Computer audio is no different than CD players. There is a large range of performance depending on price and manufacturer.
Because USB and networked streaming is new to most manufacturers, the majority of them use off-the-shelf USB modules from one or two third-parties rather than designing their own. The performance is therefore limited to how good these OEM Modules are. I have evaluated some of these and IME, they are fairly poor as a rule.
The experience you have with computer audio is therefore completely a function of the manufacturer/designer that you choose. Choose the right one and you will dump the CDP.
Steve N.
Empirical Audio |
Alex wrote:
"Steve, but I thought you are using one of those third-party off-the-shelf modules in your current products too, thought with improved clocking and power supplies?"
You would be mistaken. All of my interfaces are custom designed by me.
I have modded many CD transports in past years. Some of these became references. None of them hold a candle to my computer audio interfaces. It helps to have 30 years of digital design experience in the computer industry.
Steve N.
Empirical Audio |
"According to Steve Nugent, wireless/ethernet is supposedly a superior interface in terns of jitter"
True, but still only as good as the designer designing it.
Steve N.
Empirical Audio |
"Does having a asynchronous USB help with jitter and obviate the need for reclocking or SPDIF converters, as claimed here?"
It does eliminate some sources of jitter, namely the computer clock, however jitter can still be introduced by common-mode noise and RFI. The master clock and the associated clocking circuitry in the interface is also critical to getting a good result. The design will determine whether Async USB interface is good or not. There are a lot of mediocre ones out there.
Steve N.
Empirical Audio |
Audiolabyrinth, I have a nice 1905 Gramaphone, which is really easy to use. Does not even require power. I dont listen to it much though because the SQ sucks.
Your CDP will also be relegated to the other antiques soon, trust me.
Steve N.
Empirical Audio |
Audiolabyrinth - S/PDIF and USB cables run the gamut and the price is no indication of the sound quality you will achieve IME. I make a S/PDIF cable that is closely matched to 75 ohms, uses silver and expanded Teflon and only costs $250. No customer of mine has found a better cable, and they have tried. Based on my experiments, I believe that it is the termination of the coax to the plug connector that is most critical, not so much the plug-jack interface. The termination of my 75 ohm cable to the 75 ohm BNC is virtually flawless.
The USB cable is another thing however. There are a lot of companies making exotic versions and most don't match the 90 ohm spec. With new XMOS-based USB interfaces on a number of DACs (I'm doing one as well), I have found that the cable makes an even bigger difference. The signal integrity and power delivery are both important for XMOS interfaces.
Gordon Rankin claims that the error rate is high on many USB cables, and this shows up in SQ degradation. I believe that the SQ is mostly impacted by the power in the cable based on my experimentation, at least for XMOS based interfaces. I have a software tool now that logs errors over USB that I can use to compare cables. I have several USB cables and I'll be testing these over the next month or so to see which ones cause a lot of errors.
Steve N.
Empirical Audio |
I have a LOT of customers comparing USB cables. Some have had literally 10 of them. The best one currently reported is the TotalDAC USB cable/filter.
Steve N.
Empirical Audio |
"That being said, can you state in specific terms what measurable variables are responsible for the improved SQ you perceive?"
Differences in digital sources are primarily due to jitter in the digital source, however reduction in common-mode noise and RFI can impact the DAC or interface indirectly by adding jitter. To fix the latter, one needs either common-mode filters like the Short-Block or galvanically isolated interfaces like many XMOS USB interfaces.
Galvanic isolation is not a panacea however. To do this, you need separate power for the input section of the USB interface, either power from the cable or another isolated AC supply. If this power is not really perfect (di/dt, regulation and noise), then it may actually sound worse than the non-galvanically isolated interface. I have experienced this myself. Therefore, with my new Off-Ramp 6, which is based on XMOS, I will be offering a "Long-Block" combination linear Hynes technology power supply and filter. The long-block can improve the performance of virtually ANY XMOS interface with galvanic isolation.
Steve N.
Empirical Audio |
"there is a 600% to 1000% mark-up on the retail prices out there anyway"
That may be true from a few big companies, but most of these high-performance cables require huge investments to get the first batch made or they are completely hand-made and very labor intensive. Exotic materials like silver and expanded Teflon are generally not available in off-the-shelf cabling from any manufacturer. Certainly not in a USB cable. I know because I have looked. The designer must actually custom order 10,000 feet just to get these made. Fabricators will not even touch a 1000 foot order.
