Hi everyone,
Recently, I am testing the IIP2 performance of my 12GHz down-conversion mixer.However, I have met some trouble about how to test IIP2 correctly. Here is my setting.
In this Gilbert down-conversion mixer test bench, LO port has a LO-frequency at 13GHz. Then, RF port has two frequencies as input. They are 12.3GHz RF1 and 12.9GHz RF2. I utilize "IPN curves" function in PSS analysis to test IIP2 performance of this circuit. At the output, I select |12.3GHz-12.9GHz|=600MHz as IM output, then |12.3GHz-13GHz|=700MHz as Linearity output. After using IPN curves analysis, IIP2 result can be seen from the figure directly. At the same time, we can see that IM curve fits well with 2dB/dB extension line in both I_path and Q_path from Prf=-40dBm to Prf =-20dBm. I think this result is convincing with such setting.
However, problems came when I have changed the 2 frequencies in RF port. Specifically, when RF1=12.2GHz and RF2=12.8GHz, IM frequency equals 600MHz, while linearity frequency equals 800MHz. Figure plotted is shown below, where IM curve starts to deviate from 2dB/dB curve.
What's more, when RF1=12.1GHz and RF2=12.9GHz, IM curve still can't fit 2dB/dB extension line.Figure of this is not shown below.
I guess some other high-order inter-modulation products which equal to this IM frequency come to the output. So, IM freq curve appears greater than 2dB/dB.But the voltage amplitude of these possible high-order inter-modulation must be too small to affect IM curve so much.
Hoping for your help^.^Thanks very much~

You should always use the small-signal analysis approach to computing IP₂ or IP₃ described in Accurate and rapid measurement of IP2 and IP3, it is more accurate than applying two large test tones.

-Ken