Article: Audiophile Reference Music Server For A Song

[darrenwm]

Coops,<br />

<br />

Your jokes never age <br />

<br />

Chris,<br />

<br />

Can I read into this that:<br />

<br />

1) The Lynx card is your favourite-sounding one out there?<br />

<br />

2) You just need a cheap tower PC with PCI slot to operate it?<br />

<br />

3) It is desirable to spend extra to get some silent parts?<br />

<br />

If so, would you say that the Lynx is responsible for the vast majority of the sound, rather than the other bits in the system?<br />

<br />

Darren

[The Computer Audiophile]

Hi Darren - <br />

<br />

1. The Lynx is my go-to digital I/O right now. It's used in the pro audio industry extensively.<br />

<br />

2. Be careful with the statement about just needing a cheap PC with a PCI slot. PCs are all over the board in terms of quality and performance. That's the reason I have to test this specific Dell before even mentioning it on the site. In theory you are correct, but it may be an over-generalization in reality.<br />

<br />

3. Yes. I don't think I will build or purchase another music server for myself unless it is silent or I can make it silent. The fact that the server is silent really eliminates any thoughts about having a computer in the listening room. It's far less distracting even if your current PC is pretty quiet.<br />

<br />

I will say the Lynx is a great part of the sound, but everything matters.

[kristian]

Regarding Steve Nugent's comments about the Lynx card, I, too, have those concerns. I don't see any particular reason why the Lynx card should be so good at outputting digital audio at it's rather steep price, unless it has superb jitter reduction and outputs a very low jitter digital signal. It doesn't look like anyone here has ever actually tested the Lynx's jitter output, perhaps apart from Mr. Nugent. Or?<br />

What about the Lynx, then, makes it better when combined with a regular DAC than asynchronous USB DACs that control the computer's clocking, like e.g. Wavelength, Benchmark (as I understand it), and Empirical do?

[The Computer Audiophile]

Hi kristian - You have very valid concerns that will likely only be addressed by you personally. I think you would have to listen to one and make your own decision.<br />

<br />

<br />

<i>"...I don't see any particular reason why the Lynx card should be so good at outputting digital audio..."</i><br />

<br />

I'm not sure what to tell you on this one. Lack of marketing material or evidence is certainly not proof that it does not exist. I don't know your background so I'm not sure what exactly you mean when you say you don't see any reason why it should be good. Nonetheless you have a very valid concern and one that more than a few readers on the site share as well.<br />

<br />

I can tell you that the Lynx card is used in post production for some of the most "audiophile" albums every made. Engineers who can pick and chose any system they want o achieve the best sound quality have selected Lynx for years.<br />

<br />

The Lynx jitter spec is 20 ps.<br />

<br />

<i>"What about the Lynx, then, makes it better when combined with a regular DAC than asynchronous USB DACs that control the computer's clocking, like e.g. Wavelength, Benchmark (as I understand it), and Empirical do?"</i><br />

<br />

The adjective "better" is very subjective. Better in terms of features and capability is easier to determine. The Lynx supports 24/192 natively and I can easily use an external clock to take it one step up. In terms of better sound quality, that's totally up to each individual. At 24/96 and under maybe they all sound similar to most people. <br />

<br />

<br />

Anyway, yours is a solid concern. Thanks for the post.

[kristian]

Thanks, Chris, for the response. When I say I don't see why it should be so good, I really mean that it would be nice to see actual testing of the jitter output to evaluate the claims. Otherwise, I'm sure you're aware that ProTools and other computer-based digital mixing software is no panacea to good sound; contrarily, I'd offer the opinion that the current overwhelming dominance of music recording by way of ProTools has reduced sound quality over good analogue recording/mixing methods. Lynx stuff is used only in combination with ProTools etc. as a digital I/O facilitator. The sole fact that it is a pro-device doesn't make it superior. And this from an active ATC owner!<br />

