Outboard DAC solutions
Scouring DIY audio forums proved beneficial as I found plenty of attractively priced DAC kits based upon some of the most revered DAC chips available. The availability of USB to I2S and S/PDIF converter boards opens the doors to team commercial outboard DAC solutions as well as complete DIY kit builds to provide a high-end sound experience. DIY solutions are still confined to 16-bit 44.1KHz-48KHz resolution via USB due to hardware/software limitations in the DIY small scale market. It is expected that 2009 will see the availability of 24-bit USB to I2S and S/PDIF chips while Vista should provide native output support for these formats. For now, 16-bit resolution and two-channel playback is adequate for most audio seeing as practically all of our music is recorded with CDs in mind.
Two kits stood out from the pack and present complete solutions that are available in either a modular approach or a single PCB with all components in place to convert data from the digital realm to an audiophile grade audio output. The first is the Doede Douma USB to I2S converter. Doede's solution uses the Texas Instruments 2707 USB to I2S and audio out chip. The venerable TDA1543 is the centrepiece of this kit, favoring a purist mindset as it does not use any kind of oversampling of the digital signal. Audio output from the TDA1543 is configured to use a minimalist passive I/V approach with a final output of 0.6V, which may be a little low for power amplifiers that require higher input voltage to reach full output power. PCM2707 audio output is also available if required, although most people put its subjective abilities behind the TDA1543. Pricing options start at the €49 for the basic kit, while dropping an additional €29 will allow you to experiment with a low jitter master clock based upon a 12MHz VCXO from Tentlabs.
The best two features on this particular kit (aside from cost) are the asynchronous clocking that frees us of the sonic confines imposed by CPU and system loads at the USB end of things. Secondly, Doede does not rely on the USB bus to power any of the circuitry on the PCB. This leaves us with the chance to explore various low noise linear PSUs and batteries with relative ease. Either way, the sonic performance should be in excess of designs that rely solely on internal PC power. Also worthy of a mention here is that the I2S output allows us to try out various DAC kits in the future. For example, we could use the Doede kit solely to provide the USB to I2S conversion and proceed to add a separate I2S input DAC kit as an alternative to the TDA1543. This element of flexibility is especially important to me as I already have the outboard tube stage that I used to provide the audio zest from the Pioneer player. Most modern DAC chips available today feature balanced voltage or current outputs. In my case, a balanced voltage output DAC chip allows me to transplant the tube based audio output stage and enjoy its sonic merits at my leisure.
Doede's kit is available either pre-stuffed and ready to go, or with all of the thru-hole parts loose, leaving you to stuff the PCB and do the soldering. As the PCM2707 is an SMT device, it comes pre-mounted and pre-soldered regardless of the kit option you choose. I decided to purchase the kit without the thru-hole components soldered in, as I felt confident enough to tackle this element of construction myself. I must note that there are no part-by-part instructions supplied with the kit. However, basic instructions and pictures on the DDAC homepage are clear enough to provide a good point of reference should you wish to confirm device orientation or placement in some situations.
If you've never built or soldered a kit before you're best advised to go for the pre-built option at additional cost. Either way, you'll have to provide a suitable case to house the kit. Do make sure to buy a case large enough to house additional PCBs for discrete regulated power supplies or additional DAC boards should you get the bug to upgrade things later. Audio output from Doede's module is configurable via the TDA1543 or from the output of the PCM2707 chip itself. In either configuration, the audio performance is quite good, although the TDA1543 is clearly the better performer in most situations (there will be more subjective stuff later).
Facebook
Twitter
RSS
114 Comments
View All Comments
johnYks - Monday, December 1, 2008 - link
Hello Rajinder - Excellent articleAnyone interested in pursuing similar projects
is advised to go to
DIYAUDIO.com and read the thread in the speaker section
titled A How to for a PC XO
http://www.diyaudio.com/forums/showthread.php?s=&a...">http://www.diyaudio.com/forums/showthread.php?s=&a...
Another excellent source of information can be found at
AUDIOASYLUM.com under the section Digital PC audio
http://www.audioasylum.com/cgi/etv.mpl?forum=pcaud...">http://www.audioasylum.com/cgi/etv.mpl?forum=pcaud...
look for the following threads
cMP the open source memory player
http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&am...">http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&am...
and cPlay the open source high end player
http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&am...">http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&am...
Rajinder Gill - Monday, December 1, 2008 - link
Thx John, excellent links.CSMR - Monday, December 1, 2008 - link
I'm afraid the author is just the typical novice audiophile who has been taken in by an unscientific way of thinking; it's a shame to have an article like this on AnandTech.It touches on interesting issues but without proper discussion. (Discussion of DRC with no mention or use of complementary/alternative use of room treatments.)
-Producing an analog line out with fidelity well beyond the range of human hearing is no problem these days. You don't need a fancy solution. There is no need to fiddle with I2S, or external DACs at all. You have semi-pro sound cards with analog out well beyond 16/44 fidelity.
