From your measurements it's quite obvious that this amp uses a combination of voltage and current feedbacks of some kind:

Voltage negative feedback keeping the output voltage stable, and a current positive feedback via a current sensing circuit (maybe just a resistor), counteracting the voltage feedback to create some "current source" effect.

So - no matter how the output stage is build, it does not react much like a high impedance current source.

Interesting is the the slight increase in output impedance with lower load resistance. This can make a little difference on low impedance headphones, it's a little bit kind of "hyper current source" effect.

Two other measurements would be interesting:

1st comparing to a voltage amp, the frequency response with a headphone having a low impedance with strong variation over frequency, like the "Enigmacoustics Dharma Production 2015", to see if the changes in frequency response follow it's impedance curve.

According to your measurements I would expect a drop of 3.4dB@2.5khz relative to 500Hz.

If you would use the transfer function you could very much improve the accuracy of your measurement.

Here's a describtion how that works:

http://www.studiosixdigital.com/audiotools-modules-2/acoustic-analysis-m...

A bit of smoothing would make the curve clearer readable.

2nd it would be interesting how the maximum unclipped voltage is on different loads, if current or voltage is the limiting factor.

Here's what Apogee claims:

225mW into 30 Ohm (= calculated 2.6Volts, 7.5mA).

40mW into 600 Ohm (= 4.9 Volts, 8.1mA).

Not that much IMO, a standard NE5534 as output stage can do 12V, 35mA.

Combined with your measurement I'd say their output stage is current limiting, say not very strong.

But enough to hurt your ears with most headphones.

Best Kai