Computer Audio Asylum

Music servers and other computer based digital audio technologies.

Return to Computer Audio Asylum

Message Sort: Post Order or Asylum Reverse Threaded

Femto Clock in a Streamer - why ??

73.229.29.71
Posted on October 27, 2015 at 14:29:22
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
 If certain DACs come with precision femto clocks (including the Auralic Vega, for example), is there a benefit to the femto clock in the streamer, like in the Auralic Aries?

Which one acts as the master clock, the one in the streamer or the DAC? If one is the master, then isn't the femto clock unnecessary in the other?



 

Hide full thread outline!
    ...
Good question..., posted on October 27, 2015 at 16:33:24
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

Would REALLY like to know whether adding two Crystek 'femto' clocks to the AMANERO USB board of my new Audio-GD Master 11 DAC/Headphone amp would make any difference.




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

Which one acts as the master clock, the one in the streamer or the DAC, posted on October 28, 2015 at 02:45:18
Roseval
Audiophile

Posts: 1846
Joined: March 31, 2008
 I do think it is protocol dependent

In case of traditional audio protocols like AES/EBU or SPDIF the source is also the generator of the timing. Hence the DAC is slaved.

In case of async USB the timing is decoupled hence the DAC is in control.

As we know "In theory, theory and practice are the same. In practice, they are not."
An example, the Vega can't often be run with the clock in highest precision when connected to a PC with async USB.

The Well Tempered Computer

 

RE: Femto Clock in a Streamer - why ??, posted on October 28, 2015 at 04:41:33
soundchekk
Audiophile

Posts: 2426
Joined: July 11, 2007
 IMO no need for femto clocks. I'm not sure if anybody ever proved the value of a femto or a 1-digit pico clock.
Those specs don't tell you what's left of femto in front of the DAC.
And meanwhile we ALL know that no DAC is immune against timing and noise issues. There can't be much left of "femto".

I do think that the side effects of having better supplies and lower noise levels in this femto/pico system makes the actual difference.
That's why such a clock or better - its peripherals - might also cause changes even if put into a streamer.
We ALL know - every upstream optimization still matters.

They're IMO all fishing in the dark. To me it seems the industry is working in Trial and Error mode.

The "Audiophiles" pay the bill.

-----------------------------------------------------------------

blog latest >> The Audio Streaming Series - tuning kit pCP

 

Interesting take..., posted on October 28, 2015 at 09:05:23
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 Can't seem to find it now that I would like to reference it but...

Seems someone upgraded a AMANERO USB>I2R card in their DAC to Femto clocks and said that the difference was no longer did they have any need or see any improvement with the USB Regen.

So a Regen is the same as a Femto clock on the USB input card/circuitry?

Don't have any intention of trying to replace clock chips at this point and if a $175 Regen will do the job just the same, why not?

Plus if it doesn't work? Sell the Regen for about what it cost?

Not much risk.








First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: Interesting take..., posted on October 28, 2015 at 09:18:02
soundchekk
Audiophile

Posts: 2426
Joined: July 11, 2007
I'm saying this since years. The whole thing can only be solved inside the DAC.
Trying to fix an extremely complex upstream environment is impossible and will only benefit one party - the audio industry.
They can't be interested to fix their DACs as long they can sell all kind of Jitterbugs, USB cables, asf.

I know that people report that the Regen improves by adding better PS, better clocks etc.. The Regen respectively the DACs still respond to cable changes.
And yes, reclocking and rebalancing the stream doesn't mean that changes on the PC doesn't play a role anymore (We have just tested it over here).
A never ending mess.

-----------------------------------------------------------------

blog latest >> The Audio Streaming Series - tuning kit pCP

 

RE: Interesting take..., posted on October 28, 2015 at 10:31:51
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
A never ending mess.

I wouldn't go quite that far. I'd say there are tiny incremental improvements here and there. And right now, these little 'add-ons' are all the rage as they're relatively inexpensive.

It's not like we were all complaining about how horrible our computers + DACs sound before these little tweaks arrived on the scene.

Some work IS being done within modern DACs that were once only attainable with these add-ons. For example, some DACs generate their own clean 5V internally for their USB receivers. In these DACs an external 'injection' of clean 5V on the USB VBUS will have no effect. It's already taken care of within the DAC itself. Similar work has been done for galvanic isolation including transformer or optical coupling.

