It's natural to focus on the "toys" at CES -- the products which are the result of all the new technologies being developed. But underneath these products are the critical chips that make possible the small form factors, low power consumption, reliable data communication and consumer-priced products.
Many key chip makers come to CES, and we spent much of our time visiting some of them to look under the covers and judge what's coming next.
After Bill Gates' speech, we visited with Megan Kidd, Product Manager in Microsoft's Embedded & Appliance Platforms Group, to see and talk about Smart Displays and SPOT. Megan showed us the demonstrations in a house Microsoft built in the convention center parking lot. We talked about how she uses Smart Displays at home, and we're looking forward to having one to test in our home soon.
The next day, we visited National Semiconductor's suite to look under the covers. National was a key player on Smart Displays and the primary technology company behind SPOT, so we were delighted to discuss both with the people working on these products.
National has been working for many years on portable wireless touchscreen technologies under the trademarked name "WebPAD". They brought all this experience to bear while working with Microsoft as a partner in the Smart Display initiative.
Marion Clary Sr. Marketing Manager, Consumer Access Business Unit, worked on the project and tested a Smart Display at her home. She described how she used one while working with her children in the evening, and told us about the National chips used in the design.
National worked with Microsoft to develop the chips underlying SPOT. Joe Montalbo VP, Custom Solutions Division, displayed a chart showing the seven National chips in the SPOT prototype watch Gates was wearing.
National's collaboration with Microsoft has given it a lead in providing chips for Smart Displays and SPOT devices. But we think Microsoft wants to avoid another "Wintel" and that other companies will compete with National on performance, size, power and price.
We've been watching Transmeta for several years. Their Crusoe chips consume much less power than competitive processors. Since they therefore generate less heat, most don't require fans. They're fully compatible with Intel chips and run with all versions of Microsoft Windows. Many vendors are building light-weight portable devices around Crusoe.
At CES, we met at Transmeta's suite with Jon Grodem, Manager Program Marketing. We looked at several devices that intriqued us:
We've long believed in the potential for portable devices, both in the home and on the road. We came away from the Transmeta suite convinced that the competition has just begun to build faster, smaller, lighter, and quieter hand-held devices.
Intersil is the largest vendor of chips based on the IEEE 802.11b standard, claiming a 60% market share. It has been a leader in establishing the 802.11 standards, and also in the Wi-Fi Alliance responsible for marketing "Wi-Fi"-branded products based on 802.11. We've written before about Intersil and their work on reducing costs and developing higher-speed chips.
At CES, we met with Chris Henningsen, VP of Marketing. At the beginning of our discussion, Chris said that Intersil believes that analysts have underestimated the growth potential of wireless technology. Current estimates include chips in access points and network adapters, and built into notebook computers and PDAs. But Intersil thinks they've missed the even larger potential built-in to desktop PCs and other portable devices such as cell phones and DVD players.
We were especially interested in Intersil's views on the evolution to higher-speed 802.11a and 802.11g standards. Readers will remember that 11a is an approved standard running up to 54 Mbps at 5 GHz. 11g is a standard in process, also operating up to 54 Mbps but in the same 2.4 Ghz band as the current 11b. These higher-speed chips are available now and starting to appear in consumer devices.
Everybody agrees that the higher-speed devices will displace 802.11b, but disagree as the whether the market will be dominated by 11a, 11g or some combination of one or both with 11b. Some have speculated that 11a will dominate in the enterprise since it provides more usable channels, while 11g will win in the home since it provides greater range.
In discussions with other vendors, we had been told that the cost difference between 11g and 11b chips would be small enough so that vendors would quickly discontinue 11b products in favor of 11g. Chris told us a different story. He said that the largest cost difference between the chips comes from implementing the OFDM modulation scheme responsible for the high speeds in 11a and 11g. The cost of supporting multiple radios to handle both 2.4 and 5 GHz is comparatively lower.
The near-term BOM for Intersil's mass-market 11b chipset is less than $20. For an 11g/b set, the BOM is less than $30. For an 11a/b/g set, the BOM is less than $35 - about the same as an 11a-only set. There's about a 2.5 to 3X factor from the BOM cost at the chip level to the retail price of devices based on these chips, so 11g will add $25-30 to the cost of devices such as APs and NICs and 11a will add another $13-15.
Intersil believes that stand-alone 802.11a is dead. Enterprises will want to support all three standards and will buy "tri-mode" access points. With 11b devices already in many homes, notebook PC users in enterprises won't want a different NIC for the office and the home, so will want NICs capable of at least .11a and .11b. If there's little or no incremental cost to add .11g (since the chip already includes OFDM modulation for .11a and a 2.4 GHz radio for .11b) they'll buy "tri-mode" NICs.
