Why are the S11 and S22 dips different in my microstrip resonator measurement?

[uWave_Matt]

I've fabricated a simple half-wave microstrip resonator with gap-coupled microstrip feed lines. It's designed to operate around 10GHz, but due to slight over etching, the half-wave portion is 5-10um shorter than I intended. What might be more significant is that the gap, which was originally designed to be 5um, is now closer to 10um.

I'm not sure if any of the aforementioned is relevant to my question...when I characterize the device on the "trusty" HP8510, my S22 dip (9.7GHz) occurs at a slightly lower frequency than my S11 dip (10.5GHz) and has a significantly greater (more negative) magnitude.

So far, I've double checked all of my calibrations.

I suspect feed line length (where my contact touch the feed line), coupling gap variation, or perhaps a problem with the 8510. Does anyone have any other thoughts?

I've posted this question on another forum and it was suggested that I use "stepped" mode instead of "swept" mode. Does anyone know where these settings are?

 

[f95toli]

If the resonator is symmetric S11 should be indentical to S22, meaning your result doesn't make physical sense which in turn would point to a problem with the VNA.
Have you tried switching ports? I.e. connected port 1 to "port B" of the resonator and vice versa. The result of a reflection measurement should obviously not depend on which port of the VNA you use.

Also, it can't be a calibration issue as such; when you calibrate a VNA you calibrate the magnitude and not the frequency (although if the calibration if way off the position of the peak might appear to move; simply becaue the magnitude is wrong).

What is the Q (roughly) of your resonator?

 

[uWave_Matt]

If the resonator is symmetric S11 should be indentical to S22, meaning your result doesn't make physical sense which in turn would point to a problem with the VNA.
Have you tried switching ports? I.e. connected port 1 to "port B" of the resonator and vice versa. The result of a reflection measurement should obviously not depend on which port of the VNA you use.

I've rotated the jig that the resonator is in which effectively causes S11 and S22 to switch. This is just the room temperature measurement. I'm dealing with HTSC material so the S-parameters should be significantly different after I cool the chamber.

Also, it can't be a calibration issue as such; when you calibrate a VNA you calibrate the magnitude and not the frequency (although if the calibration if way off the position of the peak might appear to move; simply becaue the magnitude is wrong).

But you have to calibrate the magnitude over a certain frequency range.

What is the Q (roughly) of your resonator?

Since it's an HTSC, the Q-should be just about humongous. Another HTSC resonator fabricated in this lab had Q>10000 and I've calculated the Q of this one to be close to that.

I should also mention that one of the coaxial lines from the VNA to the jig is 4-5 time longer than the other, although I'd assume calibration would take care of this.