Bit Perfect, Sound Cards, and Windows 7...

[Paul R]

I know little of what I speak, so if this is a silly question, please excuse it.

 

I am going to play this weekend with setting up a PC as a music server again. I seem to have very little luck with PC's, but hey, I'm gonna give it a try.

 

I don't entirely understand why people would use a "sound card" for "bit perfect" music. I mean, wouldn't just hooking up to a DAC, either through USB, Firewire, or a digital out connection from the built in sound card just pass pure digital data? If you are sending an uncompressed stream, such as an AIFF stream, would it not by definition be "bit perfect?"

 

Now, if the DAC upsamples it, I guess that no longer applies. For example, I am going to hook the PC up to a DacMagic, which will up sample but has never, at least to my ears, done any damage to a song.

 

I was thinking that if I connected the DacMagic via USB, it would appear to the Windows 7 PC as a soundcard anyway... right?

 

I'm going to install the James River Media Center software, though to be honest, all our music is stashed away in iTunes. I suppose I can copy over some to test with.

 

By the way, when iTunes sends out lossless or uncompressed audio, over a TOSlink or Coax connection, isn't that "bit perfect", so long as the music was ripped from CD with a lossless or non-compressed codec?

 

Like I said, these might be silly quesitons, but I would sure appreciate some education.

 

Thanks

-Paul

 

P.S. using Windows 7 on the machine - should I load Home Premium or Professional?

 

 

[The Computer Audiophile]

Hi Paul - Good questions. I'm sure you're not the only readers who wants to know this information. Have you read my Windows 7 & J River Media Center article? It may help give you more information similar to your questions. Here is a link -> http://www.computeraudiophile.com/content/Windows-7-Audio-J-River-Media-Center-14-Configuration

 

Not all digital outputs are built the same. They vary in quality depending on design and implementation. Some require special software/drivers to function while other can use native operating system drivers. People use sound cards because they like the sound of them and some have very advanced features. Dual wire AES and word clock in/out are two features a good sound card like a Lynx or RME can provide.

 

Don't confuse the file type, AIFF, WAV, M4A, WMA, FLAC, with how the audio is output. Any lossless file type can be output bit perfect or can be mangled and output nothing like the original. The critical piece is how the playback application handles this digital output.

 

Yes, a DacMagic will appear in Windows 7 like a sound card or audio device.

 

iTunes is not the best application to use on a Windows computer. Its tough to get bit perfect audio from iTunes on a PC. Macs are a totally different story.

 

I recommend Windows 7 Professional because it allows one to connect to the music server via Remote Desktop. If you don't ned this feature then pick up Home Premium.

 

[Paul R]

Without a doubt I prefer the James River Media Software sound over iTunes when coming from the PC, especially on any material above 24/48. Some of the user interface was beautiful, as you mentioned in your review. I was unable to get one of our iPods to control JRMC 14 in an acceptable manner though. I guess I am just not so patient as you are!

 

The "zones" capability was nice, and so, being the nutcase I am, I tried it with a small, headless, thin client device. It worked, but to get acceptable sound, I had to connect a USB based sound device. The onboard audio was - eh - horrible.

 

With a USB sound card, the device could handle standard definition material just fine, but high-def material overwhelmed the poor thin client.

 

We bought a copy and it will be installed for more "critical" listening experiences.

 

Here is the thin client I used if anyone is interested. http://www.10zig.com/thinclient/xpthinclient.php

 

I think perhaps a small under $200 device to remote music at quality is what is needed. This would work for Pure Music as well, using their NetSend based protocol. Have to try that.

 

Question: assuming issues of network latency, jitter, etc. could all be worked out or minimized, would something like an Audiophile Grade Airport Express be worth doing? It can probably be done with some of these little thin clients, though I expect the output signal from 'em would have to be digital only to hold the price down. A decent DAC would add a great deal of cost - and nobody would agree on the one they liked anyway.

 

-Paul

 

 

[JeffH]

Chris, Sorry, but I'm just learning too. I would like to ask you a little more about this statement you made:

 

Any lossless file type can be output bit perfect or can be mangled and output nothing like the original. The critical piece is how the playback application handles this digital output.

