AudioNervosa
Systemic Development => Digital Audio Devices => Hardware => Topic started by: Carlman on May 12, 2010, 07:32:27 AM
-
So, I've enlisted the help of my friend, Nick to build a new audio PC... Goals are simple: be the best transporter for flac or wav files. I don't care much about hi-rez but it should be able to play those too.
I'll let Nick chime in on his hardware choices... right now I'm looking at sound cards.
I found a USB to SPIDF converter ...here... (http://www.audiophileproducts.com/usbtospdif)
I thought it'd be best to get the digital path started outside the PC, rather than doing a traditional PCI card. I wish there was a PCI to PCI extension cord so I could physically move it out and still use a PCI card.
Anyway, what's the hot solutions for sound cards these days? Just trying to get a bit perfect and super clean digital signal out of that noisy PC...
-C
-
Facinating topic, Carl!
I'm scratching my head a bit - when you say wanting to start the digital path outside the PC, what do you mean? I assume you want the DAC outside the computer (and not on the sound card) such that the PC is serving up a digital source via either USB or SPDIF or some other method but those are all digital paths.
There's much I want to learn about this so thanks for starting it here! Are you still going down the Linux path?
-
Carl,you might want to look at getting a USB to S/PDIF Converter using the PCM2704 codec.
I started out with the USB to S/PDIF Converter that you linked to and it is inferior in everyway to the PCM 2704 codec. The DIGITALPCLINK using the Phillips UDA1321 has a smaller sound-stage,reduced dynamics and an overall grungy sound compared to the PCM 2704. Here is a link to a USB to S/PDIF converter using the PCM 2704. http://cgi.ebay.com/MUSE-HIFI-USB-S-PDIF-Converter-USB-DAC-PCM2704-/120563258242
Good luck with your project.
Scotty
-
Mike, I'm not trying to get the DAC out of the case, I'm trying to get the digital source (spdif processing) outside the case. The DAC is already its outside...
One option I found was that digitalpclink that I linked to that could accomplish this... Scotty found something else that worked better... but using a USB/spdif is just one option. I'm looking for all ways to get the bit creation process physically removed from the PC as much as possible. And that's just my theory...
There are a lot of soundcards that claim to be 'bitperfect' yet only the Lynx sounds notably better to me, as a digital transport.... and the others vary a little also. It 'shouldn't' be the case but it is.. So far the USB DAC has sounded very good to me.. very smooth and whole.. but I'd like to have a reference level spdif out too so I can compare.
-
I am looking very hard at the ART Legato USB-SPDIF Asynchronous Converter.
http://www.analogresearch-technology.net/
This may be very close to as good as it gets,the drawback is an inability to playback HiRez files.
The Hi-Face converter also looks promising but I am waiting for a Linux driver.
Scotty
-
I'm looking for all ways to get the bit creation process physically removed from the PC as much as possible.
Carl,
Just curious, is there a specific technical reason why you want to get the bit creation process outside of the PC case? If the reason is simply because I've heard it that way and it seems to sound better, that's OK. (There are plenty of things that we as audiophiles do that conventional wisdom says should make no difference, but does anyway. :shock: )
Of course the overall goal is better sound, but I was wondering if there was a specific situation or interaction that you were looking to facilitate or avoid by placing the bit creation hardware a spearate physical space from the rest of the PC? :-k
-
With the PCI cards I've tried there is a background noise to them that you don't notice until it's not there.. Which has been my experience comparing a PC transport to a high-end cd-spinning transport.
My PC was better than most cd and dvd players when I used the PC as a transport. However, that wasn't the case with the Oracle and some other (Raysonic? BAT?) can't remember.. anyway, 3-4,k players seem to do transport as well or better than my somewhat nice PC with $250 pro audio card.
I have since changed things around.. I reconfigured and re-purposed the old audio PC to be the audio server. Now I use a cheap throw-away PC for the sound room that is very quiet but requires a riser board to do PCI... which I don't have. So, I'm using USB directly to DAC.
I just thought about it.. and I think using USB as the digital out is like using a PCI card really... The bits are generated in the noisy environment. I think a PCI expansion board would work, putting the audio card in its own case... but I wouldn't know unless I tried.. and compared it to one that's inside the case.
