Part of charge development for carbohydrate molecular dynamics simulations requires an ensemble charge set. This charge set is built from PDB files which were extracted at regular intervals from a preliminary MD simulation using initial charges from a single QM optimized structure. These PDB files are consistently built however, bond lengths and angles vary because they're oscillating around a equilibrium at 300 K. The atomic order in the babel z-matrix is consistent between each converted structure but the bond, angle and torsion definitions vary depending on how far one atom is from another. If possible, it would be beneficial to create a z-matrix from either a template setup where bonds, angles and torsions are defined consistently or to force multiple inputs of the same molecular structure to be built consistently. I have already tried varying the bond and angle limits for the atoms and this doesn't correct the issue.