This paper shows that a well-known causal PD controller plus feedforward solves the global output feedback tracking control problem of robot manipulators, by requiring only the existence of the robot natural damping, no matter how small. To this end, we first demonstrate that a robot controlled by a causal PD is globally input-to-state stable (ISS) with respect to a bounded input disturbance. Then, we prove that the addition of a feedforward compensation renders the error system uniformly globally asymptotically stable. Furthermore, we present a possible extension to more general nonlinear systems and also to uncertain systems.