In the past years the observational evidence that s-process elements from Sr to Pb are produced by stars ascending the socalled Asymptotic Giant Branch (or AGB) could not be explained by self-consistent models, forcing researchers to extensive parameterizations. The crucial point is to understand how protons can be injected from the envelope into the He-rich layers, yielding the formation of 13C and then the activation of the 13C(a,n)16O reaction. Only recently, attempts to solve this problem started to consider quantitatively physically-based mixing mechanisms. Among them, MHD processes in the plasma were suggested to yield mass transport through magnetic buoyancy. In this framework, we compare results of nucleosynthesis models for Low Mass AGB Stars (M<=3Mo), developed from the MHD scenario, with the record of isotopic abundance ratios of s-elements in presolar SiC grains, which were shown to offer precise constraints on the 13C reservoir. We find that n-captures driven by magnetically-induced mixing can indeed account for the SiC data quite well and that this is due to the fact that our 13C distribution fulfils the above constraints rather accurately. We suggest that similar tests should be now performed using different physical models for mixing. Such comparisons would indeed improve decisively our understanding of the formation of the neutron source.