Orthogonal transformation has been used for State Estimation, since it is hard to solve by normal WLS method when weights of some measurements differ much from others. Though the numerical stability of orthogonal transformation has been generally accepted, it is reported that the efficiency might be another problem for Orthogonal SE. An efficient engine for orthogonal SE has been proposed in this paper, which is based upon fast Givens rotation and sparse matrix techniques. Further considerations including column & rows ordering, block matrix, and fixed factor table are also explained in order to further improve the performance of this engine especially on large-scale systems. Performance testing has been made on a desktop computer using multiple power systems of different sizes, and the feasibility and efficiency of this engine has been verified.