Computer audio (CA) refinement marches on- this really is another golden age of audio as the progress made every 6 months is breathtaking. If you have the desire and persistence to overcome a little difficulty in understanding the process and implementation of CA you will be very rewarded, especially if you have a high resolution system that is capable of revealing the extremely high information (and "better" music) of hi-rez, upsampled digital.
If anyone tells you that 16 bit 44.1Khz is as good as 24 bit 384Khz , well you can believe their outdated views or you can listen for yourself. But please don't do this half-cocked or half interested, the results of a fully performing CA system with good digital files is beyond belief, but it is not simple to get everything firing and working in unison to unleash the potential. You dont have to spend megabucks , but it is essential to look beyond what the mainstream is believing , do a little research, try a lot of different things, and most importantly believe your ears. You're better off spending an hour listening and tuning than 10 hours of reading opinions on the forums. There's too many people talking rather than doing, if someone cant give specific explanations of why they have their opinions , then the result of following their advice is at best, questionable.
So the following should also been taken in context, its my experience with my equipment and knowledge, but has given me an extraordinary increase in performance over the last few months.
DACS
I wouldn't consider any dac that doesn't work at 384Khz. Firstly it shows the designer has the expertise and knowledge to build something that is state of the art, future-proof and forward looking. Secondly the advantages of using state of the art software to upsample to 384Khz (and beyond) is now beyond doubt , even mp3 is considerably improved when upsampled to 384Khz by software. Upsampling, or oversampling in the dac (hardware) is a completely different process and imo has no real advantage. Software programs which can accomplish this are JRiver, XXX High End, PureMusic and Audirvana Plus.
Spdif is outdated and incapable of getting the best from digital- I'm presently using two dac systems: the EXA U2I which is a usb-I2s convertor, matched with an Ackodac board with internal 32 bit/384Khz Sabre dac chip www.exadevices.com and the EXA E18 dac which takes usb from computer and passes an I2S signal to the same 32 bit/384Khz capable ESS Sabre dac chip . www.exasound.com
Beware of I2S implementations- it is unquestionably the best way to pass a digital signal to a dac's input, but also very difficult to implement properly- there are many more bad I2S implementations than good .
The EXA E18 is an 8 channel dac which opens up the very exciting prospect of active crossovers in the digital domain- a very exciting direction.
The exa/ackodac/tube output combination is my current reference and is the basis for the Supratek dac of which I am building very limited numbers, 2 units only.
Prototype exa/ackodac board with one of those old crappy sounding CD for comparism.
Exa U2I usb-I2S convertor locked to 352.8KHz
DAC OUTPUT STAGES
I simply don't like the sound of opamp output stages in dacs, they can sound very good, but always have a signature that imo isn't as natural or dynamic as a good tube output stage.
Unfortunately there are very few good tube output stages, most digital designers have little tube design experience and the tube output stages are usually based on the same tired designs using boring tubes like the 12AU7, ECC88 ,6922 . These cheap and nasty designs are usually fed by inadequate power supplies and squeezed into the dac chassis with more consideration to making it fit rather than good sound.
Towards the end of the tube production era some really fine tubes were released with quite incredible performance characteristics, the 7788, E280F, D3a and others are capable of high gain, have lots of drive ability, low distortion and are really quite superb in tube output stages which need highly accurate and resolving tubes to pass on the hi levels of resolution available from digital now.
A ESS Sabre dac chip with a 6922 cathode follower tacked on behind it is a waste of time and money.
For my output stage I'm using the E280F tube (the 7788 and D3a can also be used) with a special loading/bias circuit and transformer coupled using Magnequest nickel core transformers. Capacitors are copper in oil or silver/gold in oil.
The power supply is fully tube shunt regulated , with a bare minimum of electrolytic capacitors, a large over-specified R core transformer, two power supply chokes for a CLCLC super quiet noise free HT supply.
The weight and size of this power supply is considerable and because of the size of the exa/ackodac dac board the dac and output stage are in seperate enclosures.
So why go to all the trouble of building a tube output stage that is virtually a state of the art preamp when most tube output stages for dacs are nothing more than a simple power supply, with a single dual triode tube in a very simple circuit dating back to the 1940's.
The simple answer is that a simple cathode follower or grounded cathode tube circuit is simply not good enough for the high resolution of the latest sota dacs. Just as you wouldn't use a 40's circuit for a high quality tube phono stage, a quality dac needs a very sophisticated and evolved tube design . A single 6922 after a dac is an insult to a quality dac chip.
Shunt regulation of the power supply is just one of the contributing factors to the amazing 3D imaging that a modern 32 bit/384Khz capable dac with tubes is capable of. Now we can hear depth imaging of very specific positioning- if a vocalists microphone is located 12 inches further back than another vocalists mic , and your system is capable of resolving it, you will hear it. Put one weak link in the system (a poor tube output stage) and it is lost in the mud.
