I am attempting to de-embed an evaluation board (sound simple enough). The board is FR4 for a differential amplifier and I will need data from 1 MHz to 1 GHz. The differential amplifier is has a 200 ohm differential characteristic impedance (100 ohm single ended). For this I decided to use automatic fixture removal (using either the M9372A AFR option or PLTS) using an open circuit (I do not have a 2x through line). So I calibrated my ENA (E5071C) from 10 MHz to 20 GHz with 2000 steps (which allows the use of the low pass mode in the time domain). After applying the de-embedded file to the measure data of the open, the return phase is flat around 0 degrees (+/- 3 degrees out to 8 GHz). The issue is that below 2 GHz the return loss is not 0 dB, but around -0.6 dB. I can also see this error in the measured data of the amplifier.
I do have port impedance conversion on (single ended 100 ohms).
I have tried a single ended measurement (S1P -> S2P) and a balance measurement (S2P -> S4P file).
Within PLTS I also adjusted the impedance of the port since the automated process said it was 73 ohms (and it should be closer to 100 ohms).
Any ideas. Thanks.