This article proposes a passive and an online estimator of the dc bus parameters (series resistance $R_s$, inductance $L_s$, and capacitance $C$) used in doubly fed induction machines for wind applications. This kind of machines is widely adopted as generators for wind applications, where predictive maintenance is an important cost-reduction driver. Two implementations are described, plug-in or parallel, depending on whether the algorithm runs in the application digital control platform or as a standalone diagnostic device. Neither requires maintenance shutdown or dc current measurements to estimate bus parameters. In the plug-in operation, the proposed method relies only on the electrical quantities required by the control system, i.e., stator-side and rotor-side voltages and currents. Conversely, in parallel mode, dedicated external voltage and current sensors are used. In both cases, the series $RLC$ estimation is based on a recursive least squares algorithm that works for the full speed range of regular operation required in wind energy applications. In this way, the replacement of deteriorated capacitors can be performed by monitoring the estimated parameters. The validity of the proposal is discussed based on a theoretical analysis, simulation results, and experimental validations.