The paper presents the development and implementation of a new sensorless algorithm for estimating the rotor position of switched reluctance motor (SRM) over a wide range of operating conditions. The algorithm is designed to overcome difficulties in estimating the rotor position at starting as well as at the normal operation region. The algorithm is successfully implemented experimentally using SRM fed from a photovoltaic source which has a nonlinear operating nature.  The results show improved performance at starting as well as when step changes in converter duty ratio or insulation levels occur. The results demonstrate clearly that the new algorithm ensures smooth starting without initial hesitation and reduces the torque ripples. The results presented are of prime importance to manufacturing companies leading to low cost reliable SRM drive systems for various industrial applications.