Don't Expect Turnkey ATE Semiconductor ATE vendors offer SoC test platforms to address digital and RF test; however, finding a turnkey test system that will test all digital-baseband ICs with RFICs is unlikely. (Figures 2, 3 and 4 illustrate test systems suitable for Bluetooth testing.)
Bluetooth can use RF frequency hopping under the control of Bluetooth digital protocol commands. Subsequent use of rack-and-stack instrumentation with digital pattern generators for the frequency-hopping commands while using RF instruments to monitor responses may work. Yet, delays inherent in rack-instrument control schemes, including IEEE 488, can cause timing problems that may make it impossible to correlate digital and RF events. One solution is for test companies to become system integrators and provide all of the instruments. Integrate these instruments as a complete system for the end user controlled by a single software package that allows users to develop and perform tests quickly and accurately. Test Strategies Manufacturers using mixed digital/RF configurations such as Bluetooth for the first time should familiarize themselves with various aspects of the product life cycle, from design through test-platform selection, test-program development, wafer-level test, device-interface-board design and package tests. On-wafer measurements require planning and coordination throughout the design and implementation phase. Under-standing the test requirements and post-production processing requirements are key to determining which ATE and wafer probe scheme makes sense for the device. The wafer probe assembly must meet device requirements. Consider issues such as signal grounding and signal impedance matching, appropriate probe card and DUT board configuration to provide optimal signal integrity. Then calibrate the test system to provide accurate measurements. Wafer Test Once a test system, probe station and probe assembly are selected, developing the test program for the DUT is next because complexity is added whenever a probe is put onto a wafer. This saves time during the on-wafer development phase because if there are any measurement problems, you will know the test problem is not the cause. Testing Costs for Mixed Signal Designs With greater integration, OEMs should be aware that the cost of testing mixed signal chips can outweigh the advantages of integration. Test costs are directly proportional to the capital cost of equipment divided by the overall throughput. Bluetooth transceivers have a 2.4GHz high frequency interface that requires analog testing such as power, noise figure, adjacent channel interference and intermodulation, unlike a purely digital interface that tests using vector scans. If the radio and controller sections are implemented separately, the baseband controller IC can be tested at very low cost. The radio IC, which requires high-frequency measurement, must be tested on more expensive equipment. If the two functions are combined on the same chip (radio and controller), then both functions must be tested on a mixed signal tester, driving up test costs for the digital baseband portion. Printed Antennas Complicating the testing process are radios and integrated modules with printed antennas. At times, it is only possible to make the connection through the air to a test instrument input antenna. With this approach, users must characterize the path loss for each of the 79 Bluetooth frequencies. If the radio IC package has a RF output connector, direct connection to the test instrument simplifies calibrated power and sensitivity measurements. Even with a direct connection, measure and correct for the path loss at each frequency. Digital Parallel Tests Digital testers can test several components in parallel, reducing test time and minimizing indexing time. Indexing is the time to remove and replace one set of components. With Bluetooth's complexity, testing four units in parallel with an overall throughout of over 4000 units per hour typically means testing at less than $0.02 per device. Extra testing costs for mixed signal testers with RF capability can be expensive. For example, a Teradyne Catalyst SoC tester typically costs over $1M. While RF multi-site testing can be performed, it is usually limited to four devices in parallel and requires careful shielding. For multi-site testing on a mixed signal tester, the cost is around $300,000 per site, significantly higher than digital tests. With the highest throughput at some 2750uph, with four site tests and a test time below 4.5 seconds, test cost is approximately $0.08 per device. Testing single chip configurations can reduce overall fabrication yields if the more problematic radio transceiver fails, the whole solution must be scrapped. With two-chip solutions, users can get yields as high as 98% for baseband and 80-90% for the radio portion. One-Pass Testing Processes Typically, semiconductor ATE systems are built to test RF or digital devices, not both. With the RF and digital functionality on single ICs, test manufacturers are offering one-pass processes for testing Bluetooth. For example, Advantest's T7610 8GHz RF test system includes 32 individual 20MHz digital channels for testing baseband logic circuits integrated on RF chips. Teradyne has added microwave capability to its SoC testers for baseband digital and low-frequency analog circuits. Its MicroWave6000 wireless RF and microwave instrumentation, with an option for its Catalyst SoC ATE system, provides the functions for testing Bluetooth RF. Pushing the Test Envelope RF ATE systems seem to have the edge in testing Bluetooth RF components. Testers can ensure the production of known-good RF semiconductors, which can be integrated with known-good baseband die in SiP configurations. However, not all ATE vendors are convinced that RF-plus-digital SoC will dominate the Bluetooth market. Credence Systems touts its RFx RFIC tester for testing Bluetooth RF components, echoing concerns expressed at this year's IEEE Microwave Theory and Techniques Symposium. Credence contends that it will always be more expensive to implement relatively low-performance digital circuits in RF-ready silicon processes, and they expect a significant market for RF-only test systems for Bluetooth components. Conclusion RF testers and digital testers can ensure the production of known-good RF and digital ICs, respectively, that can be assembled into multi-chip SiP configurations. No matter which strategy wins in the end, it is unlikely that one test strategy will fit all applications. * See "How Bluetooth Packaging Choices Impact System Cost" by Douglas J. Mathews, January-February 2001 Chip Scale Review, p. 67 for some invaluable background and insights.
|
Copyright © 2003