I've been seeing similar problems lately. Here's what I (think) I know now.
BTW, I'm running S3D on a Windows 7 Pro, 64-bit PC.
If I look at all of my exported (saved to local disk) .fff files, I find that if they aren't zero-length, they have one of three lengths:
-- 8K - 9K (these turn out to have been created in S3D 3.0).
-- 21K (these turn out to have been created in S3D 3.0.1).
-- 22K (these turn out to have been created in S3D 3.0.2).
Those lengths also apply to .fff files that I'd derived from .fff files of those lengths, also.
Only the 22K-length files represent profiles that are stable when editing their contents (they don't crash S3D).
When I make a copy of a 22K-length profile by using "Save As New" in the profile-editing window, I can then change the copy all I want with no crashes.
BUT, I've found that if I've run a "crasher" .fff file with S3D, to get things back to normal I have to shut down S3D, bring it back up, immediately load a "good" profile, then I usually have to shut down S3D again. When I restart it, it automatically restarts with a "good" profile as the current one, and it's in a good state.
After that I can copy the profile to a new name using "Save As New" as mentioned above, and it always seems to be a clean file that I can edit, save, export, etc. And if I do export it, it'll be a proper 22K-length file.
So it appears to me that ever using any profiles created with earlier versions of S3D, or copying and modifying them, generates .fff files that will -- sooner or later -- crash S3D. But following the process above, in my limited testing (I just figured this all out this evening) SEEMS so far to give stable operation again.
I'd be interested to see if this is your experience, also, if you decide to try it.