Mike,
Your Lumin i
s just a computer and a DAC in a very nice box with a linear psu. Don't know about the Linn, but the Bryston BDP-2 is also a computer in a nice box with a linear psu(just reviewed here:
Bryston BDA-2 and BDP-2 | AudioStream).
That's what I was thinking (all be it a very good computer and a very good dac in a very nice box).
This may not be the right place to ask this but a lot of you have a lot of insight.
Regarding the best digital to analog architecture, I am trying to figure this stuff out. As I understand it you gotta go from raw digital data to analog in the way that sounds the best (i.e., with as little added noise from required components, the lowest digital artifacts, time distortions etc.) To do this you need:
1) Storage to hold the raw digital data: a Hard Drive (loud) or Solid State (quiet).
2) A motherboard with a clock and a chipset to decode the raw digital data.
3) A DAC with it's own clock to take the decoded digital data from the chipset in the motherboard and make analog out of it.
So it sounds like keeping the raw storage separate from everything else and thereby isolating the hard drive noise through the an ethernet connection is a great start.
Next making a purpose built motherboard that runs the simplest, quietest software with a quiet, linear power supply is a good next step.
Then the question is whether you keep the DAC in that same box with the motherboard or you make it a separate unit. The benefits of it being in the same box is that you can easily synchronize the DAC clock with the motherboard decoding clock without having to go through an asynchronous usb which can sound less than optimum.
The benefit of putting the DAC in a separate box is that you avoid any noise from the motherboard and chip set, you can sell it and upgrade as DAC technology evolves and you have access to the higher end of the DAC market which all seem to be stand alone (msb, dcs, trinity etc.). You can use balanced interconnect for the best sound and you can still synchronize the clocks with a bnc or a buffer at the dac can be used for re-clocking.
So if I have something really wrong please comment but if this captures the essence of whats happening is there an inherently preferable approach that people prefer and why?