cps port in hfss
Placing ports on CPW/CPS transmission lines can be a trick, particularly if you are using differential lines. You now get into an issue of whether they are tightly or loosely coupled as well.
Read the notes on impedance calculations (Zpi, Zpv, etc.) in the HFSS help sections. Those transmission lines are "mode rich". I would examine the fields and modes at each port to make sure you have the excitation you expect and you are using the appropriate impedance calculation. Be sure to ask for multiple modes when you start the simulation and then back off when (if) you get the mode you want as the first or second mode.
Simon and Wadell have helpful books on the subject and if you can get to an impedance tool like Polar Instruments as a rough check on things I would recommend that too. It is in no way a substitute for HFSS but it may catch a mistake if things go wrong.
I am currently messing with a similar problem using HFSS 11 (beta) and am not convinced that I have it in hand. Apparently there has been significant work in this area and it is easier to manage than in HFSS 10.3. Maybe I will get a little quiet time to work with this soon.