It's not like the add-on makers have a lock on incremental improvements. It's not proprietary rocket science that the DAC makers cannot wrap their heads around and implement internally..... but these add-ons are cheap, trendy, and popular so the window for PROFIT opportunity and market share is still there to make it worthwhile for these tweak manufacturers to take advantage. ;-) Just my cynical 2-cents worth of course.


 

RE: Which one acts as the master clock, the one in the streamer or the DAC, posted on October 28, 2015 at 11:01:48
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
Thanks, I thought it was something like that but I was thinking mostly along the lines of USB i/o.

In case of async USB the timing is decoupled hence the DAC is in control.

So it would be 'wasteful' to pay the up-charge for the femto clock in the Aries streamer vs the cheaper Aries LE if you plan to interface via USB. I understand there are other minor benefits to the Aries over the LE model but there is significant cost in the unneeded femto clock in this case.


 

RE: Inside the 'DAC'..., posted on October 28, 2015 at 12:30:20
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 The 'DAC' box or the actual 'DAC' which is, for all intents and purposes a chip or a grouping of chips?

The USB > I2R conversion is one area where there is a lot of focus and rightfully so. IF the 'DAC' box responds positively to a Jitterbug or a Regen then on can point the finger at the maker of the 'box', no argument there.

That said, 'garbage in-garbage out' still applies in most all of these cases and perhaps these add-ons will continue to enjoy a deserved following if the indeed can clean up the garbage.




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: Femto Clock in a Streamer - why ??, posted on October 28, 2015 at 19:38:27
Archimago
Audiophile

Posts: 821
Joined: January 18, 2002
 Well, femtoclocks are great at resolving engineering needs of low-jitter digital devices like multi-GHz CPUs, SATA interfaces, PCI Express, etc...

But we're talking audio devices here with sample rates in the low MHz (DSD) and most often <192kHz in terms of timing demands of the final output.

The real question is therefore - what engineering question is the femtoclock "solution" supposed to address? What "jitter" in audio requires this resolution? What device can anyone demonstrate femtoclocks materially improving the output quality in?

Hey, I'm all for technological improvement and it's great that products are out there to be purchased for anyone who wants the tick-box checked when purchasing a SOTA DAC... But I think it's quite clear that this is unnecessary over-engineering whether in the streamer or DAC side. And logically the OP is correct in suspecting that a better clock in the DAC is more important.
-------
Archimago's Musings: A 'more objective' audiophile blog.

 

what engineering question is the femtoclock "solution" supposed to address, posted on October 28, 2015 at 23:45:32
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 With the right power supply and circuit topography, provides lower phase noise at the low frequencies that count ie 100Hz downwards where high SQ gains can be had.

In a usb system, it is no use having a Low Phase noise clock in say a ESS dac but with poorer clocks for the 44.1 and 48 k related streams.

 

RE: Good question..., posted on October 29, 2015 at 00:04:14
analog_sa
Audiophile

Posts: 390
Location: Europe
Joined: December 26, 2002
 The Crystek clocks will certainly improve the Amanero up to a point. Unfortunately good clocks need really good regulators and layout as well. Sold the Amanero after a very brief encounter and don't quite remember - is there sufficient physical space for the Crysteks?

 

RE: Femto Clock in a Streamer - why ??, posted on October 29, 2015 at 07:33:39
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
Well, 'femto clock' sounds cool and everyone needs it... even if it's not used in the scenario I described.

If you want to spend even more, there are outfits that make ridiculously expensive atomic clocks for high-end audio.... rubidium oscillators to be more precise with questionable benefit for this application.


 

IT's been done already...., posted on October 29, 2015 at 07:34:18
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

but I'm not sure to exactly what version of the AMANERO board.

The board in my Audio-gd Master 11 is not exactly like the OEM as Audio-GD has 'licensed' it somehow from AMANERO.

See above.

Those big caps are clearly NOT on the version AMANERO sells as OEM. And they MIGHT be in the way if the CRYSTEK clocks are bit larger in size.

That said, King Wa uses a well regulated linear supply to power the board and all of those big caps MUST be doing SOMETHING, so....

Haven't tried a REGEN as yet. Want to get used to the sound before I try any tweaks.




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

"MOUSER' sells them for about $25 each in onesies...., posted on October 29, 2015 at 07:46:16
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

So not a great big tweak.

