Most variable step-size algorithms for the photovoltaic（PV）generation systems use the derivative values of power versus voltage P′（u）as the basic step-size references.However，due to the fact that the mathematical properties of P′（u）are variable on different sides of the maximum power point（MPP），some essential parameters，such as the maximal step-size and scaling factor，are difficult to be estimated accurately.In order to solve this problem，a variable step-size MPP tracking（MPPT）algorithm based on the contingence angle of the P-U curve is proposed.First，the arctangent of P′（u）is calculated to uniformize the tracking characteristic of the algorithm on both sides of the MPP.Afterwards，cosine function is adopted to normalize the arctangent，simplifying the design and debug of the maximum step-size.On this basis，a new scaling factor is further proposed.The convergence of the algorithm proposed is proved using the D′Alembert criterion and the theoretical convergence scale of the maximum step-size is calculated.The rate of step-size change is a monotone decreasing function of the distance to the MPPT，which is different from the traditional algorithm，making the convergence faster.Simulation and experiment results show that the algorithm proposed has faster convergence speed and higher tracking efficiency.