I play music at least 12 hours every day in my study, 365 days a year. Even when I am not at my desk, music suffuses my house. Whatever I can do to make my music better I therefore do. I am not famous for my tenderness towards a buck that can be spent on hedonism instead of giving a banker a thrill. But I use a standard Quad 67 all the time, and have a modded Quad 66 as well on which the modification is entirely to the power supply rather than the signal chain. Many people find my cavalier attitude to the CD player surprising. They shouldn't. I trust my ears which are well-educated, proven servants of my taste. I have in an emergency used a common cheap Sony Walkman personal CD player as a demo player for a collection of golden ears and none of them spotted the difference, though they were quick to spot the SE amp being rung into the chain behind the curtain.
The CD player and its associated electronics is the part of his audio chain where the audiophile can make the least difference and easily waste the most money. (Idiots who spend thousands on cables to tune the stray capacitances between incompetently matched components come a distant second...) This, on one view, is surprising, as many CD players are cheap and nasty, regardless of boutique prices that can reach five figures. But turn the viewpoint on its head, look for the expensive or at least good parts in cheap players, and suddenly you can see why the improvements at the margin will always be marginal. This is quite unlike the tubey's switch from transistors to tubes, or the discovery of ultralinear circuits, or deserting negative feedback for more honest circuits, or discovering directly heated triodes, all of which can be blinding revelations on the road to Emmaus.
First of all, forget improving the physical player itself. The Philips CDM-12 is the standard and the -9 is also pretty good. The old Pioneer laser disc player was also superb. There are many other CD players made for computers that are also very fine but are problematic in installation because the electronics or even the capability of installing the electronics for audiophile use are missing; this can be as simple as a lack of remote control. The best you can do with the rotating mechanism and laser frame is to bolt the unit down tightly to something solid. There are opportunities to improve the power supply both for stiffness and regulation but that is a standard electronics job one does for many other components in the signal chain, not special to CD turntables and laser controllers. The power supply may be the most significant upgrade you perform in terms of results. In any event, until the power supply is perfect to all parts of the CD player/DAC chain, nothing else you do will have noticeable effect.
Secondly, forget any idea that even a huge improvement in the DAC chain will be perceptible in any audio chain which has not already been very, very expensively developed to a peak of discrimination in every single one of its other components and in particular the speakers, always the weakest link in the chain. In any lesser system any shortcomings of the DAC are simply swamped in the noise floor. Let me say it again: the likelihood is that the DAC and the surrounding parts in even a cheap commercial CD player are the most highly developed, most nearly perfect, and most silent parts of your entire audio chain. You're trying to improve on near perfection. The rest of your chain needs to be perfect to hear your improvements. The implication is of course that you should come to the CD and its associated electronics last in your search for audio nirvana.
Having built various high definition systems, I was always surprised at how little DAC experiments added to audio enjoyment in relation to their cost. Eventually I built my Type 199 Millenium's End amplifier, which was so powerful that it required a complete 300B as a pre-amp to provide drive voltage and current to it. This amplifier puts out 75 gloriously undistorted single-ended watts and drove my ESL63 to ear-damaging volumes on a fraction of that. The Millenium's End was designed so that it could at the flick of a switch be tuned down to 25W to be less of a threat to my first series Quad ESL (the retrospectively named ESL57). I had other high-definition speakers if required to test whatever advance I could gain from a better DAC.
Separation into various boxes (stereo, separate power supplies) has never really worked for me in amps but I was determined that the DAC should be utterly silent. I thus decided to build a three box DAC chain, one for power, one for the CD motor and laser mechanisms and the third for the DAC itself. Parts availability is no problem.
You can get a CDM12 out of a cheap Philips 723 player or you can order it brand new from Holger Stein in Germany. The double handful of chips audiophiles need to consider are generally available from commonplace mailorder houses. (The one exception is the rare and odd single-ended DAC CS4328 as used in the exemplary DAC sold as a kit by the Audio Crafter's Guild.) Information is no problem. The Crystal site is a treasurehouse of high quality information; whatever you cannot learn there is on the Burr-Brown site. DIYers are very well served by serious experimenters like Andrea Ciuffoli and Thorsten Loesch.
