IM3 with imbalanced sideband

To test the inter-modulation distortion of a PA, a two-tone test is done. The two input tones (f1, f2) have same power. Because of the nonlinearity of the PA side bands (IM3) are generated, 2f1-f2 and 2f2-f1. But the two side bands have different power level and the left one is larger than the right one. What causes this problem?

Regards,
rfcn

The cause is "memory effect", and/or poor AM-to-PM response of the amplifier.
Usually careful design of the bias decoupling network at baseband/video frequencies (or f2-f1) solve somehow this issue.
Generally if the sidebands even they are asymmetrical, if they doesn't overtake the spec limits, they are not harmful for the amplifier.

Thank vfone!

My WiFi can't pass spectrum mask specification because the one sideband is higher than the other one. I think this is caused by the "memory effect". Wish your method can help me. But I can't modify the bias decoupling in the IC. What I can do is to add more decoupling capacitor to the power supply. Wish it works. Thanks!

Regards,
rfcn

What happens if you change the distance between f1 and f2? Will this change the level of 2f1-f2 and 2f2-f1?

How to measure the IP3 if the PA has memory effect?