I work in a screening lab and have these 1:1 xmission line xformers:
http://www.futureelectronics.com/wir.../ETC1-1-13.pdf

I need to measure their insertion loss, per spec to 3 GHz. But I've
never used one before and am not sure how to connect it.
I connected it as shown in this tutorial:
http://www.bytemark.com/products/tlttheory.htm
using an HP8753D in CW mode as a source and an HP power meter across
the outputs as RL. Does that seems ok? I'm not sure about using the
power meter across the output. Should I load one output and connect
the power meter to the other, then measure insertion loss? Then switch
load an p.m. and I would check amplitude balance also?

My problem is after "cal'ing out" my fixture, I get a bad dip in the
insertion loss around 2GHz, like there's parallel C across the
'inductor' and it's resonating and opening up.

I made a short fixture on regular copper clad with 1/8inch wide
traces for 50 ohm, short runs anyway, with SMAs in and out. Using the
HP8753 in CW and a power meter, I took reference readings it with
copper foil across where the DUT would be, then put in a DUT and took
readings.