Converter line in tests

I did another set of tests between the Apogee Mini-Me, MOTU UltraLite (first generation), Focusrite Scarlett 18i6, and the M-Audio 2626. All recording was done at 96kHz, 24-bit. As per a suggestion by termites2 on Reddit, I did a test of a line-level signal coming into each converter.
Now one of my big problems is that I don't even own a CD player anymore. I think my parents might have one. So what I had to do was take the headphone output of one of the few Macs in my studio which actually has a working optical drive (this is a long-running frustration of mine) and in iTunes play a CD through the headphone output.
Yeah. Not exactly the highest-end way to get audio into a computer. But it's what I have.
So I took the headphone out and injected it into the high impedance instrument jacks on the front of a pair of Neve 1272 mic preamps (BAE). The balanced output from the Neves fed the analog inputs of the different converters.
Note: the converters all take about the same line in level EXCEPT for the Scarlett. So I turned the Scarlett down 9.8dB in Samplitude in order for it to match level-wise.
So now the analog path of this test is... well it's funky. Good preamps but lousy source 'cause that's all I've got. Such is life. (The recording is noisier than it should be, and the overall audio quality of what is otherwise a stellar recording and mix is not as good as it could be with, er, a better D/A converter.)
Listening through, with full-range and fully-mixed material, it's kind of interesting to hear the differences in converters. The tests all play through as an "album" in Bandcamp (embedded above).
I lined up the waveforms so that they're sample accurate (well, +/- a sample because there's a bit of argument where each conversion is concerned). Then, for my amusement, I inverted the polarity of the MOTU, the Scarlett, and the M-Audio 2626, and played them along with the Apogee's track (which was not inverted.)

Top to bottom: Apogee, MOTU, Scarlett, M-Audio 2626.

Interestingly the Scarlett (which is the third track) goes so perfectly out of phase with the Apogee at one point that it actually disappears. (Note that the inversion was done at the track level so it does not appear in the waveform view above. Also, there's some obnoxious noise evident in the inversion.)

That's it. I'm tired of testing things. Tests. Meh. It's time to make some records.