Thanks Dave,
Your simulation is the response I expected. Unfortunately i's not the response I'm measuring. I re-calibrated and then used the Smith Chart as a sanity check. I got the expected results for open, short and 50 ohm load.
When I built the fixture I connected the ground permanently so that the ground is connected between CH0 and CH1 even when the LPF is not in place. I've re-tested without this link and got the same results. In general terms I'm not sure if this link is a good idea or if I should remove the link.