<br />

Further, there are all kinds of reasons to get jitter redution etc. out of the computer due to power supply issues. It's tough to get a Lynx card good, clean power. Off the bat, I'd think that Wavelength's asynchronous method of controlling the computer clock and Empirical's reclocking/i2s interfacing is where it's really at for best sound. <br />

<br />

It would be beneficial (and fun) if you did some basic testing of the stuff you review. This is critically important in computer audio in particular, as it is in so much flux, with so many shibboleths waiting to be debunked. What is the true jitter output of the Lynx cards? They say 20 ps; test it. What is the jitter reduction capabilities of DACs on test? What does internal computer power really look like? Is it truly noisy? Scope it. What are the jitter figures of various computers' TOSLINK outs? Bit perfect? Oh yeah? We'll see about that. Etc. etc. You don't have to be an engineer to learn how to do this either. <br />

<br />

Best,<br />

<br />

Kristian

[The Computer Audiophile]

<i>"...I'd offer the opinion that the current overwhelming dominance of music recording by way of ProTools has reduced sound quality over good analogue recording/mixing methods. Lynx stuff is used only in combination with ProTools etc. as a digital I/O facilitator. The sole fact that it is a pro-device doesn't make it superior. And this from an active ATC owner!"</i><br />

<br />

I agree for the most part. I'm thinking of applications like Soundblade and Sequoia Digital Samplitude. The digital 24/176.4 Reference Recordings masters are doing just fine. Good analog is certain welcome in my book as well. Shelby Lynne's new album comes to mind for good analog. I agree that Pro does not equate to good sound quality. I'm thinking more of the people who master some of the best recordings available, not the pop Pro Tools stuff. One also must think about the fact that Pro gear is what all our music is made with. I avoided Pro gear for years but now I am a <u>little</u> more open to it.

[audioengr]

There are a couple of things I like about the Lynx AES16:<br />

<br />

1) it allows multi-channel for those that want to do software crossover<br />

<br />

2) The word-clock sync works and it's progress is very visible in the control panel - you know when it's synced<br />

<br />

Apart from the good customer service that Lynx evidently provides.<br />

<br />

Steve N.<br />

Empirical Audio

[audioengr]

"The Lynx jitter spec is 20 ps."<br />

<br />

The oscillator on the board maybe with a perfect power supply, but not the system. Also, jitter specs must include several measurements, P-P, RMS and others. Even these are insufficient IMO.<br />

<br />

There is no way in the world that they can spec a clock that is powered from an unknown power supply, such as the PCI bus power in some computer.<br />

<br />

Steve N.

[audiozorro]

Is our niche market too small to support specific and optimized solutions for our needs? It seems that some the recommended hardware and software solutions are more than what we need in our quest for better music fidelity. It seems that I would only use a fraction of the capabilities of the Lynx cards or what seems to be coming from Sonic Studio. The need to strip down Windows or make-up simplified custom cables is additional examples.<br />

<br />

I believe most of the present focus here seems two-channel stereo playback of digital files. These may be some of the best HW/SW solutions now but it seems that we are paying for only a fraction of what we will use and introducing a lot of complexity that probably have some performance penalties, maintenance or setup difficulties. <br />

<br />

Firewire might be dying as an interface but using my MacBook Pro and an excellent FW DAC sure seems like a simple clean solution. I am still trying to learn about the Lynx cards, I2S, Off Ramp and Pace Car and trying to weigh the pros/cons of these solutions without spending a lot of money or being disappointed with the results.<br />

<br />

Steve - please explain in lay terms your comments on the Lynx AES16:<br />

<br />

1) it allows multi-channel for those that want to do software crossover<br />

<br />

2) The word-clock sync works and it's progress is very visible in the control panel - you know when it's synced<br />

<br />

[audioengr]

The Lynx is multi-channel and some of my customers are using four or six channels to drive systems with multiple amplifiers driving speakers with no crossovers. They drive the individual drivers, either 2-way or three-way systems. In order to get low jitter, they are using multiple Pace-Car reclockers that are synchronized to each other to deliver two or three stereo channels of digital signal to 2 or 3 DAC's. In order to get multiple-channels, they are using special crossover software on the computer.<br />

