Firewire vs. PCI vs. _ _ _ _ _ vs....

[mark4man]

People,

Right now I'm running an Echo Audio Layla 24/96 thru SONAR on a Wintel system.

If I up the ante & go to higher grade converters, I can go the Firewire route (e.g. RME Fireface 800, etc.)...the ADAT route...up to a mastering grade via AES/EBU, feeding into a 25 pin D-Sub from a mixing board (e.g., Apogee Rosetta 800, etc)...or back to PCI.

My question is: Which of these platforms are better; & why?

Is the main concern conversion jitter?, or is this not a concern at all (with any platform) since the interface converts as a standalone unit & then simply feeds the PC a digital signal?

Are there problems associated with Fireware & ADAT? Are they as fast or efficient as PCI?

If I go up the hill to a Rosetta, Mytek or Benchmark, how is AES/EBU as a secondary interface (ADC > DAI > PC)?

Thanks,

mark4man


Also: Are there any good informationals out there comparing these systems for native DAW recording?

 

[LeeRosario]

Wow, lets see if I can offer some help where I can:

People,

Right now I'm running an Echo Audio Layla 24/96 thru SONAR on a Wintel system.

If I up the ante & go to higher grade converters, I can go the Firewire route (e.g. RME Fireface 800, etc.)...the ADAT route...up to a mastering grade via AES/EBU, feeding into a 25 pin D-Sub from a mixing board (e.g., Apogee Rosetta 800, etc)...or back to PCI.

My question is: Which of these platforms are better; & why?

mark4man

You're gonna get alot of different answers to this, the reason: personal opinion, bias or experience.

I can only tell you about the Rosetta 800 from my experience with it. First, it depends on what exactly you're to do and what you want to accomplish.

The Rosetta 800 isn't a magic wand. Subjectly, it improves overall clarity maybe 10-20% at most. The producer I worked for had an older mix system, so it had more impact on quality and use.

I don't really have any real complaints about it, but it still left me so so. I honestly beleive there is something better out there. However, if you choose to go with the 200 or 800, you can probably survive a good 3-5 years easily with it.

People,

Is the main concern conversion jitter?, or is this not a concern at all (with any platform) since the interface converts as a standalone unit & then simply feeds the PC a digital signal?

In digital recording, jitter is ALWAYS going to a problem. What you have to understand is that timing and audio data go seperately. You can never expect an interface to do both, well. It's like your heart trying to pump blood and trying to process brain information at the same time.

Let the wordclock process clock and the audio interface process audio data.

Thats why you might hear alot about clock. Sample rate coverters and Masterword clocks are two different things.

A wordclock is what sycronizes your equipment (depending on how many feeds you can run from it). It's the hearbeat of your studio.

A good word clock makes the difference between jagged waveforms and clean, smooth waveforms. That's another 10% improvement to your sound. Plus, increased depth and overall tightness.

This is where you start going pro. Inherently, all audio interfaces come with an internal clock. They have to. No human being can live without a heartbeat, same with all those interfaces out there.

However, the internal clocks are very cheaply designed, which is the whole reason you need external clocking in the first place. The absolute best senario is a recording interface with converters the quality of something like a rosetta and a wordclock the quality of a Big Ben (for the sake of argument).

But that would mean an 002 rack would cost you at least 5000 clams, plus the company would lose power of it's customers. A do-it-all box that has everything to give you crystal clear audio is bad business for companies like digidesign.

However, in terms of masterclocks, there are better options than Apogee. I've heard Manley come up a few times. Drawmner also comes up. However, just be sure to research that as well.

People,

Are there problems associated with Fireware & ADAT? Are they as fast or efficient as PCI?

I'm not too tech savy about computers, but I beleive PCI still has the advantage. Thats why Pro Tools HD systems still run off of expensive PCI cards instead of making HD firewire systems.

100 + tracks at 192 k each on firewire? Oh god no.

You could never get that speed and power with Firewire or Optical Adat *yet*.

However, if they keep improving on firewire and it surpases PCI, then you can pretty much imagine how Digidesign is going to handle that.

People,



If I go up the hill to a Rosetta, Mytek or Benchmark, how is AES/EBU as a secondary interface (ADC > DAI > PC)?

