With the demand for more functionality at less cost, the RF testing landscape is evolving and morphing into a new frontier.
In the cellular space, convergence of services continues to occur at an accelerated rate. Consumer service providers and product manufacturers are releasing an abundance of new products that crossover between the PC and handset markets.
Take, for example, some recent announcements from industry leaders like Cingular in the infrastructure market and Samsung and Nokia from the handset market. Notice that the directions of their new products stray away from their mainstream past.
Cingular is launching 3G high-speed data networks in major cities across the United States. BroadbandConnect is the marketing name for the high-speed downlink packet access (HSDPA) technology that will provide subscribers with data rates up to14.4 Mb/s. Surprisingly, its intended customer initially is not the cell phone user but the laptop user. As a result, the service-provider barriers that consumers were accustomed to are broken and merged.1
Consumer electronics manufacturers also are creating crossover products. Samsung, the third largest provider of handsets, recently announced that it will release 3.5G service handsets with an aim toward capturing market share through mobile TV.
Mobile TV broadcasting could be worth $8.4B by 2010 according to Informa Telecom & Media of London. Samsung believes its products are well positioned to capitalize on the future mobile TV market. This is yet another example of the blurring of traditional barriers by consumer product manufacturers.2
Nokia recently announced that it is investigating Internet-only products that could leverage WiMAX to deliver high-quality services to the end consumer. Although premature, the Finnish company is hoping to leverage the success of its Linux 770 product into the next-generation high-speed Internet services. Traditionally, Internet access has been a PC-dominant market but now is challenged by the likes of Cingular, Samsung, and Nokia.3
Supporting Technologies
Zooming in from the macroscopic consumer services level to the microscopic device technology, integration is made possible through high-tech semiconductor processes. The continually shrinking CMOS processes such as 90 nm and even 45 nm open the door for more robust RF devices in wireless applications.
At UCLA, researchers are creating applications in the 60-GHz world using advanced semiconductor processes that once were reserved only for expensive GaAs and III-V processes. The advent of a 60-GHz silicon process will extend semiconductor models that now are only valid to about 5 GHz to higher frequencies that should, in turn, lower the design and manufacturing cost of current cellular-band RF system-on-a-chip (SOC) products. As a result, lower costs to integrate multiwireless applications in SOC products with denser semiconductor processes are on the horizon, forming the foundation for crossover and merged wireless products.4
In the semiconductor packaging world, convergence is achieved because multidevice system-in-a-package (SIP) technology allows for more applications to be integrated. Advanced packaging technology such as SIP combines traditional board-level elements like passive components and hybrid circuits along with the SOC to create solutions that appeal to larger market segments. The higher-density packaging of SIPs coupled with higher integrated SOCs from shrinking processes present an environment that potentially merges WLAN and cellular technologies.5
From a review of recent changes in the consumer market and technical trends, it�s clear that service providers and consumer product manufacturers are venturing into new space and investing in crossover products. The crossover products and services made possible by improved semiconductor processing and packaging technologies are broaching traditional service and product-line boundaries, and differentiation is definitely blurred.
Already cable and telecom service providers are engaged in an ongoing price battle to gain consumer market share. Price slashing and multiservice bundles can be seen in local advertisements across the nation. The convergence of technology will present a myriad of choices for high-end consumer services.
What does all this high-technology convergence of services mean to the testing industry? Semiconductor producers expect�and in some cases demand�that increased integration and added complexities will not translate to higher test cost. Manufacturers want the testing cost to reduce at the same rate as the end product to preserve their gross margins. Consequently, the RF testing landscape is evolving and morphing into a new frontier.
Multisite RF Test
As with the trend of most commodity products, multisite RF testing, once thought of as an unbreakable technical barrier, will become the norm. Industry experts will recall that multisite RF is not new. However, previous multisite test solutions were limited to single-function devices such as power amplifiers and receivers and single-mode, low data-rate receivers such as Bluetooth applications (Figure 1).
Figure 1. Serial Radio Block Diagram
The new RF SOC will contain multimode wireless communications that feature functions such as Bluetooth and GPS as a subset (Figure 2). With multiple service functions come the multiple frequency bands that will challenge the ATE to comply to a multitude of industry test standards while maintaining the downward price trend.
