We have a Tektronix MSO73304DX oscilloscope (firmware 10.14.1 build 15) running TekExpress DisplayPort (version 10.3.9.13) and have a few questions regarding the digital filters used in some tests.

Running the eye diagram test on HBR2 with the pre-emphasis set to 6db or higher almost always caused the zero-length eyes fail.  The waveform gets so distorted that DPOJET cannot even make an eye out of it.  Interestingly, the worst-case eye would pass without issue on the same waveforms that failed the zero-length eye.  After doing some digging into what the software was doing, I found that HBR2 applies a continuous-time linear equalization (CTLE) filter for the zero-length eye, and then applies an additional "cable emulator" filter on top of the CTLE filter for the worst-case eye.

Taking the waveforms saved during a run of the TekExpress software, I manually ran the same DPOJET tests on the waveforms with and without these filters to see what would happen.  What I found is that the waveforms without the CTLE filter performed as expected and passed just fine.  I have attached eye diagrams generated for 0db and 6db (both with SSC and at 400mV swing).  The top left is just CTLE, top right is CTLE + cable emulator, bottom left is raw waveform, bottom right is just cable emulator.

Is this the expected behavior of the tests and filters?  Or is something else going on here?