Re: ustrip to 1.92 conector at 40 GHz
I assume you typed a little error on subj. "...1.92...." do you mean 2.92 mm ?
If yes, we have found a factory, namely Southwest Microwave, that produce a particular 2.92 mm connector that is more easy to mount on jig wall, so the transition became more easy and (we hope) more matched.
The difference between the original Wiltron (Anritsu) method is that Anritsu use a glass bead, wich require to be brazed (indium? low temp SnPb alloy?), the Southwest solution is a PTFE bead, it don't require heating, the pin is lillte larger (because PTFE has lower εr than glass) so mechanical alignement is less critical. Finally the pin is properly designed and supplied with an "ad hoc" step to help in impedance matching.
To connect the pin to microstrip, i've used Epotek H20E rather than a brazing alloy because the "drop" is mutch little.
Another method that i've seen inspecting inside an old Miteq LNA, is 2 simmetric and radial ribbon bonds.... People at Miteq are artists