Noise Figure of Gilbert down-conversion mixer

[mhtcas]

i design a differential input balanced Gilbert down-conversion mixer.The current bleeding technique is adopted in the mixer.
The simulated SSB NF@1MHz=21.5dB and the input 1dB compression point=13.7dB, when the supply cirrent equals 1.8mA.
Is the NF specification a little bad?
how can i improve it?

 

[jiripolivka]

Hello there:

In any mixer, its noise figure includes its conversion loss and then the noise figure of the following device (IF amplifier).
Mixers including Gilbert cell are lossy devices. The noise figure is the essential parameter of a receiver, so to improve receiver noise figure, the best way is to use a low-noise RF preamplifier and a similar IF amplifier following the mixer.
Read a good textbook on receiver design, there are answers to your questions responded.

 

[mhtcas]

Hello there:

thank you for your remind, firstly
yes , there is a LNA preceeding the mixer. but isn't a NF of 21dB of down-conversion mixer too high for the receiver, especially when the gain of the LNA restricted to 15dB or less?

 

[jiripolivka]

Dear mhtcas:

Your mixer's noise figure depends on its conversion loss- you did not give any information on it! It does depend on the current as well as on input/output matching, and LO power input.
You should be able to optimize your mixer for the lowest conversion loss. Then, if you use a good IF amplifier, your receiver AFTER LNA may be better. Otherwise, your LNA gain of 15 dB is too low to give you a low-noise RECEIVER.

 

[snafflekid]

The input referred noise of the system will divide the NF of the mixer by the gain of the LNA.

Did you optimize the NF on the mixer transconductor?

The chopping mosfets produce noise during the non-zero power switching transitions. Maybe it is switching too slowly.