That said, they won't fit on my AMANERO board as Audio-gd is using a differing layout from stock, licensed from AMANERO somehow (see above).

Looks like I'm stuck with those crappy MEC clocks. :-(






First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: "MOUSER' sells them for about $25 each in onesies...., posted on October 29, 2015 at 08:09:00
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
 For $25 why not? But DAC makers would like you to believe that these femto clocks are magical and very expensive, often charging $125 to nearly $500 for the upgrade.

Even worse are the DAC makers who try to sell you on the idea that an atomic clock is the cat's meow charging the unsuspecting audiophool several thousands of dollars for something that is known for it long-term accuracy vs its short term phase noise. Long term accuracy is not needed for an audio DAC. Short term phase noise can be achieved with a modest crystal oscillator.

There are certainly valid applications for these atomic clocks (rubidium oscillators) but audiophile DACs is not one of them. They typically cost around $1000 to $2000, but some DAC makers will charge upwards for $10,000 for this option.

For some, I'm sure it's fun to boast that their atomic clock in their DAC is the same caliber as those found in orbiting satellites! LoL



 

I think the true test might be..., posted on October 29, 2015 at 09:44:47
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 Does the REGEN improve the sound of your DAC or not?

But as I have no REGEN as yet, I really have no idea.

I CAN say that the 'Jitterbug' has little effect. It does stay put in the rather loose USB port on the left side of my MacBook Air that the USB cable I normally use, so there's that. =:-0

OK, not very 'Audiophile' but still...


First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

Bargain!, posted on October 29, 2015 at 11:27:45
Roseval
Audiophile

Posts: 1846
Joined: March 31, 2008
 Is it over-engineering or as it is a part common used in computers hence mass produced a way to get extremely low intrinsic jitter at a bargain price :)
The Well Tempered Computer

 

It's not a simple matter, posted on October 29, 2015 at 11:52:39
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 If the streamer is sending USB it has a clock that generates the 8 kHz packet timing. This clock affects when packets are sent and hence when packets get to the DAC. If this clock is erratic than packets will arrive at the DAC at varying times.

In theory with Async USB the effect on this will be minimal, because of the buffering. Even if the average rate drifts out, then the buffer will start filling up more or start emptying out. Long before this creates a problem, the async protocol will send a message to adjust the number of buffer samples. Also, in theory, data will be clocked out of the buffer by a separate local clock, and this is the only clock that will affect the timing of the audio samples.

In practice, any changes in the clock in the source and other changes in the USB cable will affect the timing of events in the USB receiver circuitry in the DAC. This will be true even if all the same feedback signals are generated. The exact timing of current consumption will vary. There will be details of how buffering in the memory buffer work and with different timing of memory loads there may be different timing of memory reads. If the output of the memory buffer is reclocked using the DAC's master clock in theory any timing variations would not matter, in practice they may.

If you feed a different signal into the DAC (even with the same bits) it is reasonable to expect that you will get a different signal out. Ideally, the difference will be perceived as "small". In this case, it would be a waste of money to have sprung for a more expensive clock in the streamer. However, in the real world, the differences may not be "small" they may be "moderate" (and some enthusiastic audiophiles would call these differences "huge"). So it is entirely possible that going for the more expensive clock in the streamer may be worthwhile.

However, it is also possible that streamers with identical circuitry with the same quality clock may still sound different, if the streamer clock sends packets at a slightly different speed. This will affect the logic of the async USB receiver in the DAC. With different logic sequencing the DAC may sound different because of different internal noise generated by the USB receiver circuitry coupling into the analog circuitry.

In other words, you can't predict what is going to happen. In the end the decision would seem to be more affected by the size of one's wallet and the hassle factor of swapping components until one is happy.

If you want a serious headache, consider that the clocks may be operating more or less harmoniously frequency-wise as a function of temperature. This means that the system may even have good days and bad days. If you play the same recording twice the DAC may not produce the same output. After repeated listening the listener may be come completely confused, not able to decide whether everything sounds the same, sounds different, or if the listener is going crazy. :-)


Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: "MOUSER' sells them for about $25 each in onesies...., posted on October 29, 2015 at 12:00:30
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 For $25 why not? But DAC makers would like you to believe that these femto clocks are magical and very expensive, often charging $125 to nearly $500 for the upgrade."

