I'm somewhat partial to using PCF (via isometric generation) over directly generating CAESAR II files, as the iso drawings are a useful immediate "sanity check" on what the configuration should be.

Aside from connectivity, have to watch pipers getting a "case of the clevers" i.e. trying to work around the program without considering the data implications. Things putting multiple intelligent components into the same group, faking in components (once had someone try to pass off two overlapping reducing tees as a basket filter...).