1W @ 50 Ohms = 7.07V (140mA).
For this you need 24V power supply rails (+/-12V) and a power supply that must be able to supply 0.4A.

The Groove can provide:
225mW into 30 Ohm = 2.6V
40mW into 600 Ohm = 5V
So most likely the internal supply voltage around is 16V (+/- 8V) with a current limit of 85mA/channel.

As it is powered by USB and the device draws around 150mA max from a USB bus which is well within spec.

  Thanks solderdude ... I'm an idiot, I always fuck up the zeros   My calculations reflect FIVE ohms resistance, not FIFTY.

I did some volume matched comparison against the Geek Pulse and the O2/ODAC. My usual method of volume matching is to plug the two amps into an A/B switch and then feed that into the line-in of my Zoom H2n, where I visually match the graphical audio meters on the recorder. This has worked in the past but completely failed to work with the Groove - I could volume match with the meter but end up with very different levels with the headphones plugged in. There's probably something that I'm not understanding here because of my limited understanding of the electrical stuff. I ended up volume matching the 1khz tone by ear with each headphone I tried, so this is far from scientific. In tests I can usually identify differences in levels of about 1-2dB.

Firstly I can say that the differences between all the sources was surprisingly small on most of the headphones. Just about the only obvious difference I could immediately pick out was that the Groove has a lower noise floor than the ODAC/O2, which itself has a lower noise floor than the Geek Pulse. That's actually pretty good because the O2 is the blackest sounding source I've heard.

I tried it with my M70x, GR07, ATH-IM02, Shure SRH-1540, AKG K712 and DT770.

I can't really say I can hear any strong advantage of the constant current drive's supposed ability to more accurately control the voice coil, nor can I hear any any real obvious harm from the supposedly low or negative damping factor the Groove has. Most of the times I thought the Groove maybe sounded a little sweeter or maybe a tiny bit cleaner, but it could well have been placebo and I wouldn't trust myself to pick it in a blind test.

The equipment that I could actually hear the most distinct difference with was the IM02, which ended up sounding more upper treble forward with the Groove. I'm assuming this is because of the way current drive amplifiers have output that follows the impedance curve - but on this point, I'm a little confused. If the Groove has something crazy like a 2.5 mega-ohm output impedance like the Bakoon, shouldn't the impedance curve of most headphones be almost irrelevant? I sent another email to Apogee asking what the actual output impedance of the Groove is.

From some of my reading it seems that the most appropriate headphones to pair with the Groove are highly efficient, low impedance headphones with minimal impedance variation across the frequency range. In my collection that would be the Audio Technica ATH-M50x and the V-Sonic GR07. I think they sound great with the Groove (again, Placebo?) but I just wanted to make a comment - as much as a lot of people hate the M50x, it seems like Audio Technica did try to prioritise making a headphone with a near ruler flat impedance and phase curve so that it could match with the widest variety of output sources.

Imma do some listening but at this point I am decently happy with the Groove. It's quite expensive for the power, but it's small and compact, doesn't look like some DIY science project on my desk like the O2/ODAC stack, has proper digital channel balance at low frequencies, doesn't make me nervous that it will eat my headphones like the Geek Pulse from power on/off pops or otherwise, doesn't need its own power brick - and has a cleaner output than both.

I'm assuming this refers to the same idea of the current source amplifier ignoring electrical damping oscillations from various sources:

The trade off is that frequency response now follows the headphone's impedance curve?

Yes, that is exactly the trade off.

Some MA drivers may sound quite wrong from this amplifier.
Most headphones will only react mildly as the impedance is quite flat (= do not need/expect/want any electrical damping) only some headphones will react more audibly.

Headphones are designed (voiced ?) with (close to) 0 Ohm in mind, perhaps some older headphones are excluded.
NONE of them are designed to be current driven which is something that may be worth mentioning.
The fact that some headphones sound 'better' probably has more to do with euphonic coloration of the FR/phase than with being closer to the 'intended' signal.

Most headphone drivers break-up at higher frequencies, the diaphragm is thus not making the same movements as the voicecoil does so that type of distortion isn't reduced.
Phase distortion (when caused by a voicecoil impedance rise) is lowered but at the expense of an increased SPL which may be beneficial to the sound ... or not.

In the end, the only thing that matters is if you like the sound, regardless if it has 'changed' somewhat.
Yes, the bass and treble might be different somewhat compared to 'normal' amps with some headphones but headphones themselves usually are 'all over the place' in the FR department all over the frequency range anyway. Most of them well over 5dB some even up to 20dB.
None of them accurately reproduce the applied electrical waveform, regargdless if they are voltage, current, or both (output R between 0 Ohm and a few hundred Ohm) driven anyway.

I can make a prediction for you.
+4dB in the bass (as if you pull up a 100Hz slider on a 5 band equalizer by 4db) + 1dB extra upper treble energy.
So more bassy/warmth than it is stock.

You could also search for experiences people have with the Bakoon + HD600/650 to get an idea.

The HD650 was NOT designed to be current driven.

The HD650 was NOT designed to be current driven.

Neither was the HD800. I don't care how many people say they 'like' it. Poor HD800s....