You would need to look at typical business models to get an understanding of the relationship between parts cost and sales price. Small volume manufacturers will need better gross margins to be viable. Also, the distribution chain overhead has to be taken into account, as it adds an additional multiplier over cost.

In addition, there are additional costs to have two model numbers vs. one, both fixed costs and variable costs. And obviously, if an "upgrade" requires rework (e.g. unsoldering) then there will be additional labor costs involved.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

REALLY?, posted on October 29, 2015 at 15:10:30
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 Wow!

Who knew?

The things you can learn from reading an audio chat board.


First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: "MOUSER' sells them for about $25 each in onesies...., posted on October 29, 2015 at 15:52:02
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
I'm not talking about a labor intensive manual upgrade where one sends his DAC in so the manufacturer removes and replaces your clock.

I'm seeing a couple manufacturers who give you the option of including a femto clock at the time of purchase charging you upwards of $500. It's just a good $25 clock so why not include it to begin with? But "femto clock" sounds new and cool so they'll squeeze what they can out of you until the coolness factor wears off. I'm sure these will become just another standard marketing datasheet checkbox item in the future and not a costly audiophile up-charge.





 

RE: "MOUSER' sells them for about $25 each in onesies...., posted on October 29, 2015 at 16:12:28
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 Instead of sending my W4S thousands of miles, I fitted one fro Mouser and it took 30 minutes.

The difficulty will be to change the two XOs on the Amanero based usb to I2S board which will call for a major stripdown.

 

labor intensive ???, posted on October 29, 2015 at 16:14:52
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 post only when you have some knowledge. Typically, with a hot air gun and practice, perhaps 10 minutes.

 

RE: "the two XOs on the Amanero based usb to I2S board"..., posted on October 29, 2015 at 17:15:57
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

Will they fit easily, or more importantly, are they the same size?

And if so, will they make an 'audible' difference. At least as much if not more than just adding a REGEN?

And last but not lease, from my point of view, will the fit in the space available on the Audio-gd version of the AMANERO board, link below:

http://www.audio-gd.com/Pro/dac/Amanero/Amanero1.JPG




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: YES labor intensive...., posted on October 29, 2015 at 18:27:38
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
 It would be "labor intensive" for a manufacturer to do it as an after the fact upgrade when you factor in hourly wage for personnel including shipping, receiving, billing, accounting, unsoldering, soldering, testing, packing, reshipping, etc. Not to mention health insurance for employees, facilities overhead like lighting, heat, air conditioning, etc.

Oh I forgot, you do most of your business with the Chinese on eBay.

Get a clue you simpleton twit.

 

RE: "the two XOs on the Amanero based usb to I2S board"..., posted on October 29, 2015 at 19:18:37
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 The W4S board is not as accessible.

China products do not contain low phase noise XOs even if they look impreeive. The smaller Femtos will fit but skill will be needed so as not to damage other components.

 

Another Fabrication, posted on October 29, 2015 at 19:20:59
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 'you do most of your business with the Chinese on eBay'

appears to be your only posting skill.

 

Thought the AMANERO board was Italian?, posted on October 29, 2015 at 20:03:58
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

"China products do not contain low phase noise XOs even if they look impreeive"

And you can say that categorically for Chinese products, even if they use parts made in Asia by US companies?

Or are you saying the MEC TCXOs in the AMANERO board (Mercury United Electronics, Rancho Cucamonga, California) are fake? Stranger things have happened, of course.

Or just not really 'Low Phase Noise'?

More likely but I assume the AMANERO board you have is a 'real' one or is it one you sourced from Ebay?








First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: Another Fabrication, posted on October 29, 2015 at 20:44:12
Bob_C
Audiophile

Posts: 2667
Location: NY
Joined: July 31, 2000
 His post above illustrates the fact he would not need a "hot air gun."

He is able to generate his own hot air. LOL

 

RE: Thought the AMANERO board was Italian?, posted on October 30, 2015 at 00:06:10
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
 Yes, but I think they allow licensed manufacture. The board you showed may well be oem as the MEC XOs have been used by some Italian manufacturers.

 

RE: "MOUSER' sells them for about $25 each in onesies...., posted on October 30, 2015 at 05:41:47
Dave_K
Even worse are the DAC makers who try to sell you on the idea that an atomic clock is the cat's meow charging the unsuspecting audiophool several thousands of dollars for something that is known for it long-term accuracy vs its short term phase noise. Long term accuracy is not needed for an audio DAC. Short term phase noise can be achieved with a modest crystal oscillator.


