Unicode characters showing up as Hash marks

We have a BI Netweaver 7.0 sp 15 installation. As we extract FI information from our R3 installation (6.40 Sp14 ) we have noticed that our Language specific text character are loading as hash marks. Both systems, BI and R3 are Unicode and the RFC's are Unicode enabled. The text display fine on the R3 side but somewhere in the extraction process it changes to the hash mark character. Have any of you seen this issue and what did you do to solve it?