Hi Mike,
That didn't solve my problem, and I'm still investigating. I think I'm going to have to iterate through ADI's commits to find my problem.
My HDL is in line with this commit (the commit just before the increase in rev to v8): analogdevicesinc/hdl at 89961c8dd7bcc169077f4af13ef5fd9f73da1bbe · GitHub
When I use the following 3.13 kernel from late April, I have good results: analogdevicesinc/linux at 5aa6400960d9f4033bc6a2e27bb8e2814627b67d · GitHub
I just tried this kernel (analogdevicesinc/linux at 1c2312f91ec131e446ee8393f0e81024ddc7c65c · GitHub) from July 2, and I have bad results. Something weird happens to the IQ. On my first tune, the IQ looks somewhat reasonable, but if I retune the LO, the data is garbage and I get the same value for every IQ sample.
For now I'll probably stick with the late April kernel that works for me. If I get more hours to look at this, I'll probably try some more kernels after the working kernel but before July 2.