<subjective opinion>
So...I finally got time to try the 3.0 firmware, replacing it from 2.14b. I'm listening to 24/96 and 24/192 FLAC from HDTracks, using V-Moda Crossfade M-100s. I hear what people are saying about what they perceive as improvements in the SQ, but I beg to differ. I think I know why. Sure, there seems to be increased detail, stereo separation, soundfield, whatever...but it feels to me like this is all being achieved by the same process as a "widen stereo" channel processing that you can do in any decent audio software. What that basically does is some subtraction of L-R and R-L to each channel, and it results in partial phase cancellation, and significant reduction in your "phantom center". Normally, vocals are centered, and vocals tend to be around the middle of the frequency spectrum, so at first, I thought what I was hearing was a bit of a "classic V" EQ coloring the sound, so I started playing with treble and bass settings. Nothing helped, and that was when I figured out what was going on. I know that not everybody loves Rick Ocasek's voice, but I'd still like to be able to hear it properly in the mix, along with everything else that's supposed to be in the center of the mix...so I dropped back to 2.14b, and I'll be staying there.
</subjective opinion>