lna+mixer noise figure problem

[rg3812]

I have designed a LNA block with high gain and low noise figure and now i want to co-simulate it with an ideal mixer (HB simulation in ADS). LO frequency initially is higher than RF frequency. The result show that chain NF is much higher than expectation from friis Equations at the end of the frequency band. when i change LO frequency to a value that is less than RF frequency, the NF is fixed and i cant find any reason for the problem. appreciate any help that anyone could offer. note that image frequency is completely rejected.

 

[G4BCH]

I am not familiar with ADS, but I would first of all try simulating just the mixer on its own. Is your simulated mixer only looking at conversion with low side LO?

 

[BigBoss]

Which type of Mixer you have used ? If it's ideal, check RF-LO isolation out first. It may disturb the LNA.
Can you share some system block scheme here ?

 

[vfone]

In the RF chain you cannot change only the LO frequency and see the NF change. Have to change the image filter also.

If LO>RF, always the image frequency will be higher than RF (image=RF+2*IF).
If RF>LO, always the image frequency will be lower than RF (image=RF-2*IF).

So, if use the same image filter (between RF and Mixer) and just move the LO frequency, you will get different image noise characteristic.

 

[rg3812]

I am not familiar with ADS, but I would first of all try simulating just the mixer on its own. Is your simulated mixer only looking at conversion with low side LO?

I am not familiar with ADS, but I would first of all try simulating just the mixer on its own. Is your simulated mixer only looking at conversion with low side LO?

yes. when I simulate it alone or with an ideal amplifier, the result is as you said with no problem.

--- Updated Jun 28, 2022 ---

In the RF chain you cannot change only the LO frequency and see the NF change. Have to change the image filter also.

If LO>RF, always the image frequency will be higher than RF (image=RF+2*IF).
If RF>LO, always the image frequency will be lower than RF (image=RF-2*IF).

So, if use the same image filter (between RF and Mixer) and just move the LO frequency, you will get different image noise characteristic.

you are right. when i change the LO frequency, I also change the ideal mixer properties. when LO>RF, the mixer mode is on "upper image rejection" and when I change LO frequency so that LO<RF, the mixer mode is also changed to "lower image rejection"

--- Updated Jun 28, 2022 ---

Which type of Mixer you have used ? If it's ideal, check RF-LO isolation out first. It may disturb the LNA.
Can you share some system block scheme here ?

I have used "mixer2" from "system-Amps and Mixers" palette. The isolations are ideal. actually i changed my simulation setup a lot in order to find the problem but here is the last setup that i used. And after simulation, I plot "nf(2)"