Bingo!

It doesn't just lure audiophiles though. I think most of these super stable master clocks are sold to the pro audio market.

I always thought it was stupid design to create a very expensive Rubidium or temperature controlled crystal clock and then put it in a separate box with it's own power supply and grounding scheme, and then route it out to another box with a different power supply and grounding scheme through a single ended coax cable to feed a VXCO. Seems like a great way to ensure high jitter for the benefit of low drift which doesn't even matter.

 

RE: "I think they allow licensed manufacture", posted on October 30, 2015 at 07:06:41
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

Yes they do as AMANERO board in my Audio-gd does not look like the one on the AMANARO web site.

The board is mounted vertically in my DAC so I can't easily read what it says on the chips to be certain they are the same brand as on the stock AMANERO card (MEC).

But I will pop it out and look at some point.




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

Truth is in the listening..., posted on October 30, 2015 at 08:24:32
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

IF a given DAC responds positively to the use of a 'REGEN' type product and sounds even BETTER with 'femto' TCXO clock chips in the USB > I2S card, who can argue?

OK, anyone can but...

I will be able to argue the point when I get my hands on a REGEN, and say even more if I can source an Audio-gd version of the AMANERO card with Crystek TCXO.




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: "I think they allow licensed manufacture", posted on October 30, 2015 at 09:00:15
Bob_C
Audiophile

Posts: 2667
Location: NY
Joined: July 31, 2000
 How about... How does it sound?

And not worrying about the clocks yet.

Are you really going to tear apart your brand new DAC?

 

RE: Truth is in the listening..., posted on October 30, 2015 at 09:36:39
Dave_K
 I wasn't commenting on either of those things. A low jitter clock is a good thing when used in the right place. And a USB > I2S converter or USB > SPDIF converter is a good place to put one.

Abe made a very good point though about high stability/low drift clocks like Rb clocks and oven clocks. They are best for keeping time, not clocking an audio system. And the way they are implemented, jitter performance is likely to suffer.

 

Tear apart a brand new DAC?, posted on October 30, 2015 at 11:54:20
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 The Audio-GD Master 11 is controlled by jumpers on the main board so 'tearing it apart' is 'part' of the process.

But no, I do not intend to start moving jumpers around for a while and that explains why I have no clue the brand of TCXO being used on the AMANERO USB > I2R adapter.

And really not even ready to screw around much with REGEN/Jitterbugs. However this piece of kit is HIUGE, 18" x 18" with the USB connector in the rear and a LONG way away, that I'm currently using the Jitterbug to add a couple inches to my USB cable. =:-0

How does it sound?

Pretty darned good! About as good as I have yet to hear digital in my system, which isn't saying much as I'm not much of a fan of digital.

But good enough for me to be torn between keeping it as a DAC/Headphone Amp next to my easy chair and lugging it upstairs and installing it in my main system

Guess that's pretty good!


First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

Like my daddy used to say, posted on October 30, 2015 at 13:34:42
Jay Buridan
Audiophile

Posts: 10283
Location: Michigan
Joined: January 21, 2004
 A Femto is a woman's toe -- prolly pedicured and painted real pretty :)

"Horse sense is the thing a horse has which keeps it from betting on people. "
― W.C. Fields

 

RE: It's not a simple matter, posted on October 30, 2015 at 15:37:05
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
 If the streamer is sending USB it has a clock that generates the 8 kHz packet timing. This clock affects when packets are sent and hence when packets get to the DAC.

I thought the DAC determines when the packets are sent, in the case of Async USB. At least that was my understanding and why I am asking these questions.

You might be making your answer much more complex than my question, Tony. Let me rephrase....

- If the Streamer is attached to the DAC via USB, whose clock is acting as the Master? The Streamer or the DACs?

- If you HAD to allocate budget for a very high-quality clock, where would you choose to put it? In the DAC or in the Streamer?





 

RE: "some DACs generate their own clean 5V internally for their USB receivers.", posted on October 30, 2015 at 17:22:49
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 "In Audio-gd products, the Amanero Combo 384 have not power supply from computer for the less disturb."

So there!





