Transcript
designfeature By Howard Woo, ADMtek Inc FOR BLUETOOTH TO BE FEASIBLE, THERE MUST BE AN UNDERSTANDING OF PRODUCT VALIDATION AND PRODUCTION TEST.
Bluetooth-IC testing meets chip design luetooth is similar to other communication devices in that its elemental form consists of a transmitter module, a receiver module, and a controller module. You can find these elements in modem, serial-line-card, codec, and Ethernet applications. Validating these communication devices is a two-step procedure in which a design team verifies the performance of the device, and a third party then verifies both performance and interoperability. Production testing of Bluetooth ICs uses similar methods that are common to all the previous transceivers. However, because of the cost constraints of Bluetooth, designing for “cheap”-IC testing is necessary.
B
BASIC BLUETOOTH A Bluetooth system comprises an RF unit, a baseband-link-control unit, and link-management software (Figure 1). In general, the receiver section is the most difficult section to design and to test. Receivers must downconvert and demodulate the RF signal under frequency-hopping control. You use a VCO (voltage-controlled oscillator) to recover the clock and the data. The receiver then forwards the signal to the baseband unit. The transmitter upconverts the baseband information to the frequency-modulated carrier, passing it through frequency-hopping and bursting circuitry before transmitting it to the outside world. This Bluetooth design emphasizes low power, low cost, and robust operation. The link-con-
trol unit and the link-management software determine the device status and establish the network connections, error correction, and encryption. The link-management system also lets you define transmit and receive frequencies, manage power control, control RF frequency hopping, and place the device into a test-loop-back mode, which facilitates frequency, power-control, and packet testing. At the system level, testing takes minutes to exercise all of the air-to-air functions. The test setup consists of establishing a piconet, in which the tester acts as a master, and the system under test acts as a slave. The link-manager protocol controls and configures the system. A series of prescribed tests exercises the transmitter under various conditions: varied packet types, pseudorandom-sequence-function verification, hopping-sequence verification, transmit-parameter-control verification, power control, and frequency switching. A loop-back test follows these tests, essentially testing the receiving section of the system. At the chip level, testing takes seconds. Chip testing focuses on verifying the correct power and frequencies of the transmitter and the sensitivity of the receiver, using a variety of test modes to verify the functionality of the silicon and obtain a measure of its performance. However, from the total-systemcost point of view, it is best for you to align the chip testing with the system testing. Doing so permits betRF
BASEBAND TRANSMITTING AMP RAM
HOST
I/O
ROM
mP
BLUETOOTH LINK CONTROLLER/ DSP BASEBAND PROCESSOR
DAC
FREQUENCYHOPPING CONTROL
FM/BURST MODULATOR
VCO
CLOCK GENERATOR LOW-NOISE AMPLIFIER
Figure 1
CLOCK
CLOCK RECOVERY
FM DEMODULATOR
A Bluetooth system comprises an RF unit, a baseband-link-control unit, and link-management software. www.ednmag.com
May 3, 2001 | edn 77
designfeature Bluetooth-IC testing ter correlation of results and reduces test-coverage issues.