<br />

The Pace-Car 2, when used in synchronous mode with the Lynx utilizes the word-clock output from the Pace-Car to drive the word-clock input of the Lynx card. The Lynx card then synchronizes its clock to the word-clock. This is what is shown in the Lynx Control Panel. A short clock cable is connected from Pace-Car 2 to Lynx. Then, the digital data from the Lynx is transmitted via the Lynx AES cable and temporarily stored in the memory of the Pace-Car and then clocked out using a local clock inside the Pace-Car 2 over S/PDIF, AES or I2S cable to the DAC of your choice. Three cables involved. <br />

<br />

The clock inside the Pace-Car is superior to the Lynx clock, as is the power supply, internal cables and board design, so the output jitter from the Pace-Car 2 is much lower than the Lynx. Same data only lower jitter.<br />

<br />

Steve N.

[cfmsp]

<br />

I'm not sure that Firewire is 'dying', but rather - the mass-market is moving to USB.<br />

<br />

Like you, I use Firewire to connect to my DAC - directly and simply. My DAC also supports Toslink (which I use via Apple TV connection), as well as S/PDIF (AES and coax), basically everything BUT USB.<br />

<br />

The hysteria over death of Firewire is just that, in my opinion. For starters, Pro audio still uses Firewire to a significant extent, and as the Pros have been utilizing computer audio interfaces longer than 'audiophiles', I think they are more knowledgeable on this than, say, the mass-market electronics producers.<br />

<br />

If you listened to all the recent articles you would think that Apple dropped Firewire, but they did not - they only dropped it on the mass-market MacBooks. They upgraded to Firewire 800 on the MB Pros, e.g.<br />

<br />

I personally have a hard time understanding why audiophiles don't pay more attention to Firewire as the least complicated to implement digital audio interface.<br />

<br />

S/PDIF, as I understand it, is the worst in theory - but somehow it became the de facto standard among audiophile companies. Makes no sense to me. The move of the mass-market to USB does make sense for computer interfaces.<br />

<br />

UBS done right seems quote promising, I'd love to listen to Gordon's products.<br />

<br />

I bought a Metric Halo ULN-2 at B&H - utilizing their liberal try-and-return policy, and haven't looked back. I think I'd be equally happy with Gordon's Brick, and certainly a Cosecant.<br />

<br />

FWIW, I also agree on the extra functionality. I didn't buy the Metric Halo just as a DAC, but I think it's worth every penny of it's price as just a DAC, compared to Benchmark DACs and the like.<br />

<br />

all for now,<br />

clay<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

[VincentH]

I'm using a RME FireFace 400 firewire DAC from Windows. It is a well-respected pro audio device, capable of multiple channel 192/24 with excellent anti-jitter built in. The Fireface was recommended to me by some experienced members on this forum. I compared side-by-side to the Weiss Minerva (also firewire, also from a company well known in pro-audio circles) and although the Minerva was better, the FireFace stood its ground. So firewire is both well established and used by some of the latest and greatest DACs.<br />

<br />

Death of firewire? I've heard nothing like that in PC world - isn't this a bit of a generalization of the Mac perspective? When all hardware is created by a single company, a single design decision may seem to change the world...

[The Computer Audiophile]

Hi Guys - FireWire is Apple's baby. When Apple drops FW400 on its newest product that is not a good sign in terms of longevity of the interface. Fortunately there will be ways around this for many years. The FW800 to FW400 adapter and an ExpressCard for FW400 are two examples. Since serial port adapters are still prevalent today I have a good feeling that FW400 will have an extended life through third party accessories.