First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: It's not a simple matter, posted on October 31, 2015 at 08:47:24
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 The situation is more complex than your understanding. There are (at least) the following clocks operating

In the USB 2.0 class 2 async audio computer:

1. A USB frame clock that dispatches a USB frame every 8 kHz.
2. A USB bit clock that times the sending of the bits that compose a USB frame (480 MBPS, typically generated by a frequency multiplier off a lower speed crystal)

In the DAC:

3. A USB receiver clock, used to receive the incoming bits of a packet (480 MBPS, typically kept phase locked to the USB transmit clock so that the bits of the packet can be received)
4. An audio clock running at the sample rate used to time the switching circuits that generate the analog output.

There is also control logic associated with the USB receiver circuitry and the DAC circuitry, and these may involve additional clocks. It is possible to design the DAC so that all of the control circuitry is synchronous with a single master clock, but my suspicion is that most DACs have separate clocks for these sections.

Packets are received by the DAC at (more or less) the time they are transmitted, which will depend on the timing at the computer and the transmission time over the wire. In addition, there will variable delay at the receiver, involving synchronizing the receiver clock with the bits coming in and this may depend on the relative timing and frequency of the computer and receiver bit clocks as well as the signal amplitude. At some point, the receive will recognize the start of packet and will begin accumulating data bits in the USB packet and later, the end of packet. The timing of the end of packet will depend timing of the packet start by the computer and the number of bits in the packet which depends on the audio data. At the end of packet the DAC receiver logic will verify that the packet was correctly received (CRC check passed) and then some number of audio samples are available for playback in a buffer.

Clocking of data out of the buffer is controlled by a clock that should be synchronous with the audio sample clock, ideally isolated from it. (But this may not be the case in some designs.) Because the computer 8 kHz clock is asynchronous with the DACs audio master clock a feedback mechanism is invoked. This allows the DAC to control the number of audio samples in the packets it will be getting. This enables the receiver circuits in the DAC to control the buffer load, preventing underflow or overflow despite a rate difference between the computer and DAC clocks.

In no case does the DAC control the rate at which the clocks in the computer operate.


My experience has been that most computer programmers do not understand the details of hardware clocking, even those who design I/O drivers. Furthermore my experience has been that most hardware designers do not understand the operation of systems involving multiple clocks that operate asynchronously, especially when these are connected by wires that may have noise on them. If you find my explanation difficult to understand, then it is quite possible that (1) you do not have sufficient understanding of how these types of systems really work, or (2) I made a mistake in these posts, which is likely due to the length and complexity of the USB specifications, or (3) this system is too complex for me to have a complete and accurate understanding. :-)

If I had to allocate budget for a USB DAC, I would allocate all of my budget for the DAC, buying the cheapest possible streamer, and spending all of my money on the DAC, including the best possible clocking architecture, clock circuitry, power supplies, USB receiver circuitry and isolation circuitry that attempts complete separation of the USB receiver circuitry from the actual DAC portions.


Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: It's not a simple matter, posted on October 31, 2015 at 09:23:12
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
 Your last paragraph sums it all up and is all I was looking for. I'll paraphrase:

The clock in the DAC is much more important than the clock in the streamer. Thanks.

 

RE: It's not a simple matter, posted on October 31, 2015 at 10:03:09
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 "The clock in the DAC is much more important than the clock in the streamer.
Thanks."

I would agree with that, provided that the DAC has "proper" design. This requires it to correctly implement the async USB architecture. I would not be surprised if any number of "well regarded" DACs to not work that way. It may be that even the DAC designer is not aware that his design is incorrect, because the error may appear inside a proprietary chip that comes with an incorrect or ambiguous specification of its operation.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: It's not a simple matter, posted on October 31, 2015 at 10:23:30
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
 This requires it to correctly implement the async USB architecture. I would not be surprised if any number of "well regarded" DACs to not work that way. It may be that even the DAC designer is not aware that his design is incorrect, because the error may appear inside a proprietary chip that comes with an incorrect or ambiguous specification of its operation.

But how is one to know? As a consumer we are at the mercy of the designer, and possibly a reviewer with decent test equipment.



 

RE: It's not a simple matter, posted on October 31, 2015 at 10:33:40
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 You tell by listening. There is no other way to tell.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: It's not a simple matter, posted on October 31, 2015 at 16:06:27
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
So the only way to tell if Async USB is working correctly is by listening. Hard to believe, but OK.



 

A Computer Audio system is not simple like a Victrola, posted on November 1, 2015 at 08:14:05
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007

"So the only way to tell if Async USB is working correctly is by listening. Hard to believe, but OK."