Computers are not obsolete either by any means. I actually use a win2000 workstation now for 15 years with the same hard-disk drive. The likes of Intel and the PC makers would have you believe that you MUST HAVE the latest technology and operating systems. They actually have marketing/development groups dedicated to developing apps and uses that require more powerful CPU's, more memory etc.. in order to get you to upgrade your PC.
The fact that most peoples PCs are fine for what they use them for and therefore they are not upgrading is a testament to the fact that they are not obsolete. This is precisely why PC sales are down.
Steve N.
Empirical Audio |
Audiolabyrinth - I was a design team lead on the Pentium 2 many years ago. It was a "slot 1" processor.
Steve N.
Empirical Audio |
"My cables do a staggering 4 pico/farads per foot of capacitance, what does you cables do?"
Got that beat. My own cables are 3.8pF/foot. I agree that this makes a big difference, but so does the silver metallurgy. Just as important if not more.
Steve N.
Empirical Audio |
Cerrot - USB stands for "universal Serial Bus". Not limited to disk drives. When are you going to give up this stupid quest? You obviously don't have a clue as to the latest technology and how good it is.
Steve N.
Empirical Audio |
Remember when CD spinners were a bust? It was not that long ago. They have gotten a LOT better just over the last 5 years. Many of them are simply computer audio inside too. These tend to sound the best has deliver the lowest jitter.
Same thing will happen with USB, and it already has for some manufacturers. Most are lagging in the performance of their designs, just like what happened with CD players. They will eventually catch-up.
Steve N.
Empirical Audio |
Audiolab.. wrote:"most world class dacs sound best with transports that you do not use an inferior cable like usb"
Its not the cable that is usually the limiter here. Its the USB interface design and associated clocking circuitry. Most USB DACs have inferior USB interface designs, but a few are world-class. Its a lot like vinyl cartridges. Some are great and most are just mediocre. The Empirical Audio USB interface has won many shootouts and best of shows, including this latest shootout:
http://forum.audiogon.com/cgi-bin/fr.pl?ddgtl&1398132150&&&/Absolute-top-tier-DAC-for-standard-res-R
Good cables for USB include the Revelation Audio Labs and the TotalDAC. Not cheap, but great performance. Buying a good USB cable alone, is not sufficient however. In addition to the USB cable, the common-mode noise must be addressed on the USB cable. This is the noise on the ground wire, not the signal wires. This can be addressed with a USB cable filter like the Empirical Audio Short-Block, or for USB interfaces that use the power in the cable, a filter/supply like the Power Block from Empirical Audio.
BTW, an excellent cable for S/PDIF coax is the Empirical Audio BNC-BNC with RCA adapters. Beats all comers at only $250.
Several of my DAC customers have actually sold their vinyl setups when they got my Overdrive SE DAC. Its only a matter of choosing the right equipment.
The question of whether computer audio can be great or not is silly. The answer is that there are no technical roadblocks that cannot be overcome, just like other technologies. Just requires a clever, experienced designer.
Steve N.
Empirical Audio |
A-L - most differential signaling uses a ground, just to insure that the source and destination ground potentials are close, including firewire. The problem with most of these is that the common-mode noise rejection provided by the receiver chips is not stellar. Therefore, it is still an issue. Its not that USB is inferior, just that most receiver chips are not great designs. Some are worse than others. Case in point is the XMOS chipset, which uses a really poor receiver chip IMO. Very sensitive to cables.
Steve N.
Empirical Audio
Steve N. |
Because it will sound better with the Synchro-Mesh, that's why.
Bit-perfect is less important than low-jitter, and the resampling in the Synchro-Mesh is the best available now. Minimal impact on SQ.
It's 30-day money-back, less shipping, so the risk is low. If your system is resolving, it will make a big difference.
Steve N.
Empirical Audio |
Also avoid ALAC, AIFF and FLAC files. Only the Antipodes server plays these as well as .wav. On all other servers and computers I have heard, the SQ is inferior to wav.
Steve N.
Empirical Audio |
"Barry Diament (recording engineer of some renown) did blinded comparisons of master files in either AIFF or WAV, and there was no discernible difference"
So what? Steve Nugent did the same comparison and found a significant difference.
This is entirely system and track dependent. Recording studios are notorious for compression and using inferior playback systems for their mixing.