 

This is one thing I have been trying to understand since I joined a few weeks ago. I understand some use sound cards like the Lynx for the features(Dual wire AES and word clock). But, if the playback application is the critical piece, why can't the built-in S/PDIF coax output on a PC's motherboard be used to provide a bit perfect data stream to an external DAC. Assuming, of course, that I am using a playback application that is configured correctly. Seems like most people use a high end sound card or USB but never the built in S/PDIF coax.

 

Thanks,

 

Jeff

 

[dbdog]

One potential benefit of higher end cards is the AES/EBU connection which seems inherently better:

1. It is 75ohm shielded wire throughout it's entire length

2. It send the signal at I think 10X the voltage as S/PDIF thus reducing the potential for noise to corrupt the signal.

 

Can anyone verify or correct these point?

 

Cheers

 

[JeffH]

As I said in my previous post "I understand some use sound cards like the Lynx for the features(Dual wire AES and word clock)" I was hoping Chris or some of the other regulars here could help me understand why it is not a good idea to use the the built in S/PDIF coax port on the MB.

 

[The Computer Audiophile]

Hi Jeff - I've never hear a built-in S/PDIF coax output sound anywhere near as good as an add-in audio card. This doesn't mean it's not possible, but I've never experienced it.

 

[elcorso]

In understand that AES/EBU was designed for professionals use, and S/PDIF for consumers.

I do remember Levinson first expensive consumer DAC was the first to use AES/EBU in a consumer product.

As I know tech specs for both:

AES/EBU (AES3) could accept a maximum cable length of 100 meters, versus 10 in the S/PDIF

AES/EBU runs in 4v logic, versus 0.5-1v in S/PDIF, for this reason, AES is advantageous in very long runs.

AES/EBU is 110 ohm, S/PDIF is 75 ohm.

 

Coax (unbalanced) is advantageous over balanced transmission for two main reasons:

 

1. Balanced cables, as in AES/EBU, has much higher capacitance and higher impedance which will round-off digital transitions (aka jitter).

2. Coax, as in S/PDIF, has better shielding to protect from EMI.

 

For multichannel and long runs, I'll use AES/EBU. For two channels and 'normal' runs (5 meters), as my case, I'm happy with S/PDIF for 24/192, and even with Toslink (optical), if the receiver DAC has a a properly designed clock recovery system. Optical also offers protection against ground loops between different pieces of equipment.

 

Cheers

 

[Audio_ELF]

I think the first level of improvement using a dedicated sound card over the SPDIF on the motherboard is in the clocking.

 

Altmann Micro Machines recently published an article on their take on the situation (scroll down to "True Audio Clocks" section. To summarize the article, he basically says that good PCI add-on cards have 2 fixed clocks which are multiples of the 48 and 44.1 base rates, where as a motherboard just derives the clocks required from the existing computer clocks. The former making the frequency more stable.

 

Eloise

 

[MR]

I am building two music servers, one for a "reference" system at home and one for the lake. The home system uses a Lynx AES16e; the lake system uses a popular PCIe/1x card with S/PDIF (RCA jack) output. While I have not listened to the motherboard outputs, I have had a chance to A/B compare the two soundcards installed in the same server.

 

There were marked differences that could be due to 1) slot in the PC; 2) cables to the DAC; 3) different inputs on the DAC; and 4) the sound cards. Thus, what I am about to say cannot reliably be generalized except to say that the items in the list above can make a difference.

 

The S/PDIF card, cable, and input on my DAC were very detailed; the presentation was somewhat forward and aggressive, however. Sibilance, particularly on female vocals, was slightly exaggerated. My speakers don't do anything to hide this artifact. I wanted to turn the volume down a little.

 

The AES/EBU card, cable, and DAC input offered a more laid-back sound. The toughest trick in audio is to preserve the detail without becoming pushy, but this path did it, including a warmer, deeper bass. I wanted to (and did) crank 'er up.

 

The point is simply that every element in the playback chain can make a difference. The only generalization I can make is that the sound cards sound different, be they bit-perfect and all; I suspect the MB outputs would sound different as well.

