Thank you, Aaron.
Quote:I'm not sure if you figured it out, but I'm sure there's something wrong in your simulation setup.
I don't know what is wrong with it. I tried various times and all of them give the same result.
Quote:In terms of whether or not Lg and Ls have an effect on NFmin, first you have to understand what NFmin is. Essentially we are finding the NF for every possible reactive network (call it network A) in between the source and the input, and calculating NF. Then we are selecting the minimum of those values and calling it NFmin.
Yes, I think I have understood it correctly.
NFmin the the NF of the circuit with network A component's values appropriate for noise matching.
Quote:For Lg, its quite obvious then why it has no effect on NF. Since it is in between the source and the input, it can be considered part of network A, and only adds a series offset to the impedance.
That makes sense. I will accept it for now. As for my understanding, input referred noises will change as network A component's values (let's assume that they are ideal L, C). So, it seems to me that NFmin also changes correspondingly. However, I know that is wrong.
Quote:For Ls, I think its not as intuitive. You could do the math to prove it. Alternatively, if you think of inductive source degeneration as a form of negative feedback, you could follow Razavi's reasoning in his Analog IC book and say that the input referred noise is not affected by the feedback as long as the feedback network itself doesn't add noise (in this case it doesn't). Maybe somebody else has a more intuitive way of looking at it.
I took a lot of time to prove it but no success. I like the idea you said in Razavi book. However, I just look it up in the book "Design of Analog CMOS Integrated Circuit" and can't find it.
Could you tell me the name of the book and which page/section/chapter in which that part is mentioned.