[audioengr]

Firewire devices can be just as good as USB, but no manufacturer has implemented an async interface yet. The DiceII chip with JET jitter-lowering technology is good, but not great IMO. The Off-Ramp 3 beats it hands-down. I have had the latest Firewire design using the DiceII in my system about a week ago and I was not impressed. It had custom drivers that bypassed kmixer, but that didnt seem to help it.<br />

<br />

The primary advantage of Firewire is for studios that must have 24/192 capability and multiple-device firewire connections.<br />

<br />

In 2009, you will see USB interfaces that are both Async and support 24/192. This is one thing I am working on.<br />

<br />

Steve N.<br />

Empirical Audio

[The Computer Audiophile]

<i>"In 2009, you will see USB interfaces that are both Async and support 24/192. This is one thing I am working on."</i><br />

<br />

This is going to be fabulous Steve.

[Wavelength]

Steve,<br />

<br />

Actually it's pretty easy for them to get less than 20ps of jitter on their 27Mhz master clock. But I imagine they are running this into one of those Audio PLL's to create the rest of their working frequencies. These are pretty bad and even the low jitter ones claim less than 50ps of jitter but again it really depends on implementation.<br />

<br />

I have the Wavecrest setup now which can measure jitter to better than 500fs RMS at 1Hz. It's pretty daunting and I ran a bunch of measurements for some companies in the last couple of days. It's really amazing the better my test equipment becomes the better I can understand why this stuff does what it does.<br />

<br />

Like for instance did you know jitter is always lower the higher the Fs rate is. I find it a little silly that most of the companies that make dac chips that go to 24/192 measure their jitter rejection at 48k??? It seems to be true for both SPDIF and USB as I measured some Cirrus SPDIF units, AKM SPDIF and the TAS1020 running in 3 modes: Adaptive running the Centrance and then my version, Async internal clocking and Async using the external clocking (what I do).<br />

<br />

I then ran the same tests using the Prism dScope III's jitter analyzer and also an idea a couple of us have been talking about using my SRS Spectrum Analyzer. All this stuff is pretty cool and will be printed up at some point.<br />

<br />

~~~~~~~~<br />

<br />

Firewire does have an ASYNC protocol for audio but it is not supported in any operating system. Steve the DiceII looks to be the Oxford part. Oxford developed a part just like this and then got out of the whole Audio Firewire thing all at ounce. <br />

<br />

~~~~~~~~~<br />

<br />

I think the MetricHalo would be a great little unit to use as an audio dac. We are their main beta test site here and Dave Davies and Rob Fetters both use these gems all the time. I would dump the switching supply and go with a linear regulated. I made one for John Atkinson and he uses that with his ML2 all the time.<br />

<br />

Thanks<br />

Gordon

[audioengr]

Gordon,<br />

<br />

I believe the Lynx uses some sort of VCXO at 40MHz to sync to the word-clock, but they also have 49.152MHz and 45.1584MHz XO's on the board as well. I doubt if ANY oscillator run off the PCI bus in a desktop computer will have 20psec of jitter, even measured at the terminals of the oscillator. Besides, when we are talking jitter is this P-P or RMS or what? What's the spectrum? There is no good standard for measuring/characterizing this IMO.<br />

<br />

I have spec sheets on the old Oxford parts. I believe the DiceII was developed by TC Applied Technologies out of Canada.<br />

<br />

Steve N.<br />

Empirical Audio

[Wavelength]

Steve,<br />

<br />

The one I had was a single 27MHZ into a PLL that can be used as a VCXO. There are many of these devices out there that handle all the audio frequencies.<br />

<br />

With regard to firewire these are all using block mode so they require the driver to allow the device too look like an audio device. I am still unsure why it needs jitter reduction at that point because it would be async to anything else.<br />

<br />

The Oxford part I am talking about is one that was never released. It supported 16 channels of audio in the small version and even more in the large. It was adaptive in nature and had a unique jitter reduction circuit on it. After they closed that door I tossed the data sheets.<br />

<br />

As for 24/192 USB just make sure your endpoint can handle the excessive frame size. Most USB based processors can only handle 1024 and that won't cut it for 24/192 as it requires 1152 bytes per frame minimum for Adaptive and more so for Asynchronous.<br />

<br />

But there are other ways to make people happy and I think some of you will be trilled with what we are working on.<br />

<br />

Thanks<br />

Gordon

[sastusbulbas]

