Problems with geometry convergence in dihedral scan

I’m happy to help, but I can’t look in detail until tomorrow. It may help to clarify the terminology/keywords. A “FROZEN” coordinate remains unchanged. A “FIXED” coordinate is on that the user specifies (fixed by the user input) that may not be satisfied by the input geometry. If you are able to generate for your molecule reasonable guesses (e.g. in the case of a simple dihedral scan) then FROZEN is probably what you want. The capability of fixing a coordinate is primitive (the code as I recall imposes an increasing harmonic potential) but may allow you to push a system toward a more complicated endpoint.