As we have already discovered, after building my three-box CD player and DAC, I returned to my old Quad player. I didn't find the real estate demands of a three box player justiied by improved quality of my sound. (I used Hammond 17x10in boxes of various depths because I always do; if you use only one size, your projects stack tidily!) I gave the mechanical and DAC boxes to another experimenter and reused the expensive power supply. You can't win 'em all.
But, as always, if one has spent months gathering information and building a prototype, you've learned something worth passing on if it prevents even one DIYer going down a blind alley for even an hour. My loss may be your gain. I therefore present two block diagrams that illustrate my thinking of making the DAC with the smallest possible component count, which is always my first design imperative, and is rarely proven wrong. My diagram is several years old so there may today be new components to be considered by audiophiles.
Albert Einstein said, "Everything should be as simple as possible but no simpler." I was therefore prepared to tolerate a component to reduce the jitter I heard so much about. It proved to be unnecessary. I never heard any jitter without the filter nor any improvement with it. That a component does not actively harm the sound is no reason to leave it in. I looked into discrete Burr-Brown components, particularly the famous PCM63 (expensive by itself, more so when you've paid for all its non-optional companions) and eventually decided the Crystal all in one CS4390 does precisely what it says on the tin, does it better and does it cheaper. With the CS4390 I eventually settled on the CS8412 receiver. Those were the only two chips in the signal chain from CD to input tube. No other chips are required.
For input from the turntable and output analogue filter I used transformers. I looked into various tubes at the output but there was no sonic advantage for a quite a bit of circuitry (well filtered HV supplies, other necessary circuits in a place where you want to keep the wiring to millimetres). More specifically, I decided that soundshaping should be done in the driver of the amplifier following. Too many soundshaping elements simply become an uncontrolled mess. One purpose of component reduction is to recover control over soundshaping into the hand (or mind) of the designer of the sonic chain.
I found my first input transformers in the RS catalogue but soon replaced those with ones a radio ham designed and wound for me; he also gave me a pair of output transformers he wound for his own use for a DAC project he scrapped when he saw how little advantage I got from my expensive and time-consuming project. Good quality dedicated transformers for both ends of the DAC are available from Lundahl or Sowter. Some of these input transformers and/or porting arrangements may not need the 75 ohm load resistor. Whether you use the bypass filtering on the primary of the output transformer would depend on the transformer and your desires. The components shown are the maximum required with the least efficient (in relation to component count) of the transformers. If the transformers were perfect, no extra components would be required. Dream on!
Thus the entire DAC box consists of input transformers, Crystal's CS8412 receiver and CS4390 combined filter and DAC, and output transformers, voila! Throw in a minimum number of resistors and caps as required by the shortcomings of the transformers, load the second transformer up with a stereo pot, and there is your minimum DAC. Two channels of digital sound translated to analogue sound.
Even with so few parts, I found soldering a pain and wrecked a couple of (admittedly inexpensive) Crystal creepy crawlies. Unless you can solder well, buy a development board (usually less than USD200 except the Burr-Brown ones which are notoriously pricey), cut the links to the bits you don't want, then outboard the transformers. Andrea Ciuffoli, who most assuredly can solder, uses development boards, and sometimes commercial DAC kits, and on his site shows where to make the necessary cuts and links.
If you already own an ultra-high-definition chain and you're searching hard for even the tiniest margin of added value to your sound, my simple trx-8412-4390-trx scheme may give it you. I wish you luck. But if your system can still benefit from development anywhere else, spend the money anywhere else. In my opinion, at least, the DAC is the last place where you will get a decent return for your time and your money.
All text and illustration is Copyright © Andre Jute
and may not be reproduced except in the thread KISS xxx on rec.audio.tubes