The existing passive synthetic array algorithms all make too ideal assumptions on the way of sensor moving and cannot endure the influence of phase noise very well. So the current research firstly presents a general processing algorithm of passive synthetic array. By carefully analyzing the phase noise model, a new estimation means of maximum coherence time and valid synthetic sensor number is proposed and then the steering vector of passive synthetic array is reconstructed. Furthermore, an improved passive synthetic array algorithm based on the phase noise model is presented. Considering the effect of phase noise, the theoretical lower bound on the variance of DOA estimate of passive synthetic array is given by deducing the Cramér–Rao Bound. Simulation results show that the proposed algorithm can enhance the precision of DOA estimate effectively comparing with the existing algorithm.