Also: Are there any good informationals out there comparing these systems for native DAW recording?

I don't think I fully grasped that question.


in terms of comparisons, Pro Sound Web is really the only internet source I can think of for now. I don't really spend all that much time looking up that information on the net.

In the end, companies that manufacture these things don't include vital information anyway, so I rely on personal experiences more than word of mouth.

 

[Tonio]

Great synposis Lee!

The problem I have understanding is teh clock issue. Now mfg and top dog digital guys say that an external clock will not add benefit to the convertors of another device being slaved, yet folks say that adding one does make it less jittery. Then you have folks say that with the all in one boxes couldn't benefit-i.e. If the convertors are marginal, then the PLL would be marginal, yet people say with experience that the mid to upper devices are better with external clocks.
Granted there are a few exceptions, but it appears that a external clock does indeed benefits the sound or "cloud being lifted".

I've questioned about this, and users say its better, but the tech folks, and mfg's says it could not be possible.

Lee or someone , please set us straight on this???

OK, I hear : Ya gotta try for yourself to see.

 

[cshannon1077]

More Generic question -

If the sound card is the audio threshold into the PC, how do the different platforms compare? i.e. PCI vs USB vs Firewire vs (any other options??)
Is there a universal test we can use for comparison? i.e. 48k to 196k ?
Since firewire is the new kid on the block, you'd think they would be taunting this comparison, if they are/were the best. right?

I've been into various areas of sound and music, and am contemplating getting into PC recording. I have download several demo s/w packages hoping to find something useable. No luck yet. I thought I'd get somewhere with Cakewalk, but it doesn't seem to be useable. I'm still trying, but I would appreciate any input on my direction.
I have several goals in shopping. Obviously I would like to record live to disc and be able to edit, and remix, and add tracks. Sounds simple enough.
Next, I have audio tracks I need to edit and repair missing audio.
(I'm hoping for some feedback here)
I'm sure there's more to add, but let's get started there.
Any ideas??

Thanks,
Chris

 

[LeeRosario]

Great synposis Lee!

The problem I have understanding is teh clock issue. Now mfg and top dog digital guys say that an external clock will not add benefit to the convertors of another device being slaved, yet folks say that adding one does make it less jittery. Then you have folks say that with the all in one boxes couldn't benefit-i.e. If the convertors are marginal, then the PLL would be marginal, yet people say with experience that the mid to upper devices are better with external clocks.
Granted there are a few exceptions, but it appears that a external clock does indeed benefits the sound or "cloud being lifted".

I've questioned about this, and users say its better, but the tech folks, and mfg's says it could not be possible.

Lee or someone , please set us straight on this???

OK, I hear : Ya gotta try for yourself to see.

oh yes, definitly personal judgment is far better than anything I have to say.


But generally, alot of engineers argue about the external clock thing. Unfortunatly, there's such a big learning curve that goes along with it, so there's alot of bad information out there. Alot of engineers talk with authority, so it's "gotta be right", right?

Sadly, not always.

Generally, if you have a shitty internal clock (like all the home quality interfaces have), then a seperate heartbeat can't do you wrong.

However, it's sorta pointless cause there's so much more that goes into excellent audio. It's like trying to fit a Yugo with 10000 dollar rims.

It's like the common saying great engineers have, "your studio is only as good as its weakest link".

Nothing could be more right in the recording world.

Another thing is, now with the advent of insane sampling rates, this actually makes the problem worse.

The higher in sample rate you go, the more need for stable clock and the more difficult/expensive it becomes to control jitter.


And thats the key, not how accurate, but how stable an external clock is.


And not to mention the connection you end up using. AES/EBU has a tendancy to lose vital clock information through badly designed or unacceptable quality cables.

 

[dgatwood]

I'm not too tech savy about computers, but I beleive PCI still has the advantage. Thats why Pro Tools HD systems still run off of expensive PCI cards instead of making HD firewire systems.

I've always assumed Digi just didn't want to retool their hardware. The core units do huge amounts of processing load, but bandwidth-wise... standard 33MHz 32-bit PCI is only slightly over a gigabit total bandwidth. Their cards don't require PCI-X or PCIe and there are only about two PCI busses in most modern computers. There really isn't a huge win to using PCI.

