FWIW I measured one of the 60 dB.com bridges at a directivity of ~40 dB using TDR with a Tek 11801 & 20 GHz SD-24. That was the ratio of the step on the incident output vs the pulse on the reflected output with a Padiall 50 ohm SMA load on the DUT port. The step was 118 mV and the reflected pulse was under 1 mV. That's really pushing the 11801/SD-24 to the limit of the vertical accuracy.
I plan to make additional measurements using a 438A and 8648C.
Have Fun!
Reg