rier frequency places severe constraints on the cabling and test fixtures you place on the device. RF LOW-NOISE DESIGN VALIDATION pc boards are characterized by low AMPLIFIER DIGITAL-SIGNAL GENERATOR: In the past, design validation complexity, precise impedance FM MODULATED/ CLOCK DEMODULATOR UNMODULATED RECOVERY of communication devices was matching, high feature accuracy, SIGNAL block-based. Each block had and low or uniform dielectric stimulus and observaThe receiver-to-transmitter-loop-back test method constant. The boards generally Figure 2 tion paths off the chip. requires a digital-signal generator to input a signal have only one or two layers. For example, you could charac- into the receiving section and loop the subsequent signal back The focus of the test fixture is terize the antialiasing, receiving, through the transmitter for analysis. maintaining signal integrity. The and smoothing filters of a momajor problems that contribute dem. However, as the frequency of com- to the transmitting section. Preferably, to signal loss are transmission-line efmunication devices increases, any at- you can generate this digital word on fects, impedance mismatching, and retempt to examine the high-speed path chip or read it from a memory device to flections. Consequently, correlating the can cause signal-integrity problems. The provide a continual transmitter signal; design validation with the production act of making a measurement can distort otherwise, you must make provisions to test is a difficult task. One major source the results. Thus, you should not direct- provide an externally generated word. of discrepancies depends on how well the ly test the transmitting or low-noise amAn added complexity of signal analy- part contacts the test fixture: In the deplifier. An indirect method, such as loop- sis is the Bluetooth requirement for fre- sign-validation environment, you can back, is preferable. quency hopping. Frequency hopping oc- solder the device under evaluation diThe Bluetooth specification describes curs over 79 channels (22 channels in rectly to the pc board, whereas in the prothe test requirements for certification Spain, France, and Japan). To simplify duction-test environment, you must (Reference 1). In general, the validation analysis, hopping must be controllable, press the device onto the test fixture. In methodology can follow one of four ap- meaning you must be able to turn it on addition, run-to-run variations in the IC proaches: loop-back of the receiver into or off. Additionally, you must be able to and packaging can lead to erroneous the transmitter, loop-back of the trans- select some different channels. conclusions. To aid in this correlation, mitter into the receiver, digital control of The Bluetooth spec also requires that the validation fixture and the production the transmitter, and analog control of the you test the transmitter’s output power test fixture should be virtually identical. receiver. Direct digital control of each of (average and peak), power density, powthese approaches, or test modes, is the er control, transmit-output spectrum, PRODUCTION TESTING preferred approach; however, you can modulation characteristics, initial carriProduction testing of any communicontrol the test using the given RF pro- er frequency, and carrier-frequency drift. cation device focuses on qualitative tocol. measurements. Bluetooth is no different. RECEIVER VALIDATION The production-test environment is TRANSMITTER VALIDATION To validate the receiving section, you more challenging because of longer conLoop-back of the receiver into the must primarily use the test approach that nections to power and ground, longer transmitter is one of the better methods requires analog control of the receiver, in connections to the instrumentation, and for qualifying a transmitter. It requires a which a digital source that can generate the fact that the surface is not ideal for digital-signal generator to input a signal a Bluetooth-modulated signal applies a making contact with various parts. into the receiving section and loop the signal to the receiver input for process- Equipment on the test floor also provides subsequent signal back through the ing. You can remove the processed signal a noisy environment that may impact transmitter for analysis (Figure 2). The after it passes through the demodulator chip performance. Often, the tests in prodifficulty of this approach is in deter- and clock recovery for analysis or after it duction are a subset of the validation mining the best place for the loop-back passes through the baseband processor tests. The emphasis, however, is on a stato occur. Typically, the loop-back occurs for analysis (Figure 3). You can also use ble and repeatable suite of tests. before the downconversion, but this the previous loop-back mode, in which You can divide production testing into method fails to exercise much of the re- you analyze the signal through the trans- two parts: testing the baseband and testceiving-and-transmitting circuitry. Loop- mitter. ing the RF blocks. (Figure 1) The baseing back after the downconversion inThe receiver measurements for Blue- band block comprises digital subblocks, creases design difficulty. The resulting tooth are sensitivity (single-slot and mul- consisting of a controller, a microprocessignal requires a spectrum analyzer, a tislot packets), carrier-to-interference sor, memory, and an assortment of I/O. vector-signal analyzer, or a power meter performance, blocking performance, in- The RF block is a mixture of analog and to perform the measurements. Instru- termodulation performance, and maxi- digital subblocks, consisting of frequenmentation must be able to trigger on a mum input level. cy-hopping control, filters, a modulator, burst packet, given that the Bluetooth burst control, a demodulator, a transsignal is a sequence of time-division-du- TEST FIXTURES mitting amplifier, a low-noise amplifier, Note that the baseband frequency of and attendant clock circuitry. plexed bursts. One alternative is to provide a means Bluetooth is 1 MHz, and the carrier freTesting the baseband block usually infor generating a digital word that applies quency of the system is 2.4 GHz. This car- cludes a combination of scan, automat-
