Brian your problem about subdivs is very common, thats because type of modeling. Let me explain. If you do your model in poly and use smooth its something different then if you use convert poly to subdivs. To model with polys and for final to convert it to subdiv is giving pretty uncertain and unpredictable results. Better is trying to model with subdivs in beginnig stage and convert back to level 0 (subdToPoly -ch off -aut on -format 3 -depth 0 -sampleCount 1 -maxPolys 1 -extractPointPosition 1 -shareUVs 1 -subdNormals 0 |polyToSubd1|polyToSubdShape1;
// Result: subdTessShape1 //
delete |polyToSubd1|polyToSubdShape1 and try if it works (you can use kind of different topologies and less faces - looking weird in polysmooth but ok in subdiv) and then convert it back to subdiv again. Of course its only my idea you can have another oppinions..they are welcome!