I've seen these things made.. but they start in the $800 range by a company called Magna.. just to move the PCI outside the case.. which is nuts to me. I can't believe no one has made a cable that just plugs into the PCI slot on the mobo and then provides an empty PCI slot at the other end of that cable. That's what I'd really like to find... It just shouldn't cost very much.
-C
-
Not sure why going out a PCI slot is any different than going out USB. It's the same signal, no conversion, just on a different bus on the motherboard. Running USB out and then converting from USB to SPDIF should be just fine.
Am I missing something?
Bryan
-
I don't know.. why do they bother using PCI or USB? Why not just have everything on 1 bus?
I'm not a PC designer... I just know that PCI slots have a lot more pins and that's what all the sound cards use. From what I've heard, USB to spdif converters aren't all that great... which makes sense... since you're changing the nature of the signal from 1 bus to a device output... and the PCI boards do the same thing but generally better.. So, what IS the difference?
-
I guess that's my point. Why even do the SPDIF step? Why not just a straight USB DAC? Less conversions, less problems. Not as many choices but they are getting more available - like the Empirical Designs stuff.
AFAIK, whether you use a PCI or a USB output, there's no difference in data. Just a change in how it connects to the memory, cpu, peripherals, etc.
Bryan
-
Jitter is the variable. Different kinds of jitter, and different degrees. A PCI soundcard like Lynx with its great onboard clock will output a nice SPDIF stream, even inside the case. If your DAC fixes jitter then just use integrated SPDIF off the mobo. The trend in state of the art computer audio now is USB DACs. The hirez limitations of USB are taken care of in the newest DACs. Anyway, the clocking is the most essential part of moving a digital stream between components. Either buy transport and/or DAC with good clocking and jitter fixing scheme, or use an external clock, or master/slave clock connection between transport and DAC.
Shane brought over a M2Tech USB>SPDIF dongle to my last g2g. We played 24/96 files through it into Buffalo32>Sol's transparent amp>Gemme Katana ceramic drivers>all JP SC3. The resolution was extreme but still smooth. Treble was too hot, but that was the speaker's fault. I would use that dongle without hesitation in my music PC. Definitely high end capable. I would not be surprised if the Lynx products are even better, but only a head to head comparison would show any limitation.
The bits are created at the recording, or at least at CD mastering. From that point not much can hurt them, error correction keeps it pure. Jitter is the only variable, and it is controlled with a clock. The nice thing about the Slim Transporter is the word clock input. Lynx has this too, or use it's own wordclock output to clock the DAC.
-
With the PCI cards I've tried there is a background noise to them that you don't notice until it's not there.
Much of the noise associated with PCs is generated by the switching power supplies that they use. Unfortunately this means that the noise is distributed to every component in the PC along with the DC output. :( I would think one of the best ways to deal with this noise is to try and cut it off at the source. Have you given any thought to trying some of Alan Maher’s CBF devices inside the PC? I would suggest putting (at least ) one onto the outside of the power supply case with double stick tape and another one on the bundle of DC output leads using tie wraps. These little boxes do an admirable job of pulling the nasty artifacts from refrigerators, air conditioners and the like of your AC lines; I don’t see why they wouldn’t do the same for internally generated power supply noise in a PC. And heck they are only $25 a piece, so what do you have to lose?
-
I don't know.. why do they bother using PCI or USB? Why not just have everything on 1 bus?
I'm not a PC designer... I just know that PCI slots have a lot more pins and that's what all the sound cards use. From what I've heard, USB to spdif converters aren't all that great... which makes sense... since you're changing the nature of the signal from 1 bus to a device output... and the PCI boards do the same thing but generally better.. So, what IS the difference?
PCI and USB are fundamentally two different animals with totally different functions within a PC. PCI, (whose full name is the Peripheral Component Interconnection Local Buss) is a set of specifications defining standards that enable different components within a motherboard to “talk” to each other. These components can either be integrated circuits soldered right onto the motherboard or separate card that the plugged into a connector that has commonly become known as a PCI Slot.
