Help please - streamer question

[paradiziac]

IMO. Also isolation of components (ie. it's effectiveness) and quality of power supply.

Yep. Much more so with digital sources than amps, IME.

 

[CnoEvil]

IMO. Also isolation of components (ie. it's effectiveness) and quality of power supply.

Yep. Much more so with digital sources than amps, IME.

Would now be a good time to run? :shifty:

 

[andyjm]

Hi all,

If you are running a streamer into an external DAC (high-end) via a digital out, does it make any difference to the sound quality which streamer you use (assuming the same type of file is used)?

Because if it doesn't, why, if you have good DAC, would you buy anything other than a Squeezebox Touch? It would seem pointless to spend many times more than that if all the DAC is reading is a bitstream.

Forgive me if this is daft, but I'm only just dipping a toe in the streaming water at the moment.

Thanks for any advice you can give.

As Mr Coates rather long-windedly explains above, jitter at the point the digital signal is turned back into analogue (the D2A chip in the DAC) really matters.

It is how the DAC generates that clock that decides whether the streamer has any impact on the overall sound quality.

A DAC clock that is slaved to the clock in the streamer via the S/PDIF link will be very sensitive to streamer quality, a DAC that has a local clock and an asynchronous link to the streamer (async USB or wordclock output) won't be sensitive at all. Halfway in between are DACs with jitter mitigation circuitry - input filtering using multiple PLL steps, FIFO buffers and whatever else designers can think of.

So like much in life, the answer is not a simple yes or no - it depends on the topology of your system, and the design of the DAC.

At what level is jitter audible and how does that figure compare with the levels of measured jitter in AV products?

Well, jitter isn't a single number, nor is it measured in the same way in various tests and publications. As a random variable you have to consider the distribution, and whether you are measuring peak or RMS. One of the best summaries I have seen is this:

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

The article points out that the threshold of jitter audibility is quoted as being anywhere between 250nS and 20pS, depending on who was doing the test, when they did it and what exactly they measured.

The 'best' type of jitter to have is purely random, the 'worst' type is correlated in some way to the programme material. A quirk of S/PDIF is that it can introduce 'code correlated' jitter which has been shown to be more audible than purely random jitter.

As for AV equipment, comparable figures are hard to come by. It would be great if all manufacturers published jitter figures to the same standard, but they don't. For a couple of streamers from memory:

Sonos ZP90 200ps RMS

Slim Devices Transporter 20ps RMS

... and of course this is jitter at the S/PDIF output, by the time it has gone down the S/PDIF link and been processed by the DAC, it is likely to be much higher at the clock pin on the D2A chip.

So depending on who you believe, AV equipment jitter is either well below the level of human perception, or bang in the middle of what can be heard.

 

[oldric_naubhoff]

I never was too much interested in streamers. But seeing they are consistently becoming the standard medium for digital music playback for the few years to come I am becoming more and more interested.

I was therefore recently wondering; where is the miracle of oversampling taking place? is it within the streamer or is it just before D2A conversion? do streamers just pass on the bits as they are received or do they manipulate with the source material.

this is very interesting to me as I'm a no fan of oversampling and I don't want my NOS DAC (if I finally have one) to decode information as it not really is at the source!

 

[CnoEvil]

I never was too much interested in streamers. But seeing they are consistently becoming the standard medium for digital music playback for the few years to come I am becoming more and more interested.

I was therefore recently wondering; where is the miracle of oversampling taking place? is it within the streamer or is it just before D2A conversion? do streamers just pass on the bits as they are received or do they manipulate with the source material.

this is very interesting to me as I'm a no fan of oversampling and I don't want my NOS DAC (if I finally have one) to decode information as it not really is at the source!

Hi Oldric, you may be right, but I think it's a mistake to rule out listening to something based on how you think it's implementation will effect the sound.

Since Linn is the brand I'm familiar with (and prefer) here is some reading, in case it's of interest:
http://forums.linn.co.uk/bb/showthread.php?tid=15456 (post No.2)
http://docs.linn.co.uk/wiki/index.php/Products (check out the Technology section)

 

[shooter]

Jitter is mind boggllng:

Transmitter jitter > Line induced jitter > Interfering-noise induced jitter > Intrinsic jitter > Transfered jitter > Data dependant jitter > Sampleing Jitter > Preamble jitter

:doh:

The clock is important part but can be very expensive, some costing more than medium priced spinners.

 

[CnoEvil]

The clock is important part but can be very expensive, some costing more than medium priced spinners.

You're not wrong: http://www.kronosav.com/products#ecwid:mode=product&product=6443178

 

[andyjm]

The clock is important part but can be very expensive, some costing more than medium priced spinners.

You're not wrong: http://www.kronosav.com/products#ecwid:mode=product&product=6443178

If you want really low jitter numbers, its no good having the clock in a separate box, the cabling, line drivers and receivers will introduce more jitter than the fancy clock will cure. The place to have the clock is on the circuit board next to the D2A chip itself.

The 'best' technical solution is either to have the D2A chip and a high quality clock in the streamer, or to have the D2A chip and high quality clock in a separate DAC box, but have an async protocol (USB, wordclock output) between the DAC box and streamer so that the DAC's clock is independent.

