This work has been supported in part by the National Natural Science Foundation of China (61671205). In this paper, the minimization of the average age of information (AoI) in status update systems is studied, in which the update packets are transmitted with fixed rate subject to maximum retransmission rounds. It is assumed that a new status update packet is generated only either after the previous packet has been successfully received or the maximum number of retransmission rounds of the old packet has been reached. The AoI minimization problem is then formulated as a constrained Markov decision process (CMDP) problem. By resorting to the Lagrangian approach, the CMDP problem is transferred to an equivalent unconstrained Markov decision process (MDP) problem. An algorithm is proposed to derive the optimal power allocation policy. Numerical results on the proposed optimal power control policy show that the optimal power control can reduce the average AoI significantly in the low-power regime compared to the constant power policy.