Best Low Jitter feed for external DAC

[barrows]

I would submit that expectations such as the following:

 

"but frankly if you are going to maintain an online presence, as a manufacturer, I don't think it's too much to ask of ANY company to stand behind their claims with either subjective tests or objective evidence."

 

are a little too much to expect from the principals of small audio companies. I think we are lucky that folks such as Gordon Rankin take time out of their schedules to try and share some of their experience and knowledge on forums such as this one. As listeners/consumers/customers/enthusiasts it is our obligation to listen to the products in question, and decide for ourselves if a given manufacturers product will work for us. I do not think that audio companies need to "prove" their products value to us, the proof (or not) of a products value is already evident to anyone who is willing to listen.

Personally, I am fascinated by the theories/ and approaches of different designers, and I love being able to get a little insight into how/why a given designer might use a given approach, but I do not need or expect a designer to try and prove the superiority of their approach in a forum posting.

 

[Thomas J]

For my part, I rather appreciate Gordon's presence and contributions. He is among the most forthright of manufacturers I've run across in public boards like this, and I find his willingness to share technical details a benefit to those of us at lower rungs of technical sophistication.

 

As to ABX testing, it's for me among the most irrelevant forms of audio testing.

 

[hifitubes]

I don't expect anyone to prove anything, this is a place to discuss ideas and experiences. I don't sit around abc testing, ever. I simply thought an expert's opinion would be more revealing than, "they better wake up and smell the coffee". I know you could write a book on the subject, but I simply expected more. I apologize if that is out of line. I guess Gordon has no opinion on the specifics I posted, that's his prerogative.

 

I don't think it's too much to ask for technical details either. You test drive a car....would you buy it solely on the merits of the subjective driving experience?

 

 

 

[PeterSt]

Jitter is and remains a "funny" subject; Suppose I could produce a DAC with 2fs of jitter I guess I would be raving about that, knowing that everybody knows the lower the number, the better it is.

 

IS ...

 

But now I am not someone who cares about jitter, just because I think other matters are more important (than jitter). Still, I produce that DAC.

Sure I won't say something like "the more jitter the better it is". I even won't say "the more jitter the better it SOUNDS". The only thing I'd say is "I don't care because other things are more important to the SOUND than jitter".

 

Boy, would I have a hard time in proving my right without you listening to the product ...

 

A bit similar is mr. Lavry's, wel, shouting. Unless I first create a product which proves the opposite, Lavry can only be right (who didn't invent the whole lot in the first place).

I already knew he is wrong (the whole theory is - and I am referring to his paper on filtering now), but keep in mind that what I just wrote today in the other post can't stand if it would be theory only. Thus, it is fairly easy to state that a filter which doesn't ring will sound better than a filter which does ... but which doesn't imply that such a filter can exist and thus is such a statement worthless.

Similarly mr. Lavry can talk about those filter theories without even mentioning that filters ring and that this is bad. I mean, as long as ringing is just needed to do the filtering in the first place it is just a good thing (because the filtering is a good thing).

But what when someone comes along with a non-ringing filter ? only then it is proven that the theory is wrong, or not the best at least.

 

You know, almost everything is more complicated than seen at first glance. Let's hop back to mr. Rankin again;

 

As said, it is not difficult to state that less jitter is better, especially if less jitter can be produced by "you" (Gordon in this case). However, thinking USB, the means to get there is asynchronous. But what if there's this software player which is able to influence the DAC for the better (sound quality). And what if this software can't do its job because of the asynchronous connection ? Will the less jitter phenomenon win from the no-influence ? or may the influence overrule the jitter in the first place ?

 

I know, I may be talking abacadabra to some (because of being brief), but those who know (me) know what I mean; All is in a context, and maybe most "lies" emerge from leaving out the context.

 

DA jitter matters at ONE LOCATION - where the digital signal is altered to analog.

 

Prrrt ...

Peter

 

 

PS: I am not trying to rant here, but I just think there is much more around digital audio we don't know yet, and as long as phenomena are unknown, theories may stand. This does not imply they are correct though, as many can prove empirically already.

Sorry for the blahbah; I guess it's my mood today :-)

 

 

[Wavelength]

Jkeny,

 

First using the FitPC with a linear power supply is not good enough for a number of reason's. The first is that the supply goes into a switching power supply that on mine looks like it has 4 or 8 outputs. I am sure there is some core power, 3.3v, 5v, 12v +/-. But the big deal is the damn thing is not powerful enough. Even running Linux it struggles with high resolution material.

 

Again as the elf said using I2S outside the box is just a stupid idea. Since there are no specs, the connection between various units is at best a shot in the dark.

 

The audio oscillator needs to be next to the dac chip. There needs to be 2 of them to support a range of sampling frequencies.

 

If you are looking at the SB units look at the Duet it's your best bet to get good performance. Without the screen there is less noise in the unit and therefore better performance.