-kmixer no longer exists. There is no need to discuss it in a new system. It's a particular flaw of Windows XP, not Vista.
-In a review of an audiophile computer system it is irrelevent to list amps/speakers/etc because the computer has only has three requirements - producing good analog out (very easy), being quiet (much harder), and good software (harder).
-But if you do mention these things you should choose good ones. Tube amps are expensive, power-hungry, and often distorting. If you want any sort of alteration in the signal you can use DSP on the computer system, it's one of the advantages. And just use a good technically competent amp, with enough power, low output impedance, good s/n, etc.. Class D is today's technology; where have you been the last 5 years?
-The computer is badly designed. Apart from unnecessary complexity (a simple pro-audio sound card should suffice), the computer is not designed to be silent. X48 is not the appropriate choice for a silent PC, and other choices are also bad.
-The author is well-read like most audiophiles but has learned from the wrong sources, namely other audiophiles rather instead of people who know about audio engineering, and so is full of misinformation despite his enthusiasm.
mindless1 - Monday, December 1, 2008 - link
Your opening and closing statements are a bit ironic for having so many false conclusions.1) Analog out with fidelity beyond the range of human hearing is irrelevant, fidelity within it is. It is a fact that even switching an ompamp alone changes the sonic signature... anywhere in the chain. Also, there is no need for "semi-pro" sound cards, some of those are no better than a properly designed non-pro card. Pro tends to mean features and software, not basic audio quality beyond the cost of the DAC if talking analog output but why are we talking analog output? That is insane, the sound should leave the PC digitized still until a DAC local to the power amp converts it. No amount of money you spend on a pro sound card with analog output and no amount of modifying that card to make it even better will equal with an off the shelf reference DAC design will achieve by simply getting the analog subsection out of the PC and keeping cable losses and noise pickup minimized.
2) It is relevant to list speakers and amps because they can and do color the sound, and provide a reference of whether someone can or can't hear something due to there being no difference or that they weren't changing the problem part or weakest link.
3) I agree, tubes are a personal preference probably best left out of an article but to replicate with DSP is a bad idea if tube-like sound is the goal, it never sounds the same. It is not enough to just use a technically competent amp with all the paper specs the same, plenty of amps with very good s/n ratio that sound difference can demonstrate this. s/n ratio is an incomplete method of describing audible differences and often fails in the resolution of the sampling and order of harmonic distortions present. Class D? You must be joking that being "today's" tech is important. Good old class A biased transistors stomp all over class D in everything except efficiency. Remember, it's not about novel ideas, it's about what actually sounds best... and to the author that might even be tubes.
4) About the computer, a X6800 CPU, 4GB memory, and Vista are all ludicrous overkill. Probably a Celeron 500 with 64MB memory would be fine, but claiming the X48 chipset system is loud based on some arbitrary assumption about heat without consideration of cooling design and that it's many times higher in performance than needed? The system could probably be underclocked and undervolted as far as it can go then every fan can be unplugged if you feel so strongly about that.
5) You claim to know audio engineering. You may know about integrated circuit engineering, how to plug a chip in and follow a schematic, or you may not - this is a different tangent, but you don't know how the variables interact to change the sound in ways that are DBX audible. In some cases one change alone may not even be audible but the sum of several changes are. You might not feel it if someone dropped a grain of sand upon your head but what if it was a bucket full?
CSMR - Monday, December 1, 2008 - link
Thank you for the reply.1. I suggested semi-pro level card because they produce good analog output and maybe more importantly have good and flexible drivers which can be used for DSP and audio routing, and because they can be relatively inexpensive by audiophile standards ($100-$300). I haven't studied the range of consumer cards but if they produce good enough analog output and you don't mind using 1/8" jacks which can be unreliable, then they are a good option.
I didn't suggest modding or spending a lot of money because it's pointless if all you need is good analog output.
2. But an a review of an audiophile computer need not involve how it sounds! Relevant factors are price, noise, analog out measurements.
3. OK class D and A compete at the high-end but for budget amps you can get very good class Ds and not class As, and mostly I would recomment budget amps because other factors (speakers, room acoustics, DSP) are much more important.
4. You can't undervolt enthusiast components to low power levels, and yes they can be cooled quietly but it becomes harder and more expensive with no actual benefit, so what's the point? Undervolting and underclocking only works to some extent on some components btw. Often only the processor.
5. Maybe changing the opamps has a measureable effect then. The study I posted concluded that a combination of ADC+96/24->16/44 downsampling+DAC resulted in no audible effect, and none of these components were high end. I would conclude that all of these steps was essentially perfect, as far as hearing is concerned. Yes, maybe combined with other effects it can make a slight difference. So bump up qaulity (s/n, thd, frequency response) by a few db by using a good card and just a DAC. Now the effect is not a grain of sand but a fraction of a grain of sand. Whereas speakers, DSP, room acoustics have HUGE effects.
phusg - Tuesday, December 2, 2008 - link
Guys, feel free to discuss what you like, but personally I'd say discussions about speakers (important as they are to the sound you hear) are pretty much off-topic for a PC hardware site such as anandtech.Another interesting discussion which I missed was analog or digital out of the PC. Sure going digital with an outboard DAC is a reasonable way to go but plenty of sites cater to that route, and it doesn't have much to do with home computing anymore. I would like to hear more on the analog option.
