For the Bluetooth personal wireless standard to gain mass consumer acceptance, prices for the IC chipsets must drop significantly. For this reason, manufacturing expenses must be minimized. Since test is a large contributor to the overall manufacturing budget for these ICs, cost-effective test methodologies are required to bring down the overall cost for Bluetooth chipsets.
Increased Test Complexity
Like most chipsets over the years, Bluetooth chipsets offer higher levels of integration to help reduce power consumption, minimize footprints, and lower costs. Today’s typical chipset—two to three ICs—includes the radio modem, a baseband controller, and flash memory.
Single-chip solutions are beginning to appear on the market, with the baseband controller and radio modem on one chip or in one package. The target markets for these single-chip solutions are low-cost wireless applications such as headsets, cordless phones, and adapter cards.
But single-chip solutions will not dominate the market. The demand for radio-modem ICs will continue in applications where the Bluetooth baseband controller is built into the host appliance’s control logic. This arrangement will be the norm for more sophisticated Bluetooth-enabled applications like cell phones, laptops, and PDAs.
For all the benefits of integration, it does create challenges in testing these wireless ICs. A Bluetooth chipset contains RF, digital, and memory blocks (Figure 1). In the case of a single-chip solution, these members exist on one piece of silicon or all in one package. Substantial test time is required to verify the performance of all functional elements using traditional test equipment.
For the radio-modem IC, besides performing DC, continuity, and transmitter tests, the test system must perform a suite of bit error rate (BER) tests on the receiver end of the radio. (See sidebar: Understanding BER). With traditional tester architectures, these tests take considerable time and resources, appreciably adding to the overall cost of test.
Bluetooth Radio-Modem IC Test
An overview of the typical suite of production tests required for a Bluetooth radio-modem IC helps describe the challenge of testing such complex devices.
The first task is to measure the maximum output power from the DUT at several different frequencies, ensuring the final appliance will meet its specification to provide +0 dBm output for 10-meter links and +20 dBm output for 100-meter links. After that, the output power of the transmitter is adjusted and the output power control is measured to evaluate the power-level control circuitry. Next comes measuring the spurious emissions from the DUT, which requires looking at leakage into adjacent channels, such as adjacent channel power ratio (ACPR).
Then out-of-band spurious emissions tests check for signal leakage into frequency bands above and below 2.4 GHz, such as leakage out of the industrial scientific medical (ISM) band. Finally, the 20-dB bandwidth measurement verifies that the 20-dB bandwidth of a modulated signal is less than 1 MHz, ensuring compliance to the Bluetooth specification.
Another set of transmitter RF tests—the modulation characteristics tests—measures the performance of the modulator circuitry as well as the stability of the local oscillator. The initial frequency tolerance verifies the accuracy of the transmitter’s carrier frequency. The next step is to measure the drift of the carrier frequency over several different packet lengths.
Finally, the test system looks at the maximum and minimum deviations in frequencies across different bits in a packet. Bluetooth devices operate under frequency-hopping conditions, resulting in a fast switching time requirement. The synthesizer lock time is evaluated to see how long it takes the synthesizer to lock onto a new frequency.
Along with the transmitter tests, there are a number of receiver tests. The receiver tests essentially are a series of bit error rate (BER) measurements performed under different stimulus conditions.
First, the minimum sensitivity level of the DUT is checked by stimulating the DUT with a low-level signal to verify that it meets the BER specification. The next step stimulates the Bluetooth receiver with a strong signal and measures BER to determine if it meets its maximum input power specification.
For the co-channel measurement, a weak interference signal is used in the same channel as the desired signal. This test checks the capability of the DUT to recover data in the presence of a co-channel interferer.
Similarly, several BER tests are performed with interference signals in channels adjacent to the desired channel to check the operation of the device in the presence of other Bluetooth devices. For the out-of-band blocking test, BER is measured while an interferer signal out of the ISM band is inserted along with the desired signal.
Intermodulation measurements, which check how much the receiver intermodulates a multitone signal, are other necessary tests for Bluetooth devices. For the third-order intermodulation (IM3) measurement, the DUT is stimulated with a three-tone signal: the desired modulated signal, a CW interferer signal, and a modulated interferer signal.
The CW and the modulated interferers are spaced so their IM3 product is in the same channel as the desired signal. The BER of the receiver will depend on the IM3 product of the two interferer signals. Finally, the receiver is tested under frequency-hopping conditions.
BER measurements can significantly impact the device test time, driving up the cost of test overall. Typically, 40,880 bits are used for BER measurements. The specification for BER is less than 0.1%, which means there must be fewer than 41 error bits for the device to pass. The Bluetooth’s data rate is 1 Mb/s, so it takes 41 ms just to process the 40,880 bits of data.
In traditional automated test equipment (ATE), all 40,880 bits are measured and subsequently downloaded to the host computer to determine the BER. With so many BER measurements necessary for the Bluetooth radio modem IC, these time-consuming data downloads increase overall test times, leading to a higher cost of test.
