Two common mistakes:
1/ Making measurements with a calibration, certainly done in the right place, but not on the same frequency range. Nothing indicates the range used when you reload a calibration range.
2/ Forgetting to do a RESET on the nanoVNA box. If you forget, in the best case, it will produce a Smith that goes out of the circle but not always. RESET of the measurement bridge is essential before a calibration.
--
F1AMM
Fran?ois