Weird gm/id vs vgs result in Cadence

[kidmanbasha]

gm / vgs

I am trying to simulate gm/id vs vgs in cadence but I get a strange transition when vgs approaches zero as shown below.

How I did the simulation : sweep vgs vs id then using the calculator i get the derivative of id and divide by id to get gm/id

 

[erikl]

id vs vgs

I am trying to simulate gm/id vs vgs in cadence but I get a strange transition when vgs approaches zero as shown below.

It's not weird, it's normal: in sub-threshold (= weak-inversion) you have maximum gm/id , s. e.g. E. Vittoz' weak-inversion paper below (p. 10).

 

[oermens]

vgs vs id

you can also use dc analysis and obtain 'gmoverid' parameter.

 

[kidmanbasha]

ids vs vgs cadence

erikl...I did as oermens said and the result is as expected without the the sudden peaking at vgs--->0.

 

[erikl]

id versus vgs

erikl...I did as oermens said and the result is as expected without the the sudden peaking at vgs--->0.

So the SPICE DC calculations provide a smoother transition from weak <-> strong inversion!?

 

[kidmanbasha]

yes. actually the peaking was in the inversion region only cuz the vto of my process (0.13u) is between 0.2 and 0.3