This paper deals with an analysis and design of Dual Extended Kalman Filters (DKFs) to estimate parameters and state variables in Permanent Magnet Synchronous Machines (PMSMs) to be utilized in a control structure. A dual estimation problem consists of a simultaneous estimation of states of the dynamical system and its parameters using only noisy output observations. In this paper, the limit of an Augmented and Extended Kalman Filter (AEKF) obtained through standard state augmentation to estimate parameters is shown and, alternatively, a DKF approach which is characterized by the use of the state model descriptions in the output of an AEKF is proposed. The two different approaches are analyzed and compared. These results are supported by simulations.