Chip Stuff from Will Strauss (Forward Concepts)
Forward Concepts Wireless/DSP Newsletter
Will Strauss
Forward Concepts
July 23, 2012
fwdconcepts.com
Samsung to Become "One-Stop Shop" for Cellphone Chips?
With its recent acquisition of CSR's Bluetooth, GPS, Wi-Fi, and FM business, Samsung Semiconductor (the world's second largest chip house after Intel) has instantly moved in to the $3 billion "combo" cellphone chip business, expanding from its huge memory business, LTE modem product line and foundry service for Apple's A4 and A5 application processors, and its own new Exynos application processor product line (World's first dual-core ARM Cortex-A15 processor).
Samsung was the supplier of single-mode FDD-LTE modems employed in many of Verizon's LTE handsets and is working on multimode versions, too Now, they only need a world-class multimode, multiband LTE transceiver (are you listening Fujitsu and Advanced Circuit Pursuit?) and a great power management unit (Maxim or Dialog) to fill out their main cellphone requirements. They are clearly headed to becoming a "one-stop shop" for all major cellular chips...and that will further distance their cellphone business ahead of Apple (who may have to buy Qualcomm CDMA Technologies to stay competitive).
Admittedly, CSR has been late to integrating their "peripheral" chip portfolio on to a single-chip "combo" device, but Samsung has the muscle and capability to remedy that situation. (See chart below)
CSR No Longer a Cambridge Operation?
The company founded as Cambridge Silicon Radio plc seems to have lost its city-centric heritage. Samsung is acquiring virtually all of CSR's original product lines in Bluetooth, Wi-Fi, GPS plus 310 former CSR employees. CSR management appears to be abandoning ship to ride the company's merger with multimedia-centric Zoran, its acquisition of, DDFA™ audio technology, APT-X audio technology and the MAP-X™ audio product line it acquired from Trident Microsystems, Inc. Also three months ago, CSR acquired Direct Digital Feedback Amplifier (DDFA™) technology from Manchester-based Zetex Semiconductors.
These acquired operations are not Cambridge-centric operations. Zoran was largely a US/Israeli operation, APT-X was a Belfast (Northern Ireland) company and Trident a US operation, effectively leaving Samsung as the principal silicon company with operations in Cambridge England.
This is leaving CSR to focus on "Voice & Music, Automotive Infotainment, Indoors Location, Imaging and Bluetooth Smart." (They apparently licensed Bluetooth IP back from Samsung.)
Nvidia/Icera LTE cellular chip with Tegra to be in Asus Windows RT tablet
The eagerly awaited Windows RT operating system that is slated to run on ARM-based processors will see the first instance of Nvidia's Icera LTE modem in the announced Asus Windows RT tablet, which will employ a separate Nvidia application processor. Although the Icera LTE chip has been certified by AT&T, it will be employed as a stand-alone chip in this first "socket." Nvidia has assured us that the Icera baseband will be integrated with its application processors in future LTE deployments.
What Comes First: Chicken or Egg?
Can an operating system (O/S) startup get into the cellphone business? Doesn't it require chips, touch screens and plastic along with an O/S to make a cellphone? The MeeGo spinout of Nokia talent is taking on the task as Jolla Ltd. (I also wander how that will be pronounced in La Jolla, California?). But, isn't RIM trying to recover with its QNX O/S? Didn't PalmOS flame out too? Do we need a new O/S or are iOS/Android/WinMo7/Win8 good enough? I don't claim to know the answer, but considering the billions of bucks poured into Windows Mobile 7 at Nokia (and HTC), for only a modest market share gain, Jolla better have substantial backing.
International investors and partners are involved in Jolla, so I presume they have sources for the rest of the cellphone. To me, there appear to be three possibilities:
1) Struggling Nokia sells the worthy N9 platform to Jolla, since MeeGo already runs on it and the N9 can be a "horse to ride", or
2) Intel, as a possible backer, would certainly find a cellphone for the O/S or
3) Taiwan, Inc. would be quick source of the necessary platform.
Needless to say, promotion would have to be very strong (and expensive) to get a place at the market table. It should be fun to watch this startup unfold.
Whither Single-Mode LTE Modems?
