networkZONE Products for the week of July 7, 2003
Atheros Says . . .
New Chips, New Attitude -- Atheros' Third-Generation
WLAN Chip Sets Offer Improved Performance and Range, Lower Power and Cost
Atheros Communications, has introduced seven new chipsets that cover the gamut of WLAN applications. These chipsets support multiple WLAN standards with optimizations for different market segments and price points. A single driver and firmware code base supports all the chipsets, providing both backward and forward compatibility with Atheros' legacy and next-generation multi-standard designs. The new chipsets feature third-generation wireless technology that boosts throughput and range, while decreasing power consumption and cost.
"Atheros is supporting all the standards, covering all the international tuning ranges, setting new benchmarks for performance and range, and making WLANs both foolproof and future proof," said Allen Nogee, senior analyst at In-Stat/MDR. "Their broad range of design wins and steady stream of new products are testaments to Atheros' power in wireless LANs. We expect Atheros to maintain a position of prominence as the market for multi-mode solutions develops."
"We are announcing availability of our third-generation 802.11a/b/g and 802.11b/g silicon, while competitors still struggle to launch their first-generation products with 802.11a support," said Craig Barratt, president and chief executive officer of Atheros. "Meanwhile, our 802.11g products are dominating benchmarks by the PC OEMs, enterprise and retail vendors, and the industry press. System developers have fielded nearly 200 products based on our chipsets, and both system developers and end users will find more of what they like in our third-generation chips."
Seven new chipsets available now in volume
The new chipsets introduced today include three client solutions and four
access point solutions:
These chipsets are the most highly integrated WLAN products available today. They include all of the functionality needed to implement a client subsystem or a complete access point or gateway. Atheros has reduced the bill of materials over the company's previous generations of highly integrated chipsets as well as over competitors' chips.
The third-generation chipsets are shipping with support for draft 8.2 of the 802.11g specification -- the final 802.11g draft version. Atheros is first to market with an implementation of this final version.
All of the third-generation products announced are now shipping in volume.
Raising the bar for wireless networking
The third-generation technology in the new chipsets provides numerous improvements
in performance and range, while reducing power consumption compared to competing
products. The new chipsets also implement Atheros Super G and Super A/G
capabilities to deliver up to 90 Mbps TCP/IP throughput for 802.11a/g and
802.11g wireless LANs.
Atheros' performance enhancements provide more than 10 times the actual throughput of other wireless LAN solutions, are scalable to any number of users and are fully compatible with legacy equipment. Other solutions that offer special turbo or bursting modes are only effective under limited network conditions and adversely affect users of 802.11b or third-party 802.11g products.
Universal connectivity and security
Universal wireless connectivity means that users can establish a wireless
connection no matter which WLAN standard they are using (IEEE 802.11a, 802.11b
or 802.11g) and no matter what country they are in. Atheros third-generation
multi-mode chipsets support all of today's WLAN standards, as well as the
draft 802.11h standard, and cover all worldwide wireless networking spectrum.
The chipsets also support the provisions of the Jumpstart Broadband Act
recently introduced in the U.S. Senate. The resulting universal wireless
connectivity allows WLAN users to roam across continents and within mixed
networking environments, because the chipsets automatically self-configure
for different wireless and security standards.
A key feature of all the Atheros third-generation chipsets is security
without compromise. These chipsets uniquely support both the Temporal Key
Integrity Protocol (TKIP) and the government-grade Advanced Encryption Standard
(AES) specified by the Wi-Fi Protected Access (WPA) standard with no performance
reduction. Competitors' solutions suffer a 20-50 percent performance reduction
when WPA or AES is enabled.
analogZONE Says . . .
An early, and often controversial player in the 5-GHz WLAN market, Atheros has made its presence felt once again -- this time with the introduction of several much-improved client and AP chip sets. Sporting well-focused mixes of features and functions, they have precisely targeted nearly every major segment of the market. While this is certainly news, what makes it worth covering is the company's new attitude about disclosing some of the details of how their products work. Until now, Atheros has preferred to keep any technical details of its technology or chips hidden behind vague press releases and misleading white papers. Now, under the leadership of Craig Barratt, their new CEO, they seem to be much more forthcoming about what makes their chips tick. And in doing so, it's become a bit more apparent why they may stand a chance of holding much of their market share as several other excellent dual-band chip sets arrive on the scene.
