Building an Efficient, Low-Cost Test System for Bluetooth Devices

Developed by the communications and computing industries as a low-cost, low-power wireless technology to replace the myriad cables connecting consumer electronic devices, Bluetooth has slowly grown to address a broad set of applications in many industries. Bluetooth’s simple protocol and ease of implementation make it ideal for wireless communication across a diverse set of products, especially those with low-power concerns, such as mobile phones and handhelds. This application note discusses some of the tests and considerations that are important to consider when building an efficient, low-cost test system for Bluetooth devices.

Introduction

Bluetooth is a low-cost, point-to-point wireless technology intended to eliminate the many cables used to connect consumer electronic devices. Initially, Bluetooth defined a way to communicate wirelessly with cellular phones, PDAs, and laptop computers. Over time, Bluetooth expanded to applications in automotive, communications, home, and office. IBM, Toshiba, Ericsson, Nokia, and Intel founded the original Bluetooth special interest group, and it later expanded to included Agere, Microsoft, Motorola, and 3Com. There are hundreds of associate and adopter members as well.

Bluetooth defines point-to-point voice and data links over a nominal 10 m range, although this can extend to 100 m in an optional high-power mode. Bluetooth operates in the unlicensed 2.4 GHz industrial scientific and medical (ISM) band from 2.400 to 2.483 GHz, with exceptions in a few nations where the frequency range is limited. Bluetooth transmits using a fast frequency-hopping, spread-spectrum technique that hops 1600 times per second over 79 channels that are 1 MHz wide. A relatively simple GFSK modulation scheme transmits data within the 1 MHz wide channels.

Because of the low cost of Bluetooth devices, it is imperative to develop tests and an overall test strategy that optimize the use of instruments and reduce test cost. In production environments, this means developing automated test equipment (ATE) systems that can efficiently and quickly test the thousands of Bluetooth devices produced every day. This article discusses various test approaches and illustrates how modern test techniques and advances in modular instrumentation pave the way for an efficient, low-cost Bluetooth parametric test system.

Testing Methods

The wireless test environment typically has used a combination of protocol and parametric test equipment. Parametric testing can analyze physical aspects of a device under test (DUT) and run through a series of tests relatively quickly. Protocol testing, typically more time-consuming, looks at the higher layers of a particular communications stack or protocol. There are trade-offs to using each type of testing. Many environments use protocol testing on a sample basis to catch defects affecting a group of devices, such as a badly programmed batch of EEPROM chips. On the other hand, parametric testing is more useful for detecting device-to-device defects caused by the manufacturing process.

In highly competitive markets with great pressure to reduce test costs, the "golden receiver" test technique has also gained popularity. This form of testing can handle complex signal protocols but achieves cost advantage in exchange for limited test coverage. However, recent advances in commercial test technology have led to the development of a comprehensive, low-cost parametric test system for Bluetooth. As this article shows, this new approach offers far more test coverage than "golden receiver" techniques at a much lower cost to test than comparable solutions.


Small flaws in silicon or peripheral components can result in performance defects that are undetectable with a quick "can you hear me" test. The following table shows some typical parameters critical for good transmitter performance and the significant improvements in test coverage obtained with an instrumented parametric test system.


Clearly, a parametric test system is far superior to a "golden receiver" setup for detecting transmitter defects in crucial areas. As mentioned earlier, the primary resistance to a parametric test system is in system cost and test time. With the advent of recent PXI-based RF test instrumentation, highly capable RF parametric testing is now an affordable option.

Test time is a key consideration in a production environment. Here again, a PXI-based RF test system delivers compelling advantages with superior throughput compared to traditional instrumentation. Users can test many more devices in a given time using a PXI-based system, thereby reducing test costs. The graph below illustrates the superior test throughput of a PXI-based system when performing a typical in-band power measurement.

The NI PXI-5660 RF Signal Analyzer, combined with software designed for characterizing packet-based transmitters, delivers an economically attractive solution for designing and testing communications transmitters. AlloSys’ Wireless Analysis software, an application created using NI LabWindows/CVI, demonstrates this testing capability. In operation, the system "sniffs" the transmitted packets and captures the first packet it locates that matches an arbitrarily chosen frequency. After it captures a packet, the system analyzes it further for spectral and modulation characteristics.

Testing is relatively straightforward. The transmitting device can perform a file transfer operation for a receiving device located approximately 5 m away. An antenna is an appropriate distance (several cm) away from the transmitting device. The signal then conditions and transmits to the PXI-5660. With special triggering and packet-location techniques, users can capture a packet contiguously and process the packet waveform. In this test, most of the packets are 5-slot packets with a packet length of slightly less than 3 ms. The described configuration is typically uncommon in a factory setting because it involves open radiation. However, a modified version is appropriate for a production test environment. A production environment would typically employ a small test box and attenuate the transmitting signal to simulate the distance required for real-world applications.

The examples below show test data from two Bluetooth transmitters, both of which passed the manufacturing test process and were shipped to commercial accounts. While Transmitter A has good performance when doing a relatively long file transfer, Transmitter B typically failed to complete the same file transfer.

The examples compare the transmitted spectrum and a segment of the demodulated packet. The system can characterize many other parameters, but these two parameters clearly illustrate the performance problem. The data in these graphics resulted from a single 5-slot packet collected from two Bluetooth transmitters while performing a file transfer. The RF signal was converted down and captured with the NI PXI-5660 RF Signal Analyzer and further processed with the AlloSys Wireless Analysis software.


