In some control applications, where the reference is sinusoidal, phasor-based strategies can be considered. The main advantage of these approaches is that instead of dealing with oscillating signals, one deals with constant or low-frequency signals. However, to implement this solution, several nonlinearities are included into the control loop. Then, linearization are performed around operating points to design a controller, but with no guaranties of performance. In this work, we present a "large-signal" model for the system. Then, considering ideal measurement of instantaneous amplitude and phase, we establish the correspondence between classical control architecture and a phasor-based one. Hence, we show that a controller can be designed as in a classical architecture and implemented in a phasor-based one with a simple transformation, ensuring the same performance level. However, in practical applications, the measurements of amplitude and phase are not ideal and may quickly degrade the behavior of the system. Thus, we propose a second design approach that takes these nonlinearities into account, ensuring nominal stability and performance. Examples illustrate the effectiveness of these methods.