Digital transport brawler

[Strictly Stereo]

As I mentioned, the best place to have the clock is right next to the D2A chip. Putting the same clock somewhere else will be worse.

Slaving a DAC to a clock generated further upstream is a legacy of a time when flow control and async digital data transfer was expensive. Nowdays chipsets to do this cost pennies. How much did that last USB memory stick cost you bought?

What does a USB memory stick have to do with USB audio?

​

If you choose to buy a DAC that relies upon a synchronous approach, then of course you are dependent on the quality of the input clock, but it will never have the same jitter performance as a DAC wth a local clock.

Not necessarily. It just shifts the dependency to the source.

 

[davedotco]

This suggests to me that integrating the source (transport, whatever) with the dac, with a single local clock, no SPDIF conversion is the way to go. I know many enthusiasts like to choose their own dac but this can bring it's own issues, at sensible prices integrated devices seem to have the advantage.

I am expecting to rebuild my system in the coming month or two and the big issue for me is going to be the streamer/dac/preamp. I am very attracted to the WXC50, which does pretty much everything I need at a very sensible price. Naturally I am suffering from a bad case of 'audiophilia nervosa', predominately caused by the low cost of the unit, the nagging 'can't possibly be any good at that price' being the big problem.

 

[ellisdj]

beggars the question is everything in one box really better than seperated better quality boxes, I still think the seperated better boxes will be better regardless of whether the approach is deemed optimal.

 

[andyjm]

beggars the question is everything in one box really better than seperated better quality boxes, I still think the seperated better boxes will be better regardless of whether the approach is deemed optimal.

You are free to think what you like, the engineering truth is that the further the clock is from the D2A chip, the worse will be the jitter.

There may be other reasons for you to prefer a particular DAC, a really good synchrounous DAC may sound better than a poor async DAC, but all other things being equal a local clock is best.

 

[andyjm]

As I mentioned, the best place to have the clock is right next to the D2A chip. Putting the same clock somewhere else will be worse.

Slaving a DAC to a clock generated further upstream is a legacy of a time when flow control and async digital data transfer was expensive. Nowdays chipsets to do this cost pennies. How much did that last USB memory stick cost you bought?

What does a USB memory stick have to do with USB audio?

A USB memory stick has a microcontroller in it to handle the USB data flow. About the size of your little finger nail, it handles all of the handshaking, USB protocol and data flow management. Given the low price of memory sticks, my aim was to point out that the hardware to handle async USB was readily available and of negligable cost.

 

[andyjm]

​

If you choose to buy a DAC that relies upon a synchronous approach, then of course you are dependent on the quality of the input clock, but it will never have the same jitter performance as a DAC wth a local clock.

Not necessarily. It just shifts the dependency to the source.

All other things being equal, the introduction of a length of wire between the clock and the DAC will introduce jitter.

Clearly if you have a poor local clock, and a much better external clock then the chances are the external clock will be a better bet.

If however you have exactly the same clock circuit, then mounting it next to the D2A chip will allways be better than mounting it in an external box.

 

[Strictly Stereo]

A USB memory stick has a microcontroller in it to handle the USB data flow. About the size of your little finger nail, it handles all of the handshaking, USB protocol and data flow management. Given the low price of memory sticks, my aim was to point out that the hardware to handle async USB was readily available and of negligable cost.

Yes, but what does that have to do with USB audio?

 

[Strictly Stereo]

All other things being equal, the introduction of a length of wire between the clock and the DAC will introduce jitter.

Clearly if you have a poor local clock, and a much better external clock then the chances are the external clock will be a better bet.

If however you have exactly the same clock circuit, then mounting it next to the D2A chip will allways be better than mounting it in an external box.

I understand your point, but we need to consider jitter across the entire system, not just the DAC.

 

[andyjm]

All other things being equal, the introduction of a length of wire between the clock and the DAC will introduce jitter.

Clearly if you have a poor local clock, and a much better external clock then the chances are the external clock will be a better bet.

If however you have exactly the same clock circuit, then mounting it next to the D2A chip will allways be better than mounting it in an external box.

I understand your point, but we need to consider jitter across the entire system, not just the DAC.

Why? It is only jitter at the DAC that matters as it impacts the D to A process.

Jitter elsewhere in the system is irrelevant unless it is so extreme that it causes bit errors (which it won't be)- that is the beauty of digital data transmission.

 

[andyjm]

A USB memory stick has a microcontroller in it to handle the USB data flow. About the size of your little finger nail, it handles all of the handshaking, USB protocol and data flow management. Given the low price of memory sticks, my aim was to point out that the hardware to handle async USB was readily available and of negligable cost.

Yes, but what does that have to do with USB audio?

Apologies if I wasn't clear. Some up in the thread someone (you?) had made comments about prefering 'simpler' DACs. I was trying to demonstrate that very cheap, readily available hardware would allow the construction of a 'simple' async DAC.