As these results clearly show, Transmitter A is functioning as expected, while Transmitter B has a considerable level of out-of-channel power and unwanted high-frequency components on the demodulated signal. While it is true that a good receiver with a narrowband IF filter can tolerate this serious transmitter flaw, the shift of signal power from the desired channel results in poor performance in a noisy environment.

A cursory analysis of this data quickly leads to the conclusion that Transmitter B uses an I/Q modulator with a 1 MHz LO (local oscillator) offset – note the narrow-band leakage 1 MHz away from the desired signal, as well as the unwanted sideband 2 MHz away. The cause is most likely a flaw in one half of the I/Q modulator, as suggested by the frequencies of the sinusoidal signals added to the demodulated signal. This is an excellent example of the utility of a parametric test that can characterize a transmitter beyond a simple in-channel measurement. It is useful to note that a similar failure would probably occur before the production test stage.

Interestingly, both transmitters can function in the file transfer mode and generally would pass a "golden receiver" test. While the transmitter with good performance usually is successful at completing a 30 MB file transfer, the transmitter with poor RF performance rarely is successful at completing a similar transfer to the same receiver before losing the connection.

The transmitter parameters used in this example are an indication of some of these system capabilities. The PXI test system can also measure parameters such as initial frequency accuracy, frequency drift, and spectral information with various offsets. For frequencies in its range, the system can perform nearly all of the Bluetooth-recommended transmitter tests. The characterization time is short because a captured packet generally can process in considerably less than one second. It is also possible to integrate this system with other PXI control devices capable of controlling the DUT and other testing components to provide a highly integrated and cost-effective test platform.

Non-RF Testing of Bluetooth Devices

While these example tests are critical to passing the Bluetooth test specification, they only cover the RF portion of the device under test. A Bluetooth transceiver is more than just an RF radio -- it is a means for enabling wireless communication between a variety of devices. Many engineers are testing Bluetooth at the system level, where the RF transceiver operates as one function in a larger device.

Many of the devices that incorporate Bluetooth include laptops, PDAs, and cellular phones. In these instances, engineers must test Bluetooth, as well as LCD screens, keypads, audio measurements on speakers and microphones, and DC current and voltage for batteries.

One strategy is to add additional box instruments to the test setup -- each performing one of the tasks listed above. Using a standard PC and the GPIB bus, engineers can tie these instruments together with a Bluetooth test set or analyzer and create an automated test system. However, there are several drawbacks to this setup. First, the speed of the GPIB bus (roughly 1 Mbytes/s) limits the speed of performing large data transfer. Additionally, it becomes difficult to synchronize all these instruments when testing a complex device with diverse functionality.

An alternate approach is to deploy a modular instrumentation system based on the PXI platform. PXI delivers important timing, triggering, and synchronization capabilities that are essential to an integrated ATE system. This approach uses the commercial PCI bus to deliver extremely high data rates between instrument modules and the host PC processor. PXI is the fastest-growing instrumentation standard since the growth of GPIB in the mid-1970s, and it now boasts more than 1000 different modules from more than 60 companies. PXI chassis also accept any of the thousands of available CompactPCI modules.

In relation to Bluetooth, PXI provides modules such as a DMM for DC current and voltage measurements, dynamic signal analyzers and signal sources for audio stimulus and analysis, and vision and motion tools for testing displays and keypads in electronic devices.

With the new RF tools for the PXI platform, combined with the development power and flexibility of the NI LabWindows/CVI programming environment, engineers benefit from a fast, powerful, cost-effective solution for Bluetooth device testing. A PXI-based test solution combined with LabVIEW or LabWindows/CVI can help meet wireless test needs for R&D testing, design verification, or manufacturing.

Appendix of Additional Bluetooth Measurements

Crystal frequency alignment – measures the reference frequency of the device and ensures it is within the desired specification. This usually contains some type of adjustment, making this measurement necessary as a secondary measurement or verification.

Output power adjustment – measures the output power of the device to ensure the user is conforming to whichever implementation (0 dBm or +10 dBm) was chosen. Measures any loss in the signal chain. This usually also includes an adjustment and a verification.

Power control – many Bluetooth devices implement power control schemes by which the device reduces the power transmitted in dB steps but still communicates with the intended receiver. These power control schemes help conserve battery life, an important consideration because Bluetooth is popular in mobile devices where battery power is at a premium. These power control schemes are somewhat dependent on the receiver having very good sensitivity – greater than 80 dB – so it can successfully receive even a weak signal from the transmitting device.

Power density – provides peak power density in a 100 kHz bandwidth. Sweep or capture the entire Bluetooth band. Engineers may want to examine the equivalent time-domain data corresponding to the spectrum, and by averaging this data, they can calculate power density.

Carrier/interference performance – simulates the effects of interferers in the 2.4 GHz ISM band (WLAN, microwave ovens) and monitors the ability of the receiver to resolve the transmitted signal. Continuously modulated signals are common in these measurements in 1 or 2 MHz channels from the carrier. A BER measurement can determine performance.

Adjacent channel power – Involves measuring the main (middle) channel along with an upper and lower channel 3 MHz inside the overall Bluetooth band (2.403 -- 2.483 GHz). Typically engineers measure only the adjacent and, in some cases, the second adjacent channels in production test because of time constraints.

Reader Comments | Submit a comment »