【Editorial Discussion】

A VNA fixture calibration that “comes back clean” and a DUT measurement that will not match simulation or a known-good part is a regular lab argument. The cal kit and the DUT rarely sit on the same reference plane, and a pretty thru residual does not mean the plane you think you have.

1) Confirm the reference plane with a known artifact, not only with the cal standard set. A beadless airline, a well-characterized attenuator, or a second thru that was not in the cal is more honest than watching the original thru go to 0 dB. If that artifact is already wrong, do not debug the DUT yet.

2) SOLT is sensitive to standard definitions and connector repeatability. Wrong offset delay or C0–C3 on an open, a slightly different torque, or a worn adapter will tilt phase and walk the delay. If you changed adapters between cal and DUT, you moved the plane. Write the adapter stack into the procedure, or the next cal will not be the same measurement.

3) Fixtures with launches, vias, and connectors often need a different method. TRL/LRL or a well-made de-embedding structure can be more stable than forcing SOLT at a coax connector and hoping the launch disappears. If the discrepancy shows up mainly above a certain frequency, look at the launch and mode conversion before blaming the model.

4) Probe, cable, and drift belong in the log. A cal done at 25 °C and a DUT measured after the cables were coiled, the probe was reseated, or the room warmed up, will disagree. A quick recal, or at least a residual thru check immediately before the DUT, is cheaper than a long model debug.

Open question: the last time fixture SOLT looked perfect and the DUT did not, what actually fixed it — standard definitions, an adapter you forgot, a switch to TRL, or cable/temperature drift? Any example where the residual thru was excellent and a second known attenuator still failed? Please share the band and connector type.