USB (Universal Serial Buss) on the other hand is s specification that was developed to facilitate the connection of external devices to a PC. As opposed to moving data around the inner workings of the PC USB ports were devised as a replacement to the parallel and serial ports that used to be used to connect devices line printers to a PC.
So the two busses serve separate purposes and are not really substitutes for each other. The real question from our standpoint is, what is the best way to interface an external DAC with a PC music server?
In a PCI based solution the digital bit stream is combined with timing information on an expansion card and outputted to the world as a standard s/pdif signal which can then be interfaced with out DAC via a coaxial or optical cable.
In the case of a USB interface the digital data is transported out of the PC to a DAC where the s/pdif conversion is done before the signal is passed off to the D/A section. (This is my impression based on what I have read. There are certainly other AN members with more experience in that area than I and I welcome any comments or corrections from them. Lizard King – any thoughts?) It is interesting to note here that using a USB interface does not avoid the PCI buss as all of the internal communications within the motherboard to deliver the signal to the USB transmitter goes over the PCI buss.
So back to Carl’s original question, which is better? As usual, there is no single answer here because each of the designs involves a number of tradeoffs. The use of a PCI audio card provides a more straight forward data path , but at the same time limits us on the S/P dif side. Most audio cards are designed by computer geeks. Even under the best of circumstances, I would dare say that an audiophiles standard of what constitutes good sound is quite different from the standards of a geek. (Not to mention the difference between what we would be wiling to pay for an audio component vs. what the rest of the human race would consider a reasonable price for a similar functionality.)
In favor of the USB interface, putting the s/pdif processing in the DAC allows that conversation process to be designed and optimized by engineers that not only are used to designing to audiophile standards, but are also likely to have the budget to do what needs to be done. However; the downside of the USB interface is that the digital signal has to go through additional stages of signal processing and protocol conversion leading to the possibility that while the USB DAC might be better equipped to process the digital season, it can be faced with the possibility of getting a lesser quality digital signal to work with right from the start.
So both methods have there own advantages and challenges, each of which is wrapped up with words like “could”, “might” or “in theory”. IMO the whole thing boils down to implementation.. Unfortunately, at this point the only way to know for sure is to set the stuff up and listen for ourselves. Perhaps as more and more people do this and share their experiences her and on other forums like this we will be able to see some patterns emerging and get a better idea of which technology is starting to sound better overall.
But for now, trust only your ears. 8)
-
If anyone is interested in further info, here are the Wikipedia links for the PCI and USB Busses:
http://en.wikipedia.org/wiki/PCI_bus
http://en.wikipedia.org/wiki/Usb
-
Good info, thanks Tom!
-
Awesome! Thank you for an excellent explanation, TMAZZ!
I'm sure it will help many understand that yet again, there is no 1 right answer.. as in all of life, it's a set of compromises that yields a product.. and in the end you have to decide whether you're satisfied or need more.
I do plan on some transport experimentation (PC vs. Unix vs. USB vs. PCI etc.) as things shape up. My friend Nick is building a PC.. and I'd also like to use one that uses a linear power supply instead of the standard switcher. The little PC's I'm looking at require lower power than the standard. So, that might clean up a little... also strategic use of card placement, case design, ERS, and mu metal might help isolate things further. My first goal is to understand the principle design characteristics of each component...
-C
-
Those switching supplies are deadly when it comes to noise. My son only has one A/C circuit in his room (not his wouldn't pass code today, but that is how the houses were wired back in the 50s). I got him a PS Audio Noise Harvester to try to clean up the feed to his stereo the best we can and you would not believe how that little thing goes crazy when he switches the computer on. Without the PC it blinks at the rate of one flash every 1-2 seconds. Flip on the PC and the flash rate goes up to about 10 per second! I've never seen a linear PC power supply,(then again I never really looked for one either), that sounds like a good idea for a music server.