Computer audio is complicated. USB based computer audio is even more complex than other forms. If you want a simple system that has predictable behavior, my suggestion is to stick to a Victrola. No complex digital logic, analog circuitry, mixed signal circuitry, power supplies and wiring, electronic clocks, circuit board traces, high frequency data signals, firmware, software, etc...

If you are willing and able to open up a DAC and reverse engineer it, including opening up and reverse engineering all of the chips inside the DAC, then in principal you can determine how a few product samples probably worked before you destroyed them in the process of reverse engineering. Otherwise, all you can do is observe the DACs performance, e.g. by measurements of its output or by listening to it. So far, no one has come up with a set of measurements that will adequately test a DAC as a black box to see if it is correctly implemented. Indeed, since your DAC almost certainly has one or more processors that execute firmware you are working on the problem of determining if software works correctly. Telling if software works correctly is impossible by black box tests. (This is proven by theory and confirmed by decades of practice.)

Sorry, but that's the nature of the beast.



Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: A Computer Audio system is not simple like a Victrola, posted on November 1, 2015 at 08:42:51
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
Never suggested that computer audio is simple from a design perspective. It is fairly simple though for most users. My question was pretty straight forward, and you already answered it... I believe. I summarize:

- Given a budget to work within, where would you place the high quality clock for an Async USB setup? In the DAC or in the Streamer?

As I had thought, you and another inmate said in the DAC. That answers my question. Thanks!

 

That would only be the case if the 'high quality' clock was..., posted on November 1, 2015 at 09:09:00
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001

prohibitively expensive, which while expensive relative to most cheap digital parts...

Link below:




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: That would only be the case if the 'high quality' clock was..., posted on November 1, 2015 at 13:17:54
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 I would not suggest changing clocks on a DAC circuit board if you don't know what you are doing. The physical layout of the parts on the board and the circuit traces are all controlled impedance (or should be) and the layout must be optimized for best performance. It is entirely possible that replacing a good clock with a better clock might produce worse performance if this is not done right. For example, if the wiring is not optimal then the result may be less jitter at the clock but more jitter inside the DAC chip where it matters.



Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

That is not what I was suggesting...., posted on November 1, 2015 at 22:14:22
Ivan303
Audiophile

Posts: 48887
Location: Cadiere d'azur FRANCE - Santa Fe, NM
Joined: February 26, 2001
 I was just suggesting that the $25 Crystek TCXO, while more expensive than say a $5 TCXO from MEC or SCTF, is not what I would call 'prohibitively' expensive.




First they came for the dumb-asses
And I did not speak out
Because I was not a dumb-ass

 

RE: That is not what I was suggesting...., posted on November 2, 2015 at 09:40:51
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
 A product being manufactured has a bill of materials. Multiplying the cost of one part 5 times may not matter. However, if this argument is used on many parts the end result will be a parts cost many times what it would otherwise be, and in the end this will mean a large multiple (perhaps 4 times or more) on the product sales price.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

Wait... are you saying it's not as simple...., posted on November 2, 2015 at 11:27:54
Beef Guinness Cutie Pie
Audiophile

Posts: 136
Joined: December 29, 2011
 .... as plugging one end of USB cable into a Mac, another into a DAC, and hitting "Play" on the iPad, or whatever Apple product they use to control their Macs?

Looking at discussions here, I somehow thought that hardest, or at least most time-consuming, part of it is going to the Web forum afterwards. To pontificate endlessly, relentlessly, how nothing else makes any difference, and even if it does, it's not worth "stressing over". And everyone who does - care, not "stress", because that's a just stupid concoction - is not into "listening to music".

 

RE: "some DACs generate their own clean 5V internally for their USB receivers.", posted on November 2, 2015 at 19:33:58
AbeCollins
Audiophile

Posts: 46296
Location: USA
Joined: June 22, 2001
Contributor
  Since: February 2, 2002
Exactly!

 

RE: Good question..., posted on August 25, 2019 at 11:50:34
joebajoe@mac.com
Audiophile

Posts: 1
Joined: June 9, 2011
 Hello,
Yes I am really interested to know. I plan the same upgrade on my Amanero and was wondering if anyone did that already and how much is the improvement. I will really appreciate any info.
Thanks,
Vlad

 

Page processed in 0.043 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]