between the thickness of the substrate and the radius of pro
The radius of the probe should not be related tothe thickness of the substrate. You pick a probe radius in design and then you can use the dimension to simulate it on IE3D/FIDELITY.
I saw some simulation files and I guess they used the probe radius as half of the substrate thickness, I try like that I have better results but not the best. May be the trial and error way can help you.
There is no specific requirement that I am aware of like the 2:1 ratio you cite. I would look at insertion loss and matching as issues to be considered rather than letting the dielectric thickness drive the coax dimensions. If there was a thickness to coax dimensional requirement, what would happen as the dielectric constant changed?