Computer transport, the right way

Assuming the computer is build in a way to reduce interferences and noise and/or the audio equipment is located far enough (5m+), which would be the ideal way to build the source around it?

The pieces of the puzzle would be:

1) Computer with external firewire soundcard (I have already determined that the best so far is the RME fireface 800). It will play high-resolution master lossless files.

2) External master clock, such as apogee Big Ben. it must accept 24-bit/176.4 kHz. Needed? Waste of money?

3) A very good DAC. The most important piece, IMO. Here there are a few that I'm considering: APL DAC (coming in 3-4 months), Berkeley Akpha DAC (coming next month), and Acoustic Arts DAC-I mk4 (already available).
Plus something DIYed based on tda1541a (EC designs units are limited to virtual 20bits, which will create dithering playing 24bit files). It must accept 24-bit/176.4 kHz.

Comments? Suggestions? Ideas?
 
1) Firewire, S/PDIF or USB is you choice. My sound card has a S/PDIF out which I use.

2) Big Ben is probably a waste. This is usually used in the pro world to provide a 'house' clock that can be distributed to multiple pieces of gear so that they all remain int he same audio clock domain. Seems like overkill to me for your application. Besides, any of the above sources are isochronous with the clock is embedded within the signal. They must be the clock master. Use of an external master would need that you would then also need to use an ASRC ebwteen the source and the DAC.

3) A very good DAC is key to good sound. There are a number of them available. I use a Grace m901. When running FLAC files to the S/PDIF out on my computer to the Grace, the sound is phenomenal.

If you want to build your own, Twisted Pair makes a nice DAC and they also have an ASRC module as well called the Metronome.
 

Telstar

Member
2007-12-26 3:49 pm
Italy
3) A very good DAC is key to good sound. There are a number of them available. I use a Grace m901. When running FLAC files to the S/PDIF out on my computer to the Grace, the sound is phenomenal.

If you want to build your own, Twisted Pair makes a nice DAC and they also have an ASRC module as well called the Metronome. [/B]

I havent found a good 24/192 diy DAC, or even 32bit. I was thinking of a modification or newer version of EC designs DI8. I would really like to use the TDA1541. I'm not sure how many I would need to attain real 24 bits.

Grace DAC? Never heard of. I'm also considering to get an used Muse Model 2 and to mod it. It can even be upgraded from the manifacturer (but that would not be cheap)

I have a TwinDAC+ (spdif and usb in), but it is NOS, and I want to be ready for high-resolution music files that I'll be buying in the next months.

Among the machines already able to accept those signals there are the three i mentioned above, plus the dCS Elgar. Everything else either do not accept higher than 96khz or is too expensive (let's say 4k€/5k$ is my price limit).
 
The Grace is sold primarily as a headphone amp for professional monitoring and mastering. That is its intended use. But at its core it is a first class DAC with AES/EBU, S/PDIF, TOSLINK, USB and also line inputs. If your budget is 4k euro then you can easily afford it and have money left over. Many people also like the similar Benchmark DAC. I think the Grace is better, that's why I have it and not the Benchmark. You can also get a very elegant remote control for it.

BTW, there was a typo in my earlier post, it is an m902, not m901.

http://www.gracedesign.com/products/m902/m902.htm
 

wakibaki

Banned
2008-01-08 11:51 pm
Computers are very noisy. Best to put it in a cupboard (can get v. hot) or a different room. If you have a desktop, you can probably get optical out cheaper in a regular PCI card and you don't need 5m of firewire cable, just the optical will do.

http://www.saverstore.com/productinfo/Product.aspx?product_id=20013132&rstrat=12702

That's the first one I found... obviously it's in the UK, but prices must be comparable in the US.

The home recording guys are fussy about their DACs, but they don't seem to differentiate between S/PDIF I/Os. Mostly they're running 2496 or better. Still,this one is so cheap, you could try it against another and see if you can hear any difference.

w

Oh, sorry, I see you're in Milan
 
The very good quality DAC is the key. If you have external DAC with ASRC and decent sound card with S/PDIF out, than the rest really does not matter as much.

I have regular desktop with E-MU 1212 (It is a great sound card by the way) and using it for S/PDIF out. My DAC is RAKK DAC MKII.

Combination beats any other digital source I ever heard.
 

