Suggestions? You bet! Get out your RFSim99 and build a model of what
you are putting on the RT test set test port, including any connecting
cable. Play with the model to see what will get you the sort of
performance you are seeing.

Note that you haven't really told us what you are plotting. I'm
assuming it's a plot of the complex reflection coefficient on a display
that's scaled to a unit circle. Are you using a standard Smith chart
overlay?

The normal interpretation of the result you describe, however, doesn't
make much sense, since the load seems to go from something like a
nearly pure inductance with reactance equal to the reference resistance
at the center of the plot (50 ohms?), and wanders around to become a
very low impedance (at the left edge, -1+j0 reflection coefficient),
and then around to become capacitive with moderate reactance (35
ohms?). Could you possibly be using some other scaling? What happens
if you are trying to display a value beyond full scale on your display?

I suspect something is not what you think it is in the measuring
instrument, and from past experience, I'd say it could be just about
anything. Careful visual inspections often reveal problems.

Cheers,
Tom