Real-time measurement capabilities in today’s ATE promote more efficient transmitter and receiver testing, significantly lowering the time to production test Bluetooth ICs. By using a real-time BER measurement and real-time RF data-processing architecture, significant improvements in time can be realized for BER measurements in receiver tests and the RF measurements in transmitter tests.
Let’s look at BER measurements first. Real-time BER measurements rely on a unique, real-time data-processing architecture that incorporates a digital signal processing (DSP) engine into the BER measurement channel. The digital data used to create the modulated Gaussian frequency shift keying (GFSK) stimulus also is sent to the DSP of the BER measurement channel.
After the receiver of the radio-modem IC is stimulated, the digital data output is sent to the DSP that compares the output data from the receiver to the original data. This comparison is done in real time, enabling the DSP to quickly determine the BER for tens of thousands of bits and send the final result to the host computer (Figure 2). To make a BER measurement with 40,880 bits takes less than 66 ms. More bits could be tested to achieve better accuracy without significantly adding to the overall test time. As the graph indicates, using as many as 511,000 bits still gives a test time of 542 ms. Note that these measurement times do not include stimulus setup time.
A real-time RF data-processing receiver architecture with onboard DSP offers the lowest time to test the transmit path of the Bluetooth radio-modem IC. With this architecture, the RF signal from the DUT is down-converted to an IF signal.
The IF signal is oversampled for maximum performance by a digital down-converter. The DSP then filters, decimates, and corrects the data received and performs an FFT. After the FFT, the final result is sent to the host computer.
As a result, data manipulation and comparison are done in the RF receiver rather than the host computer, eliminating the need for time-consuming downloads of data. Instead, only the final result is sent to the host computer, helping to minimize test times.
Besides real-time capabilities that directly improve the throughput when testing a Bluetooth IC, general techniques for testing complex ICs also can be applied to further lower test times. For example, multisite test can be implemented to test multiple Bluetooth ICs in parallel rather than in serial. (See sidebar: Case Study)
Multisite testing can help drive down costs where the price tag for test is large compared to the cost of manufacturing. Ideally, multisite testing would provide 100% efficiency; the multisite implementation would be completely parallel so that the test time is the same as for a single-site test.
In the real world, some overhead is associated with multisite test, which depends on tester architecture and the device under test. For example, larger data transfers are required when testing multiple devices at one time. However, the overall reduction in test time still is significant enough to warrant multisite test.
Lower Cost of Test
Although Bluetooth IC test is demanding, there are many ways to lower test times and reduce the overall manufacturing cost of these devices. A real-time test architecture addressing the demands of BER and RF tests can lower the test times for the dozens of transmitter and receiver tests needed for Bluetooth IC production test. When this real-time test capability is combined with more general-purpose techniques such as multisite testing, the overall test times can be reduced, resulting in a lower cost of test for Bluetooth ICs.
Case Study: Multisite Test Reduces Cost
As proven in one case study, multisite testing using automated test systems can reduce the cost of test in today’s complex devices. In this study, a Bluetooth radio-modem IC was tested on a system with a traditional architecture and on the Agilent 93000 SOC Test System with real-time test architecture and quadsite test capabilities. The case study included the following tests:
- Maximum output power over frequency—7 tests.
- Output power control—8 tests.
- In-band and out-of-band—10 tests.
- Synthesizer lock time test.
- 20-dB bandwidth test.
- Maximum sensitivity test.
- Maximum input power at two preamplifier gain levels test.
- Out-of-band blocking—5 tests.
- Adjacent channel—2 tests.
- IM2 and IM3 tests.
The cost of test on the Agilent 93000 decreased from $0.14 to $0.072, down 49% compared to standard ATE architectures. However, pairing the 93000 with quadsite test reduced the cost from $0.14 to $0.027, a decrease of more than 80% compared to the performance of standard ATE architectures.
Since BER is such a critical measurement for Bluetooth radio-modem ICs, it’s important to understand BER before delving into how to measure it.
First, the bit sequence to be sent to the radio-modem IC is coded into an RF modulated signal. The DUT demodulates the signal to recover the I and Q signals. Then the I and Q signals are decoded to obtain the digital data (bit sequence) that originally was sent to the device. Next, the digital data from the DUT is compared to the original bit sequence to determine the BER. Basically, BER is the number of error bits divided by the number of transmitted bits.
The specification for BER is <0.1%. So if 10,000 bits are transmitted, then the number of error bits must be less than 10. The more bits transmitted, the more error bits that are allowed and still have a passing device.
About the Author
Gina Bonini has been employed by Agilent Technologies for more than seven years. During her career, she has worked in business development and product marketing, mainly addressing wireless applications. Ms. Bonini holds a B.S. from the University of California and a M.S.E.E. from Stanford University.
Mandy Davis is involved in product marketing supporting wireless applications at Agilent Technologies. She earned a B.S. in electrical engineering from the University of Wisconsin. Agilent Technologies, Wireless Semiconductor Test Solutions, Automated Test Group, 1400 Fountain Grove Parkway, Santa Rosa, CA 95403, 707-577-3065, e-mail: [email protected]
For more information:
Return to EE Home Page
Published by EE-Evaluation Engineering
All contents © 2002 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.