Dave does indeed have a unique amplifier section - it is actually a 2nd order analogue noise shaper (a conventional amp is a first order noise shaper).
The reason I do this is two fold - firstly, whenever you load a headphone amp you get more distortion, and its audible and all amplifiers suffer from this. By using a 2nd order noise shaper approach allows me to eliminate this problem - when I load the outputs there is zero change in distortion (apart from a small increase in 2nd harmonic). This means when you load the OP, there is no change in sound quality (all other amps harden up and to mask this they add a lot of 2nd harmonic to fatten the sound up).
The second reason this is done is high frequency distortion. All conventional amplifiers have higher distortion as frequency rises, and this has important SQ consequences, in terms of making things sound hard and un-musical. This is due to insufficient feedback available at high frequencies with poor open loop distortion. Now the distortion with modern high performance op-amps is down to the output stage - and that's one reason why all my DAC's have discrete OP stages, so I can eliminate this weakness. But my second order system has excellent open loop distortion performance plus no HF feedback problems too. A indicator of this is the gain bandwidth product - the best audio op-amps are 100 MHz, but with Dave it is 1 GHz.
You can see this HF distortion problem with the 19k/20k test. All DAC's make a mess of this but Dave has the lowest distortion of any other DAC using this test:
This level is actually the residual performance of the APX555, so Dave is probably much better than this (one reason why Davina is happening - I need better ADC's to measure).
By using the second order approach does not affect transparency - it is still a single global feedback path, so the effective number of passive components is the same as before.
Rob