 

I listened to the recommendations here for the Lynx card and I am glad I did. The new server beats the pants off of my CD transport (AES/EBU) which will most likely soon have a new home.

 

[Audio_ELF]

MR ... what DAC were you using to compare the two digital signals? Can you say what the other sound card was?

 

I think it depends what level of DAC you have - to me it would be rather a waste to spend £600 on a card for a (for example) Cambridge Audio DAC Magic DAC at £225. A bigger improvement would be seen spending £700 on a higher end DAC and £125 on the ESI Juli@ for example.

 

Eloise

 

[earflappin]

Elcorso, excellent post. My experience echos yours - I prefer SPDIF BNC as an interface now for my DAC after starting off with AES/EBU.

 

I want to thank Gordon Rankin who posts on this forum for taking the time to educate me on these differences between the AES/EBU and SPDIF BNC interfaces. I think there are many who may believe AES/EBU is better than SPDIF for all applications since it is balanced, runs at higher logic voltage and was designed for "pro audio".

 

As I understand it, AES cannot be 110 ohms and since this is a transmission medium it will not perform optimally. Second since the AES specification runs at 5V and at that level with speeds up to 12.288MHz the drivers will have to be very powerful which means they will typically overshoot and cause reception errors. BNC SPDIF on the other hand is always capable of 75 ohms impedance and since the level is only 0.5vac then the drivers are only required to push about 1/10th the power.

 

AFAKI, AES/EBU was pushed by pro audio due to their need to routinely deal with long cable runs and their preference for defacto standard balanced XLR connectors. Assuming a DAC has equally well designed AES and SPDIF input circuitry, I think audiophiles with short cable runs are likely to find SPDIF BNC will be sonically superior.

 

[MR]

I think your comment about putting money into the DAC first makes perfect sense.

 

Since I am not a reviewer, and since I think differences can as easily be attributed to other parts of my system as to the sound card, I don't feel comfortable naming the card. That is, I have not been thorough enough to rule out everything else but the card, so I don't think it's OK to seem to bash it. It is very musical, and in the other system, which has less resolving power, it will sound great. All I have to offer is an anecdotal observation: different stuff, post-transport, often changes the sound. Many such changes are complex: better in one respect and worse in another. The key elements seem to be timing (jitter) and the DACs themselves.

 

The DAC is a Theta Casablanca III (Xtreme 96K DAC cards) in two-channel mode with no filtering or DSP. This is better than my balanced Theta Gen Va and probably a bit inferior to the Gen VIII.

 

[earflappin]

MR, I've been down the same path as you and unless your AES16e and S/PDIF sound cards have identical levels of quality in terms of their on-board clocking and noise isolation, then part of what you were hearing were these differences in addition to the sonic differences of AES vs S/PDIF....this assuming, of course, that the DAC you used for your comparative testing has equally good AES and SPDIF input circuitry. The reason the Lynx AES output beat your transport's AES output was likely for the same reason.

 

I own the Lynx AES16 PCI card and use it as an interface for my DAC. The Lynx has very good on-board clocking, but I have been able to improve sound quality by using an external word clock and a noise isolator. That being said, when I recently tried a leading async USB-to-SPDIF interface it was better in every way than the ext clocked Lynx.

 

Lastly, if you want to hear S/PDIF at it's best you need to use BNC connectors at both ends. BNC will give you 75 ohms whereas the RCA connectors will not. This both AES and S/PDIF are transmission mediums impedance mismatches will impact sonics in sufficiently revealing systems.

 

 

 

[MR]

I agree that when the clocks are right the music is much more likely to be right. I do not propose that either interface type is always better than the other. The differences included 1) the cards; 2) the interface type; 3) the wires; 4) the receivers in the DAC; 5) the motherboard slots. Couldn't say without a lot of testing which of them caused the difference. My experience in audio, however, suggests that they all did, probably some for the better and some for the worse; given what I paid for the Lynx card, I guess I hope the card was a critical factor. Remember AT&T glass?