Telstar

Member
2007-12-26 3:49 pm
Italy
wakibaki said:
Computers are very noisy. Best to put it in a cupboard (can get v. hot) or a different room. If you have a desktop, you can probably get optical out cheaper in a regular PCI card and you don't need 5m of firewire cable, just the optical will do.

http://www.saverstore.com/productinfo/Product.aspx?product_id=20013132&rstrat=12702

That's the first one I found... obviously it's in the UK, but prices must be comparable in the US.


I am building a silent htpc and in alternative i can use one of my desktops with a long cable (haven't determined yet which cable can be long without problems, usb can go to 5m, not sure about firewire, probably yes). But an optical spdif cable shouldn't be long.

About the spdif card, it may work, but it will retain the noise from the computter (unless i build it without moving parts, but i wont).
So, as stated in the first post an EXTERNAL FIREWIRE soundcard is the best digital transport.

I wish some of the pro manufacturers like dCS make a firewire DAC of the same quality. I am quite confident of Berkeley, because the guys are from Pacific Microsonics.

So, mostly my thread is asking suggestions for an up/oversampling dac that will actually not perform any OS, but will retain all the resolution of the master audio files without any manipulation.

4. The software that I will be using is XXHighEnd that uses exclusive mode on Vista and that in my listening tests surpassed the SQ of ASIO.
 

Telstar

Member
2007-12-26 3:49 pm
Italy
Irakli said:
The very good quality DAC is the key. If you have external DAC with ASRC and decent sound card with S/PDIF out, than the rest really does not matter as much.

I have regular desktop with E-MU 1212 (It is a great sound card by the way) and using it for S/PDIF out. My DAC is RAKK DAC MKII.

Combination beats any other digital source I ever heard.

Quoting myself from an above post:
I have a TwinDAC+ (spdif and usb in), but it is NOS, and I want to be ready for high-resolution music files that I'll be buying in the next months.

You guys probably dont know the TwinDAC:
http://www.twindac.com/TwinDACSpec.htm
It is an EXCELLENT DAC, the only problem is that it accept only 16/48k from usb input and 18/96k on spdif. I need 24(true)/176.4k

I'm looking for a similar quality but that can handle higher resolution audio files. Eventually a DIY project. Or some mods to an old Muse Model 2 that I could get for 500€.
 
I think a fundamental possibility is being missed here:

Use an audio interface with a clock input!
You won't need a nasty ASRC and the transmission won't add any jitter beyond what the DAC's xtal produces.

If you're not desperately after a DIY DAC project, then I recommend simply buying a Lynx Aurora8 (and the firewire interface card for it) and you're done.

Alternatively, any DAC (maybe Apogee, Benchmark, whoever) that has a clock output will be able to sync a decent sound card (the Lynx AES-16 is fantastic) to it perfectly.

Originally posted by wakibaki
Computers are very noisy.

That's really not true if you put any effort into the design at all. There's a multitude of cases designed for minimal airflow turbulence and whatnot, plus a whole range of super-quiet fans and power supplies. CPUs and graphics cards can be cooled passively. Hard drives can go in quietening enclosures.
The result is a very quiet machine that doesn't make any more noise than a CD player - perhaps less.
 
Wingfeather said:
I think a fundamental possibility is being missed here:

Use an audio interface with a clock input!

You won't need a nasty ASRC and the transmission won't add any jitter beyond what the DAC's xtal produces.

If you're not desperately after a DIY DAC project, then I recommend simply buying a Lynx Aurora8 (and the firewire interface card for it) and you're done.

Alternatively, any DAC (maybe Apogee, Benchmark, whoever) that has a clock output will be able to sync a decent sound card (the Lynx AES-16 is fantastic) to it perfectly.

How does it compare to the Fireface 800?
SQ should be similar, and btw, I'm considering it only as digital transport.

Hmm... a 24/192 DAC with word clock output is even more difficult to find...
 
How does it compare to the Fireface 800?

It's similar, and aimed at a similar sort of market. The FireFace is more general-purpose recording (it has mic preamps and suchlike), whereas the Aurora is pure A/D and D/A.
Lynx are generally regarded as having the edge in sound quality over most other soundcard manufacturers (but that could be for any number of reasons, their higher prices being a definite possibility. I think RME is going to be roughly equivalent). I use a TWO-B soundcard to do all my D/A conversion (yes! Inside a PC! Heresy!) and find it to sound thoroughly excellent.

Hmm... a 24/192 DAC with word clock output is even more difficult to find...