Of course, if you want a "tube" sound that is romantic, warm and ultimately inaccurate, then one of the old 40's designs will suit you well- I still wouldn't use a 6922 or its derivatives the 6DJ8,ECC88 etc - it tends to have a steeliness in the upper treble that can be fatiquing when used for a dac.
It is quite amazing how many people prefer "tone",or "timbre" rather than dynamics, transperancy, 3D imaging and the other attributes of a well designed tube circuit.
This "warm" romantic sound is usually the result of a tube design with inadequate drive ability and you will often find a 3db accentuation to the midrange .
Seems a bit silly to be using something like this and worrying about perfect bit output, but the audiophile world can be very strange indeed.
COMPUTERS
A month ago I prefered the Mac OS over Windows7 for audio, which was a bit frustrating, as the windows based media player JRiver Media Centre has the best overall integration of file management and audio implementation , however macs always sounded more organic and natural to me- in the digital world that usually means less jitter. Then I read an article about optimising windows7 solely for audio operation.
It claimed that a brand new computer is severely limited in performance- it is set up for ease of use, rather than performance and optimising it as a music server makes a big difference.
I admit I was a bit skeptical, but bought a brand new Toshiba computer with windows 7, and followed the instructions , and was very surprised to see how throttled back it was. With everything set to "high performance" the change was very pronounced, with a substantial lift in sound quality, very audible, very worthwhile, a fantastic upgrade. Although I concede it is subjective I prefer it over the mac with PureMusic and Audirvana Plus.
So a computer set up as a transport is not really as good as it can be until optimised solely for audio.
Once done though, there's not a digital transport at any price that can compete with it.
Another great advantage of using windows7 and Jriver is that an iPad makes a great interface for controlling music from your listening chair- there's a very nice satisfaction from being able to have cover art, info and the ability to choose any album, genre, playlists or song from your listening position.
THE CABERNET DHT PREAMP WITH INBUILT TUBE OUTPUT STAGE
Putting the EXA/Ackodac board with a Supratek tube output stage into a single package proved impossible - the exa/ackodac board is 300x200mm , and with transformers, output caps etc would not fit into the standard Supratek copper chassis.
I decided to take a different approach and build the highest quality preamp complete with an integral tube output stage for an external dac. The preamp is a line stage using 6SN7 input tubes and 71A direct heated triode (DHT) output tubes utilising Lundhal amphorous output transformers, with a special circuit that minimises capacitor requirements. The audiophile fashionista's love to talk about exotic capacitors, but it is much more rewarding to remove as many capacitors as possible. Capacitors always produce filters , filters produce phase shift, and phase shift smears imaging.
The preamp chassis containing the line stage and the dac output stage has a total of 6 capacitors- compare that to the conventional preamp which can have as many as 30 or more capacitors.
This reduction of capacitors is much more beneficial than a chassis full of exotic caps- however i still use very good quality caps- Ampohm copper in oils, and Mundorf Supreme oils.
Supratek DHT preamp with integral tube output stage for dac (the two small tubes in middle)
EXA/ACKODAC BOARD
At 300x250mm the dac board is 3-4 times larger than the average dac circuit board and i decided to put it into its own enclosure, build is underway at the moment- the pix shows the board and transformers, and output caps on a prototype board - the enclosure will match the good looks of the preamp.
One of the advantages of this board is that a multiple of adjustments are possible, bandwidth, roll-off, inputs, outputs etc. These adjustments can be done manually with switches, or even finer control can be achieved by connecting to a computer.
DSD playback is automatic (and sounds wonderful) The best playback I've heard is native 352KHz wav files
from the www.2l.no download site. Half a Gigabyte for 4 minutes music, but what a sound, the closest to real music we have to date. If you download these free test files- make sure your dac can handle 325.8Khz- most cant.
With Supratek being famous for its tube phono stages you would expect I would be a champion for vinyl rather than digital. My Luxman PD444 with DV-505 arm, Expert Stylus re-tipped Ortofon SPU and Micro-Seiki RX5000 with FR64, Wilson Benesch, Expert Stylus retipped DL103's (highly recommended) are very good turntable rigs, and combined with my best tube phono stages sound very good, very organic, very seductive.
The problem is you know your also listening to wow, flutter, tonearm resonances, uneven cartridge response (no cartridge is perfect), loading issues, and an RIAA filter trying to replicate the inverse of a difficult filter curve. No matter how good a vinyl rig is it remains colored. It sounds damn nice , full of "timbre" but it is not a faithfull representation of the recording.
Digital done state of the art (and improving at a frightening pace) is as close as you can get to the real thing.
Whether you really want to hear the real thing or "timbre" is your choice.
Actually, you can have the best of both worlds- JRiver and PureMusic players have excellent 64 bit graphic equalisers, that do a very good job of fine tuning a system to "taste" . They dont output bit perfect, but do sound very good, and quite acceptable if you want to, for example, overcome a bad room reaction, and can be turned off and on in seconds.