When using the rectangular display, you can see that results may be compromised by the limited resolution of the magnitude/angle data, which is 0.001 in magnitude. You can see that the magnitude doesn't change between several adjacent points. You might try using an RI Touchstone file. The data is unlikely to be more accurate, but it may have much greater resolution.
I see no problem using the imaginary part of S11 to locate the exact null.
Brian
