I am attempting to use a mosfet as an RF varactor in an SOI process. The process I'm using has a few different modelling options for the standard nfet device.  There are BSIM and PSP models, and for BSIM the rgatemod switch can be either 1 or 3, but is set to 0 for extracted simulation. I'm concerned mostly about Q and capacitance range at the moment, and am observing the following:

Q@1GHz	Range	Model
8369	5.36	BSIM, rgatemod=1, schematic
36	5.36	BSIM, rgatemod=3, schematic
1025	4.39	BSIM, Calibre RC extracted
11900	5.79	PSP, schematic
1081	4.59	PSP, Calibre RC extracted

The post-extracted PSP and BSIM are not too far off, but I'm worried about the BSIM with rgatemod=3. This includes a bias-dependent intrinsic gate resistance which I'm not sure is being modeled in any of the other setups. Does anybody know if that is true or not? If so, is there any way to extract it with Calibre? Any help or comments is very much appreciated, thank you.

Aside from the Q=36 case, the Q values are much higher than I would expect to see at 1GHz.  You would be better off using a designated varactor model instead of trying to use a MOS transistor.  My understanding is the modeling and manufacturing steps are different if the device is designated as a transistor or varactor, so I would try to obtain a specific MOS varactor model from the foundry.  

Tim