 

Thanks

Gordon

 

[hifitubes]

http://www.cappuccinopc.com/ has quite a few choices and details.

 

I was looking at some linear PSUs, but Wayne at Bolder Cable said they weren't really audiophile and would only improve performance a little, and advised not spending more than 100 dollars (they cost about 300 for 12v 10A which is what the 1616M uses).

 

What about the picoPSU? I use an 80w version. If you can't go linear, low power may be the next best thing. You could possibly modify a linear supply to use as I think they do 12v at 6A?

 

 

 

[Wavelength]

HifiTUbes

 

I don't expect anyone to prove anything, this is a place to discuss ideas and experiences. I don't sit around abc testing, ever. I simply thought an expert's opinion would be more revealing than, "they better wake up and smell the coffee". I know you could write a book on the subject, but I simply expected more. I apologize if that is out of line. I guess Gordon has no opinion on the specifics I posted, that's his prerogative.

 

Look I could post some nice pictures and graphs from my Prism dScope III, Stanford or my Wavecrest DTS. But the thing is I would piss off every engineer who is designing this stuff. I have not made an SPDIF or I2S dac in many years as well told that Asynchronous USB is significantly better at jitter than SPDIF every could be. I have a number of companies dacs here. Some because they claimed things that frankly were not possible. Some because people just sent them to me. Believe me I have a lot of data with regards to jitter and jitter testing.

 

Simple statement that no engineer will refute...

 

A set oscillator will always outperform a moving oscillator with respect to jitter.

 

SPDIF, Adaptive, Firewire streaming all require moving clocks.

 

Thanks

Gordon

 

[cfmsp]

 

"SPDIF, Adaptive, Firewire streaming all require moving clocks."

 

Gordon, it's my understanding that Firewire does not REQUIRE 'moving' clocks, and that Firewire can be implemented using fixed oscillators (in the DAC) as the master clock.

 

Perhaps you are referring to 'native' (i.e. no drivers rquired) Firewire?

 

clay

 

 

 

[hifitubes]

Cool. Thanks GR. I really didn't mean to jump on your back. So is a set oscillator found only in async usb, and not adaptive mode I assume?

 

I'm considering going USB soon for playback. Currently I am using an external sound card with BNC pushing to a Nova with BNC.

 

I recently was tempted to go with a USB ADC for a project but it was cheaper to have a mod done to my external sound card for BNC and get a Black Sparrow. Maybe someday I can compare this particular USB ADC I have in mind with my current setup.

 

[hifitubes]

I'm curious if using a word clock capable DAC for example would cure the most, some, or all of the ills of S/PDIF.

Is async USB just the "new" and easier way to do this? More possibilities with async for sure.

 

 

[jkeny]

Interesting to see all the comments!

 

Thanks for the suggestion Gordon.

 

Here's what I went for - a SB receiver (Santa left one at the bottom of my bed ) - due to be modded over the holidays with an external 3.3V & 1.2V PS (batteries) - I'm only interested in 44.1KHz wired operation so I'll be providing a 11.289MHz clock directly to the Xilinx CPLD - this will then be feeding I2S (short 1-2" leads) to an ES9022 DAC located inside the box (possibly fed by it's own low jitter clock?). Basically what I'm aiming at, initially anyway, is a minimal configuration with only the PS supplies & chips necessary for my 44.1KHz wired ethernet operation - so I'm turning off anything that is not needed!

 

[barrows]

that are interested in how/why an Async USB solution delivers low jitter vs. SPDIF the following paper gives a pretty good run through of how/why Async USB gives good results with regards to jitter performance:

 

http://www.ayre.com/pdf/Ayre_USB_DAC_White_Paper.pdf

 

This is written such that anyone can understand it.

 

[Wavelength]

Jkeny,

 

Here's what I went for - a SB receiver (Santa left one at the bottom of my bed ) - due to be modded over the holidays with an external 3.3V & 1.2V PS (batteries) - I'm only interested in 44.1KHz wired operation so I'll be providing a 11.289MHz clock directly to the Xilinx CPLD - this will then be feeding I2S (short 1-2" leads) to an ES9022 DAC located inside the box (possibly fed by it's own low jitter clock?). Basically what I'm aiming at, initially anyway, is a minimal configuration with only the PS supplies & chips necessary for my 44.1KHz wired ethernet operation - so I'm turning off anything that is not needed!

 

You would be better off using say a 22.5792 into the ESS and then divide that by 2 to the CPLD using a Flip Flop. This way you would be using the ESS in sync mode with the I2S interface. The big deal will be feeding the ESS supplies. This is no trival matter and using the EVAL schematics are not the ticket. That and you really need to set a couple of internal registers to make the part really go.

