DACs with internal master clocks: less importance of transport?

DonH50 said:

The jitter problem that is usually stated is when the transport's clock is not clean, or is corrupted by noise and distortion, and is used directly as the clock for the DAC.

Click to expand...

would you say this is an area where digital cables can make a difference?

DonH50 said:

I cannot remember who made it, but years ago I heard a CD player that first pulled the data off into a memory then began playback. After endless outcry from those who did not like waiting minutes for it to start playing, the designers were apparently made aware of dual-port memory and released a new version that started playing after a short period to load a few seconds worth of data so the music could play on whilst the rest of the disc was being read. It was still a bit distracting as the rapidly-spinning CD transport could still be heard if you were near the player until it pulled all the data off.

Click to expand...

Right, we have discussed products like the http://www.thememoryplayer.net/ here in the past...

microstrip said:

As far as I remember some Meridian, Mark Levinson and the PS Audio do it. I will have to look for the specific models. We can expect it from transports that use ROM drives, as they are able to read up to x56 the normal speed. I got this From a Meridiam post in Audiogon : Take Meridian as an example, they use CD-rom drives, often with multiple read passes, and buffer the data through RAM, before using a separate low jitter clock to pass the data through the DACS. The key words for a search could be "multiple read".

Click to expand...

Yes, but simply using a CD-ROM drive does not make the player go the whole way, with buffering, reclocking, etc.
Sure, the Meridian, as you stated, does it, as well as the PS Audio transport (and the MSB, of course). But these are the vast minority. When it comes to CD players (and transports) the overwhelming majority is still doing it real-time, interpolating away.

To get back to the topic's question, I think the transport is still important, but as a whole, not only the physical, disc-reading part of it. With all the advances in software, disc players are operating more like single-tasking computers these days. I believe Parasound has a cheap(ish) CD player (http://parasound.com/pdfs/CD1Whitepaper.pdf) that's actually a computer, that runs embedded Linux!

"The Parasound CD 1 demonstrates a new method for playing Compact Discs. To begin with, itemploys a CD ROM drive instead of a conventional CD drive. The CD ROM drive is connected to apassively-cooled Intel ITX computer. It uses the Linux operating system kernel and Holm’sproprietary software that improves the reading of CD disc data. The CD ROM drive is set to run atfour times the speed of a conventional CD player drive in order to accumulate a vast amount of dataquickly. Because the CD spins at four times the normal speed, the CD 1 has the advantage ofbeing able to read sections of a disc as many times as needed to significantly reduce read errors. "

alexandre

asiufy said:

Yes, but simply using a CD-ROM drive does not make the player go the whole way, with buffering, reclocking, etc.
Sure, the Meridian, as you stated, does it, as well as the PS Audio transport (and the MSB, of course). But these are the vast minority. When it comes to CD players (and transports) the overwhelming majority is still doing it real-time, interpolating away.

To get back to the topic's question, I think the transport is still important, but as a whole, not only the physical, disc-reading part of it. With all the advances in software, disc players are operating more like single-tasking computers these days. I believe Parasound has a cheap(ish) CD player (http://parasound.com/pdfs/CD1Whitepaper.pdf) that's actually a computer, that runs embedded Linux!

"The Parasound CD 1 demonstrates a new method for playing Compact Discs. To begin with, itemploys a CD ROM drive instead of a conventional CD drive. The CD ROM drive is connected to apassively-cooled Intel ITX computer. It uses the Linux operating system kernel and Holm’sproprietary software that improves the reading of CD disc data. The CD ROM drive is set to run atfour times the speed of a conventional CD player drive in order to accumulate a vast amount of dataquickly. Because the CD spins at four times the normal speed, the CD 1 has the advantage ofbeing able to read sections of a disc as many times as needed to significantly reduce read errors. "

alexandre

Click to expand...

I could have been more clear - it is not only they use CD ROM drives, but also manufacturers use parts and techniques that are used in CD ROM drives for faster readout.

Anyway, the important aspect is that, irrespective of number of reads, most of the times the bit content of the SPDIF is almost bit exact. This was measured by several authors - I remember that John Atkinson of Stereophile published a test with several CD transports, capturing the SPDIF output and the differences were either non existent or minimal. The few interpolations in poor CD disks can not explain the sound differences. However the jitter spectra were different from each other.

