For RF-measurements at discrete components (resistors, inductors, capacitors or even transistors etc.) a vector network analyser must analyse the RF-response from the device under test (DUT). For that reason a directional coupler is necessary to distinguish the sent generator signal from theand reflected DUT-response. I build my own with a transformer I bought from Mini Circuits.
Absolute directivity of the coupler is not even that important, as the VNA is anyway OSL-calibrated using calibration standards of open, short en well defined 50Ohm load.? But the better the bridge derectivity the more sensitive the measurement will be - important for especially if the response from the DUT is very small in magnitude.
As I couldn't find any smaller RF-enclosure in stock, the directional coupler finally became almost as big as the whole main VNA. In fact the coupler could be made much smaller, but I simply didn't want to buy another smaller enclosure than the one used - the enclosure is almost empty...
I followed the strategy to use SMA connectors for precision, easy availability and consistency with the main VNA unit.