Does NFmin in cascode LNA depend on the negative feedback inductance Ls?
This is a cascode LNA. Please help me with the question in picture. Thank you.
As I told in many times here..
NFmin is a intrinsic parameter of the active device and it's absolutely independent from external elements.
External components are used to get this active devices NF closer to NFmin.
It's generally defined by Source Pull measurement technique and it's also given by manufacturer.
In RFIC design concept, nonlinear models have NFmin value but the best method is to measure.
Thanks. Is there a decent proof for this?
Read those pages "The Design of CMOS Radio-Frequency Integrated Circuits-Thomas H. LEE;Cambridge Press 2004;pp364-370"
The NFmin of the circuit will not change, but the Γopt will change. That, as far as I know, is the point of inductively degenerated LNAs.
Hi.
I just simulated optimum noise impedance Zopt = Ropt + jXopt as Vgs changes. And just realise that Zopt changes a lot as Vgs varies. However, this makes me really confused.
According to the simultaneous noise and input power matching procedure below, after steps 2 and 3, Zopt will be fixed.
However, in step 4, until at this time Vgs is chosen. And this will change Zopt.
It seems useless.
Please explain.
As BigBoss has (so clearly) indicated, NFmin is not something you can do anything about short of cooling the whole thing to the kind of temperatures where the noise of molecules colliding together becomes less, but I sense you are not entirely persuaded.
The trade-off when designing LNA is that the input impedance where the lowest noise NFmin is to be had is also at a place where the VSWR is poor, and a designer's matching attempts are constrained. The better the VSWR, the more the match point has to be placed further away from NFmin.
The pros and cons of which LNA circuit, and how to get it stable with a good noise figure is one of the most fraught and difficult things to attempt. Those who do it well are extremely skilled, have probably spent years learning how. The design process you showed lets you design one kind of input match. There are many other approaches. There is not one "right" input circuit. In all of them, you can only approach NFmin, and nearly every trick trades increased noise, stability, bandwidth, and input VSWR.