Two common specifications for a digitizer are the signal-to-noise ratio and the number of effective bits. To measure either specification, one first inputs a sine wave to the digitizer. The next step fits an analytic sine wave to the digitized sine wave. The fit parameters of the analytic sine wave are frequency, amplitude, phase and offset. The fitting process varies the fit parameters until the analytic sine wave matches the digitized sine wave. The final step calculates the signal-to-noise ratio and the number of effective bits from the fitted sine wave. It is well known that the signal-to-noise ratio and effective bits specifications exclude errors in the fit parameters. What does this mean? The usual answer concerns the digitizer's frequency response. Specifications based on sine wave fits brush aside errors in amplitude flatness and phase linearity. This paper investigates specifications for digitizers that contain undesired square and cubic nonlinearities. How does the absolute accuracy of a nonlinear digitizer relate to signal-to-noise ratios and effective bits?
展开▼