 

[Strictly Stereo]

Why? It is only jitter at the DAC that matters as it impacts the D to A process.

Jitter elsewhere in the system is irrelevant unless it is so extreme that it causes bit errors (which it won't be)- that is the beauty of digital data transmission.

Not at all. What matters is jitter across the entire recording and playback system.

 

[andyjm]

Why? It is only jitter at the DAC that matters as it impacts the D to A process.

Jitter elsewhere in the system is irrelevant unless it is so extreme that it causes bit errors (which it won't be)- that is the beauty of digital data transmission.

Not at all. What matters is jitter across the entire recording and playback system.

Nope. Jitter matters at the point the A2D conversion takes place. It matters again when those digital samples are turned back into analogue.

It does not matter apart from that - as long as the jitter isn't so extreme that it introduces bit errors.

Why do you think it matters apart from the ends of the chain?

 

[Strictly Stereo]

Nope. Jitter matters at the point the A2D conversion takes place. It matters again when those digital samples are turned back into analogue.

It does not matter apart from that - as long as the jitter isn't so extreme that it introduces bit errors.

Why do you think it matters apart from the ends of the chain?

Here...

http://www.thewelltemperedcomputer.com/KB/BitPerfectJitter.htm

 

[andyjm]

Nope. Jitter matters at the point the A2D conversion takes place. It matters again when those digital samples are turned back into analogue.

It does not matter apart from that - as long as the jitter isn't so extreme that it introduces bit errors.

Why do you think it matters apart from the ends of the chain?

Here...

http://www.thewelltemperedcomputer.com/KB/BitPerfectJitter.htm

Indeed. At the risk of leaning on my walking stick, when I graduated (albeit in the dark ages) I designed high speed data communication test equipment. Unless jitter has changed a lot since then, I think I have a decent handle on it.

Why do you think it matters apart from the ends of the chain?

 

[Strictly Stereo]

Indeed. At the risk of leaning on my walking stick, when I graduated (albeit in the dark ages) I designed high speed data communication test equipment. Unless jitter has changed a lot since then, I think I have a decent handle on it.

Why do you think it matters apart from the ends of the chain?

Because jitter can be introduced at various places in the recording and playback chain, not just during AD or DA conversion but in the digital domain as well. What you eventually hear is the cumulative effect of the entire chain, not just the DAC. Why do you think that it only matters at the ends of the chain?

 

[nopiano]

Indeed. At the risk of leaning on my walking stick, when I graduated (albeit in the dark ages) I designed high speed data communication test equipment. Unless jitter has changed a lot since then, I think I have a decent handle on it.

Why do you think it matters apart from the ends of the chain?

Because jitter can be introduced at various places in the recording and playback chain, not just during AD or DA conversion but in the digital domain as well. What you eventually hear is the cumulative effect of the entire chain, not just the DAC. Why do you think that it only matters at the ends of the chain?

Your own link said so! Here it is:-

“As long as you stay in the digital domain, jitter is not a problem. Computers are designed with jitter in mind and it should be low enough to avoid bit flipping due to timing errors.

Copy a audio file from one device to another is a matter of a bit perfect copy not a matter of a jitter free transmission. Jitter in digital audio only counts when it is converted to analogue by the DAC. A DAC which is input jitter immune saves a lot of problems. In this case the only source of jitter are the components of the DAC.”

That seems to me to be what Andy is saying.

I can can get that we might feel anything that keeps jitter low throughout is ‘better’, because there are often arguments about processors not having to work so hard, etc., but I’m not sure how relevant to the resultant analogue signal that might be.

 

[ellisdj]

do the bits all sound the same - if they come off a cd, usb drive, hard drive nas drive or the cloud into the same dac - does it all sound the same - nope not in my experience so jitter before the dac must matter

 

[Strictly Stereo]

Your own link said so! Here it is:-

“As long as you stay in the digital domain, jitter is not a problem. Computers are designed with jitter in mind and it should be low enough to avoid bit flipping due to timing errors."

Copy a audio file from one device to another is a matter of a bit perfect copy not a matter of a jitter free transmission. Jitter in digital audio only counts when it is converted to analogue by the DAC. A DAC which is input jitter immune saves a lot of problems. In this case the only source of jitter are the components of the DAC.”

That seems to me to be what Andy is saying.

I can can get that we might feel anything that keeps jitter low throughout is ‘better’, because there are often arguments about processors not having to work so hard, etc., but I’m not sure how relevant to the resultant analogue signal that might be.

Read those sentences in their original context and it is pretty clear that the author is referring to file transfers in this section. Jitter is not an issue here, except in theoretical extreme circumstances which are unlikely to be encountered in the real world. Even then, there are mechanisms in place to ensure that file data can be copied from one place to another reliably.

However, DACs do not deal with files. They deal with streams. Jitter does matter when handling streams, which the article goes on to explain. Just to clarify, I am talking about bit streams here rather than streams from TIDAL, Spotify etc.