When you say you are using a PC as a transport do you mean that you are playing CDs through it in real time or ripping the CDs to the hard drive and playing back those files. One of the big advantages of using a PC as a music server is that by not trying to read the optical disc in real time you have the opportunity to get better error correction and therefore a cleaner signal.If you are playing CDs in real time it does not surprise me than audiophile dedicated transport could do a better job in that regard. The transport was of course optimized for realtime playback and will (or should?) have better mechanical isolation than a $30 stock computer CD drive. However the CD drive can produce excellent results when not saddled with the obligation to perform in realtime. Optical disc are notoriously unstable in their dynamic rotation introducing all kinds of data errors that the CD playback system has to either figure out how to correct or replace with an interpolated value. I think we can all agree that the less number of times you are forced to interpolate (guess) the better your source will sound. When you are not operating in real time the CD drive will have the time to go back and re-read any bad sectors and try to get a better read on the corrupt data. With less unrecoverable data you need less interpolation and get a more accurate reproduction. So, a cheap drive can perform very well if given the time to run more complex error correction, but if it as to get it right in realtime, an audiophile designed transport will have a better (and more expensive) designed that will help lower those errors before they ssrt through better structural design.
In the case of a PC drive the unit will, for the vast majority of people spend very little time playing back music so there was no real reason for the design engineers to build in (or pay for) a level of structural integrity that ill be used only by a small percentage of owners and even within that small group, how many of them will have the system, the time , or the inclination to worry about the sound quality of their PC's sound system.
-
My friend with the modified mac mini made an external regulated linear power supply for it.
Can foobar play a CD live from the optical drive as Windows Media player does? That would be a nice feature for us music lovers who are not enraged if 1 bit is missing. Slim software could not do that. Requires ripping a CD to library and scanning before it can be played. A 5 second buffer would be acceptable to sooth the perfectionists' outrage. Or just let unreadable sectors pass (muted.) If you want perfect, rip it. If you want it right now, accept the errors.
A great PCI sound card like Lynx or RME, or CardD, etc, will have engineered the power supply on the card so the sensitive chips are well served, even by a standard PC smps. But every change is audible, everything helps. You'll hear less improvement going linear with one of these cards, compared to a $200 hifi wannabe in a gamer's box. But the good stuff can be quite expensive as we all know!
-
My interest is only in building a computer that runs a player, like Foobar. The music files are ripped and located on a server outside the room.
My goal is to get bit perfect, jitter free output from it... without buying a $600-700 sound card if possible. Buying a Lynx is like buying JPS Aluminata.. It'll be perfect but costly.
I'm thinking I can do it, we'll see.
-
My interest is only in building a computer that runs a player, like Foobar. The music files are ripped and located on a server outside the room.
My goal is to get bit perfect, jitter free output from it... without buying a $600-700 sound card if possible. Buying a Lynx is like buying JPS Aluminata.. It'll be perfect but costly.
I'm thinking I can do it, we'll see.
Carl:
I think we have a similar goal and I know I am not going to spend a bunch of money for a sound card. It would seem to be going down a similar highway to the one I traveled with the Modded Transporter. Before I would do that I would run ethernet cable to the Touch and feed a DAC by Spif.
Ken
-
There's no free lunch...
I'm all about finding the diamonds in the rough or trying to do something on the cheap if it is the same result or similar...
I've had dozens of consumer level cards here for spdif out.... most recently the Asus one that stereophile or one of them raved about.... the Lynx with JPS UC2 custom dongle was better... better detail retrieval, lower noise, more natural and real sounding.
$500 MSRP for a top shelf card is not what I consider outrageous pricing... not everyone will agree... I'd love to not have to pay it. I'd love to find an even better performer. So far, I haven't been successful. The HiFace USB to BNC dongle is convenient, but the Lynx is better...
I hope folks realize that when I post an impression it's an honest opinion based on my experience, not motivated by any financial bias. But just in case, I don't sell Lynx sound cards, nor do I sell any of the others that I've tried. I'm purely an audiophile consumer in that area paying retail prices like everyone else. Just sharing my experiences.
If you do find something better for less, by all means, post about it here. I want it! :)
-
Shane and Others:
Yes, I am really confused - of course this is no new news! On the Computer Audiophile Site and on other forums folks are raving about running USB or USB to Spidif Convertors to DACS. They are using Foobar or ITunes or something else on their computers as music software but I do not get the feeling they are using hi dollar aftermarket sound cards. I know I am not totally on another planet as I have talked with 3 people that have converted to computer audio in the last year. One of them is using DACs that cost $3000+ and I know he is connecting via Firewire.
