Measuring Differential Trace Impedance with 2 port vna

[electronx]

yes this is the question .I'm surprised it hasn't been asked before
I want to measure trace impedance with a 2 port vna. I am thinking of designing a demo board for this. It will be connected with smd sma connectors. The differential lines I want to measure are 90 and 100 ohm. What kind of termination should I use for this?
I think I can extract it with mixed mode syz parameters.

[virtualparticles]

There are some interesting things you can do with this. If you terminate the far end with 50 ohms loads and attach ports 1 and 2 to one side, you can calculate differential S11 from SDD11 = (S11+S22-S12-S21)/2. You can then switch to time domain and measure return loss vs time/distance. If you don't have time domain on your VNA, you can use SciKit-RF to do the calculations for you in Python. You can calculate all differential parameters by performing a full 4-Port measurement, 2-ports at a time, with other ports terminated. See the attachment.

[electronx]

thanks for your feedback.
Is it possible to provide a simple visual about the termination you mentioned?

[virtualparticles]

Sure, here is a measurement setup of a differential line with the port numbers annotated. It's a four-port measurement with a two-port VNA, so pretty tedious. Connectors on the test board are female SMA and the cables are male SMA. I used 50 ohm male terminators on any port that was not connected to the VNA, which is required. The way I numbered the ports on the board match the article I attached earlier for calculating the differential parameters.

I'm using a 9 GHz VNA from Copper Mountain Technologies here. This one is extendable to 330 GHz if needed.

I should mention another thing. You have ground on each side of the traces. If you don't have stitching vias attaching top ground to bottom ground, those two ground planes will resonate and take a big bite out of the trace response.