Hi guys,

I've gotten something unsure about noise factor. Hope for help here.

The Noise Factor is defined as below.
N = (Na+G*Ns)/(G*Ns).
Different Ns produces different N. And in Cadence spectre, the noise source is usually a port. The impedance of the port provides the thermal noise 4kTRB V^2, which is related to the value of R.
Q1: What value of R should be set?
Q2: If R=50 Ohm, why?
Q3: If R=Rout of the preceding device, then noise factor of the DUV(Device Under Test) is dependent on different source impedances. So noise factor is not a general performance factor of DUV for comparison, isn't it?

Anyone can help me?
Thanks in advance.

With different source resistance u get different NF since the gain from the voltage source to the output changes.

For NF simulation the impedance should be set to 50 Ohms since this is a standard in the industry and when u measure something, you will use 50-Ohms measurement equipment. If you know NF of a device u can estimate the NF with any source impedance so do always a NF simulation with a 50 Ohm port and you will never confuse yourself or somebody else in the future.

If you calculate not a 50-Ohm system, it is better to calculate voltage and current noise at the output and use SNR term because this is what is needed at the end of a day.