This thread basically sums up my frustration with the last 1/3 of my project -- some of the substrates are so highly crystalline that dry loading leads to a complete disaster, yet for some of them the separation is so difficult that if I wet load with DCM (even if just a small amount) onto a column packed in 100% hexanes, I kill any hope of getting clean fractions.
Eventually for the really difficult cases I resorted to running 2 columns. First one I wet load in DCM and get the bulk of the crap out, then dry loading and running a second column with a hexane/DCM system. The one good thing about this system is that because DCM is such a good solvent, stuff crystallizing on top of the column no longer becomes an issue.
In a few cases I found that hex/DCM actually gave a far superior separation compared to hex/EtOAc. One thing about a hex/DCM system though is that the trailing problem can become really annoying. Had a couple cases where Rf was about 0.5 in the system I was running it in, yet it took forever for everything to come off. After running the initial TLC plate a couple more times, I found that while the product moves fast initially, it starts tailing off, and it took about 6-7 runs before the product hit an Rf of 0.8.