 

[MR]

As with everything audio, extended listening shows something-lost, something-gained (if we're lucky). I am trying to settle down a music server in a relatively high-end rig. Given what I am hearing, I am tempted to just jump off the Lynx and on to the Legato, but I kind of want to take it in steps and learn something in the process. It takes me a lot of listening to come to terms with what I am hearing.

 

My comments are with regard to my system/cables and their interactions and should not be taken as a product review or a conclusion using a review reference. I have a second system in which the Xonar card sounds fantastic.

 

I have been A/B'ing the Lynx via AES/EBU and a Xonar Essence via S/PDIF in the same machine to the same DAC. With the Xonar, I get a slightly congested soundstage and a sort of fat, bloomy mid-bottom.

 

The Lynx card sounds ever-so-slightly etched on snare drum/symbols; symphonic brass get a little buzzy; there is a touch of sibilance on vocals. The soundstage is expansive, precise, with plenty of ambience/decay. The etch is enough for my brain to say "not-music," however.

 

I use Windows 7, WASAPI, and the direct, exclusive connection (output 1/2 on the Lynx), S/PDIF passthough on the Xonar.

 

I can't hear a difference between Foobar and J River media on either card. I am going to try the Black Lion external clock, since what I am hearing from the Lynx may be jitter.

 

If the word clock doesn't settle things down, I'll try the Legato USB. I'd sure be interested in any comments or suggestions regarding the word clock, since I know a number of people on this forum use it. Common sense suggests that trying to convert the Lynx signal to S/PDIF won't help, but in audiophilia, common sense isn't always useful.

 

 

[The Computer Audiophile]

Hi MR - Does your DAC have a word clock output? I've picked up some sonic gains by sending word clock out to my Lynx card from a DAC.

 

[MR]

No, no DAC word clock out or in.

 

Part of the problem is that I fell for software attenuation; I de-installed and re-installed the drivers and set all of the gain controls to unity. The sound improved a lot; the buzzy horns sound like brass, and sibilance is reduced. Still don't have all the body on voices or violins (things are a bit too bright) but the raggedness is much reduced. The edges are now much more cleanly-cut. I understand better why the manual says to add dither if the mixer controls are used to reduce gain. Better to set unity and skip the dither.

 

If it doesn't sound like music when listening from the next room over, don't bother sitting in the "sweet spot."

 

I have a Black Lion clock, waiting on the cable from Redco.

 

 

 

[xiddox]

This is really interesting. By way of background, let me say i am both new and not new to computer based audio. I have been using a Slim Devices (now Logitech) Transporter (and a couple of squeezeboxes) for the past couple of years, with my music served from a Freelinked Buffalo Linkstation Pro NAS running squeezebox server. As most folks will know, the squeezebox family of devices accepts audio in packets via Ethernet (or Wi-Fi) and as I understand it this essentially eliminates jitter - or at least the jitter is fully internal to the squeezebox/transporter, since the bit stream generator and DAC are in the same chassis and presumably on the same board.

 

About 2 weeks ago, I started thinking of upgrading from the transporter to something like the Ayre QB-9 or Bryston DAC. I have browsed Computer Audiophile from time to time, but just reading reviews, not really hunting for technical information.

 

At the same time I have decided to build my own HTPC, and so it made sense to me to move the music server component from the NAS to the new HTPC, and possibly to sell the transporter and replace it with the HPTC streaming data to a standalone DAC.

 

And so i started reading Computer Audiophile to see what Chris and other folks were recommending. What struck me immediately was the same point raised by Paul in this thread: i had expected that i could use any (sp/dif) standard motherboard output, and would have no need for a dedicated sound card. Yet Chris (and others) had indicated that the sound card made a difference to the bitstream, even recommending a ~ US$600 sound card for this purpose!

 

After the shock wore off, I started hunting for some explanation and was amazed to find that there was none - or none that i could find - until i stumbled onto this thread! As Chris indicated in his first comment in the thread, many will be wondering these questions, and yet - with all due respect to Chris, who does a fantastic job with this site, but he didn't really answer Paul's question, stating only that the interfaces vary in quality.

 

It struck me that it would be worthwhile to have a FAQ or dummy's introduction to Computer Audio with answers to questions such as:

 

Why do I need a dedicated sound card when all i want to do is send digital data from my PC to a DAC?

