All network analyzers that I have worked on (repaired, calibrated and use nearly every work day) since 1977 (8409, 8505, 8753, 8510.......) have the test ports connected to and thru the same front panel (built in or on the S-parameter test set). I see absolutely no reason to isolate them on the nano. I have not used the isolation part of the cal yet but isn't that the reason for doing isolation?
I thought about mounting my unit in an enclosed box as well but I had this old HP radar test set with everything right there and ready to go plus I do every thing at HF and have not seen any leakage problems there.
K6JCA's set up looks nice but I again I see no reason to separate the connectors. It might even lead to cross talk when doing S21 creating a ground path loop.
Last night, after finally getting a second HF antenna lashed up before the winter snows, I checked the isolation between the two antennas using S21 for the first time (except for a quick functional check when I rec'd the nano in late August). On 40 meters there is only 20 dB of isolation and on 20 meters and 160 meters there is about 29 dB of isolation. For each band I retuned the transmatch on the one antenna that needs a transmatch. When I de-tuned the transmatch even by a small change, the coupling dramatically decreased showing the impedance match was critical for max radiation from one antenna into the other.