If anything, PCI is a big disadvantage because PCI (at least standard parallel PCI) is an internal bus that is being phased out in favor of PCI Express. In the long term, I would hope that Digi will move to something a little more sensible like a card with four FW800 controllers and everything of value outboard.

100 + tracks at 192 k each on firewire? Oh god no.

True enough. 800 Mbit firewire would only handle 99 tracks at 192k/24. Of course, FireWire is spec'ed out to 3200 Mbit/second. Nobody builds the hardware yet, but it's in the spec and has been for years.... I'd imagine Digi would love those sorts of speeds....

 

[LeeRosario]

Great synposis Lee!

The problem I have understanding is teh clock issue. Now mfg and top dog digital guys say that an external clock will not add benefit to the convertors of another device being slaved, yet folks say that adding one does make it less jittery. Then you have folks say that with the all in one boxes couldn't benefit-i.e. If the convertors are marginal, then the PLL would be marginal, yet people say with experience that the mid to upper devices are better with external clocks.
Granted there are a few exceptions, but it appears that a external clock does indeed benefits the sound or "cloud being lifted".

One more note I forgot to add:



Clock information and audio information are correlated. Not the same thing, just correlated *hint*. So if the A/D converstion process is using a solid clock vs a wobbly drunk-ass clock as a guide rail, then I think that gives you an idea of how your conversion is gonna sound.


You don't want the drunk guy to guide you along the highwire, right? You want the sober and steady acrobatic genious to make sure you get to the other side in 1 very desirable piece.

 

[altitude909]

I'm not sure where gta gets his info:

there are only about two PCI busses in most modern computers.

Huh? Are you saying that there are 2 pci slots in most computers?

Their cards don't require PCI-X or PCIe

No. HD CORE Accel is PCI-E and the HD CORE is PCI 64

Possible spec'd speeds and future predictions of whats going to happen in the computer industry have little real value at all and should be taken with a grain of salt.

Intel has already submitted a PCI-E2 spec and as of now no-one other than digi makes any audio interfaces based on PCI-E (or even announced plans to)

 

[mark4man]

Holy s_ _t !*!*!

Guess I came to the right place on this inquiry. Thanks very much for everybody's input...good info, all the way around. And the graph from altitude909 is spot on.

Now...in evaluating all the posts, I get the feeling that some of you feel internal clocks aren't all they're cracked up to be. My understanding is that the crystals & their accompanying oscillator circuits are easily & cheaply implemented...making even pro audio level converters sound good. Is this not right?

But if one wanted to test already recorded audio for conversion jitter...can this be done? If so how...& what type of software?

Also: 909 mentioned digidesign gear; & I have two questions:

Do any of the front ends work with native DAW PC's? (Could I use one with SONAR)?

How are their internal clocks?

Thanks again, everybody.

mark4man

 

[altitude909]

Do any of the front ends work with native DAW PC's? (Could I use one with SONAR)?

I know the 001 has WDM drivers out and they are pretty weak. For the money, non-protools options are vast and good; so if you do not want to run PT, then get something else since its probably not worth the hassle.

As far as clocks go, I know its pretty important when running lots of digital gear but I'm not sure how that impacts a single interface (Lee?)

 

[LeeRosario]

Holy s_ _t !*!*!

Now...in evaluating all the posts, I get the feeling that some of you feel internal clocks aren't all they're cracked up to be. My understanding is that the crystals & their accompanying oscillator circuits are easily & cheaply implemented...making even pro audio level converters sound good. Is this not right?

But if one wanted to test already recorded audio for conversion jitter...can this be done? If so how...& what type of software?

----
How are their internal clocks?

Thanks again, everybody.

mark4man

To my knowledge, I don't know of any software that checks for exsisting jitter on audio material.

I suppose the best thing is simply a with and without comparison using a trained ear. After a while, you can sort of see and feel what a clean wave looks like compared to a jagged wave. However, don't take my word on that. I might of heard of that kind of software, but mastering engineers use it. I'll have to check.

Afterall, the numbers are important up to a certain point, but your ears are really the analytical tool in this situation.

So the clock on things like an OO2, the motu HD/96 interface, the saphire, the mackie interface, and everything else in that species of audio interface will have marginal internal clock. It's a design limitation that most audio companies just haven't addressed in the current day.

