I was just doing some serious listening on 1.2.8 and did some AB testing between the DX50's HO and the LO/UHA-6SMKII.
At the same volume level in DX50 (210) and matching the HO and amp volume using the SPL Meter (iPhone app) at around 77 dB, I can say it is really difficult to hear any difference (in terms of SQ, soundstage, instrument separation and depth). You will need tracks with a lot of instruments playing to even hear the smallest of difference. The bass might be a little bit tighter with the LO/amp but on casual listening you will not notice any difference.
On a proper blind testing I don't think anyone can tell a difference. I'm now 100% convinced 1.2.8 truly is the best FW to date.
However, there is still an advantage using the LO/amp( in this case with UHA-6S). If I turn the volume level in DX50 all the way up to 255 and then attenuate the volume knob on the amp so that the output volume is still around 77 dB the difference gets a little bit more obvious. On this setup the bass is much more tighter, slam more forceful, the mids a bit cleaner and more details being resolved in the highs it seem. Also the instrument separation is more obvious and each instrument can be heard clearly and does not interfere or mask another. Overall, the sound is a bit better.
Of course, if you want the pristine pocketable DAP setup, DX50 + 1.2.8 is tough to beat and an amp is not needed unless you want to use a full size headphone that requires more power than the internal amp of DX50 can provide.
I did the same thing before between DX50’s amp section and O2 and came to the same results, as posted in #6414.
I really don’t think the internal amp is that bad. It’s the digital volume control that makes the HP out bad. We are not supposed to hear much difference but I believe I did and many others could as well. Ibasso should have made the lower gains even lower so that we can crank the volume up to ~240 or higher when using sensitive IEM’s on HP out.