Well I hope JW does not mind.. I've grabbed the message from the avsforums about the construction of the DV88/DV27, it was in a thread about using DVD players as Cd players...
I quote :
"Since someone has mentioned Arcam's name I'll come in with our point of view, which is based on actually developing and listening to real products, and monitoring their success in the marketplace.
To do proper D/A conversion in audio you need, among other things, a very clean crystal generated clock signal, with as little jitter on it as possible (we are talking very small numbers, sub 200 picoseconds for good results, as measured at the output of the DAC by the Paul Miller/Julian Dunn technique). It helps to have the master clock physically right next door to the DAC chips too.
There are many other factors involved of course, including the DACs themselves, the digital filters, post conversion analogue filters and so on. The layout of the PCB, especially with respect to ground currents is also critical and mistakes and less than optimum layouts here are definitely audible, even though the obvious measurements may not really change.
4 or 6 layer boards help. We've done lots of listening inside the company and are absolutely certain of the huge amount of attention that needs to be paid to these things. It's reflected in our sales too.
It's not too difficult to do all of this inside a CD player, where you only need one master clock and where the transport is slaved to this. DVD players need multiple clocks however, to get 27MHz for video and integer multiples of both 44.1 and 48 kHz for audio, as a minimum. In nearly all DVD players I have seen, the master clock is 27MHz and the audio clocks are derived from this via additional phase locked loops. Now this can be done well or badly - there are a couple of pretty good if somewhat expensive parts out there - but mostly it is done using the on-chip PLL of the MPEG/DVD decoder chipset (the lowest cost solution) and the jitter results are from mediocre to horrendous. As you might imagine this compromises the audio DAC's performance, even if all else were perfect (which is usually isn't by a long way). It also affects the SPDIF digital output stream.
At Arcam we overcame this in our house designed DV88 and DV27 DVD players by using 3 separate master clocks, one for video and two for audio, the latter being adjacent to the audio DACs. The Zoran MPEG/DVD processors we use support asynchronous clocks, AFAIK most others do not and it was one very good reason why we chose to use Zoran parts in our DVD players. We do a lot of other stuff too, including phase locking our power supply switchers to the audio clock running at that time, but I am sure you get the idea.
It is interesting to note that Perfect Vision magazine rates Arcam's DV27 as "one of the few DVD players that is truly a reference quality CD player in audiophile terms, and is the best combination of absolute performance with DVDs and CDs this side of the Ayre's mighty D-1". (March/April 2003 issue, page 46).
Back to jitter. In practice the SPDIF sysytem is not very good at transmitting the master clock signal and most receivers and many AV processors are not very good at recovering the DAC clock from this stream, so you get jitter on the DACs inside the receiver/processor. The standard SPDIF receiver chipsets are definitely not good enough here if you want the best results. This limitation can be overcome, for example by fitting a crystal based PLL clock recovery circuit (as used in the best processors - Arcam, Meridian, Lexicon, TAG; this list is not exhaustive) - which is quite expensive and difficult to do. But the results are clearly audible and speak for themselves. To prove my point another way, we initially tried to avoid doing this in the Arcam AV8 processor on the grounds of expense and time to market, until we did the auditioning - then we rushed back to the drawing board to implement the circuitry!
I don't know if the Krell Showcase has this circuitry or something similar (I suspect not but am happy to be corrected). What I can say with some confidence is that if it does not then (IMO of course) you will be better off with an analogue connection using the DACs inside a good CD player or that rare beast, a properly implemented - in audio terms - DVD player. Of course if a processor doesn't have a true analogue bypass then all bets are off again.
HTH.
John Dawson (Arcam)"
Makes intersting reading.. and makes u realise that a chipset does not make a player !
Cheers
Matt