We construct bounded globally asymptotically stabilizing output feedbacks for a family of nonlinear systems, using a dynamic extension and a converging-input-converging-state assumption. We provide sufficient conditions for this assumption to hold, in terms of Lyapunov functions. The novelty is that our construction provides formulas for the control bounds while allowing uncertainties that prevent the use of classical back-stepping, and cases where only part of the state variable is available for measurement, without requiring the time lagged states in the feedback control that were required in the artificial delays approach. We illustrate the relevance of our work to engineering in an application to a single-link direct-drive manipulator.