Low-cost testing of high-precision analog-to-digital converters
Kook, Se Hun
MetadataShow full item record
The advent of deep submicron technology has resulted in a new generation of highly integrated mixed-signal system-on-chips (SoCs) and system-on-packages (SoPs). As a result, the cost of electrical products has sharply declined, and their performance has greatly improved. However, a testing throughput still remains one of the major contribution factors to final cost of the electrical products. In addition, highly precise and robust test methods and equipment are needed to promise non-defective products to customers. Hence, the testing is a critical part of the manufacturing process in the semiconductor industry. Testing such highly integrated systems and devices requires high-performance and high-cost equipment. Analog-to-digital converters (A/D converters) are the largest volume mixed-signal circuits, and they play a key role in communication between the analog and digital domains in many mixed-signal systems. Due to the increasing complexity of the mixed-signal systems and the availability of the new generations of highly integrated systems, reliable and robust data conversion schemes are necessary for many mixed-signal designs. Many applications such as telecommunications, instrumentation, sensing, and data acquisition have demanded data converters that support ultra high-speed, wide-bandwidths, and high-precision with excellent dynamic performance and low-noise. However, as resolutions and speeds in the A/D converters increase, testing becomes much harder and more expensive. In this research work, low-cost test strategies to reduce overall test cost for high-precision A/D converters are developed. The testing of data converters can be classified as dynamic (or alternating current (AC)) performance test and static (or direct current (DC)) performance test . In the dynamic specification test, a low-cost test stimulus is generated using an optimization algorithm to stimulate high-precision sigma-delta A/D converters under test. Dynamic specifications are accurately predicted in two different ways using concepts of an alternate-based test and a signature-based test. For this test purpose, the output pulse stream of a sigma-delta modulator is made observable and useful. This technique does not require spectrally pure input signals, so the test cost can be reduced compared to a conventional test method. In addition, two low-cost test strategies for static specification testing of high-resolution A/D converters are developed using a polynomial-fitting method. The cost of testing can be significantly reduced as a result of the measurement of fewer samples than a conventional histogram test. While one test strategy needs no expensive high-precision stimulus generator, which can reduce the test cost, the other test strategy finds the optimal set of test-measurement points for the maximum fault coverage, which can use minimum-code measurement as a production test solution. The theoretical concepts of the proposed test strategies are developed in software simulation and validated by hardware experiments using a commercially available A/D converter and designed converters on printed circuit board (PCB). This thesis provides low-cost test solutions for the high-resolution A/D converters.