We worked on a new scheme of Quasi-Phase-Matching (QPM) based on the negative first-order of the spatial modulation of the sign of the second-order nonlinearity. Applying this scheme in the case of Angular-Quasi-Phase-Matching (AQPM) in a biaxial crystal reveals new directions of propagation for efficient parametric frequency conversion as well as "giant" spectral acceptances. The experimental validation is performed in a periodically-poled Rubidium-doped KTiOPO4 biaxial crystal. This new approach naturally extends to other periodically-poled uniaxial crystals such as periodically-poled LiNbO3. Phase-matching is the privileged way to optimize the performance of nonlinear optical frequency conversion. Usually, it can be obtained by using the birefringence of anisotropic crystals, which corresponds to birefringence phase-matching (BPM) [1,2]. However, BPM conditions are strongly limited by the refractive indices of the material and they usually forbid the exploitation of the highest coefficient of the second-order electric susceptibility tensor. To overcome this limitation, quasi-phase matching (QPM) is a powerful alternative. It is based on a periodical modulation of the sign of the second-order optical nonlinear coefficient of anisotropic or isotropic non-centrosymmetric crystals [3,4]. We propose here the development and the experimental validation of a new scheme of QPM that allows us to enlarge the spectral range of phase-matched generation and to get access to giant spectral acceptances. For that purpose, we studied Second-Harmonic Generation (SHG) in the ferroelectric periodically-poled Rubidium-doped KTiOPO4 crystal (PPRKTP) [5]. The key parameter of QPM is the periodicity Λ of the spatial modulation of the second-order nonlinearity along the direction of propagation of three interacting waves such as their wavelengths fulfil: 1/λ ! − 1/λ " − 1/λ # = 0. Actually, the goal is to compensate the #% ' , is the grating vector oriented along the propagation direction, and m is the order of the Fourier series. Usually, m is chosen as a positive integer (m ≥ 1), as it is the case for instance while targeting the largest nonlinear coefficient χ !! (#). It is shown in Fig. 1 (left) in the optimal case, i.e. for the lowest QPM positive order m = 1. This is the classical scheme of QPM that we called QPM-A. However, it is also possible to consider propagation directions where m ≤-1. This configuration has been previously theoretically considered, but without any phenomenological interpretation or experimental verification [6]. We called here this new scheme QPM-B. As for the case of positive orders, the lowest QPM-B order (namely m =-1) shown in Fig. 1 (right) leads to the best efficiency compared to higher< Réduire