cst microwave integral solver problem with metalic loss

Hi,
I'm using cst microwave integral solver to study lens antennaes. When i tried to analyze a PEC feedhorn its all ok, but making it a lossy metal one makes Directivity < Gain :S and that's clearly wrong
for example taking the horn from the macros, MACRO>Construct>Online>Horn antenna and changing the PEC material as a lossy metal with 5E7 conductivity produce this...
any ideas are welcome :)! and sorry for my poor english xD

mm, it gives a radiated power > stimulated power, how thats even posible :S