I really am trying to understand this before I jump in and wish I had waited or done something else.
Ken
-
Ken, you are wise to attempt to understand the problems inherent with computer audio before buying equipment. There are a lot of opinions and well marketed solutions.
There is a lot of good reading about jitter on the internet. After a while it will make your head spin and you will be looking for clocking solution that "just does it." And you are in luck, there are many competing solutions, and more on the way.
http://www.tnt-audio.com/clinica/diginterf1_e.html (5 parts)
http://www.empiricalaudio.com/computer-audio/technical-papers/
http://www.stereophile.com/features/1208jitter/
http://stereophile.com/reference/1290jitter/ Note the date: 20 years ago!
http://www.stereophile.com/reference/1093jitter/
http://www.stereophile.com/features/368/
http://www.computeraudiophile.com/content/What-does-jitter-sound
http://www.esstech.com/PDF/sabrewp.pdf
http://www.tctechnologies.tc/index.php?option=com_content&view=article&id=29&Itemid=28 (see white paper link)
After all that, you still need a solution in a good looking box with superior analog design, good human interface and software design and reliability, and a reasonable price. With source component being so important to our pursuit of audio excellence, this topic is very important, and timely considering all the new super-DACs emerging now to solve some of these problems. I don't know of any current offerings that are a slam dunk fix for all the problems in computer audio when price remains a consideration. Pro audio is far ahead of audiophiles in this area of high quality digital audio products and research.
-
Thanks Rich!
Okay, if I understand the basics a clock signal from the PC will be embedded in the audio signal it sends to the DAC. You will likely hear this "jitter" unless you process it out of the signal from the computer with something like the Lynx PCI Card? Damn! Convenience certainly comes at a high cost!
Ken
-
Thanks Rich!
Okay, if I understand the basics a clock signal from the PC will be embedded in the audio signal it sends to the DAC. You will likely hear this "jitter" unless you process it out of the signal from the computer with something like the Lynx PCI Card? Damn! Convenience certainly comes at a high cost!
Ken
I have the Empirical Audio PaceCar2 with my Transporter that does this. The PC2 provides a clean word clock input to the TP (on BNC) and the re-clocked SPDIF out of the PC2 drives the Bryston BDA-1 DAC. Plug and play to 24bit/96khz.
-
Thanks Rich!
Okay, if I understand the basics a clock signal from the PC will be embedded in the audio signal it sends to the DAC. You will likely hear this "jitter" unless you process it out of the signal from the computer with something like the Lynx PCI Card? Damn! Convenience certainly comes at a high cost!
Ken
Ken,
Jitter is not a signal in and of it self that gets added or can be taken out. Jitter is basically a measurement of the instability of the timing signal. that can cause timing slips and ultimately data errors in the D/A process. A better PC card does not remove jitter per se but rather generates a higher quality clock signal that has less jitter in it in the first place.
Tom
-
The freq of PCI buss is many times faster than the SPDIF output sample rate. So there are two different clocks at work, each timing their own domain. From the file on the HDD to the Lynx output is probably a dozen different clock domains. None of it matters. Error correction at each conversion assures data integrity. The SPDIF output is clocked by the Lynx card and the PCI buss has no effect. But even if the Lynx output a perfect stream, travelling on copper wire and being reclocked again by a crappy clock serving the DAC would ruin it before conversion. If the DAC relies on SPDIF stream for timing, then the Lynx will have a larger effect. IMO it is best to worry about the clock quality at the DAC, or IN the DAC chip itself. ESS Sabre DACs have clock inside the chip. The stream timing has very little effect on analog output quality. But I will admit, JPS SPDIF cable made an audible difference. I have to assume it was due to reduction of reflections in the wire length that is common with plain copper, and another source of jitter.
-
I was reading the thread (http://www.audiocircle.com/index.php?topic=62364.0) on AC started by ny_paramedic and followed some of his links. He uses MPD (Music Player Daemon) on the server. Looks pretty good, small, adaptable. Can be controlled remotely by almost anything. Here is an article describing actual steps to make a Linux music server work.
http://www.rainbowcyber.co.uk/html/modules.php?op=modload&name=News&file=article&sid=404