Novel method of model-based measurement for A/D and D/A converters

It is known that the traditional fullcode test for A/D and D/A converters is time consuming and costly. The currently used model-based test method has only limited applications in low resolution converters (below 10 bit, for example) with the lack of taking account of the superposition errors. Facing the challenge, a test strategy of model-based measurement for A/D and D/A converters proposed in this paper is applied to solve the above-mentioned problem. In order to predict the converter specifications in full-code states with a few test data, an error model of A/D and D/A converters has to be built. Through building the error model, the measurement problem can be converted to the system identification problem of which variables are bit states, and estimated parameters are significant bit errors and superposition errors. The test codes are well selected with statistical experimental optimal design method. Using the selected test codes, we can get a set of sufficient and necessary test data with which the significant bit errors and the superposition errors can be accurately estimated and the full-code-state specifications of A/D and D/A converters can also be predicted. Results of laboratory experiments verify that this test strategy can well solve the contradiction between measurement cycle and accuracy. The proposed test method can provide a new approach to a highly efficient product check and specification test for high resolution A/D and D/A converters.