|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
85.19.92.6
Hi !
sorry for the very trivial questions
I have a big confusion about clock in general and reclocking in particular
A dac can use 3 different clock:
1) the one embedded with the input digital signal
2) one internal
3) one taken from an external clock generator
If i have a dac with a internal quartz and no way to select the clock source i guess the dac always reclock the incoming digital signal, if not what would be the reason of the presence of a quartz inside ?
Am i right ?
So in the end what should really matter is the quality of the dac's internal clock and not the quality of the incoming spdif signal.
Am i right again ?
Then the sound quality out of very good dacs should be quite independent from the quality of the spdif input because there will always be a "reclocking" of the digital signal in the dac.
It is very important for me to understand this because it will free me to chose even a very cheap source and focus all the attention and allocate most of the budget to get a good dac.
Thanks a lot for any kind advice.
Kind regards,
bg
Edits: 01/27/15 01/27/15 01/27/15 01/27/15Follow Ups:
Hi !
sorry for the very trivial questions
I have a big confusion about clock in general and reclocking in particular
A dac can use 3 different clock:
1) the one embedded with the input digital signal
2) one internal
3) one taken from an external clock generator
If i have a dac with a internal quartz and no way to select the clock source i guess the dac always reclock the incoming digital signal, if not what would be the reason of the presence of a quartz inside ?
Am i right ?
Usually such an internal clock is there with the purpose of resampling the data to reduce jitter. This is a PLL type clock, so it is still affected by incoming signal jitter.
So in the end what should really matter is the quality of the dac's internal clock and not the quality of the incoming spdif signal.
Am i right again ?
They both matter.
Then the sound quality out of very good dacs should be quite independent from the quality of the spdif input because there will always be a "reclocking" of the digital signal in the dac.
In theory yes, but in practice it is not independent.
It is very important for me to understand this because it will free me to chose even a very cheap source and focus all the attention and allocate most of the budget to get a good dac.
Thanks a lot for any kind advice.
Kind regards,
bg
Hi and thanks for the valuable advice
So the dac's independence from the transport is just a dream
Sadly because it would simplify the life a lot
Kind regards,
bg
Edits: 01/30/15
Yes, just a dream for most manufacturers.
Hi,
> 1) the one embedded with the input digital signal
Yes. This one MUST be used if the DAC cannot control the information flow rate from the source. This clock can however be filtered to have a similar level of phase-noise as a decent crystal clock. All you need is the right kind of tech.
An example is the TC Audio DICE/DICE2 chip which is found in some Firewire DAC's where it imparts in effect the jitter performance of something close to an asynchronous USB DAC, while using adaptive Clock Firewire or SPDIF.
> 2) one internal
In this case the transport (CD-Driver, Computer etc.) must be slaved to the clock from the DAC. In USB this is taken care of by using the asynchronous protocol, for SPDIF a clock link of some description back to the transport is needed.
> 3) one taken from an external clock generator
This is mostly limited to pro-audio, generally as word clock or AES3id Sync. Usually external clocks offer little advantage over the internal clock generators (actually, they often make things worse). Much depends on the detailed design of the DAC.
> If i have a dac with a internal quartz and no way to select
> the clock source i guess the dac always reclock the incoming
> digital signal, if not what would be the reason of the presence
> of a quartz inside ?
The DAC may use a microprocessor to control a number of functions in the DAC without being at all related to the audio clock. Some receiver chips can use a crystal clock to detect the incoming sample rate.
Other DAC's use Sample Rate converters that convert from all external sample rates to an internal one. Generally using sample rate converters will embed at least some of the source jitter in the output signal. Some recent DAC Chip's have sample rate conversion build in and hardwired into the audio path.
> So in the end what should really matter is the quality of the
> dac's internal clock and not the quality of the incoming spdif
> signal.
No, this is generally not the case. There are many ways to design things. But as in principle you cannot directly replace the SPDIF source clock with a fixed rate crystal clock.
If the local crystal clock is for arguments sake 50ppm different from the source, you will get either 50 dropped samples for every million source samples or 50 repeated samples. This repeat/drop is generally audible as a more or less faint click or tick.
> It is very important for me to understand this because it will
> free me to chose even a very cheap source and focus all the
> attention and allocate most of the budget to get a good dac.
A few DAC's indeed incorporate such internal clocking, but they are extremely rare. I can show you what such a system implemented well is capable of though.
All traces show a 24-Bit/48kHz J-test signal from an AP2 testset via coaxial SPDIF, with the AP2 set up to add 50,000pS (50nS) of jitter as 400Hz squarewave.
For the red/green traces the "reclocking" system (in effect a digital Frequency Locked Loop - FFL - wrapped around a memory buffer) was switched off and the DAC was was allowed to behave like most current DAC's using generic receiver IC's (AKM, CS, BB/TI,Toshiba etc). As one can see, the level of distortion is gross.
For the yellow/cyan traces everything is the same, except the FFL system was enabled. As we can see, 50,000pS of jitter vanish as if they had never existed.
So,the proof is in the pudding. It can be done, it just is not done very often.
Ciao T
At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to untolerable pain. Really, what more could we need?
Hi and thanks sincerely for the very interesting explanation quite beyond my ability to understand sadly
But my question was very very basic
If i am not wrong the very reason of reclocking is to reduce the influence of the transport on the overall sound
So if a dac that reclocks, when well designed and built of course, is used with different transports it should give a very similar sound.
If not it reclocks badly just that.
By the way i stop here ... i will try with different sources and listen for any change in sound.
Thanks again.
