An Alta Vista search on "manchester encoding" reveals a simple explanation of two different types of encoding. It really doesn't have to be complicated if you are simply using it to DC-balance an RF data stream. If a bit to be transmitted is a 1, follow it with a zero, and vice versa. It doubles the amount of data sent, but it's very simple to implement. I am sending byte-width messages in my case, and just wrote two routines - one to convert an 8-bit value to a 16-bit encoded one, and one to convert a 16-bit encoded value to an 8-bit decoded one. If using the Keil compiler, I recommend using bdata to easily perform the bit operations and assignments.