Most of these tests were carried in the 90's - it is not easy to get them nowadays.

I believe the differences we hear are boil down to jitter levels and the amont of digital grunge each transport is injecting into the DAC. The later is the reason we still hear differencies between transports on DACs that reclock the signal (like the MSB DAC IV) - chence should be immune to transport quality. Any yet - they aren't.

The bit-perfect (or the lack of) theory can be safely put to rest.

I suspect noise injection, including ground noise, rather than jitter is the culprit in most cases.

ack said:

would you say this is an area where digital cables can make a difference?

Click to expand...

Disclaimer: As you know I am a cable skeptic. A decent cable can be had for little money and I think in the vast majority of cases that is good enough. That said I think I have talked about cable issues in one of the tech forum threads.

If the cable is not properly matched, and/or source and load (receiver, DAC input) are not properly terminated, then reflections could induce signal and clock dependent jitter. A bad connector (or connection) can also cause discontinuities that can corrupt the signal. A cable that is not shielded well can pick up (and generate) noise. Some type of galvanic isolation, like an optical link, could help isolate transport noise from the receiver (DAC).

A really poor cable that exhibits high loss and dispersion can corrupt the signal, increasing jitter and even causing bit errors (though at the low rates audio streams use I would think dispersion is not a significant factor, nor skin effect, loss tangent, and other RF/mW/mmW issues). Cable sensitivity to temperature and mechanical vibration are issues at very high frequencies and very high resolutions but these effects are tiny compared to the distortions present in other components.

If we can really hear 0.0001% distortion (-120 dB, something else about which I am skeptical but I have not had time to research that claim) then maybe some of these effects are contributors. In a real system the speakers alone generate several orders of magnitude more distortion than that.

DonH50 said:

I suspect noise injection, including ground noise, rather than jitter is the culprit in most cases.

Click to expand...

Very perceptive. This is exactly the reason why Async USB interfaces are sensitive to cable choice. It is actually common-mode noise created by the ground-loop. USB interfaces can be galvanically isolated or filtered to minimize this effect.

Steve N.

DonH50 said:

Disclaimer: As you know I am a cable skeptic. A decent cable can be had for little money and I think in the vast majority of cases that is good enough. That said I think I have talked about cable issues in one of the tech forum threads.

If the cable is not properly matched, and/or source and load (receiver, DAC input) are not properly terminated, then reflections could induce signal and clock dependent jitter. A bad connector (or connection) can also cause discontinuities that can corrupt the signal. A cable that is not shielded well can pick up (and generate) noise. Some type of galvanic isolation, like an optical link, could help isolate transport noise from the receiver (DAC).

A really poor cable that exhibits high loss and dispersion can corrupt the signal, increasing jitter and even causing bit errors (though at the low rates audio streams use I would think dispersion is not a significant factor, nor skin effect, loss tangent, and other RF/mW/mmW issues). Cable sensitivity to temperature and mechanical vibration are issues at very high frequencies and very high resolutions but these effects are tiny compared to the distortions present in other components.

If we can really hear 0.0001% distortion (-120 dB, something else about which I am skeptical but I have not had time to research that claim) then maybe some of these effects are contributors. In a real system the speakers alone generate several orders of magnitude more distortion than that.

Click to expand...

It is true that speakers create massive distortion, however high-levels of distortion are at the very low frequencies. High-frequency distortion on the other hand is generally the realm of jitter and other harmonic distortions, often created by digital circuits and clocks.

Digital cables can make a HUGE difference, but it depends entirely on the design and materials used. I used to sell cables and had a $600+ 1.5m digital cable (all digital cables should be at least 1.5m). All pure silver with expanded Teflon dielectric. I stopped selling cables, but my customers were always asking me where to get a good cable for less than $1K. Recently I started experimenting with a BNC-BNC cable and found the right materials to make an optimum cable. I discovered that the termination of the shield to the connector is more important than the connector itself. If you have a well-terminated BNC-BNC cable and use 75 ohms RCA adapter or adapters, it blows away 99% of cables out there at any price. Trying to terminate a true 75 ohm cable to an RCA plug is futile. Don't even bother trying.

BTW, here is my white-paper on why it should be 1.5m long:
http://www.positive-feedback.com/Issue14/spdif.htm

Steve N.

