iip3 measurement

I designed a differential LNA with the output resistance about 400 ohm, When I measure the IIP3, I used a balun (1:4) to transfer the differential signal to single-ended, then connectted to the 50 ohm equipment. However, because the output resistance does not matching, the measured IIP3 extrapolation curve shows a gain less than I designed.

I think that because the 1st and 3rd harmonic are both attenuated at the output, so the IIP3 measured is the actual value of the LNA. And the simulated results are the same before the balun and after it.

But how could many papers give the IIP3 measurement results figure with the correct gain? Did they used an off-chip matching network at output? Or just calibrated the measurement result according to simulation?
Thanks a lot!

somebody?

the iips result is the measurement result in common

The problem is in balun. It can be used only when it is matched. Simulation suggested ideal balun, but you have strongly mismatched one. Even slightly mismatched balun degrades significantly.

Best regards

Thanks for your replies.
I know the measured iip3 is the correct result, and the gain attenuated by the mismatch of the output balun. The problem is that how can I get a perfect iip3 figure with the correct power gain? Or I just need to explain the figure?

First: it seems that the real load of your LNA is 400Ohms // 200 Ohms, so you are measuring a lower power gain also for this. Please confirm.
Second: are you using two tones at input or only one tone? One tone measurement is always not desiderable.

The way to obtain the correct IP3 is to put between the LNA output and the balun a 400->200 Ohm lossless differential matching network. Or design a test buffer with lower output impedance and higher OIP3 of LNA.

I hope it can help.
Mazz

Your 50ohm terminated 4:1 balun will present differential 200ohms to the amplifier and you designed it for 400ohms so the the load line presented to the last stage will not be what you want it to be. This is where the gain compression and nonlinearity usually happens, at the largest signal point, the output. Also since the IP3i is IP3o-G, the more gain you have, the lower the IP3i for the same IP3o. How does the simulated IP3o compare with measured IP3o?

There is test equipment which provides 2 channels to get differential measurements with 50 ohm coaxial unbalanced probes, or 4-port measurements. Maybe this is how they measured the performance of differential amplifiers without baluns in your references.