Essentially, in a one piece CD/DVD player, the tray takes up some of the room, a couple of solid state circuit boards a little bit more, and the insides are mostly air. The D to A conversion, digital filters, analog filters, and analog output stage are normally done with a couple of large scale ICs the size of your fingernail and a few surface mount passive parts. Cost, maybe a dollar or so. Maybe a couple of transistors thrown in.
This would be just fine if this kind of engineering worked well musically. It doesn't. To preserve the music, we must translate all of the digital data stream from the disk to music faithfully. We must also filter out all of the switching frequencies and their harmonics, we must drive loads downstream. The typical IC solution for this has neither the bandwidth or current drive capability to do the tasks required. The residual digital switching noise is going to be at 44KHz (or 176KHz with 4 x oversampling) and its upper harmonics. An audio stage that barely works to 20KHz does not cut it. The typical linear IC barely has 10 mA of current drive and that is not adequate to drive the analog filters at the frequencies involved, let alone a downstream load (cables, switching networks, etc.). The typical "one bit" D to A converter can only operate at the internal clock speed, maybe 10 meg or thereabouts, and thus can only retreive about 1 percent of the digital data on a "one pass" basis at high frequencies. A pure 16 bit parallel processor is needed to retrieve all of the data at all frequencies. Nobody does that any more (except us). Why not? Because 1-bit processors measure better on steady state test signals. Of course they only provide 8-track cassette resolution at high frequency on music. Oh well.
The Ultra DAC requires a fairly complex multi-stage regulated power supply, a hefty transformer to support the tubes and the digital stuff too, a complex logic board doing things our way with many carefully selected discrete parts and circuits rather than dinky little surface mount stuff to make it tiny and cheep (and sound like "who cares"). The Ultra DAC also requires a hybrid tube analog filter and audio output board to make sure the music can actually get to the rest of your system undisturbed. Although we were able to squeeze it all into the previous 12" wide chassis, it was really nice to have more real estate in the new 17" wide chassis to provide more distance to the power transformer and make a few other useful enhancements.
Of course we will plead "guilty" as to having lots of air in the box when the new chassis is used for the OmegaStar DAC, where the analog filters and audio output stage are done with very high speed linear ICs and current amplifiers (a great way to maintain the necessary bandwidth and current drive without the significant expense of the hybrid tube section). The OmegaStar does not have either the price or the absolute "air" of the Ultra DAC, but it does have the approval of The Absolute Sound (a Golden Ear award there a couple of years ago). They liked the Ultra DAC too in a recent report.
I guess the real reason for a high quality external DAC is that those of us who build them want a lot better music than what the mass market designers can give them. They can get the digital data stream perfect, easy to do that with the proper error code execution, and turn out perfectly working transports like popcorn for under $100, but when it comes to analog music, well, I believe the designers really don't care. As long as they can sell jillions of I-Pods and compressed downloads, they are happy. I want more than that, that's why we do it.
Frank Van Alstine
