Sine-wave parameters estimation - the second source of inaccuracy

Two IEEE standards which is IEEE Standard for Digitizing Waveform Recorders and IEEE standard for Terminology and Test Methods for Analog-to-Digital Converters approximate the ideal rms deviation by ()/spl radic/(12), where () is the average code bin width of an ADC under test. It has been already accepted in J.J. Blair and T.E. Linnenbrink (2002) that the exact rms deviation formula for the sine-wave input is much more complex based on M.F. Wagdy and W.M. Ng (1989) and K. Hejn et al. (1998). However, this approximation is only the first source of inaccuracy in ADC testing. In this paper, we identify a second source, namely the fact that the input sine-wave parameters must be assessed from the digital output signal, rather than from the analog input. We show that the DC offset and the amplitude of the input analog signal evaluated on the base of the digital output differ from their true values. This intrinsic property of mixed-signal circuits have not yet been analyzed, and remain a reason of errors in testing. In particular, the measured values of the effective resolution, calculated according to the IEEE definition, have low repeatability and may vary from test to test for the same ADC. A new method of quantized sine-wave fitting is proposed that removes the bias and thus overcomes this problem. Low repeatability of the IEEE effective resolution measurements contrast with highly stable results obtained for the new method, especially for ADCs characterized by small differential nonlinearity.