Figure 2. MIMO and Multiband Radio Block Diagram
ATE suppliers are scrambling to be prepared for RF multisite tests. In the future, the ATE supplier that can produce RF test solutions which meet the multisite price pressures will emerge as the market winner.
A key technology enabler to multisite RF test will be access to multiple RF receivers to realize parallel test efficiencies. Most RF testers today can only test RF parameters serially, which negates the value proposition of multisite testing. However, enabling multiple receivers needs to be low cost to meet the eroding prices for cellular handsets.
The cellular/WiFi market size is estimated to be somewhere in the area of 500 million units by 2010. 3G cellular handsets, early nonstandard 802.11n and 802.16, and ultra wide band (UWB) products already have begun shipping. Many of the new wireless standard products are targeted to ship in larger volumes in 2007. This leaves little time for the ATE manufacturers to introduce next-generation RF/wireless SOC test systems into the marketplace.
Multiport RF Test
Next-generation wireless products will have even more radios integrated into a single product to support multiple bands and multiple wireless standards. In conjunction with multiple-input multiple-output (MIMO) growing in popularity due to promised increased range and higher data rates, the RF pin-count is expected to at least double and, in many cases, quadruple.
Additionally, the new products must have the same or reduced cost structure as their predecessors. To remain competitive, the ATE companies will have to completely rethink the RF system�s architecture.
Historically, RF/wireless testing was an expensive operation where each radio was tested serially. A typical serial RF radio block diagram as shown in Figure 1 shows only two RF pins, one for transmit and one for receive, and one set of baseband inputs/outputs.
The market no longer will support a serial testing paradigm. In fact, due to the emerging multibanded radios, serial testing is not commercially feasible. The specification testing for these products historically has been much more difficult for cellular vs. WiFi.
Cellular phones must operate over larger areas of a few miles per cell site while short-range WiFi products like Bluetooth, Zigbee, UWB, and WLAN operate over short distances of approximately 10 to 100 meters. Cellular�s longer distances necessitate higher transmission power. Higher transmission power dictates that tighter masks be imposed on the cellular devices to minimize interferences with other RF bands. Conversely, short-range WiFi need only be designed to operate within a small radius.
The up and coming long-range WiMAX standard (802.16e), however, brings the convergence of the testing requirements even closer to cellular. Mobile WiMAX basically is WiFi with cellular range, so the cellular and WiMAX testing profiles from an ATE viewpoint are very similar.
To compound matters, both cellular and WiMAX products will be using MIMO antennas. The current popular MIMO architectures are 2�2, 2�3, and 2�4, which mean instead of just one RF pin per radio, each device will have four to eight pins per radio.
An example 3�3 WiMAX block diagram with 3G triband, GPS, and Bluetooth is shown in Figure 2. This places an enormous challenge on ATE suppliers because the current RF/wireless testers were designed to test single radio architectures, not larger pin requirements.
Additionally, the bandwidths of both cellular and WiFi products have increased. With UWB also on the horizon, this places higher performance requirements on the ATE�s baseband digitizers and arbitrary waveform generators.
Lastly, if that weren�t challenging enough, due to the low average selling price of these devices, there is no commercial way to test them with existing serial RF ATE architectures. As a result, ATE companies are forced to develop a system that can address the wireless market�s needs of high RF pin-counts, multiple parallel digitizers, arbitrary waveform generators, and high device throughputs to minimize test cost.
In the last several years, wireless/RF price erosion has burdened the ATE companies to think outside the box to develop newer and better test strategies to remain competitive. Perhaps you can recall that the industry price necessary to enable Bluetooth as a viable commercial technology was $5 several years ago. That price has long since come and gone.
Now, the buzz price to enable WiMAX, which is basically WLAN on steroids, also is $5. WiMAX devices are much more complicated in terms of RF pin-counts, broader bandwidth, and more stringent RF specifications. A typical RF radio bill of material currently is approximately $2.01.6 The testing cost of these devices historically has been in the 10% to 30% range.
Much of the packaged device testing is outsourced to test houses in Asia, which purchase testers from various ATE companies and then sell time by the second. The rates run $0.03 to $0.08 per second for a typical ATE system equipped with RF/wireless testing capabilities. A single RF pin radio chip, whether it is cellular or WiFi/WiMAX, typically requires 3 to 8 seconds of testing, so the test cost of a single radio ranges from a $0.09 to $0.64 per device.
