Intel Works To Develop Monolithic Integration Of Wifi And RF, Part I

Jun. 6.13 | About: Intel Corporation (INTC)

Introduction

In my first two articles for Seeking Alpha, I discussed my thoughts for why Intel (NASDAQ:INTC) has seen relatively flat stock pricing over the past 10 years. One of my chief conclusions was that Intel has suffered from an untimely development and/or partnership in cellular communications technologies, particularly the 3G and now 4G standards based components necessary for modern smartphones.

Historically, these communications products lay outside Intel's traditional focus area in processor architecture and manufacturing. However, Intel has long developed and sold wireless (WIFI) chipsets, a different communications standard, roughly distinguished by being higher data bandwidth and shorter spatial range. Intel could have easily partnered with some of the other mobile players (e.g. Qualcomm (NASDAQ:QCOM)) to develop smartphone solutions, but for better or worse -- or perhaps more accurately, for better and worse -- Intel has decided to develop its own products in house.

What is necessary now is to understand the current status and future prospects for Intel's mobile communications product developments. Towards this end, in this article I will begin reviewing the ongoing work and reported progress from Intel, in particular, a recent 2013 paper by Hasnain Lakdawala et. al. [1] out of Intel, Hillsboro, Oregon.

This reference is a technical paper, with 26 authors and over 15 Ph.D.'s contributing to the work. I am not an electrical engineer -- I am a physicist -- so it is a struggle to understand it, but I hope this should help educate myself as well as the investment community at large about this important topic that may be an essential factor for Intel's future financial success. And aside from its benefit for investors, this stuff is just interesting -- at least to me, as a guy with a relatively obsessive engineering and science interest.

Background: Towards CMOS integration of RF capabilities

The basic motivation for the work in [1] was discussed by Intel Chief Technology Officer Justin Rattner at IDF 2012 [2]. Radio components have traditionally been based off of analog circuitry, which has begun reaching severe scaling limitations, unlike digital circuitry. As radio standards and protocols (e.g. GSM, 3G, 4G -- in order of increasing bandwidth -- and their various subsets such as GSM, HSPA+, etc.) have evolved, more and more separate circuitry is needed to accommodate each standard. To get an idea of how complicated this gets in a modern device, one can refer to the excellent ChipWorks technology blog [3], where their teardown of an iPad 3 reveals 19 separate RF packages, many with more than one die a piece.

Ideally, one would like to integrate these components or functionalities into a minimal chip size, ideally onto a single die, with as small an area as possible, since the area of the chip greatly correlates with its cost. Since Intel is all about manufacturing, this has presumably been a particularly important issue for its engineering management. But the functionalities aren't shrinking, and building a huge, single wafer with all these functionalities poses problems.

In order to shrink the designs, as discussed in [2], it has been necessary to actually re-engineer the radio as a primarily digital system. This article intends to begin penetrating into some of the details of this re-engineering. I will assume the reader is familiar with basic background knowledge such as the benefits of Moore's law, and I will include wiki-references as hyperlinks but separate page references at the end of this article.

Overview: A 32 nm SoC With Dual Core ATOM Processor and RF WiFi Transceiver

Now we begin reviewing the work in [1], and in particular, trying to pick out the points most relevant to investors. In particular, I will basically be going into details wherever I meet something important with which I am less familiar with what it means.

The paper presents an x86 compliant -- e.g., compliant with existing Windows software -- SoC based off a dual core ATOM processor and an integrated WiFi transceiver, in a 32 nm process technology first released by Intel in 2010. The WiFi standard makes use of RF in the 2.4 and 5 GHz bands. Wifi is notably different from cellular standards such as 3G and 4G [3].

WiFi permits higher bandwidths on the order of 10 Mbps, whereas 3G is on the order of 100 kbps. Cellular networks are typically deployed and managed by a service provider, with access ranges on the order of 1 km, compared to 100 m for WiFi, which are not necessarily managed by a large service provider. 3G networks achieve this longer range by operating at lower frequencies, in the 400 Mhz to 3 Ghz range -- as lower frequencies means longer wavelengths and longer wavelengths are less attenuated by the environment. Another key differentiating feature of 3G networks is their intrinsic management of service hand-off as users travels between distant access points.

The paper concentrates on several novel or important features of the device:

  • Building blocks for integration of RF subsystems in future SoCs, such as a standardized SoC interface and a system interconnect fabric called Intel on-chip system fabric (IOSF).
  • Integrated voltage rails on the SoC, consisting of circuit elements called linear voltage regulators (LDO) and charge pumps
  • a 32 nm process with separate designs for high performance transistors, high voltage transistors, and RF passives, such as inductor elements, which is noteworthy since characteristics of silicon tend to be lossy for these elements
  • Clock generator with spread spectrum clocking, a type of mechanism for reducing electromagnetic interference
  • And the WiFI transceiver, as already mentioned, including a Low Noise Amplifier (LNA), power amplifier (PA), Transmit/Receive (T/R) switch, and on-chip calibration engine

A layout of the SoC is shown below:

In a next article, I plan to begin analyzing the body of the paper and some of the details of these subsystems -- most of which I am admittedly unfamiliar with. Questions or comments, or guidance from electrical engineers, is most welcome.

References

[1] Lakdawala, Hasnain, et al. "A 32 nm SoC With Dual Core ATOM Processor and RF WiFi Transceiver." (2013): 1-13.

[2] Rattner, IDF 2012 video, and slides

[3] Gary Tomkins and Jim Morrison, Chipworks, "iPAD 3 LTE/3G - Multi-band support is very complex." (2012)

[3] Lehr, William, et. al. "Wireless internet access: 3G vs. Wifi?" (2003)

Disclosure: I am long INTC. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.