column, by mixing technology and general observations, is thought provoking and “couples” with your thinking. Most of the time I will stick to technology but occasional crosstalk diversions may deliver a message closer to home.
Moore has Left the Building!
Unlike Elvis, Gordon Moore, the co-founder of Intel, is still with us. Although the debate continues among very smart people as to whether Moore’s Law is “truly dead”, this argument is now purely academic. As the electronics industry has moved to the “Post Personal Computer (PC) Era”, Moore’s Law which accurately predicted price over time for complementary metal–oxide–semiconductor (CMOS) integrated circuits, is no longer relevant.
Heresy! Did he just say that Moore’s Law doesn’t matter?
In 2011, the International Technology Roadmap for Semiconductors (ITRS) identified that increasing semiconductor value was being created by incorporating functionalities that did not necessarily scale according to Moore’s Law. They named this innovation trend as “More-than-Moore” (MtM) and the continuation of improvements via scaling as “More Moore”. The diversification for MtM was forecasted to come from analog / radio frequency (RF), passive, high voltage power, sensors & actuators, and biochips devices. The basic difference in functionality is that More Moore devices focus on information processing and digital content in system-on-chip (SoC) configurations while MtM devices allow interaction with people and the environment through non-digital content integrated in system-in-package (SiP) configurations.
What has really changed is the end user doesn’t care (much) about what is inside their device rather they care about what it does for them. In the PC Era, many users could tell you which microprocessor was in their PC along with the clock speed, number of cores, and amount of system memory. While today, in the Post PC Era of mobile devices such as smartphones and tablets, most users have no idea of the particular configuration of the internals of their device. Apple sells next generation devices by promoting iPhone features such as the number of mega-pixels to provide bigger and sharper pictures not by talking about processor speed.
Moore’s Law & Dennard scaling
It probably doesn’t help that many people – especially those outside of the semiconductor industry – have oversimplified or attempted to “interpret” Moore’s Law to suit their purposes. Nor the fact that many in the semiconductor business have confused Moore’s Law with Dennard scaling or have otherwise lost touch with the economic basis. Of course, fifty years in the world of technology seems like eons due to the fast rate of innovation enabled by Moore’s Law.
“The complexity for minimum component costs has increased at a rate of roughly a factor of two per year…” – Gordon E. Moore, “Cramming more components onto integrated circuits”, Electronics (Volume 38, Number 8, April 19, 1965)
Mr. Moore predicted that the number of transistors in an integrated circuit that could be produced at the minimum price per transistor would double every two years. Most semiconductor manufacturing costs are a function of area, so the smaller the transistor the more per wafer and lower the cost per transistor.
At the same time as transistor sizes are reduced they also become more energy efficient with increased performance as predicted by Robert Dennard. The combination of Moore’s Law and Dennard scaling have led to end customer expectations of electronics becoming constantly cheaper with significant performance improvements. So much so that “Moore’s Law” has become a technology consumer addiction. Who wouldn’t like half the price with twice the performance every two years?
Today the most popular features of our mobile devices are enabled by MtM technology. Everything from wireless communications to motion sensing to optical imaging. That turn-by-turn directions from your smartphone while you drive? The Global Position System (GPS) radio receives the positional data stream from the satellites while your wireless modem retrieves the data for the maps and the micro-electromechanical system (MEMS) gyroscopes and accelerometers track your position. And of course, the audio digital to analog converter drives an amplifier that powers a speaker to provide you with the navigation voice. And let’s not forget the wireless data stream that updates the status of the traffic along your route or the beautiful liquid crystal display (LCD) that shows the map. True, the processor and memory are very important in this use case, however performance improvements beyond the minimal functionality required in either device do not translate to noticeable improvements in the user experience.
Looking at a teardown of a current smartphone tells a similar story. One will find three large integrated circuit (IC) packages: microprocessor stacked with dynamic random access memory (DRAM) in a package-on-package (PoP) configuration, Flash memory, and baseband (radio) processor. These ICs are very large and even though the process technology improves every year the die sizes continue to grow as the designers continually incorporate as much as possible into their CMOS circuit design.
The printed circuit boards (PCB) of smartphones are also crammed with a multitude of smaller devices that that are manufactured in non-CMOS technology. Devices include wireless power amplifiers, RF filters, audio codecs, audio amplifiers, power management ICs (PMICs), MEMS, optical sensors, and a lot more. These devices are built in an “alphabet soup” of fabrication processes everything from III-V semiconductor materials including gallium arsenide (GaAs) to compound processes such as BCD. [BCD is a mixture of bipolar for precise analog functions, CMOS for digital design, and double diffused metal oxide semiconductor (DMOS) for power and high-voltage elements.] And MEMS devices contain micro fabricated parts typically built with silicon that aren’t even semiconductors!
Needless to say these devices are what enable the “smarts” in our phones. Without these technologies the phone will not function or would be unable to interact with people or sense the environment. It is no wonder why Qualcomm, an industry leader in mobile processors and mobile baseband processors, formed a three billion dollar ($3B) joint venture with TDK EPCOS earlier this year. This joint venture “RF360” will focus on RF modules and filters for all kinds of mobile electronics including smartphones, Internet of Things (IoT) devices, and automotive applications.
As I began explaining to clients as early as 2012, as the “bang for the buck” of More Moore (continued scaling) decreases, product companies will be forced to seek solutions using More-than-Moore technology to add new features, increase performance, and reduce prices to satisfy our Moore’s Law addiction. Constant innovation is required to develop new and improve existing MtM technologies since they do not scale in a similar fashion as transistors do per Moore’s Law.
Substantial innovation is neither a quick nor an easy process. Nor does it usually happen randomly without encouragement of forward-thinking sponsors. Most of the wireless technologies in use today for our mobile devices started out as military or aerospace applications. And the time scale is many years if not decades from initial concept to high volume application. Hedy Lamarr and George Antheil patented the basis of spread spectrum radio technology in 1942 which wasn’t commercialized for cellular phones until the mid-1980’s. Similarly, many early MEMS research and development startups were funded by the Defense Advanced Research Projects Agency (DARPA). More often than not, the timeline from initial academic research to large-scale applications for MEMS has also been on the order of 25 to 30 years.
Even though the US has previously funded much of the fundamental work for MtM technology in MEMS and wireless technology, current levels of applied science research has been significantly reduced. In particular, the US activity has fallen behind that of the European Union’s Horizon 2020 program which is spending eighty billion euros (80 B€) over seven years on research and innovation.
Such innovation is highly predictable due to the long development time during which there are typically academic publications, government grants, and patent applications. However, it still catches many companies unaware when it shows up in products. These “black swans” (events or innovation that in retrospect were predictable) often go unnoticed when organizations are not proactively searching research areas, exploring adjacent market technology, and thinking about how to drive their product forward. It is critical to provide your marketing, product management, and engineering teams with the resources to help them look ahead and think “outside the box”.
The addiction that Moore’s Law has created will continue and successful companies will find ways to satisfy this demand. Wouldn’t you prefer your business to be proactive and innovative instead of reactionary?
As always, I look forward to hearing your comments directly. Please contact me to discuss your thoughts or if I can be of any assistance.