Harmonic cancellation strategies have been recently presented as a promising solution for the efficient on-chip implementation of accurate sinusoidal signal generators. Classical harmonic cancellation techniques consist in combining a set of time-shifted and scaled versions of a periodical signal in such a way that some of the harmonic components of the resulting signal are cancelled. This signal manipulation strategy can be easily implemented using digital resources to provide a set of phase-shifted digital square-wave signals and a summing network for scaling and combining the phase-shifted square-waves. A critical aspect in the practical implementation of the harmonic cancellation technique is the stringent accuracy required for the scaling weight ratios between the different phase-shifted signals. Small variations between these weights due to mismatch and process variations will reduce the effectiveness of the technique and increase the magnitude of undesired harmonic components. In this work, different harmonic cancellation strategies are presented and analyzed with the goal of simplifying the practical on-chip implementation of the scaling weights. Statistical behavioral simulations are provided in order to demonstrate the feasibility of the proposed approach.