How to prevent Bluetoothaches

"Testing regimes are diversified," says Mike McCamon, executive director of the Bluetooth Special Interest Group (SIG; Overland Park, KS), to accommodate the different "test philosophies of companies as varied as Intel, Nokia, and Logitech." Complementing the chipset maker, cell-phone maker, and mouse manufacturer in that list is now automaker Saab, whose 2003 Model 9-3 includes a built-in Bluetooth-enabled cell phone.

Furthermore, the end use of a Bluetooth product can affect the optimum test strategy. Bluetooth is not like Wi-Fi or Ethernet, McCamon says, for which physical-layer and lower-software-layer parameters are well defined. Bluetooth is more flexible, and "we have to accommodate everything from keyboards to cars to phones."

Karsten Beckman, a Rohde & Schwarz product specialist for Bluetooth protocol test, explains how different applications require different tests: For example, point-to-point performance, such as mouse to laptop or headset to handset, is relatively straightforward and may require minimal test. In contrast, adds Beckman (who is based at the Beaverton, OR, headquarters of Rohde & Schwarz's partner Tektronix), a manufacturer wishing its product to operate well with other firms' products may require more extensive "stress tests" to see how it reacts as data traffic and interference increase (for example, a vendor's laptop communicating with other vendors' laptops in a piconet—Bluetooth's implementation of a wireless LAN).

Not surprisingly, a wide range of Bluetooth test equipment has emerged to serve in prototype design and debug, product qualification, manufacturing test, and field service. The equipment ranges from dedicated Bluetooth test sets to general-purpose instruments that users configure to make Bluetooth measurements.

Complementing instrumentation approaches to Bluetooth test are the unplug fests that the Bluetooth SIG sponsors. These events provide engineers with a chance to try out pre-production designs and learn if their prototypes can communicate with those of other engineers, says McCamon, adding that the unplug fests also help identify conflicts or ambiguities within the Bluetooth specification.

Successful communication with other prototypes at an unplug fest doesn't earn a design formal Bluetooth qualification. Qualification requires that a Bluetooth Qualified Body (BQB), which can be an independent organization or an employee of a manufacturer of Bluetooth products, develop a test plan and documentation. Subsequently, manufacturers seeking top-level (called "Category A") certification must subject their products to tests at a recognized Bluetooth Qualification Test Facility (BQTF). Manufacturers pursuing Bluetooth qualification must join the Bluetooth SIG and can find information on BQBs and BQTF on the organization's www.bluetooth.org members-only Web site. (The www.bluetooth.com site is open to the public.)

The MT8852A performs full Bluetooth audio-layer through RF-layer measurements. As part of its audio test capabilities, it can handle CVSD, m-Law, and A-Law codec algorithms. Courtesy of Anritsu.

Peter Cain, Agilent Technologies' Bluetooth program manager, agrees that using modules that implement full Bluetooth functionality are an effective way to build Bluetooth products, and he says that the practice naturally leads to the use of dedicated instrumentation that can implement full, automated Bluetooth test on the production floor. "Some engineers will want to develop their own test set-ups, but the need is to move on rapidly to manufacturing test, and the fastest way to do that is with a turnkey Bluetooth test system."

Rohde & Schwarz's Beckman explains that, when developing tests, it's important to determine the optimum mix of test thoroughness and production throughput. Engineers may choose to run full RF and protocol tests at the production stage, but such testing could be too expensive and time-consuming.

"Bluetooth chips cost about $2 each," says Beckman. "How much can you afford to spend testing them?" Beckman adds that lower-level physical-layer RF tests are the critical ones to perform. His firm's CMU200 can perform those tests as well as higher-level protocol tests and stress tests in which it simulates multiple devices with which the device under test can attempt to communicate.

If test cost is your overriding concern, you might be considering a "golden radio" test, in which you employ a reference Bluetooth radio with which your device under test attempts to communicate. This test, which Carl Petersen, National Instruments' product manager in charge of high-frequency measurements, calls a "Can you hear me?" test, can do the job if your performance requirements are minimal.

Petersen says a better test will ask "How do I sound?" To provide quantifiable answers to that question, Bob Hay, president of AlloSys (Boise, ID), is developing a Bluetooth test system based on AlloSys Wireless Analysis software and the NI PXI-5660 RF signal analyzer (Ref. 2).

Of course, Bluetooth functionality within a product doesn't stand alone. It's an adjunct to a phone, computer, or other consumer product that offers many more functions in need of test. Petersen points out that additional test requirements can cover LCDs, keypads, audio circuitry, and so on.

Manufacturing test throughput could be enhanced if Bluetooth functions could be tested in parallel with other product functions. Petersen suggests that the PXI architecture in which NI's PXI-5660 operates can readily accommodate the additional instruments required to test the non-RF components of Bluetooth-based products.

Such parallel tests could become a matter of necessity, not just speed. Agilent's Cain hypothesizes that if Bluetooth functionality is integrated with a cell phone's baseband processor, then it becomes impossible to test the Bluetooth function without exercising the cell-phone function.

Cain says that cell-phone and Bluetooth tests typically occur separately. Yet if combined cell-phone and Bluetooth test eludes today's high-volume manufacturing process, vendors of bench-level instruments are taking aim at performing as many tests as possible in one instrument. The most recent effort in that regard is Anritsu's MT8852A, introduced last December. It performs full Bluetooth audio-layer through RF-layer measurements. As part of its audio-test capabilities, it can handle continuously variable slope delta modulation (CVSD), µ-Law, and A-Law codec algorithms.

Other dedicated test instruments include Advantest's R4870, Berkeley Varitronics' Mantis, and Frontline Test Equipment's FTS for Bluetooth. Serving lab and production tests, the Advantest R4870 can measure RF specs such as power, frequency, modulation index, and receiver sensitivity; the FTS for Bluetooth focuses on protocol analysis. Serving installation and field-optimization applications, Berkeley Varitronics Systems' Mantis is a handheld Bluetooth receiver that can measure RF power and packet error breakdowns.

Are there specific benefits to dedicated Bluetooth test sets? Says Bluetooth SIG's McCamon, "My goal is widespread availability of high-quality, interoperable Bluetooth products. I don't care which tools engineers use to achieve that result."