How to measure noise of an amplifier?

A noisy network model of an amplifier is below
http://imgur.com/gwSqf
I want to know how measure In En and C which is correlation between In and En?
Thanks.

I don't see "C" in the diagram. en and in are assumed uncorrelated in usual noise models. Generally speaking, you can measure both quantities by operating the amplifier in a high gain feedback circuit with high and low source resistance.

P.S.: I thought about the correlation point. As said, the usual noise model assumes unrelated noise sources. If we consider an additional correlation coefficient, which will be most likely small, would it change much to the overall noise? Technically, BJT Ib and Ic shot noise are independent noise sources that can be related to in and en in a rough estimation.

If you treat the amplifier as a black box with unkown correlation of in and en, you should be able to extract en, in and C from three measurements with different source impedance. If you get a reliable C value is a different question.

Hello

the picture you upload is not valid
If you would know how to simulate the noise I can help you

http://imgur.com/sJWhV,nXgwu,fEltE,RcbUs,hYuur,t2xvw#0
I have read a textbook above about noise measurement.
It depict that we measure Vso(the output voltage of the amplifier) with Rs=0 and get En, and with large Rs get In.
and then C with Rs=Ro .
But I don't understand why and how we get C with Rs =Ro.May anyone explain it?
I think it is important to know what value C is to design optimal Rs for the min noise figure .

You'll get it by applying statistical caculation rules, if your application actually requires to determine a correlation coefficient. I never came across such an application, neither I'm aware of a definition of "C" in electronics text books. I understand it's meaning similar to the concordance correlation coefficient discussed here http://en.wikipedia.org/wiki/Concord...on_coefficient.

What's your intended usage of the correlation coefficient? It's not present in the standard transistor models used by circuit simulators. I already explained why I expect a low value of correlation coefficient for BJT.

Because in many textbook,the optimum source impedance is caculated as following

http://www.maximintegrated.com/app-n...ex.mvp/id/3169

1.bmp

for the lowest noise factor(this is also depicted in appendix L in Guillermo Gonzalez's "Microwave Transistor Amplifiers: Analysis and Design " ),so I want to know how to find it (by measurement ,caculation or some method) or if we don't know Yc ,how do we design an optimum source impedance of a low noise amplifier for the lowest noise factor?

I believe, considering correlation of noise sources might be more important in microwave amplifier design than it is for low and medium frequency OP circuit design. The suggested measurement method in post #4 will hardly work for microwave, you'll rather use different reactant source impedances in this case to extract the noise quantities.

The Maxim application node is presesenting a consistent noise model and calculation method. So if you need to consider correlation, you can base your calculation on this model. Maybe others can hint to references in literature, otherwise you need to figure it out yourself.