A couple of comments,<br />

<br />

First I would be interested in what upgrades have been done to Kana813's Genesis Digital Lense?<br />

Quote, AIFF files on my Apple TV/iTouch combo running through an upgraded Genesis Digital Lens. No PC required for playback.<br />

<br />

Secondly regarding the original subject,<br />

<br />

Are there not more affordable alternatives to the AES16? Pretty much all it's features are going to waste if only using two channels, and are flying leads the only alternative? What 2 channel alternatives did you try? And what differences are you going to get in comparison to digital direct from a PC motherboard with basic 16bit.44.4khz WAV CD files?<br />

<br />

A £700 soundcard seems a little steep when a second hand Mac can be bought on eBay along with the likes of an Apogee or Benchmark DAC to utilise the Firewire output. <br />

<br />

And as for a £5k Dac, well I would rather see alternative inputs utilised in comparison also, as I am under the impression that most DACs these days seem a little less bothered about input quality, as they are designed to alleviate such problems and offer consistency regardless.<br />

<br />

So come on, what about some real world ancillaries with required features which offer high performance per pound here, not some highly recommended expensive kit that is rumoured and opinionated to sound better?<br />

What about cheaper alternatives to your server, build a basic PC maybe even running Lynux, with a more down to earth 24/192 capable soundcard, comparisons with a firewire PCI card outputting to a suitable DAC would be worthwhile, as would utilising a laptop with an external DAC or such?<br />

<br />

I mean seriously, can we expect a £700 multi channel do it all card to offer value for money performance in comparison to a simple bit perfect stereo output via the likes of a PCI Firewire interface or more basic 2 channel only digital output 24/192 capable soundcard?<br />

<br />

How much of the sound per pound do you get when you remove all the unnecessary channels and features of the Lynx? How much would it cost with only two inputs and two outputs? Are there 2 channel 24/192 alternatives in that price bracket that you have listened to? <br />

<br />

[The Computer Audiophile]

Hi Steve - Thanks for the post your questions are very valid. There are certainly many different ways to piece together a system and nobody can please everybody all the time. There are cheaper alternatives to virtually every piece of this music server. My goal was not to put together the absolute cheapest PC that would produce sound rather it was to put together a PC that is capable of audiophile sound quality for a very reasonable price. Can this be done other ways? Most certainly. If I remember correctly the cost of the Lynx card is not because of all the channels. Making a multi channel card is not much more expensive than a two channel card.<br />

<br />

As far as your comments about using more interfaces like FireWire and laptops etc... I think you'll find these topics addressed all over the site. This article was a little more focussed.<br />

<br />

Thanks again for the post. Even if I disagree with much of it I do respect your opinion.

[francis]

Hi Criss, i managed to get the Lynx card but not the firmware and drivers that you mentiond, at the Lynx portal the only have newer versions, can you give a hint?<br />

<br />

thank's<br />

[The Computer Audiophile]

Hi Francis - When you search for the drivers and firmware on the Lynx site you have to check the Archives box so it looks at the old versions. If you can't find them I'll upload them to the site here. I'm on my iPhone right now, so let me know if you need the files and I'll get to a computer.

[odyssey]

Chris,<br />

<br />

What is the reason that the older driver and firmware are recommended for the AES16? What is it about the newer version(s) that make them less desireable?

[The Computer Audiophile]

Hi John - I can't remember the specific language but I believe newer drivers use an upgraded "engine" or something like version 2 of the core of the software. I know this is a terribly bad worded response but I wasn't paying much attention in the last conversation I had with people who knew exactly what the differences are.

[jriver]

[Disclosure -- we publish J. River Media Center, mentioned above]<br />

<br />

WAV Out has been an output choice for many years, but the Wave Out plug-in above may not deliver bit perfect sound. <br />

<br />

On XP, this would normally be done by using ASIO, and on Vista by using WASAPI. I've summarized more info here:<br />

http://wiki.jrmediacenter.com/index.php/Audiophile_Info<br />

<br />

Chris, if you had a problem playing a particular format, we'd like to hear about it.