78 edn | May 3, 2001
TRANSMITTING AMP
DAC
FM/BURST MODULATOR
ANALYSIS TOOL
www.ednmag.com
designfeature Bluetooth-IC testing ic-test-program generation, and IDDQ. LOW-NOISE BLUETOOTH AMPLIFIER DIGITAL Embedded memory can cause complicaLINK SOURCE: FM CLOCK tions, so some form of built-in self-test CONTROLLER/ BLUETOOTHDEMODULATOR RECOVERY DSP BASEBAND MODULATED exercises the memory. The embedded PROCESSOR SIGNAL memory (depending on the memory size) may also require a burn-in to ensure BIT-ERROR TESTER reliability. You should also implement test modes so that they continualFigure 3 ly cycle. Such cycling in combination with the RF block generates a given You can test receivers using the analog-input-digital-output method. frequency signal that aids in determining TRANSMITTING AMP the functionality and performFM/BURST ance of the Bluetooth system. DAC MODULATOR Figure 4 Testing the RF block is interestBLUETOOTH LINK ing and difficult. In the days of modem CLOCK CONTROLLER/ testing, concern arose that the noise from VCO GENERATOR DSP BASEBAND the faster digital circuits would affect the PROCESSOR L0W-NOISE performance of the slower analog cirAMPLIFIER FM CLOCK cuits. Now, there is equal concern that the DEMODULATOR RECOVERY noise from analog circuits will affect the performance of the digital circuits. Consequently, designs must contain adequate Testing the receiver through analog control makes use of the loop-back path from the transmitter to the receiver. It applies a given digital word to the transmitting section, upconverts it, loops it VSS connections. Testing the receiver through analog back into the receiver section, and then downconverts it back into a digital format. control plays the foremost role in these circumstances. It uses the loop-back path block intellectual property and then per- er-to-transmitter mode, but all of the test from the transmitter to the receiver by forming the integration, or a combina- modes facilitate verification and producapplying a given digital word to the tion of both. For example, you may de- tion testing.k transmitter section, upconverting it, sign an FM modulator and demodulator looping it back into the receive section, on the test chip. Once you debug these References 1. Bluetooth RF Test Specification,Verand then downconverting it back into a blocks, you can integrate them onto one digital format (Figure 4). You must make chip with the remaining components. Al- sion 1.0B. 2. “Performing Bluetooth RF Meaallowances, because this test mode may ternatively, you can purchase and intenot reflect what you see in the field op- grate modulator or demodulator blocks. surements Today,” Application Note eration of the chip. In theory, it is a simple matter of putting 1325, Agilent Technologies, 1999. 3. Robinson, Angus, “On your marks Again, loop-back occurs at the base- the respective blocks together. However, band frequency because signal-integrity in practice, the interactions and coupling for testing Bluetooth,” Test and Measureeffects may cause erroneous conclusions between the blocks, the silicon substrate, ment World, September 2000. 4. Dewey, Mike, “In process functionif the loop-back occurs on chip. To avoid and the package itself pose a difficult this situation, the tester itself may con- challenge. Using the following test al test cuts cost of RF products,” Test and tain a Bluetooth demodulator and mod- modes helps you with validation and Measurement World, December 1998. 5. www.bluetooth.com. ulator. One benefit of this arrangement production-test development: is that, although you conduct the system ● loop-back from the receiver to the Author’s bio graphy test in a similar fashion, you use two sets transmitter before demodulator, to determine connection and transmis● loop-back from the receiver to the Howard Woo is manager of product and test engineering at ADMtek Inc, where he sion. At this point, you could implement transmitter after demodulator, has worked for four years. In his current tests that the Bluetooth specification ● digital control of the transmitter, position, he performs IC-product validamentions. ● analog control of the receiver, tion, arranges and implements IC testing You can supplement the loop-back test ● loop-back of the transmitter to and product support, and supports qualiwith a transmitter-section test followed the receiver, by a receiver-section test. Implementing ● on-chip generation of test words, ty and reliability engineering goals and objectives. Woo has a bachelor’s degree in scithese tests verifies that the device can and transmit and receive over the Bluetooth ● PLL measurement via a divide-by- ence from the University of Toledo (Toledo, OH) and a master’s in physics from San specification. Judicious testing results in 24 signal. designs that can hold their ground in the These test modes do not shut off any of Jose State University (San Jose, CA). In his field. the major blocks, so they provide a rela- spare time, he enjoys traveling and has vistively accurate picture of the silicon per- ited Egypt, Turkey, Greece, Italy, France, IC-DESIGN RAMIFICATIONS Germany, the Netherlands, Spain, Japan, formance. Chip design may generate test chips for The Bluetooth specification recom- and China. each of the IC blocks, buying the IC- mends only the loop-back-from-receiv-
80 edn | May 3, 2001
www.ednmag.com
