double balanced mixer design (problem on inut matching work)

[xiaosongshu]

double balanced mixer

Hello , everyone.
I've designed a mixer based on Gilbert cell, and plan to realize this mixer and test this mixer alone. But I am confused on the input matching work of the mixer now.
The RF signal is connected to the gate of the MOSFET whose input impedance is infinite, so I think I can parallel a 50Ω resistor so that input matching is OK.
However, I think this does work for conversion gain test, but when it comes to noise figure test which seems to be based on power calculation(?), formentioned matching method does not be effective any longer.
Is there any suggestions?
Thank you ~

 

[E-design]

double balanced mixer design

You are right, the 50 Ohm on the input of a MOS device will add at least 3dB to the noise figure. Simple LC match network won't do much better because the real part of the MOS input impedance is too small to be effectively matched to 50 Ohm without ruining the noise figure. Inductive degeneration can help to increase this real part so that you can use a L matching network. In packaged chips these bond wire inductors (high Q) in series with source and gate comes free and is commonly used to overcome this problem. You can also add small external inductors if the wire inductances are too small.