 

~~~~~~

 

HIFITubes,

 

So is a set oscillator found only in async usb, and not adaptive mode I assume?

 

Correct, Adaptive has to use a Frequency Synthesizer which has tons of jitter and on top of that it has to be adjusted ever 1ms which adds more jitter to the I2S feed.

 

Also Word Clock will not help SPDIF at all. It merely synchronizes the clocks. Word Clock still requires a PLL/VCXO to match the receiving WCLK. This in some cases adds more jitter to the solution.

 

Thanks

Gordon

 

 

 

[jkeny]

You would be better off using say a 22.5792 into the ESS and then divide that by 2 to the CPLD using a Flip Flop. This way you would be using the ESS in sync mode with the I2S interface. The big deal will be feeding the ESS supplies. This is no trival matter and using the EVAL schematics are not the ticket. That and you really need to set a couple of internal registers to make the part really go.

 

So Gordon, you reckon the Sabre Dacs are best operated in synch mode - why?

 

I'm not using the full blown Sabre 32 bit DAC but the ES9022 (24 bit) with internal op-amp Vout - no registers to set & simple, single 3.3V PS - maybe not the best of the ESS DAC range but a fine performer nonetheless & easier to locate close to the CPLD. I might go with the 32 bit DAC in time?

 

[Elprior]

Wow Clay, Gordon has just ruined your entire year

 

Elp

 

[cfmsp]

 

Elprior,

 

My girlfriend just asked - "what's so funny?" Apparently I really did LOL!

 

 

Actually, I'm very confident in 'my understanding', and was essentially asking Gordon if he wanted to rephrase his comment, since it is incorrect as stated, unless he's referring to Firewire's capabilities sans custom drivers, which is probably the case.

 

thanks again for the laugh,

clay

 

 

[Elprior]

You're welcome.

 

Btw, you'll be a happy judge for I'm considering going fire(wire) : weiss int202, although not exactly your purist way (I would be using dual aes in such a case). My usb adaptative implementation does sound a bit (tiny) harsh, and thus not entirely satisfying.

Of course firewire will have to prove better (still waiting for Gordon's usb to spdif adapter to make a choice, and Sonicweld next implementation... harsh times !)

 

Elp

 

[Audio_ELF]

Elp said "Wow Clay, Gordon has just ruined your entire year :)"

 

Well at least what's left of it...

 

Eloise

 

[cfmsp]

thanks Eloise.

 

Re the Firewire discussions throughout 2009, a recent post by BJ Buchalter - designer of the Metric Halo ULN-8 - recently clarified a point that some have wondered about, i.e. whether Firewire is (almost) always asynchronous. Apparently, it is NOT.

 

To set the context, the thread was about the differences between AES/S/PDIF and Firewire/USB interfaces, and the original response gave USB much more credit than it deserved, as astute readers at CA would know.

 

Poster:

"FW/USB based playback uses pull data delivery, which means there is no clock at all, just data, being delivered to the interface. The interface then creates the clock (as vs. gegenerating the clock), and it plays the data without delay requesting additional chucks of data as needed. So playback can start virtually free of buffer delays, there is no regenerated clock, just a clock generated as close to the destination as possible, and that's that.

 

So yes FW/USB based playback is architecturally vastly superior, unless the designers of the equipment involved take shortcuts such as deriving the audio clock from the FW bus clock or similar penny-pinching engineering decisions that negatively affect the sound."

 

BJ's response:

"Actually, that is not true of "run-of-the-mill" USB audio -- which is push. IEC-61883 can also be configured as push, with roughly +/- 40-60 ns of jitter that needs to be squeezed out; that is what you get when you clock multiple devices via FireWire (rather than word-clock). So I can say with certainty that none of those considerations apply to the MIO, but you can't say that about most other FireWire and almost all USB solutions. In general it will depend on a number of factors, including how things are configured."

 

 

IEC-61883 is part of Firewire spec, so he's saying that it is quite possible to use Firewire like run of the mill USB (aka adaptive), perhaps even likely.

 

clay

 

PS, the thread was called "Clocking" on Metric Halo's MIO Forum for those interested.

 

 

 

 

 

 

[joster]

Hi cfmsp,

 

So, may I concluded that Metric Halo ULN-8 could be a suitable DAC for a high-end music playback system? Even, if I don't need any of the other pro features (recording capability, etc.)?

 

Is there any other recommneded pro audio products I should investigate?

 

Or should I stick with the usual high-end stuff - Benchmark, Weiss, Empirical Audio, Berkeley Audio Design, Grace, Lessloss, etc. - and pay what they charge, that is, a lot?

 

Cheers,

Joster

 

[Audio_ELF]

Well Benchmark, Grace and Weiss are primarily pro-audio companies. Antelope Audio is another pro-audio company developing a line of more consumer orientated products.

 

If you don't need recording facilities the Metric Halo LIO-8 comes without the ULN-8's mic preamps.

 

 

 

[Codifus]

Ooops. Old thread:)

 

CD