We found this argument pretty persuasive. 11g will augment 11b in the home, and the chip pricing may well favor a "tri-mode" approach. Whether the consumer market moves to a "dual-mode" 11g/b or a "tri-mode" a/g/b will depend more on how ODMs and OEMs price products than on the cost of the underlying chips.
The future outlook for wireless LAN (WLAN) chips for consumer electronics devices is that they will continue to be cost reduced, through what the industry calls "Up integration". Because much of the functionality in today's chips is software, especially in the MAC layer, the trend is to remove this from the chips and embed it in the application processor. The WLAN chip then provides only the PHY function. Intersil says this can reduce the BOM by 40% and will enable lots of new applications.
( www.intersil.com )
Just as we were convinced that 11a had no place in the home, we heard a countervailing argument from ViXS. We've previously reported on their XCode chip which provides real-time "video QoS" for broadcast video over home networks.
At CES, we met with Sally Daub, ViXS President and CEO. She showed us their demo of the XCode carrying multiple HD video streams over 802.11a wireless. The RF environment was far from ideal, since the CES show floor was filled with lots of potential interference. As the XCode dynamically adjusted the bit rate for the video streams, we didn't see any lost frames or any notable change in the quality of the image. The video images didn't seem to be the kind (like action sports) that would show much effect, so we felt this demo showed only that the technology was working. We'd like to see future demos that stress the real-time compression a lot more.
The most interesting part of our discussion concerned what ViXS calls their "Matrix" 802.11a chip. We had previously reported that ViXS was developing wireless networking based on 802.11a and frankly found it hard to understand why a start-up company would be simultaneously investing in innovative video QoS technology and what seemed to be a highly-competitive wireless networking technology.
Sally explained why ViXS felt it necessary to develop its own implementation of 802.11a. Standard 11a implementations have the speed to carry high-quality video, but would have a problem networking two simultaneous video streams if one were going to a nearby receiver and the other to a distant one. The distant receiver would "hog" the channel by using all of its capacity at a low speed. Neither receiver would get sufficient bandwidth for good-quality video.
ViXS solves this in its Matrix chip by operating simultaneously over two 802.11a channels. That way, one channel could be used for nearby receivers at a higher speed, and the other for more distant receivers at a lower speed. Since the channels are independent, the receivers wouldn't interfere with each other and both would get the best-quality video possible at that distance from the video access point. Matrix also operates at the highest power allowed by the FCC in the 5 GHz band, and aims at the highest receiver sensitivity.
Unlike most 11a participants, ViXS is taking an asymmetric approach, putting added complexity and cost in the access point chip to get the best possible performance for video applications. Matrix is designed to work with all standard 11a chips in client devices, which will ride the cost curve down with volume.
Together, XCode and Matrix provide the basis for what ViXS calls a "residential video area network". A "video master" device like a main set-top box or video gateway would use the Matrix chip along with an XCode chip to provide multiple HD and SD video streams. All client devices would be based on industry-standard chips. We're looking forward to seeing consumer electronics companies take the next step and go forward with the ViXs approach.
( www.vixs.com )
At Cogency's suite, we met with Ron Glibbery President and CEO. Pete Wilson VP Business Development showed us some of the latest devices built with Cogency's HomePlug powerline networking chips. Cogency and ViXS demonstrated full-motion, rate-adaptive Video over Homeplug. We liked the digital audio player with built-in HomePlug - similar to the setup we have in our home with AudioTron (see the article in this report).
We were particularly interested in seeing HomePlug used with Microsoft's XBox Live service. This worked very nicely as an easy way to connect a game machine in a teen's bedroom to a broadband modem elsewhere in the house.
We had thought of Phonex Broadband as exclusively a consumer device maker until we met with Brad Warnock, Director of Marketing, at CES. While we devoted a few minutes to discussing their latest HomePlug products, we spent most of our time learning more about their newly-announced ReadyWire chip for "powerline home communications".
Unlike HomePlug, which is designed for comparatively high-speed computer networking and broadband modem sharing, ReadyWire is optimised for telephone, audio and control applications over powerline. It is based on the TDMA protocol, which is optimised for time-critical communications.
With a BOM under $15, ReadyWire is targeted to enabling low-cost consumer devices over powerline. Examples would include multi-line phones (it can support up to 7 simultaneous full-duplex voice lines); streaming audio to powered speakers (up to 15 simultaneous channels); distribution of V92 dialup modems to remote rooms; and low-speed data networking. Brad said Phonex was in active discussions with many consumer electronics manufacturers.
Chips like this could give "plug and play" a new meaning.
( www.phonex.com )