Currently, there's only one single-chip multimode LTE baseband (modem) family shipping in volume, namely, Qualcomm's Snapdragon chips shipping in AT&T's Nokia Lumia 900 and HTC One-X models, and U.S. versions of HTC Titan II and Samsung Galaxy S III LTE smartphones (hey, those are still two modems, but apparently in a single package). Multimode means that the LTE modem can switch to 3G or 2G if it leaves an LTE base station cell. There are multi-chip LTE modem solutions shipping that generally require a 2G/3G modem along side of the LTE modem. For example, most Verizon LTE cellphones use this approach, often with Samsung (single-mode) LTE modem and Via Telecom or Qualcomm CDMA modem (necessary for voice and for fallback to 3G/2G). It'll be interesting to see if the upcoming Verizon version of the Lumia 900 employs a separate CDMA modem (but more likely two modems in a single package).
We count more than a dozen announced single-mode LTE cellphone modems (some of which are TDD versions that have evolved from WiMAX). However, to be really successful in the developed FDD cellphone market, they will require capability to handle 3G and 2G as well. Unfortunately, most of the sources for 3G/2G software stacks have dried up, and most companies can ill-afford to develop their own. Our earlier 2009 Smartphone Market Report listed the major vendors and their 3G stack sources:
================================================================
3G Stack Sources for Major UMTS Baseband Chip Suppliers (c.2009
================================================================
Chip Supplier 3G Stack Source
============= ===============
Broadcom Comneon
Freescale Motorola
Icera TTPCom
Infineon Comneon
InterDigital InterDigital
Marvell TTPCom
MediaTek TTPCom (via Analog Devices)
NEC Electronics NEC Electronics
NXP NXP/Comneon
Panasonic Adcore Tech
Qualcomm Qualcomm
Renesas TTPCom
SunPlus TTPCom (in development)
STMicro EMP & Nokia (Custom)
Texas Instruments EMP & Nokia (Custom)
Of course, NXP wireless was acquired by STMicroelectronics as was Ericsson Mobile Platforms (EMP). Texas Instruments is confined to Nokia's waning Symbian products and Icera has been acquired by Nvidia. Nokia's 3G (and LTE) stack is now part of Renesas Mobile's offering (as is the NEC Electronics IP). Infineon and Comneon have been acquired by Intel. SunPlus spun out its cellphone product line to HT mMobile, Inc., which now appears to be a shell company with no product.
The reason for the paucity of 3G stack sources is the considerable amount of development that goes into it. Based on our interviews and observations, we believe that development of a basic WCDMA stack (before adding HSPA+ or LTE functionality) is on the order of $300 million over three years, at a minimum. Even with that, licensing of basic spread-spectrum and other IP will be required because of existing patents. Few companies can afford that budget of money and manpower, essentially barring new sources of 3G stacks. Note that even after licensing a 3G stack, several man-years of effort must be devoted to tailoring it to a specific chip platform.
So, what's a poor single-mode LTE chip supplier to do? Well, an offer for HT mMobile might excite SunPlus. But, Japan-based Adcore Tech (with 3G only/no 2G) may be out of business. Rumor has it that Sasken may also be a source of 2G/3G stacks. InterDigital Communications appears to be alone in licensing 2G/3G stacks, now. Otherwise, the single-mode LTE modem house will have to partner with someone who has such a stack.
Should TI Get Back into Cellular Basebands?
In November 2010, my newsletter pointed out that the stand-alone application processor constituted a shrinking share of the smartphone market, since integrated baseband/apps processor chips were increasing in market share. In that newsletter, I advised the two major stand-alone apps processor houses, TI and Nvidia, that one of them should acquire Icera Semiconductor. Nvidia later took my advice. I expect that Nvidia/Icera will be a direct competitor to Qualcomm by mid-2013 (at least for non-CDMA cellphones).
Now, with OMAP, TI becomes the major stand-alone applications processor company. Certainly, OMAP is a great architecture, but it's now addressing a market segment that is shrinking. And tablets aren't nearly as big a market as smartphones. So TI becomes #1 in a market segment that will eventually be small.
So, should TI get back into the baseband business to pump up OMAP? If so, acquisition is the only feasible alternative, in my opinion. About the only target companies with "available" state-of-the-art LTE/3G/2G basebands are Renesas Mobile and ST-Ericsson. If TI decided to acquire one or the other, the problem becomes getting some high-volume sockets quickly. Renesas Mobile is presently stuck with no visible smartphone sockets, since Nokia has gone over to Qualcomm…the only company that could allow Nokia to quickly get Windows Mobile 7 cellphones to market. Renesas Mobile's parent, Renesas Electronics, is having financial difficulties and might be amenable to a reasonable offer. ST-Ericsson has its own set of financial problems, but acquiring a company that has government investments by governments of both France and Italy might be messy.
But, TI may continue to ride the OMAP platform, marrying it with its able DSP technology to serve markets other than consumer tablets. ### |