Steve N.
Empirical Audio |
I have an Oppo, but I only use it for DVD and 3-D movies, and occasionally for VUDU streamed movies. My reference system is driven by USB from a customized Mac Mini.
If you want the Oppo to sound better for audio, then improve the source jitter. Adding an Alpha and using Oppo as a transport will not help IME. You have not solved the fundamental jitter issue.
There are a couple of ways to do this:
1) use the CD transport output from the Oppo or another transport spinning CD's to drive a Synchro-Mesh reclocker, located between the transport output and the Oppo coax digital input. If you can loop the transport output from the Oppo back into the coax input with a cable and that works, then this will work too. Its kind of like a tape-loop for digital that allows you to reduce jitter.
2) use an Off-Ramp 5 USB converter driven by a computer using USB to drive the S/PDIF coax input on the Oppo. It goes from USB to S/PDIF coax. This will allow you to play up to 24/192 tracks, unlike the system above. This requires a bit more work, but the SQ will be even better than the Synchro-Mesh. The work involves careful choice of computer, power supply for the computer, ripping software and playback software. More expensive, but worth it IMO.
Here are some tips on rippers and playback software, as well as the computer:
http://www.empiricalaudio.com/computer-audio/
Steve N.
Empirical Audio |
What does that tell us? Its obvious to me based on 15 years of listening to systems at trade shows and in salons. Most systems are not resolving enough or low noise and distortion enough to make these differences obvious. I have heard a number of reviewers systems as well. Same thing, unfortunately. Most systems simply have a preamp that creates so much masking and compression that these differences cannot possibly be heard. Good active preamps are like hens teeth.
Steve N.
Empirical Audio |
Current DACs have difficulty reducing jitter to inaudible levels because this is a difficult engineering challenge, even for seasoned designers. It always has been. The jitter on S/PDIF inputs on DACs is generally reduced somewhat by the receive chip, which uses a PLL to recover the clock from the datastream, but a low jitter input to this receiver chip is still beneficial. Other DAC designs use resampling chips and circuits to establish a new master clock. These can reduce jitter even more than the receiver chip, but there are two downsides: 1) they inpart their own kind of distortion due to way that the resampling algorithm is implemented 2) the new master clock and associated circuitry/power supply adds its own jitter.
Thes best solution for reducing jitter in a DAC is to put a master clock front-end on the DAC. There are two types of these available now, the Async USB interface and the network renderer. Both of these effectively discard the clock in the source computer or device and generate a NEW master clock.
if the power, circuit design and clock selection is optimized, the jitter can be extremely low with these input circuits.
The thing to understand is that these are not easy to design and its really esy for lots of jitter to creep back into the circuit, even if you u a Femtoclock etc..
also, jitter is never reduced to zero as some manufacturers would have you believe.
Jitter when characterized by a single number, such as RMS jitter is an inadequate measurement. Jitter has a spectral component as well as a distribution of amplitude. these are actually more important than any single number to predict if one jitter is more audible than another.
Steve N.
Empirical Audio |
I have modded some CEC transports to improve clocks and output circuits etc.. I dont mod anymore, but I know what they sound like.
I dont even have any CD player in my reference system, and its one of the best ones on the planet.
Steve N.
Empirical Audio |
He's just never heard a Synchro-Mesh with a good power supply, or a recent Off-Ramp for that matter.
These don't continue winning shootouts for no reason.
Yes, I've heard the Bryston with the same PCI Juli@t board that Carrot uses. Its very good, but my USB interface still beats it and so does Synchro-Mesh.
Steve N.
Empirical Audio |
"Interesting. In your opinion, is jitter management or the lack thereof the stumbling block to true analog reproduction? "
Absolutely, #1 cause. Also digital filtering is a big offender, #2.
Also, what are your thoughts on grounding and jitter production?
Grounding is a multifaceted issue. There is the issue of earth ground and where this is connected to DC common. This is what star-grounding is used for. Many designers don't understand this. Then there is the issue of ground-return paths. It is generally poor ground return paths that add jitter in most designs. Then there is the issue of the power supply currents in the ground planes and how these affect the signal currents in the ground planes. These can also affect jitter. One must understand the physics of how currents flow in digital systems. This is not intuitive. Creative solutions are required to overcome these issues. This is what sets apart really creative designers from well-schooled designers.
