Hi,
I dont think there is any general method for that. If you dont want to use small signal because of higher signal amplitudes then you will have to take into account the variation in dc operating point, you will perhaps have to repeat the small signal analysis at different dc points. Still I think it would be better if you could show the schematic or tell what circuit it is?
For W/L, you will may calculate them from small sig. assuming the Mosfet as resistance but I am not sure of this for large signal.
Hi,
Kp/Kn is different than W/L, and flicker noise will depend more on L than on W ( for same values). So if you are increasing the L, then it will reduce noise as stated. Nmos has more noise than pmos but apart from that I am not aware of effect of Kp/Kn on noise but if thepaper talks about increasing Kp keeping Kn constant then its becoz of higher W*L for Pmos. ( But will perhaps increase if you compare it with a reduced Kn value and constant Kp value)
Hi
I'm not sure about the working of this circuit. But I can tell that transistors M1, M2 M3 M4 have more chance to be in linear region since its gate is connected to the highest potential, VDD.
If Vp1,2,3,4 nodes are swinging the then M1-4 will go to linear region.