In a time when so many companies expect the media to simply reprint their press releases and call it news, my conversation with Rick Bahr, Atheros' VP of Engineering, was a refreshing experience. While not able to be completely forthcoming he managed to give me enough details to give confidence that Atheros is not making any patently untrue claims about its products. He also managed to clear up several mysteries about the Atheros architecture that have bugged me for years.
For example, Bahr explained that the current products use the same modified dual-conversion architecture they've had since the beginning. While not quite a true superhet design (their channel filtering is done at baseband), it does provide better linearity than a direct-conversion system, and avoids the need (and cost) for transmit and receive calibration loops, as well as avoids PA Pull issues. Using this quasi-dual-conversion technique allows them to distribute gain throughout the system, making for a design that is more tolerant of the process variations found in foundry CMOS fabs. They also manage to eliminate costly, space-hungry SAWs from their BOM by using "slightly loose" receive filters (on-chip LCs), plus a combination of digital (adjacent channel) and analog (alternate channel) filters in the baseband.
But while Rick insists that the new chip sets represent incremental improvements to the architecture they've had all along, I'll contend that these new products are significant because they are a real break from the "all-on-chip" approach Atheros pushed so heavily in its early days. While Atheros reference designs did include the option of external components they claimed that ODMs could produce an acceptable low-cost "consumer-grade" product without them. This initial focus on eliminating off-chip PAs and LNAs in their early radio products cost them (and their customers) dearly in terms of output power, sensitivity, and overall performance.
In fairness, I should mention Mr. Barratt pointed out that some of the disappointing performance was due to ODM manufacturers doing some of their own tinkering with Atheros reference designs. I'll concede that this certainly caused some of the problems, but I also know of at least two companies who made small fortunes selling discrete PAs and LNAs to prop up the performance of the original "one-chip" radio.
When all is said and done, the new chip sets still have vestigial traces of on-chip PAs and LNAs, but the real work is now done off-chip with discrete external components. The good news is that the on-chip elements reduce the amount of gain (and cost) supplied by the external components. This, and some other incremental tweaks, should finally put to rest most, or all, of the persistent complaints I've heard about earlier chip sets.
In addition to these basic radio improvements Atheros has broken out its product line to support both g and dual-band products in a cost-effective manner. Their client side dual-band solution uses the same Tx/Rx chain through the baseband and LO, and then splits the RF path, although both paths live on the same chip. Since a client is non-concurrent (it does not operate on both bands at the same time) both RF chains can share the same synthesizer.
Depending on your requirements Atheros can supply you with a dual-band access point chip set that uses either one or two radio chips. For fully-concurrent dual-band operation they have a reference design with two radios sharing a single chip that sports dual baseband/MACs. In more cost-conscious situations you can also get a single-MAC, non-concurrent access point.
They've also added some significant elements to the MACs that differentiate the product lines and enhance performance for their intended application. One good example is its security accelerator cores, which support full-speed encrypt/decrypt operations without burdening the host system. Besides assuring full performance the hardware cores allow a host CPU to run much slower adding significant power savings in laptop and handheld clients.
Between their well-differentiated silicon, greater willingness to discuss technical details with the press, and support for both the new 802.11g standard and the ITU's initiative for a 5-GHz "World Band," Atheros seems to have abandoned the belligerent stance that characterized its early days. In adopting a more open, cooperative posture, and concentrating on performance as closely as price, they are improving the overall credibility of the 802.11 market.
While these efforts should help quell the confusion that threatened to slow consumer acceptance of all things wireless, I do worry about Atheros' recently-introduced "SuperG" multi-channel aggregation scheme. The proprietary protocol that bonds a pair of channels together for a system throughput to over 100 Mbit/s does not currently "play well with others." Not only is competing silicon excluded from using the higher rate, but the channel aggregation technique splatters enough noise across the rest of the 2.4-GHz spectrum that most other channels are useless. While this anti-social behavior might be fine in a home setting where the nearest neighboring WLAN is a few hundred feet away, it's downright unacceptable in a multi-tenant, or public infrastructure environment.
Hopefully, Atheros is working on a solution, under the new leadership of Mr. Barratt, that will reflect the company's new, more cooperative character.
|