 

[andyjm]

Actually, DACs deal with 16 bit samples (for CD quality) it just happens that the data link has a clock signal embedded in it (if the data is transferred via a S/PDIF link).

The jitter on this clock is entirely dependent on the last step in the chain, and is not impacted by anything that has gone before.

This is why digital is so powerful, it gets completely regenerated at each step. It doesn’t matter how many times, it is still as good as it started off.

 

[Andrewjvt]

do the bits all sound the same - if they come off a cd, usb drive, hard drive nas drive or the cloud into the same dac - does it all sound the same - nope not in my experience so jitter before the dac must matter 

https://youtu.be/bFN8WYpEmho

 

[Andrewjvt]

Indeed.  At the risk of leaning on my walking stick, when I graduated (albeit in the dark ages) I designed high speed data communication test equipment. Unless jitter has changed a lot since then, I think I have a decent handle on it.

Why do you think it matters apart from the ends of the chain?

Because jitter can be introduced at various places in the recording and playback chain, not just during AD or DA conversion but in the digital domain as well. What you eventually hear is the cumulative effect of the entire chain, not just the DAC. Why do you think that it only matters at the ends of the chain? 

Your own link said so!  Here it is:-

“As long as you stay in the digital domain, jitter is not a problem. Computers are designed with jitter in mind and it should be low enough to avoid bit flipping due to timing errors. 

Copy a audio file from one device to another is a matter of a bit perfect copy not a matter of a jitter free transmission. Jitter in digital audio only counts when it is converted to analogue by the DAC. A DAC which is input jitter immune saves a lot of problems. In this case the only source of jitter are the components of the DAC.”

 

That seems to me to be what Andy is saying.  

 

I can can get that we might feel anything that keeps jitter low throughout is ‘better’, because there are often arguments about processors not having to work so hard, etc., but I’m not sure how relevant to the resultant analogue signal that might be.  

To the understand club
Haha

 

[Andrewjvt]

Why is there always a hifi 'seller'
Arguing with a designer/scientist/engineer?

Dismissing anything they say based on stuff they have read or heard in a marketing brief.

Andyjm
Thank you for your contributions on here and your experience/knowledge on the subject is not wasted.

Please don't stop no matter how dismissive hifi shop men act. I suppose they have to make a living somehow.

 

[Strictly Stereo]

Actually, DACs deal with 16 bit samples (for CD quality) it just happens that the data link has a clock signal embedded in it (if the data is transferred via a S/PDIF link).

The jitter on this clock is entirely dependent on the last step in the chain, and is not impacted by anything that has gone before.

This is why digital is so powerful, it gets completely regenerated at each step. It doesn’t matter how many times, it is still as good as it started off.

Yes and those 16 bit samples are transmitted as a stream of bits. In the case of S/PDIF transmission, jitter is largely dependent on the sending device, but may also be dependent on the clock or conceivably clocks inside the DAC, depending on its design.

 

[Strictly Stereo]

Why is there always a hifi 'seller' Arguing with a designer/scientist/engineer?

Dismissing anything they say based on stuff they have read or heard in a marketing brief.

Andyjm Thank you for your contributions on here and your experience/knowledge on the subject is not wasted.

Please don't stop no matter how dismissive hifi shop men act. I suppose they have to make a living somehow.

Nobody is being dismissive here. We are having a friendly conversation. And for what it is worth, I worked in IT for 15 years before starting my business, so I do know a little about this subject.

 

[andyjm]

Actually, DACs deal with 16 bit samples (for CD quality) it just happens that the data link has a clock signal embedded in it (if the data is transferred via a S/PDIF link).

The jitter on this clock is entirely dependent on the last step in the chain, and is not impacted by anything that has gone before.

This is why digital is so powerful, it gets completely regenerated at each step. It doesn’t matter how many times, it is still as good as it started off.

Yes and those 16 bit samples are transmitted as a stream of bits. In the case of S/PDIF transmission, jitter is largely dependent on the sending device, but may also be dependent on the clock or conceivably clocks inside the DAC, depending on its design.

Just to clarify the discussion, a synchronous DAC slaved in some way to the previous device in the chain is a bad engineering idea.

An async DAC with its own internal clock, using some form of flow control to regulate the data fed to it is a good engineering idea.

That doesn't mean all async DACs are good, and all synchronous DACs are bad, but all other things being equal, an async DAC is certainly a better bet.

The other point to clarify, as it seems to have got a few confused, is that jitter is not cumulative in a digital system. A moments thought would point this out. Is the quality of the music downloaded from a server in Cupertino dependent on the jitter of the data server? Of course not. The jitter that matters is the jitter of the A2D converter at the start of the chain, and the jitter of the D2A converter at the end of the chain. The rest in between is irrelevant.

Which gets back to an earlier point, that with a decent async DAC, a RasPi for £35 is just as good as a Whiz Bang streamer for £10,000.