Cowell,

I would guess the problem is related to the parameters from BSIM (is this the model you use?). The model parameters are not the exact values you need for your simulation. BSIM has a lot of first and second order coefficients e.g. for influence of length, gate voltage and so on. If you look at the device parameters output by Spectre during an OP simulation, you will notice they are different from the ones written in the model files. Using these values, you should already get closer to the real current.

Typically, you need to somehow "extract" the values needed for a first order calculation (that's what you do by hand). You can get these either using the parameters from a compact model or by extracting them from your simulation results.

Paul

Hi, Cowell,

You should not expect to get the precise value of (u*Cox) from BSIM model directly.  Actually, it varies much beacause the real MOS transistor's behaviour is so complex. Many parameters varies due to temperature, voltage, process, width, length, etc. So it is impossible to get a constant, precise value for all transistors.

BSIM model is a empirical model. It has many parameters that havn't any physical explanation. So you need extract it using SPICE or Spectre simulation. Or you can calculate Id, Rout using the Gm, Gds, and so on listed in the report of an OP simulatioin. But Gm, Gds, etc must differ for transistor with different width, length, or temperature, etc.

Best regards,

Yawei