I'm working on a fairly complex part that 'wants' to be built in three stages, top, middle and base. The base (red) and the top (blue) are simple enough to model as solids. The 'middle' is a little more complex and I found it easier to model as surfaces. Its made up of a number of lofts, extrudes and trims, not super fancy stuff but it came together much quicker as a surface model than in previous attempts to do it all as solids. Finished up, knitted just fine and combined easily with the base...

But we when it came time to combine with the top (blue), everything blew up and I lost half the surfaces. The ones that remained (previously defined as a loft through two sketched) all converged on one point where the top sketch used to be.

Are there any best practices about working with parts created through the two modes of modeling (surface and solid)? I though I was home free when everything knitted up, although 0.05mm gap control was required. As a work-around, I have also tried rebuilding the geometry from the top (blue) portion but any feature that intersects or modifies the previously surface geometry causes it all to break/converge to a point. However, I can make extrudes, fillets etc on the base (red) at will.

Thoughts? I cannot post files, them's the rules like it or not... Suggestions are appreciated.