I'm running a GTX eye scan test similar to the method used in XAPP1198. The problem I'm seeing is that when transitioning from vertical offset -1 to vertical offset 0, the eye diagram gets "shifted" up. For example, you can see the bottom of the eye forming below the 0 offset, but then once you hit the 0 offset, you start to see the bottom of the eye starting to form again. Here's a sample of one of my GTX to illustrate.
I've verified that the correct HORZ_OFFSET, VERT_OFFSET, etc are being written to the DRP port of the transceiver. Has any one seen this before or have any idea why this might be happening? This seems to be happening on all of my transceivers, but the amount of vertical shift that occurs varries. In some cases, it seems to be shifted in the opposite direction.