 

[CnoEvil]

If you want really low jitter numbers, its no good having the clock in a separate box, the cabling, line drivers and receivers will introduce more jitter than the fancy clock will cure. The place to have the clock is on the circuit board next to the D2A chip itself.

The 'best' technical solution is either to have the D2A chip and a high quality clock in the streamer, or to have the D2A chip and high quality clock in a separate DAC box, but have an async protocol (USB, wordclock output) between the DAC box and streamer so that the DAC's clock is independent.

I certainly can't argue with you on a technical level, but I do know that introducing the clock (which is an extra) into the DCS system, makes a good difference.

DCS make very expensive "no compromise" products, so I would be surprised if they made a basic error here...could be wrong, of course.

Here is what they have to say about what they do (including Clocking):
http://www.dcsltd.co.uk/page/myths#clocking

 

[oldric_naubhoff]

I never was too much interested in streamers. But seeing they are consistently becoming the standard medium for digital music playback for the few years to come I am becoming more and more interested.

I was therefore recently wondering; where is the miracle of oversampling taking place? is it within the streamer or is it just before D2A conversion? do streamers just pass on the bits as they are received or do they manipulate with the source material.

this is very interesting to me as I'm a no fan of oversampling and I don't want my NOS DAC (if I finally have one) to decode information as it not really is at the source!

Hi Oldric, you may be right, but I think it's a mistake to rule out listening to something based on how you think it's implementation will effect the sound. Since Linn is the brand I'm familiar with (and prefer) here is some reading, in case it's of interest: http://forums.linn.co.uk/bb/showthread.php?tid=15456 (post No.2) http://docs.linn.co.uk/wiki/index.php/Products (check out the Technology section)

thanx CNO for the links. however, I spent quite a lot of time trying to understand how it all works and I came to a conclusion that only NOS is the way forward.

NOS DAC technology is not without its flaws, I know. but there's no cutting corners with NOS DACs, if you want them to measure well there's bound to be a lot of clever engineering involved. whereas in case of oversampling DACs even a DAC in some fecking iPhone will measure incredibly well compared to your Linn DS.

not to mention you get all sorts of digital artifacts, due to using of digital filtering, with outputted signal in case of OS DACs. no thank you. oversampling is not for me.

 

[paradiziac]

With the dCS, the clock signal is sent to the DAC separately via a word clock output, that's not the same as a digital source embedding clock data in the SDPIF signal and sending it to a (jittery) SDPIF receiver chip in a DAC.

 

[andyjm]

With the dCS, the clock signal is sent to the DAC separately via a word clock output, that's not the same as a digital source embedding clock data in the SDPIF signal and sending it to a (jittery) SDPIF receiver chip in a DAC.

It is certainly better to distibute a clock via a dedicated line rather than embedding it in the S/DIF stream. It is even better not to distribute it at all, but have it co located with the D2A chip.

dCS say:

"The DAC is the point in the chain where dynamic errors in clock timing (known as jitter) are directly translated into errors in the analogue audio, so this should be the best place to site the master clock. Listening tests indicate that this mode gives a very worthwhile improvement in sound quality, which will obviously vary from system to system."

Strangely, they then go on to say:

"The best option is to use a dedicated Master Clock unit to run the whole system. A Master Clock has a very simple job to do: generate an accurate, stable clock signal with minimal jitter at a frequency either equal to the sample rate or an exact multiple of the sample rate."

You would almost think they wanted to sell you another box.

 

[CnoEvil]

You would almost think they wanted to sell you another box.

It's all academic, as I can't afford any of their other 3 boxes! :wall:

 

[oldric_naubhoff]

I never was too much interested in streamers. But seeing they are consistently becoming the standard medium for digital music playback for the few years to come I am becoming more and more interested.

I was therefore recently wondering; where is the miracle of oversampling taking place? is it within the streamer or is it just before D2A conversion? do streamers just pass on the bits as they are received or do they manipulate with the source material.

this is very interesting to me as I'm a no fan of oversampling and I don't want my NOS DAC (if I finally have one) to decode information as it not really is at the source!

Hi Oldric, you may be right, but I think it's a mistake to rule out listening to something based on how you think it's implementation will effect the sound. Since Linn is the brand I'm familiar with (and prefer) here is some reading, in case it's of interest: http://forums.linn.co.uk/bb/showthread.php?tid=15456 (post No.2) http://docs.linn.co.uk/wiki/index.php/Products (check out the Technology section)

thanx CNO for the links but my mind is already set on NOS digital technology. I spent quite some time reading what it's all about and I believe NOS is the way forward to me. it's not an easy task to design a NAS DAC that will also measure well so obviously some clever engineering needs to be implemented. I'm not buying into notion that some DAC implemented in fecking iPhone is worth anything only because it measures surprisingly well compared to your Linn DS or other hi-quality DAC. but that's what oversampling techniques do. they greatly improve measured performance. and also add some digital artifacts to outputted signal, but this is another story. so no Linn nor Naim for me, thank you.

 

[oldric_naubhoff]

You would almost think they wanted to sell you another box.

It's all academic, as I can't afford any of their other 3 boxes! :wall:

it's also academic because any quartz clock used in every budget level good quality DAC will have error levels at around few ps. waaaaay lower than detection capabilities of any human.

I've seen attempts to use atomic clocks for controlling word clock timing (like the new Antelope DAC). but I can't see anything to be gained here except maybe for a huge increase in the unit price.