Does u0 and vth0 viewed in simulation result reflect default constant value or the ones already considering any Layout-dependent effects in such environment: tsmc65nm, BISM4.5 model, Spectre7.2 ?

No, model parameters are model parameters; they are not affected by the instance's conditions.

Andrew.

Hi Andrew,

If so, the model parameters keep unchanged. But those effects has to be reflected in certain way in real simulation. Which part has been changed under those effects from simulation perspective?

Is my following understanding correct? All those layout dependent issues are considered in such way that they are reflected in electrical or performance-related indicators or parameters.  For example, the mobility affected by  STI are considered in calculation of Ids. The Ids in older technology which does not consider any STI is different from the Ids in modern advanced technology whcih does consider STI effect. This impact is physically reflected in the calculation of Ids in modern simulation. While all model parameters still keep the same.

From simulation side, are there more ways to see or compare STI impact, any indicators?

Regards,

Alex. 

Hi Berndfi,

 I have reviewed the pdf file. From what I learned from this file, it seems the Cadence Assura activate the calculation of STI indicator, i.e., the result of SA,SB/SAref,SBref. However, in Virtuoso I can still measure that dimension. The STI impact is like this:

Layout Geometry => SA/SB (reflective) => mobility/velocity/Vth (physical/electrical root reason) => Performance (gain/bandwidth/phase margin) observed after simulation. The only invisiable part seems to be the root reason part. Model parameters are there without any consideration of STI. But effecitive mobility/velocity/Vth does consider STI effect because this is recorded in manual or the pdf file you provide. Are those effecitive mobility/velocity/Vth  observerable or user-editable in Cadence environment or any other ADE tools?

Regards,

Alex 

Alex,

The device model equations manual describes how the equations work (although it has a few omissions here and there, and sometimes you need to look at the Berkeley site to get the full model equations).

As I said before, you wouldn't expect to see effective information in the model parameters, because these are the input model parameters. You can however get a number of useful effective parameters for each instance by doing a dc operating point and then you'll get the "Operating Point Parameters". See "spectre -h bsim4" and you'll see there's a large number of parameters, including id, ids, vth, betaeff, gm. gds and many capacitances. I don't believe there's any output of effective mobility though.

You should see the effect of STI in the performance/behaviour of the transistors - for example, the id variation in a current mirror caused by STI (as in Bernd's presentation).

(BTW, I'm not sure why the email subject headers are ending up as gobbledegook - I'm guessing it's something to do with either the length of the subject line or the characters in it).

Regards,

Andrew.