To collect star transits data qualified for in-orbit calibration, this study derives the full error constraints to limit star trackerʼs influential error sources and computes their error boundaries from a theoretical perspective. The full constraints, including not only the minimum variance estimation of position but also the error bound prediction of scale and intensity of Gaussian-shaped starspots, are studied based on the Cramér-Rao Lower Bound (CRLB) theorem. By imposing these constraints on motion, drift in focal length, and other factors, their boundaries could be determined before launch. Therefore, the in-orbit correction accuracy is expected to be close to CRLB through suitable implementation of these constraints. The correctness of the theoretical position error of motion is demonstrated by the data-fitting procedure against test results of star tracker on dynamic performance. The simulation result shows that the drift in focal length can generate an error with the same magnitude as detector noise and thus might be the dominant error source when star tracker is working under stationary circumstance. Using the accuracy performance of some typical star trackers, this study shows that the CRLB constraint may be very effective to estimate the overall position error of a starspot or one axis, valuable data that can be used for online calibration. The overall position uncertainty analysis shows that a weighted method can be employed for calibration, a process where star data can be given a weight in inverse proportion to the CRLB value.