Can anyone who has been testing it out elaborate a little on how it compares with the original E10 and the dragonfly?
The E10K is definitely better than the E10. Subjectively, it's more open-sounding (less roll-off) and less-veiled, leading to better resolution. It doesn't have circuit noise issues when plugging in headphones anymore, and the bass boost is improved --- less boomy and a bit more subtle. With the new analog design of the E10K, there shouldn't be any clipping issues at high gain, and line-out should be a true Red Book 2 V
rms. That's one of the benefits of switching to a charge pump voltage output DAC like the PCM5102A --- you'll get a clean line-out without needing a lot of supporting circuitry. While I'm a fan of the AD8397 used in the original E10, it's probably not suitable for low-cost designs because it's a finicky opamp. Solder the AD8397 onto a DIP socket adapter, drop it into a CMoy, and you'll find out exactly what I mean. Ideally, it needs substantial short-circuit protection and clean power, necessitating a more complex PCB layout.
The theme of the E10K is simplicity --- it reduces design complexity over the original E10 while simultaneously improving performance. I'd say it's a win-win for both the consumer and FiiO.
I compared it against the Dragonfly 1.2, and my feeling is that with respect to SQ, they perform similarly. The Dragonfly sounds a little dry but has less treble splashiness, perhaps due to a decent asynchronous jitter implementation. The Dragonfly isn't stellar in the distortion department, but I didn't really hear those problems when listening to it. Overall, I feel the E10K is comparable to the Dragonfly, losing out on USB performance but having better drive capability.
My major criticism is that the E10K preserves the Galaxy Tenor TE7022L USB receiver from before, meaning that USB transfer is still isochronous and still won't do 88.2 kS/s sample rates. However, FiiO seems happy with their familiarity with the part, and seems to be confident in how to get the best performance out of it. They told me that they tested it against other, newer parts with asynchronous transfer, such as the SaviAudio Bravo SA9027 (used in the JDS C5D and Stoner UD110V2) and still preferred the TE7022L. Basically, just don't expect mind-blowing jitter performance from the E10K (the TE7022L can only be entrusted down to ~200 ps at best), but it should be no worse than before.