If the character itself is not displayed correctly then Calibre doesn't support the way the diacritics are defined. Messing with the fonts won't help.
The list of diacritic characters that Calibre has support for is here:
http://bazaar.launchpad.net/~kovid/c.../preprocess.py
Search for "# Fix Accents" - no quotes - to see the relevant part of the source. If the characters you're concerned about are already covered in that list then there isn't anything to be done, you just have a set of junk pdfs.
The only thing you could do is use the search and replace wizard to replace whatever garbage is being generated with the correct character, but this is dependent on how many you need to do it for.