The Aurora has that covered. It's not meant as a firewire interface per se, that's just one option. In the main it's simply an 8-channel A/D and D/A convertor, with AES/EBU I/O, and clock I/O. The Firewire interface card is an optional extra if you want to use it without a sound card.
- RME have their "ADI-8 QS" converter which is 24/192 and has clock I/O.
- Apogee has the DA-16X, which is a 16-channel 24/192 DAC with clock I/O.

These and pretty much any other "studio-style" converter box is likely to have clock I/O.

If all you want the PC to be is a transport, then all you need is a sound card with digital outputs and a clock input. Any Lynx TWO sound card has these features, as does the RME HDSP product line (including the AES-32, the direct rival to Lynx's AES16), and many other professional cards.

With a clock input, there's no advantage to be had in using an external interface. You just end up paying more for extra casework and a power supply. My thumbs go up in favour of either the Lynx AES16 or the RME AES-32 (I own a Lynx myself and know for a fact they are rock solid. I would presume the RME is just as good, though).
 

wakibaki

Banned
2008-01-08 11:51 pm
SPDIF is good up to 5 metres. That's what it says in that advert, anyway. Being optical it doesn't carry any electrical noise.

You're not really concerned about vibration carried on the lightpipe are you? If this is significant it's news to me... it's a plastic filament, there can't be much audio energy transmitted at the end of 5 metres, or for that matter, radiated into the air, especially if you put a loose coil in it. If you can point to an internet article or review detailing this problem, I will be grateful.

If you are talking about vibration within the computer affecting the data read from a CD, then, if this is occuring, it will occur regardless of the presence of the firewire interface. Anyway this is pretty unlikely.

Audio data is not streamed in a computer in quite the same way it is in a CD player, as the processor is required to service numerous interrupts and concurrent threads. A master and slave disk may cohabit on an ISA bus. Larger buffers are therefore required, available and used.

If you read your data from a hard disk, it is identical every time.

Firewire is a nice interface, particularly for multi-channel recording, but if you can get the computer out of earshot without it, then in terms of sound reproduction it's just an unnecessary expense and another thing to go wrong.

Clock jitter is another issue. I didn't comment on clock jitter.

Why not just try the nice M-audio Audiophile 2496 soundcard which is highly regarded and has SPDIF too if you don't like the sound...

w
 
wakibaki said:
SPDIF is good up to 5 metres. That's what it says in that advert, anyway. Being optical it doesn't carry any electrical noise.

You're not really concerned about vibration carried on the lightpipe are you? If this is significant it's news to me... it's a plastic filament, there can't be much audio energy transmitted at the end of 5 metres, or for that matter, radiated into the air, especially if you put a loose coil in it. If you can point to an internet article or review detailing this problem, I will be grateful.


I dont know about interferences in spdif digital cables, but in general, the longer a cable is, the more risks for getting.
From undirect experience I know that USB cables and likewise firewire are pretty "safe" to use at long distances, like 3m.

Audio data is not streamed in a computer in quite the same way it is in a CD player, as the processor is required to service numerous interrupts and concurrent threads. A master and slave disk may cohabit on an ISA bus. Larger buffers are therefore required, available and used.

If you read your data from a hard disk, it is identical every time.

This is correct. The problems starts when the audio stream is going to leave the computer. It shouldnt matter, but it does. In audio many things do not have a scientific explanation (i.e. cables)

Firewire is a nice interface, particularly for multi-channel recording, but if you can get the computer out of earshot without it, then in terms of sound reproduction it's just an unnecessary expense and another thing to go wrong.

To sum up what my conclusions are on this matter:
-An internal device (soundcard or spdif out or similar) gets jitter and/or noise from anything. Particularly power supply and anything connected to it, fans and the weakness of the pci bus.
-An external device is less subject to the above. But the USB bus has many issues by itself and it is NOT the ideal bus to be used to transfer audio streams (no matter what USB dac manifacturers say). Particularly problematic are bus-powered devices, because of the unclean power they get, which affect the sound.

What do we have left? Firewire and LAN (wired or wireless). Wireless has basically a bandwith problem that would cause problems with high-resolution audio, otherwise, it is a very good choice. The squeezebox sounds good, altough has a very poor dac, and cannot compete to a high quality computer transport IMHO. A Wired network connection (gigabit please), that i would consider. But I haven't seen any DAC on the market with it. That would be another topic.

