Take a look at http://www.designers-guide.org/Modeling/varactors.pdf. You can also find this described in http://www.designers-guide.org/Books/dg-spice/index.html.

-Ken

c100 - your explanation sounds fishy.  It's like saying one could use gm-based models instead of drain current-based, so you don't have to compute the derivatives of drain current.

In fact, the capacitance-based models have to compute the charge -- because it's the change in charge that gives you the current -- so they do this by numerical integration within the model, which is ugly since the model doesn't control the timestep (at least not directly) and can't guarantee accurate numerical integration.

In the beginning, people used capacitance-based models because they could measure the capacitance and construct equations to fit the measurements.  It was "hard" to analytically integrate the partial differential equations to get the charge: you have Cgs, Cgd, Cgb, which are supposed to be -dQg/dVs, -dQg/dVd, -dQg/dVb, but it might be that integrating Cgs with respect to Vs gives you a different function than integrating Cgd with respect to Vd.  Then what?

So, now, all the "modern" MOS models write the charge equations as the fundamental equations.