This paper proposes a modified dynamic minimization algorithm for parameter estimation of chaotic systems, based on a scalar time series. Comparing with the previous design proposed by Maybhate and Amritkar (Phys. Rev. E 59:284-293, 1999), two important new design concepts related to the feedback control and the auxiliary functions for parametric updating laws are introduced. Two different types of estimates can then be derived, and numerical simulations confirm their superior performances to the designs based on the original dynamic minimization algorithm or other existing approaches. Furthermore, a circuit experiment is carried out to demonstrate the robustness and practicability of the proposed design.