The aim of this work is to create an iterative precoding algorithm with a higher convergence rate compared to known algorithms. This paper considers the task of antenna system pattern formation using linear precoding for massive MIMO systems, including precoders with approximate matrix inversion methods, such as the use of the Newton Iteration Algorithm. A variant of the Modified Newton Iteration Algorithm with a variable adaptation step is also proposed. A mathematical description of signal processing in the downlink model for massive MIMO systems is given, as well as linear precoding procedures. The convergence of the algorithm has been analyzed from the initial parameters determining the adaptation step, from the number of user equipment served, as well as the angle of user equipment position.