Hi, the formulas of many transistor models can be often found in simulator reference material, e.g. for Cadence's Spectre there is a huge pdf on all models, including bsim. How different the freq response is depends on the model parameters. Using defaults like RS=0, CJ..=0 you almost switch back to simpler mos models and have little differences. I expect much more differences in subthreshold and output conductance, than in freq responses.

Bye Stephan

Dear  Kittipong Tripetch,

I believe you can find all the information you need about the BSIM4 model in the model user's guide from UC Berkeley.

http://www.eecs.berkeley.edu/Pubs/TechRpts/2000/ERL-00-48.pdf

Prof Shih Lu of National Taiwan University ever proposed his high frequency equivalent circuit in his journal paper. That circuit is exactly different from textbook but it has more node maybe 6 or 7 nodes but I think no one want to use circuit analysis with that circuit without SPICE based software if no one guarantee that the novel circuit is better than the conventional circuit compared with experimental measurement results. Please pointed out  if someone ever saw equivalent circuit in BSIM4 manual from UCB. I never saw it . Maybe I must draw parasitic capacitance from charge conservation model in that manual ?

BSIM4 simulation result and text book calculations match very well for simple circuits (2 stage OTA etc). You may expect these two to match within 20-30%. It is true if your circuit is dominated by lumped elements (e.g. big compensation or output caps) and not parasitic elements. So for high speed circuits the error could be more.

I am not aware of simple BSIM4 model as it is quite complex model with 100s of parameters.

You can also look at the C code in the latest NGSPICE release. I recently re-implemented BSIM4v5 by grabbing the open-source and hard-coding TSMC parameters. There are many AC mode equations, but they are simple to understand.