 

How can different sound cards possibly make a difference to sound quality?

 

Why does it matter what software I use to play my music files when I am outputting digital data?

 

We have some good contributions above to answer the first two questions. As i understand digital data: there are 2 aspects: (i) is the data sent without modification - i.e., is it bit perfect, and (ii) how much jitter is in the bit stream?

 

Now i would have though that even a cheap motherboard would (or could) send bit perfect data, but I accept that the jitter from such a device is probably very high. Mind you, i would have expected that a high quality DAC could reclock the bitstream and hence effectively eliminate the jitter, but i don't see that mentioned anywhere here as an option...

 

In this thread we have some interesting discussions of SP/DIF vs AES/EBU and Eloise's very informative comment about dedicated clocks on sound cards (which again reinforces the jitter issue in my mind).

 

But we also have evidence that the sound quality of a Xonar essence card (which i assume has dedicated clocks) is inferior to the Lynx card. One imagines that again this is jitter? The answer is not clear. I confess I find that rather frustrating because one of the things that attracted to computer audio was the idea that the digital source would be straightforward and low cost... now it seems that is not the case. I though opayind $200 for a Xonar card was bad enoiugh when all i wanted was diogital out ... now it seems i need to pay $600 for a Lynx card? Makes me wonder about keeping the transporter or getting a squeezebox touch and connecting it directly to an external DAC... but then how much jitter would there be...?

 

After the surprise of digital output hardware, i then read about the effect of software, which is perhaps even more of a shock. Now i accept that normal audio through Windows (for example) is normally mangled is various ways before it is send out through the audio device. But surely all that is required is to bypass all that processing...? I greatly appreciate Chris' efforts in his detailed account of J River/Windows configuration to get bit perfect output. All well and good.

 

But then I read about the $$$$ Amarra software and here we go again having spend yet more $$$ to get good sound. Huh? Surely if we have bit perfect output from JRiver, then that is all that is required. How can Amarra possibly be better that bit perfect? Again, I accept that if one wants to do digital volume control, then I can accept that the folks at Amarra have put a lot of effort into that. But if one wants to use a preamp for volume control and hence wants to send unmodified, bit perfect data to the DAC, do I really understand correctly that Amarra supposedly has some magic that improves the sound? Or is it only for volume control that Amarra has any advantage over, say, JRiver?

 

Sorry to ramble so much, this stuff has been doing my head in over the past couple of days. I can only imagine that others may be similarly confused.

 

Finally, thanks so much to Chris for the fantastic site, and all of the forum members for their helpful contributions!

 

- Richard

 

[jzexport]

Richard,

 

I couldn't have expressed my confusion on these as well, if I tried.

 

I, too, don't understand why a lot of this add-on stuff, e.g., soundcards and software, matters.

 

Worse is when the explanations are, "But, I hear the difference." IMHO, too much of high-end audio is based on suggestion, belief, and placebo.

 

I was looking forward to the convenience of having my CD collection on a giant juke box, but am having second thoughts on the cost, complexity, and potential insecurity of the quest.

 

I hope more members will come forward with explanations for why these add-ons matter.

 

Josh

 

[The Computer Audiophile]

"I was looking forward to the convenience of having my CD collection on a giant juke box, but am having second thoughts on the cost, complexity, and potential insecurity of the quest."

 

 

 

Hi Josh - You still can accomplish everything you were looking forward to with a music server. There is no need to to spend a lot of money or increase the complexity. If you want to push boundaries with a music server then cost and complexity will increase. It's like buying car tires. The standard models work wonderful for 99% of people. The very good models tend to be more expensive and may add complexity because of lifespan and what conditions are best for using them. 99% of people would have no clue is the tires were swapped out in an A/B fashion.

 

I will say there are many measurement that support using cards like the Lynx or a sophisticated asynchronous USB DAC. Many readers here on CA love the sonic benefits of these add-on components.

 

I would also like to ask you a rhetorical question. Answer if you want but don't worry. I'm not trying to push you or start a debate :~)

 

You said, "I, too, don't understand why a lot of this add-on stuff, e.g., soundcards and software, matters."

