suitable model for mixed-signal design

[arbalez]

i would like to know which model is suitable for mixed-signal design? BSIM, EKV, etc?
how does the CMOS modeling effects the overall design and its particular functionality?

thanks in advance.

 

[shpongle]

BSIM or EKV: the never ending story...
A first answer can be be: the better is the one provided by the fab... BSIM is the more widely available while EKV is the more close to the physics.
EKV has also a fair smaller number of parameters and hence it is easier to handle and to support your physical intuition. On the other hand, BSIM now include hundreds of parameters and you will (surely) get a lot of convergence problems...
So, as a conclusion, if you dispose of a set of EKV parameters try it !

Other models, like the Philips MM11 or SP are also physics based and they can model a lot of effects adequately.

It is tough to answer your question on how it will impact the whole mixed-signal design. Firstly, it depends on how much your design is dependent on second-order or third-order effects... You better first study your system to know which effect could have an impact (a big impact, I mean) and then look for a model which correctly implement these. Aside from this point, the model you will use will impact the simulation of your mixed-signal system. Mixing digital and analog may result in big convergence problems depending on the model and on the simulator you use.
Then, the use of behavioral models to describe the analog part while you are busy with the digital and the digital while you deal with the design of the analog portion may be a good alternative.

shpongle