It appears there is an impedance mismatch.

[frohro]

This graph seems to indicate there is likely a high impedance mismatch. See the figure which was taken with between two and three meters of coax between TX and RX.
We need to measure both ends with another known good VNA. :-)

[frohro]

We may be able to see this in simulation using LTSpice. We should check this out.