Well I want to make the contacts mobile, on a fixed stripline, by a screw mechanism. I think I will just try it.
Regardless of how bad it is, the normalization should take care of the problems right?
However, if there is particular resonances/etc in the test fixture, then even normalized (digital process?), the dynamic range of the measurement there will be reduced right? Or do VNA typically dynamically reduce signal power to maintain dynamic range during normalization? I have a E5100A I got real cheap, otherwise I would not have a VNA.
Like say by some weird property, the test fixture has a 80db peak (at the top of the measurement range) at 250MHz, and this is flattened by normalization. During the sweep does the sweep signal get attenuated by a dynamic gain stage to prevent saturation, or is it purely a digital offset? (in the case that the test object makes it even worse, and adds another 40db of gain at 250MHz, which in the digital case would cause saturation/gain compression(correct term?)).
I assume this could cause additional errors not cancelable just by correction though right? Like complex interaction between the test fixture impedance mismatch and the part being tested impedance mismatch, antenna effects (small?) and the mutual inductance due to geometry of the clamping system? I assume you can't get perfect results with a complete bullshit setup... or can you??
Can you clarify what factors are perfectly nulled out and which ones cannot be (if any)?