I find it hard to believe speaker distortion even at low levels approaches jitter noise unless jitter is very high. At 10 kHz, 10 ns of jitter yields about 72 dB SNR, comparable to 0.026% THD+noise. I admit I do not know a speaker is not that linear, however, it just seems pretty low. Not sure I have ever measured one that low, but the last time I tried measuring speaker distortion was a long time ago, and it was exceedingly painful to try to separate mic and speaker distortion at low levels.

Any discontinuity will cause reflections. I agree keeping the return path consistent can be painful, and of course not keeping the shield with the signal will disrupt the impedance. I have certainly seen a lot of connections where the shield is simply twisted and soldered to the connector with no attempt to keep the return (ground) loop as short as possible. In the RF world we don't use BNC's, either (except Agilent uses a very special version on some of their wideband DSOs), for various mechanical and electrical reasons.

I skimmed your paper very quickly. I may have missed some things so please accept my apologies in advance. If the source and load are properly terminated then length does not matter, natch, except for loss and dispersion in the cable itself (not usually a concern at the data rates and cable lengths audio uses). If not perfectly matched (i.e. the real world), how large a mismatch causes audible issues is, like everything else, debatable. If the interface is asynchronous (i.e. clock regenerated in the DAC independently of the incoming bit stream) it (mismatch) should not matter. One possible caveat with the "perfect length" problem is that, if the bit stream has arbitrary runs of 1's and 0's (even if limited to short runs by the encoding scheme), then the perfect length to eliminate the triple-transit problem is no longer a single length.

IMO, based upon fairly limited experience with audio cables, you have hit upon the major flaw with most cables. That is, the cable itself is not the culprit (inexpensive coax is more than adequate), but bad connectors (or rather bad connector assembly) can cause problems.

Isolation is a separate issue; I have seen transformers, active buffers, and optical links all used to good effect. Certainly USB was not designed as a low-noise link for things like DACs. Test instruments I have used always isolate the USB link.

Thanks Don and Steve. Steve's paper makes it quite clear how reflections can introduce jitter, and is worth repeating:

When a transition [ack: e.g. from -250mV to +250mV] is launched into the transmission line [ack: from the transport to the DAC], it takes a period of time to propagate or transit to the other end. This propagation time is somewhat slower than the speed of light, usually around 2 nanoseconds per foot, but can be longer depending on the dielectrics used in the digital cable. When the transition reaches the end of the transmission line (in the DAC), a reflection can occur that propagates back to the driver in the Transport. Small reflections can occur in even well matched systems. When the reflection reaches the driver, it can again be reflected back towards the DAC. This ping-pong effect can sustain itself for several bounces depending on the losses in the cable. It is not unusual to see 3-5 of these reflections before they finally decay away, particularly when using the best digital cables, which are usually low-loss.

So, how does this affect the jitter? When the first reflection comes back to the DAC, if the transition already in process at the receiver has not completed, the reflection voltage will superimpose itself on the transition voltage, causing the transition to shift in time. The DAC will sample the transition in this time-shifted state and there you have jitter.

Click to expand...

I skimmed your paper very quickly. I may have missed some things so please accept my apologies in advance. If the source and load are properly terminated then length does not matter

Click to expand...

I think you missed the main point actually. Its the risetime and the timing of reflections that makes a longer cable beneficial. A longer cable pushes the reflections beyond the time when the receiver is detecting the edge, so they are a "don't-care". This is not a VSWR problem, it is a digital problem.

IMO, based upon fairly limited experience with audio cables, you have hit upon the major flaw with most cables. That is, the cable itself is not the culprit (inexpensive coax is more than adequate), but bad connectors (or rather bad connector assembly) can cause problems.

Click to expand...

All of it matters. I have a LOT of experience, analysis, modeling and empirical measurement. I use TDR and 6GHz scope with 75 ohm internal termination. Here are the things that impact jitter on a S/PDIF coax cable:

1) losses and B/W limitations - these slow edge-rates which adds jitter at the receiver
2) impedance discontinuities or mismatch to the terminations - these can cause reflections that occur when the edge is being detected and shift it in time adding jitter
3) dielectric absorption - This is sometimes called "soakage". It is non-uniform storage of charge in the dielectric and release of same. Different data patterns can cause more or less charge to be stored, causing shifts in edges. Its kind of like having a garden hose lined with foam versus a solid one one that does not absorb any water. Turn off the water in the solid one and the water stops quickly. Turn of the water in the foam hose and it dribbles for a long time. Likewise, the initial water pressure is higher on the solid hose versus the foam hose.