Even look at the PTHD interface. They basically give you two different options:

The 192 I/O with it's own A/D conversion (which is not bad) or the best option, the 192 Digital I/O. That bypasses the A/D conversion process at the interface level with a 25-pin digital system.

It gives you the option for custom outboard A/D of your choice. By that point you would buy something like a large format digital console (SSL console or Neve console, or shit, maybe even an Amek) with prestine everything built into it.

So if you want to compare the A/D conversion and Wordclock (timing) relationship to say, pulling up a lifeboat on a cruise liner, it would be like this:

Visualize the boat being pulled up by two cables on either side. The left cable is your workclock information and the right cable is the A/D audio information.


With the current state of internal clocks, this boat is being pulled up rather unevenly and a little shaky, rather than smooth and steady. So the boat generally makes it fine, but it's passengers are a little disturbed.

That's the whole point, we don't want to disturb our signal at all.

But one really disheartening factor I'm about to mention might even further dissapoint or frustrate home users looking for purist sound:

So you get the best clock money can buy, you get the best A/D/A conversion man has ever made; theoretically you should be ready to record prestine audio, right?

Unfortunatly, no. Dithering inadequecies or misuse degrades digital audio. Poor word length calculations degrades digital audio (cheap plug ins that are mathematically limited, even they generate noise and problems). Cables, types of mics, poor CD burning, the list goes on.

In the end, people still say it's cheaper to get warmth and good clarity with analog. I'm afraid that's true.

But in that aspect, I think I'd rather bullshit myself and root for digital cause I have faith in the technology that we will see in the future.

Analog is still going to have alot to do with that.

 

[dgatwood]

Huh? Are you saying that there are 2 pci slots in most computers?

No. Busses. On average. Most machines have about one bus per two slots, give or take. BTW, I'm only talking about parallel PCI, not PCIe here.

No. HD CORE Accel is PCI-E and the HD CORE is PCI 64

My bad. Guess I should have skimmed those specs more carefully. Is that 64-bit 33 MHz or 64-bit 66 MHz?

 

[mark4man]

Lee...

Not meaning to be disrespectful here, but...

I think it's pretty much common knowledge that even pro audio front ends contain stable clocks...in concert with the fact that...since the advent of oversampling converters, great sound quality with a low noise floor is easily achievable (w/ good detail, imaging & w/o smearing.)

And a well written dither algorithm will always help retain a decent amount of high-res ambience when BRCing down to 16.

Guys like Martin Walker (Sound-On-Sound) & hosts of other digital audio experts have written extensively on this.

Are you sure you're just not giving digital a break, 'cause you love analog ???

[&...don't get me wrong...I love analog too...as in, really love it.]

In my question...I was just worried about secondary interface jitter, since this will be the first time I've gone that route.

Anyway...thanks for your input. Hope I didn't ruffle any feathers. I just have a different understanding of it, that's all.

mark4man

 

[dgatwood]

In my question...I was just worried about secondary interface jitter, since this will be the first time I've gone that route.

Sampling jitter is independent of interface technology. The interface side of the hardware just reads data from a ring buffer (a chunk of memory to which data is written in an endless loop, jumping back to the beginning when the end of the RAM is reached). The timing of reading data out of solid state memory isn't really relevant unless it gets so far out of whack that you miss an entire pass across the buffer....

The converter hardware and associated chipset write data into that buffer. Those bits are independently clocked (possibly slaved off an outside clock source, but not tied in any way to the bus clock). Those are the bits of hardware where clock stability matters.

 

[mark4man]

Sampling jitter is independent of interface technology. The interface side of the hardware just reads data from a ring buffer (a chunk of memory to which data is written in an endless loop, jumping back to the beginning when the end of the RAM is reached). The timing of reading data out of solid state memory isn't really relevant unless it gets so far out of whack that you miss an entire pass across the buffer....

The converter hardware and associated chipset write data into that buffer. Those bits are independently clocked (possibly slaved off an outside clock source, but not tied in any way to the bus clock). Those are the bits of hardware where clock stability matters.

That's what I'm talkin 'bout !!!...that's what I needed to know.

Thanks much, dgatwood,

mark4man