Hi Julia,
I am glad that both the systems are Unicode enabled. Can you also confirm if the communication type with target system is also set as Unicode. ( Please refer to the screen shot below ) This can be found in the RFC connection configuration (SM59).
The above setting in the character conversion also explains why you are getting # in case of error in character conversion.
If your characters are entered in single code page, system is Unicode enabled and you have communication type with the target system is Unicode I don't know any reason why it should not work.
Cheers
Ajay