OK, I gave up and wrote my own UDF. It ended up having exactly the same problem as the one in FreeUDFLib, namely that if the first character, which gets copied directly into the result string, is accented, you get "Malformed String" when it is returned.

This is despite the fact that in my first draft for testing, all the code did was copy the first 5 char from the input to the result string, and that worked with no errors, with the same input with an accented first char! So I don't understand why the real code, which copies in the first char then calculates 3 more digits based on further consonants in the input, causes that error.

Anyways, I gave up and changed the code to only copy in the first char if it's between ASCII 33 and 127, and now it doesn't crash, and I think will work quite well enough for my purposes.

If anyone wants a copy of the code, email me.