Ricardo Diaz is correct the issue may come from using different data types.
It may also come from invisible characters (spaces, strange ascii,...).
You need to isolate the guilty row or value and white them in an editor showing all characters. I am using Notepad++ with the option show all characters (Show hexadecimal/ASCII codes on Notepad++), but many other editors are able to show the Hex-values.
Once you know what has to be changed, you may correct it, with a JavaScript for example, or in the SQL the a Table Input [This is one of the rare case where something else than a SELECT * in the SQL of the Table Input does not hurt my eyes]. Just read the data in table, apply your change and use the Stream Lookup or the Join Rows (Cartesian product) (with the join condition). Don't be afraid for performance, those steps optimized for large volume and are pretty fast, sometimes faster that the database lookup.