I apologize for mistyping Scd21 for Sdc21 which I am more interested in.

Could I do the following as a sanity check for completing the theoretical calculations of uncertainty of Sdc21-----complete a 4 port ECAL with my test cables. Insert 2 THRUs, then perform the fixture simulator measurement of the THRUs themselves to get a feel for the residuals?

I hope this doesn't sound too dumb. The idea is to arrive at a test of the VNA itself to see how much it might add and compare the theoretical calculations.

Could I do the following as a sanity check for completing the theoretical calculations of uncertainty of Sdc21-----complete a 4 port ECAL with my test cables. Insert 2 THRUs, then perform the fixture simulator measurement of the THRUs themselves to get a feel for the residuals?

I hope this doesn't sound too dumb. The idea is to arrive at a test of the VNA itself to see how much it might add and compare the theoretical calculations.

For Sdc21 (or Scd21), you can show that for the case of 2 throughs, Sdc21 should be zero (assume balanced port 1 is ports 1 and 3, balanced port 2 is ports 2 and 4; throughs are S21=1, S41=0, S23=0, S43=1: Sdc21=0). So measuring 2 throughs will give you a composite floor to your Sdc21 measurments, which should be added or subtracted (in linear terms) to all your Sdc21 measurements to determine the uncertainty bars, much like a residual directivity is added to return loss measurments.

Some of this is in this paper which I presented at the last MTT-S conference, along with the presentation (which has the residual measurement information).