The really interesting bit to me is the DRC software. Thanks for opening my eyes to this. Unfortunately not testing the open source ACXO software because you couldn't install or find a Windows XP system, but still recommending it, is a bit lame. Now I'll have to install and test it on my own system, which is exactly what I want to avoid by coming to a site like this.
RamarC - Monday, December 1, 2008 - link
i'm not sure all of your criticism is warranted but i do agree that this article is very out-of-place on anandtech. with the somewhat exoctic amps and speakers, it should be on a home theater mag/website or some other speciality audio site, but not my "source for hardware analysis and news".aapocketz - Monday, December 1, 2008 - link
RamarC has pointed out something that struck me as well, why is this posted on anandtech (not the blogs even). Its somewhat out of place. Also I think CSMR has a point, it seems a bit like cargo cult science to me. Next we will be talking about gold foil CDs and premium HDMI cables and stickers to get your cell phone better reception.whatthehey - Monday, December 1, 2008 - link
I'm afraid the above poster is just the typical arrogant troll who has been mistaken in thinking his opinion is somehow more important than yours; it's a shame to have post like this in the AnandTech comments.It touches on some issues but without meaningful discussion. (Discussion of flaws with no real proof of how they're wrong, other than to suggest much more informaiton is needed.)
-Producing an analog line out with fidelity well beyond the range of human hearing is no problem these days. Obviously. You dimwit, Rajander! How dare you suggest that you can actually hear a difference, when clearly it's a trivial problem solved by sound engineers five years ago. Didn't you know that a $150 sound card is all you need?
-In a post in Anandtech comments, it's tradition to bitch about the choice of components. God forbid anyone use what is available instaed when he admits the core aspect (analog out) isn't that hard. You should never use X48 in anything even if that's what you have, because other better options abound. See his list of good alternatives and clear, concise explanation of how they're superior.
-Where have you been the last 5 years? All you need to do is plug in a semi-pro sound card and all your problems are solved. Just look at all the testing and results he's compiled in the above post!
-Your software choices also are terrible. Again, see his list of superior alternatives as constructive criticism.
-The poster is opinionated like most readers but has learned to believe in a subjective audio article exploring a new subject, your opinion is wrong and his is right. Who in the hell listens to any audiophiles when all you need is some people that know about audio engineering? Because obviously he does!
In short, despite his obvious disdain for your treading in his territory, he contributes nothing to the discussion other than to piss on your ideas and those of all other audiophiles, without any clear indication that there's a better way.
The only real problem I have with articles like this is that they're so subjective... and they cause the opinionated trolls of the internet to surface and tell you you're totally wrong and the article should never have been posted, but without any clearly better alternative. It seems like this might become another area similar to digital cameras if you don't back the f*** off, because you should just listen to audio engineers. Don't trust your ears or anything else; trust the guy posting anonymously without any links to back up his assertions.
Personally, I only know enough about audio to say that sound mixers and the like use all sorts of different mics, speakers, DAC/ADC, etc. because it really does make a profound difference. That's why some albums come out sounding amazing and others suck; it's an art form rather than engineering. God forbid engineers be put in charge of anything where creativity plays a role. FWIW, I'm sure the final result is highly pleasing to you, and that's what's really important. Now I'm just going back to my boring old audio setup that does what I need it to do (because I don't really listen to music much).
CSMR - Monday, December 1, 2008 - link
-Analog fidelity is measurable and there have been tests done to show how far human ears can hear.One example here: http://www.aes.org/e-lib/browse.cfm?elib=14195">http://www.aes.org/e-lib/browse.cfm?elib=14195
A 16/44 ADA loop inserted into a hi-res audio stream was inaudible to a group of pros and audiophiles with good equipment. Good semi-pro audio cards operating at 24/96 will have measured specs better (s/n, thd) much better than ideal 16/44.
So your first point which you meant to be ironic is in fact true; the irony is on you mate.
-The X48 and other chosen components have high power consumption. These components are good choices for some uses but not others. It is perfectly fine and normal for tech-savvy posters on anandtech including myself to point out that a combination of components is not optimal.
-I had no problem with the software choices. The DRC software I am sure is a good piece of software.
-There is in fact research done into audio production. Audio engineers design all parts along the recording and playback chain and it is not all magic, at least for those that aren't involved with marketing to audiophiles.
-Yes various choices in recording are important but not those of DAC/ADC because these components are so good these days. Any decent DAC/ADC will be indiscernable from any other. That is not true of mics, mic preamps to a lesser extent, and styles of mixing. And also speakers. These are places subjectivity is inevitable and important.