Method of reducing contactor effect when testing high-precision ADCs

Being able to test the intrinsic performance of a device under test (DUT) has always been the main goal of a test engineer. Achieving this goal is becoming increasingly difficult due to tester and test setup limitations. In particular, with the wide spread availability of high-precision, medium-speed ADCs in tiny packages, the properties of the contactor between the DUT and the device interface board (DIB) is becoming an increasing concern as it introduces significant discrepancies between intrinsic device performance and the performance measured in production test. This paper presents results of a 12-bit ADC case study carried out at Analog Devices. For a representative sample set of devices, the linearity is characterized using a customer-like setup and a production test setup, successively. The discrepancy between device measurements using the two setups is first determined. Using a model-based testing approach, the noise-induced test uncertainty can be reduced [1]. We present a modification of this modelbased technique that predicts the performance of a device measurable using the customer-like setup from measurements obtained using the test setup. The performance of this modification is evaluated over a sample set of devices.