Summary
Emerging and next-generation WiFi products will have three to 10 RF pins per device. One key reason for the increase in RF pin-counts is that MIMO antennas will be used to extend range and increase data rates.
Early WiMAX devices, for example, will likely be 2�2 and 2�3 architectures. The required test cost per RF pin cannot be achieved with serial RF testing. Not only will ATE companies have to increase the number of RF pins available in their systems, but they also will have to reduce the capital cost of the system and simultaneously enable RF parallel testing methods to address and reduce the test cost per RF pin.
Additionally, since RF and baseband integration continues and often the radio chip includes a portion of the baseband functions, chip manufacturers will require more system-level testing such as bit error rate and error vector magnitude in place of the traditional parametric testing from their RF ATE. This further burdens the ATE manufacturers with integrating parallel, high-precision, high dynamic range baseband digitizers and arbitrary waveform generators into their ATE systems. The new digitizers and arbitrary waveform generators will need some creative way of processing complex modulation schemes in parallel to achieve the low test times necessary.
Another necessity of the next-generation test systems will be parallel low-frequency reference clocks with excellent phase noise characteristics to act as the reference crystal oscillator for the devices. Next-generation WiFi standards like WiMAX place tighter phase-noise specifications on the reference clock due to the higher-order complex modulation schemes that they use.
This is a nontrivial challenge because the ATE�s internal system local oscillator must have better phase noise to meet the test requirements while simultaneously improving tuning speeds. Generally, these are two conflicting design parameters where one is traded for the other. Allowing the ATE system cost to increase can relax some of the trade-offs.
Due to the highly volatile nature of the market, device volumes typically start off fairly benign and, with little notice, can ramp up exponentially. For this reason, ATE purchasers will look for the new test architectures to be easily configurable to support incremental capital cost as RF sites are added to the ATE systems to support the higher volumes.
One way to accomplish this and still remain a profitable endeavor is to invent a test system that enables site-configurable parallel testing of RF products with high RF pin-counts. The next-generation systems must be able to test MIMO type architectures, but testing MIMO architectures won�t be enough.
Additionally, the next-generation test systems must test MIMO products, either cellular or WiFi, in parallel. Just as today you can purchase a digital pin card with 32+ pins per card, the next-generation RF testers will likely have RF pin cards. The density of RF pins per card, cost per pin, and feature set will be the battleground for the RF ATE suppliers.
Using a simple MIMO architecture, 2�2 for example, add �4 parallel testing capability and suddenly the RF pin count quadruples from the serial radio architecture. High-density parallel baseband stimulus and processing also must be available, and that too must be site configurable.
Unfortunately, the market will not accept a trade-off. So the ATE race is on to provide parallel RF test capability of complex RF SOC and help drive down the cost of handsets.
References
1. http://news.com.com/Cingular+launches+3G+network/2100-1039_3-5984005.html?tag=nl
2. http://news.com.com/Samsung+pins+hope+on+3.5G%2C+mobile+TV+phones/2100-1041_3-6086263.html?tag=html.alert
3. http://news.com.com/Next+Nokia+minitablet+to+get+Webcam%2C+WiMax/2100-1044_3-6086747.html?tag=html.alert
4. Moore, S., �Cheap Chips for Next Wireless Frontier,� IEEE Spectrum, June 2006.
5.http://www.amkor.com/enabling technologies/SIP/index.cfm
6. �Low-Cost Handsets Drive Single-Chip Designs,� Electronic Engineering Times, June 12, 2006, pp. 53-54.
About the Authors
Keith Schaub is an RF product engineer at Advantest America and the author of Production Testing of RF and System-on-a-Chip Devices for Wireless Communications. Mr. Schaub received an M.S. in electrical engineering from the University of Texas. e-mail: [email protected]
Anthony Lum is an SOC product engineer with Advantest America and has almost 20 years of RF/microwave-industry test experience. He has a B.S.E.E. from Arizona State University. e-mail: [email protected]
Advantest America, 3201 Scott Blvd., Santa Clara, CA 95054, 408-988-7700
October 2006