This paper reports on the conception and digital signal processor (DSP)-based hardware-in-the-loop realisation of a nonlinear observer for a grid-connected variable speed wind turbine using a doubly fed induction generator (DFIG). The objective of this work is to build some observation structure based on the extended Kalman filter (EKF), the sliding mode observer (SMO) and the adaptive model reference adaptive system (MRAS) observer to observe certain quantities of the wind power system connected to the electrical grid in order to reduce the complexity and the cost of the hardware, to achieve an increased mechanical robustness, to operate in hostile environments, to have a higher reliability and to ensure an unchanged inertia of the system. The proposed observers will be combined with robust nonlinear controls based on the backstepping approach to control the wind energy conversion system (WECS). The whole will be implemented on a TMS320F28335 DSP board. A comparative analysis of the proposed observers will be carried out. Through the results of the DSP implementation in hardware in the loop (HIL), we will prove that this improved combination increases the desired performance.