Kind regards,
bg
The term "re-clocking" has been commonly used to describe decidedly different jitter reduction methods. Some DACs utilize a technology called Asynchronous-Sample-Rate-Conversion (ASRC), which alters the original sample amplitude values in order to effectively filter jitter from the incoming signal. This is more correctly called a re-sampling technology than it is re-clocking. ASRC does produce two independent clock domains, that of the incoming data signal and that of the re-sampled outgoing data signal. The outgoing data signal is usually timed by a fixed crystal controlled local (located within the DAC box) clock generator, which probably why it is often termed re-clocking.There are other entirely different jitter reduction methods which do not alter the original sample values, yet still produce two nearly independent clock domains. These circuits typically utilize a specialized memory function known as an asynchronous FIFO, and some form of adjustable crystal controlled local clock generator. A DAC box might internally contain one or none of the above two highly effective jitter reduction technologies. In addition, there is yet another common meaning for the term re-clocking, wherein a new clock domain is not created. The data and any synchronously derived clock signals are simply re-aligned to whatever is the existing local clock domain.
In short, the term re-clocking doesn't by itself conclusively tell you whether effective jitter reduction is employed. In addition, we haven't even touched on the jitter that noise conduction across the transport to DAC digital signal interface can induce. A unusally well engineered DAC digital signal interface, such as Thorsten describes, can effectively cure all these problems, however, such is far from common.
_
Ken Newton
Edits: 01/27/15 01/27/15 01/27/15 01/27/15 01/27/15 01/27/15 01/27/15
Hi and thanks a lot for the very valuable explanation
But this is difficult. I think that in the end the only way for me is to try and listen ... with patience and good luck.
I understand that there are so many variables that can have a remarkable impact on the final sound.
Thank you sincerely anyway for the interesting advice.
Kind regards,
bg
Hi,
> The term "re-clocking" has been commonly used to describe
> decidedly different jitter reduction methods.
Thank you. This is what I wanted to bring out.
There is no clear definition here.
If I am taking the System Clock from the SPDIF receiver (which is PLL derived from the data and usually high in jitter) and the use this clock to re-clock the signals to the DAC Chip, before they enter the DAC Chip I can claim "special re-clocking".
Of course, it will not lower the transfer of source jitter and so it will not make the DAC less sensitive to source jitter, even though it may, in practice improve the sound quality of the DAC nevertheless (or make it worse, as such things may be).
On the other hand, what I showed is a case where a new FFL (not PLL) is used to create a new clock that is very low in phase-noise (in practice close to the phase-noise of the crystal oscillator serving as reference for the FFL) and is normally completely static in frequency once locked (I find several minutes between each clock update, which is normally a singe 40ppb (parts per billion) step.
So as long as the actual receiver can lock onto the source stream the re-clocking system removes all source jitter from the SPDIF stream. That too could be called "special re-clocking".
So it is definitely worthwhile understanding exactly what sort of re-clocking technology is employed, rather than to just take the ad-copy and say: "Oh re-clocking - good, now I do not need to worry about crappy sources"...
Because, you might still have to after all.
Ciao T
At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to untolerable pain. Really, what more could we need?
Beppe, this is effectively the same as asking the question " Is the transport important?" or " Which is more important, transport or DAC?".If you search the archives on this forum you will find dozens, maybe hundreds, of postings.
I am confident that you will find the overwhelming consensus to be that there is a lot more to the answer than just the accuracy of the clock controlling the DAC. Possible reasons are proffered in some of the postings.
In some cases the use of a computer as a source rather than a CD transport initially seemed to offer some kind of solution at the "cheap" end of the market. However a short visit to PC Audio in AA indicates that the folks there apparently have discovered all sorts of issues when doing this and there are many, many, postings on the need to optimise the computer before the question of a suitable DAC is considered.
Edits: 01/27/15
In a perfect world, where DAC is immune to anything that's going on on the computer side, optimizing computer wouldn't matter at all.
In real world though, optimizations on computer ("transport") side are equally important - and INCOMPARABLY more cost-effective - as selecting a suitable DAC.
Those who don't understand this fact, too lazy to do anything about it, or just don't want to bother - are crippling their system's performance, no matter what DAC they have in it.
...there are many, many, postings on the need to optimise the computer before the question of a suitable DAC is considered.
Yes, many many such postings from just a few individuals. But that wouldn't make sense to me. Why would one spend a lot of time optimizing his computer before having a 'suitable' DAC? You need a DAC in order to hear the effects of any computer optimizing so would you propose optimizing the computer around a less than 'suitable' DAC?
The truth is, a good Async DAC will require less computer optimizing. Just a short few years ago we didn't have Aync DACs in which case the effects of computer will be greater than it is today. I would get a good Async DAC first, rather than chase my tail in circles optimizing the computer first.
"so would you propose optimizing the computer around a less than 'suitable' DAC?"I wouldn't propose anything. I merely reported on what I had read from others posting on PC Audio. I have no opinion one way or the other on optimising computers as I don't use them for music.
My purpose was to point out to the OP that many threads already exist which relate to his question.
Edits: 01/27/15
Hi and thanks for the valuable reply.
yes the issue is exactly the transport influence on the sound out from the dac.
I would dare to say that a dac whose sound is depending on the transport is not a well designed dac.
There must be something wrong with its design.
Very soon i will try a dac which is marketed with the words "... The secret to the dac's unique qualities lies in a proprietary re-clocking technology ... "
I have already a pretty indecent source to use with it ( a media player just to be precise of which i like the flexibility and the fact that can be controlled from a tablet).
I am expecting no substantial differences between different transports, like it should.
Thanks again.
Kind regards,
bg
Edits: 01/27/15 01/27/15
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: