Inverse folding (IF) is a classic instance of negative RNA design, and consists in finding a sequence that uniquely folds into a target secondary structure with respect to free-energy minimization. Even when base pairs uniformly contribute -1 to the energy, and energy minimization coincides with maximizing the number of base pairs, the (exponential) numbers of potential alternatives over a candidate sequence makes IF a challenging problem to tackle. Accordingly, IF is generally believed to be NP-hard, as formally demonstrated under mild constraints by Bonnet et al. [1]. In recent work [2 ], we show that inverse folding can be solved in linear time for a large collection of targets, including every structure that contains no isolated base pair and no isolated stack, i.e. such that hmin =3 with hmin that denotes the minimum number of base pairs in an helix. This constraint has practical relevance as RNAs are known to contain few helices of small sizes and, even though, such helices are not major contributors to the free-energy/stability. Our approach introduces a notion of modulo m-separability, generalizing a separability property introduced by Hales et al. [ 3 ]. (m-)Separability represents a sufficient condition for the existence of a solution to the inverse folding problem. This allows us to propose a dynamic programming algorithm that, for any input secondary structure of length n, returns a modulo m-separated sequence in time O(2mmn) anytime such a sequence exists. Using this algorithm, we show that every structure with hmin = 3 is either trivially non-designable (if it contains a forbidden motif) or always admits a modulo-2 separated solution (m = 2), which can be obtained in linear time. A trivial extension of our algorithm allows us to uniformly generate them within the set of modulo-2 separable sequences. Further experiments show that structures featuring shorter helices (hmin = 2) are often solved with our algorithm with m ”small” and that our sequences obtained for hmin = 3 represent better candidates than random compatible sequences in the more realistic Turner energy model.