Extraction of key parameters for analog design

[leohart]

find key parameters in designing

Hi guys, what would you do if you are facing a new analog process before doing
you design?

Where do the key parameters come from?like k',Vt,lambda,gamma...etc.

in Allen, he said you should do extraction,so did you do it youself?

I only have a spice model file on hand, and the process document dont have all the parameters on it

 

[drabos]

you can find some of the parameters on your process document, you can draw the transistor characteristics, from that you can determine approximate values for all the parameters, it is possible to do and these models in the books are usually Level 1 models, nobody provides level 1 models to the tehcnology, it doesn't make sense. Anyway you will refine your circuit with simulator, so I would say usually k' is enough for the first roughly calculations and after you can simulate. Matching rules atc. are much more mportant... and nowadays you can face with many problems, when sometimes the simulator is not accurate enough (soe models in the edge of subthreshold...), to my mind nowadays you need more intuition, than paper calculations and after you can refine your idea with the simulator...

 

[layoutmaster]

1 - Extracted parameters might not be real, they depend on how accurate is the extraction deck data reflecting the actual process data...

2 - Main source for getting that parameters must be a process manual, or whatever the EDA team who defined that new process gives you for designing since, their data comes (at least should be comming) from measurements made from some silicon runs they do before any process release to design teams. What they use do to is to make a series of test chips within the new process, by placing all the available components in all their available parameters changes in order to collect data for future designs.

 

[leohart]

yes,better with special purpose test chip data,but what I can do now is virtual data from spice, at least it means something when doing design and verify with simulator,right?

In fact the Id,gm,gmbs,vth is first order right with hand caculate with extracted parameter, and if the hand caculation is failed, I can get these from ADE/print DC operating point...But the rout of the mosfet is not given in DC operating point,and nor it is easliy extracted...the early voltage varies with different vdsat even with a constant L,this is totally high order effect... Is there anyway I get rout from simulator?hand calculation doesnt work well for this parameter...

And I also found the vdsat given in DC operating point is kind of smaller than hand calculation,and from curve I simulated,seems the hand calculation is more precise...