 

Why wouldn't these pieces to an audio system matter? The common response from many people with the same viewpoint is, "Digital is digital, it's either on or off, a one or a zero." However, this is a layman's perspective and doesn't consider that every one of these pieces to the puzzle touch the audio during playback. Each one has it's own weaknesses and strengths. Audio cards for example can add noise and jitter that degrade sound quality. Also, an audio card like the Lynx supports every sample rate from 16/44.1 to 24/192. Many on-board audio chipsets will not support anything above 24/96. Even fewer support 24/176.4 kHz.

 

Again, no need to increase complexity or cost. It's more of a choice to try to squeeze every ounce of sound quality out of a system.

 

Thanks for the post :~)

 

 

 

 

 

 

 

 

 

 

[jzexport]

OK, call me a layman, but I do believe that if the bits coming off the storage media show up at the input to the DAC "bit perfect", that's all that can matter. It shouldn't matter if they came from the MOBO or a soundcard. It shouldn't matter if the media player was iTunes or JRiver.

 

I'm no jitter expert by any stretch of the imagination. But, my layman's understanding says that's what buffers are for. Fill the DAC's buffer with accurate bits and let the DAC clock them out nice and uniform.

 

All of this presumes digital out from a server to an external DAC. I agree totally that soundcards can matter if you want them to do the DAC. I also agree that media players can matter if you want them to do signal processing.

 

The practical matter for me is specifying a system. I've been looking at A-Tech's offerings. They offer a case that allows for a soundcard and one that does not. I suppose paying an extra $180 to keep the soundcard option door open should be worth it, but it is hard to understand why that should ever be necessary, assuming I will never want the soundcard to be the DAC.

 

Josh

 

 

 

 

 

 

[Paul R]

It *does* sound different, at least to some people. While placebo effect can, and almost certainly does explain some of differences people hear, it is a very long way from explaining all of them.

 

In short, there is a very high probably that there is an audible difference between even completely digital components.

 

There is a lot of evidence that "bit perfect" is, as Barry, or Clay, or Chris, or someone said - the starting point for digital sound, not the final goal.

 

Truthfully, what you should probably consider doing is starting with what you have, then trying other components until you reach a level and quality of sound you are happy with. At least for the moment.

 

I personally recommend just starting with iTunes. It is hard to beat for the pricing and quality. It also sounds good, at least to me.

 

-Paul

 

 

[Audio_ELF]

I think a lot of this depends on the level of equipment and your expectations...

 

If you are coming from the lower end of HiFi, I'm willing to bet that a HP Mini 2110 netbook connected via USB to a Cambridge Audio DAC Magic (or similar) running iTunes under Windows (or better J.River) will put up a good fight against most £5-700 CD players - straight out of the box and tweeks (sound cards, etc) will make little difference.

 

On the other hand, Chris' "reference" MacPro with Lynx card feeding a Berkerley DAC can be put up against most £10-15,000 CD players - but will take a lot more experience and tweeking to achieve. It's not that the second is more suseptable to things like jitter, but the higher hardware will allow you to hear the nuences more.

 

Eloise

 

[jzexport]

I am afraid starting with what I have will not get me very far. An old Dell laptop with no way to get iTunes in digital format into my surround sound processor. I think, as I suggested in my response to Chris, that I should allow for a soundcard in my media server should I ever find I need it or want to check it out.

 

In short, there is a very high probably that there is an audible difference between even completely digital components.

 

Do you mean the input digital stream or some downstream component? If you mean the former, I would say, in the absence of hard evidence, and I admit being ignorant, that it is very low probability, rather than a very high probability as you say. I accept two difference DACs fed the same bitstream will sound different. One DAC fed the same bitstream twice will sound the same, unless it's incompetently designed, e.g., poor SNR, high distortion, etc.

 

There is a lot of evidence that "bit perfect" is, as Barry, or Clay, or Chris, or someone said - the starting point for digital sound, not the final goal.

 

And what might some of that evidence be other than "I hear it" or opinion? If people hear it, something must be going on. The original poster and I just want to know what it is. I'll settle for what it could be.