Steve N.

Personally, I found your paper fascinating and extremely clear, short and to the point. It makes it clear that impedance mismatch between transport and DAC (with further impact by an improper cable) is the cause of reflections, which cause jitter in that interface. It is perfectly clear to me now why digital cables that aim to address reflections and noise - like those top-of-the-line from MIT and Transparent - are undoubtedly offering significant improvement (given a sufficient level of resolution in the system). What is now more interesting is, why would these improvements be audible even if the DAC buffers and reclocks. Any idea?

"False information" as in "bits are not bits"? That can't be - it would imply we can't reliably transfer, bit-for-bit, digital information from point A to point B, regardless of time variations (jitter); but not ruled out either (perhaps error correction is having a hand in this). Something else is going on, and possibly buffering and reclocking is still subject to incoming jitter on the interface.

Al M. said:

That is precisely the question that interests me as well. Can the clue perhaps be found in the following from the paper?

"When the reflection reaches the driver, it can again be reflected back towards the DAC. This ping-pong effect can sustain itself for several bounces depending on the losses in the cable. It is not unusual to see 3-5 of these reflections before they finally decay away, particularly when using the best digital cables, which are usually low-loss."

The repeated bounces would suggest to me that actually false information is added to the signal, instead of them just causing regular jitter. Or do I read this wrong?

Click to expand...

Incorrect signal is added to a particular signal edge. This only shifts the edge in time slightly, it does not change the intention of the edge, either 1-0 or 0-1.

Steve N.

ack said:

Personally, I found your paper fascinating and extremely clear, short and to the point. It makes it clear that impedance mismatch between transport and DAC (with further impact by an improper cable) is the cause of reflections, which cause jitter in that interface. It is perfectly clear to me now why digital cables that aim to address reflections and noise - like those top-of-the-line from MIT and Transparent - are undoubtedly offering significant improvement (given a sufficient level of resolution in the system). What is now more interesting is, why would these improvements be audible even if the DAC buffers and reclocks. Any idea?

Click to expand...

Sure. It depends on what kind of reclocking is used in the DAC. If it is at all sensitive to incoming jitter, it will matter, otherwise not. Many DACs that claim to be insensitive to incoming jitter actually are. I've tested lots of them.

Steve N.

Empirical said:

Sure. It depends on what kind of reclocking is used in the DAC. If it is at all sensitive to incoming jitter, it will matter, otherwise not. Many DACs that claim to be insensitive to incoming jitter actually are. I've tested lots of them.

Steve N.

Click to expand...

Yeah, I totally believe you, and that's what I concluded in post #54 - it's the only thing that makes sense.

Al M. said:

That is precisely the question that interests me as well. Can the clue perhaps be found in the following from the paper?

"When the reflection reaches the driver, it can again be reflected back towards the DAC. This ping-pong effect can sustain itself for several bounces depending on the losses in the cable. It is not unusual to see 3-5 of these reflections before they finally decay away, particularly when using the best digital cables, which are usually low-loss."

The repeated bounces would suggest to me that actually false information is added to the signal, instead of them just causing regular jitter. Or do I read this wrong?

Click to expand...

Al. M,

Digital cables are usually of small length, meaning this "ping-pong" is very fast, and the SPDIF receivers are usually very robust. You will not loose bits even with very poor digital cables - this means having characteristic impedance far from 75 (RCA or BNC) or 110 ohm(AES/EBU). But surely it can affect the recover of the clock, introducing jitter.

microstrip said:

Al. M,

Digital cables are usually of small length, meaning this "ping-pong" is very fast, and the SPDIF receivers are usually very robust. You will not loose bits even with very poor digital cables - this means having characteristic impedance far from 75 (RCA or BNC) or 110 ohm(AES/EBU). But surely it can affect the recover of the clock, introducing jitter.

Click to expand...

It's fairly difficult to actually lose bits. With a short cable, you can even use a piece of hanger wire... The jitter will be terrible though.

Steve N.