Then firewire. Theoretically IEEE 1394/a has sufficient bandwith. So, I'm pretty sure that the Lynx Aurora will not be limited by this. Why people found a difference in sound between a Fireface400 and a Fireface 800? Because the first is bus-powered and the latter has it's own power supply.
It could seem hard to believe but even the firewire chipset used in your computer does matter on the SQ in the end.

Clock jitter is another issue. I didn't comment on clock jitter.

Why not just try the nice M-audio Audiophile 2496 soundcard which is highly regarded and has SPDIF too if you don't like the sound...
w

I think I have answered to this question at this point :)

---
Links to support my findings:


about power and noise affecting digital audio:
http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&m=27041
about cable lenght:
http://db.audioasylum.com/cgi/m.mpl?forum=pcaudio&n=21105&highlight=flac+John+Swenson&r=&session=
about power line noise:
http://db.audioasylum.com/cgi/m.mpl?forum=pcaudio&n=20669&highlight=usb&r=&session=
more about computer transport noise (i was the most skeptical then):
http://www.phasure.com/index.php?topic=339.0;all
about FF400 vs FF800:
(sorry cant find but one person on asylum was saying that the ff800 sounds more airy)
about different firewire chipsets:
http://www.phasure.com/index.php?topic=55.0
 

wakibaki

Banned
2008-01-08 11:51 pm
I looked at some of your references. These guys are sufferers from obsessive-compulsive disorder. You might as well take technical advice from an astrologer.

There is little available in terms of program material in very high quality format, and there is no real sign of a successful emerging standard. Much existing music will NEVER be available in that format when it arrives because of the quality of the original material. It would be a shame to cut yourself off from that... people are still enjoying material cut on 78s.

Why not just wait till the market settles down a bit and enjoy the excellent quality already available without taking extreme measures.
 
Originally posted by wakibaki
I looked at some of your references...

I have to agree. The amount of "information" floating around on audio forums that's based on nothing more than intuition is a big problem. With adequate power filtering and solid design, an internal sound card isn't going to perform any worse than an external device. Both measurement and listening confirm this (in my own experience, at least).

And anyway, even if it does perform worse, the fact that your internal sound card isn't acting as the clock master in this case nullifies any problems that could theoretically arise. No jitter (beyond what the DAC's oscillator produces) is going to interfere with the audio if the DAC is acting as the clock master. Unless the jitter is so high that the DAC can't lock onto the incoming signal properly, then it just makes no difference.


Originally posted by Telstar
In audio many things do not have a scientific explanation (i.e. cables)

Not everyone believes this.
 

dwk123

Member
2002-02-06 4:51 pm
Telstar said:


To sum up what my conclusions are on this matter:
[cut]

Clock jitter is another issue. I didn't comment on clock jitter.


Well, I once again am amazed at the lengths people go through to complicate things. All that time spent hand-waving about how anything but firewire has problems, but you then claim that clock jitter wasn't part of it.

Lets review: Digital audio has 2 parts:
1) bits
2) clock

That's it. No more, no less. So, *IF* there are differences in digital transmission from a PC, it *must* be via one of these mechanisms. If we assume that any competently designed interface will preserve the bits, then we're left with nothing but the clock.

So, power supplies, noise, process scheduling etc are all completely and utterly irrelevant in a PC transport *except* so far as they impact the clock.

The solution to this of course is to slave the PC to an external clock generated by the DAC. If you're really uptight use a galvanically isolated transmission like optical or xformer-coupled coax spdif. This can be achieved easily with relatively inexpensive DIY means using the Twisted Pear wolfson spdif boards and a simple *synchronous* reclocker (google John Swenson's circuit for the SB). Generate I2S 'silence', feed it into one spdif board configured as an xmitter, and export this to the PC. Use a 2nd board as a receiver, and feed the resulting I2S to the synchronous reclocker circuit driven by the same crystal used to generate the original I2S silence. This will result in I2S clocked directly by a local crystal osc, and the PC will be completely out of the equation. Cost is probably $200, maybe $500 if you use an expensive boutique off-the-shelf clock.
 

steaxauce

Member
2008-01-24 9:47 pm
dwk123, are you certain that the computer has no other impact on the signal? I don't actually know, but it seems that with high bit depths (24 bit), some of the less significant bits may be obscured by interference. There are no setups out there that can get full resolution out of a 24 bit signal. As far as I know, 20 bits is as far as any DAC can go. Again, I don't know where those bits get obscured, but it seems reasonable to think that PC interference might be a potential cause. Can anyone say anything about this?