This is a lot like talking about power supplies. Power supplies are only the start of a much more complicated system that I refer to as "power distribution" that exists in all components. Its like saying that Hoover dam creates really clean power, but ignoring all of the transmission-lines and transformers and shared loads that are in between your outlet and Hoover dam.
I applaud any actions like the Vertex/Nordost tools for making better audio measurements. The field has historically been lacking of sufficient measurements to characterize these effects. I have myself tried to do exactly what they are doing, with limited success, when I used to design cables from 1996 to 2000. Correlating analog before and after signals in perfect sync is a difficult task for sure. This demonstrates differences in dynamics, which is the foremost problem with most consumer audio gear. Most gear compresses due to deficiencies in the component power distribution system as well as slow reacting DC power supplies. Cables can also be at fault, but to a lesser degree. More difficult to measure cable effects IME.
Steve N.
Empirical Audio |
"If you analyze people's opinions on SQ superiority of WAV versus AIFF or any other format, its a coin toss statistically. What does that tell us?"
Not if you are selective about who does the listening and on what systems. I do these comparisons at virtually every show I exhibit at. The differences are always obvious. Usually the listener that has ripped his entire library in AIFF or FLAC just leaves in disgust because he now knows that he has worked hard to produce something suboptimal.
Steve N.
Empirical Audio |
Resampling is never bit-perfect. There are good resamplers and poor resamplers. I use a good one. It has gotten lots of glowing reviews. I actually used to sell a non-resampling reclocker, the Pace-Car, but the source had to be synchronized to it. This means that the source had to have a word-clock input or have a master clock driven from the reclocker using mods to the source device. It is a complicated matter to do reclocking of S/PDIF without resampling, not as simple as you allude to. Even though the Synchro-Mesh resamples and is not bit-perfect, it still beats the older Pace-Car, which was bit-perfect.
Reasampling is a lot like EQ. EQ has always been a dirty word for high-end audio, primarily because the graphic equalizers that were first introduced were such low quality. With the advent of Amarra software, EQ is now not only good, it can elevate your system significantly.
Likewise, the Synchro-Mesh uses new technology that makes the non-bit-perfectness a non-issue. Lots of folks are using it with their CD transports and Sonos in particular, but even with their Squeezebox.
I'll bet it would make even your Juli@t sound better.
Steve N.
Empirical Audio |
I have compared my own Off-Ramp 5, which is bit-perfect to the Synchro-Mesh, both being powered from the same high-quality DC supply. The SM is pretty close to the Off-Ramp 5, even though OR5 is slightly better. In many systems, the difference would not be perceivable.
Steve N.
Empirical Audio |
Agear - I have an "about AE" page that describes my system and has a photo. It is not up to date though, since I now have custom 805A SE monoblocks. Soon to get Vapor Audio Nimbus speakers too.
The best way to play files is probably wired Ethernet. I am currently designing a network renderer for this purpose. My products, including the Overdrive SE can be upgraded to change out the USB interface for this network renderer, called the "Interchange".
Storing files depends on the playback. Generally the best way is to have them local to the computer, not NAS, and not USB if you are using USB interface. If you use network interface, then USB or firewire access to the drives is fine. IF you use USB, then Firewire to the drives is best IME.
Steve N.
Empirical Audio |
"You are saying that jitter is important. That jitter can also result from cable induced errors. That re-clocking at the DAC does not necessarily correct for all/any errors related to jitter that could occur during delivery of the raw digital signal through a cable."
All correct.
"Does it follow that some digital cables are better at delivering digital signals free of or with less added jitter?"
Absolutely.
"in theory or in measurement, can a digital signal be corrupted in a cable, say due to exposure to strong EMF, to the point where 1s and 0s are actually deleted or unreadable at the DAC. I.E. outright data loss?""
Very unlikely. The EMI would have to be kilowatts of power, like a radar antenna.
Corruption of data on a USB or S/PDIF cable is not usual and quite difficult. The cable must be extremely long and have poor impedance match to have significant error rate.
Signal integrity ala "eye-pattern" however can easily be affected by connectors, cable construction, impedance etc.. Also, ground loop noise can impact both USB and S/PDIF causing jitter and even errors if it is bad enough. These are currents running in the ground of the cable that should not be there. IF high enough, they will appear on the signal(s) and cause problems at the receiver. This is why I developed the "Short-Block" USB cable filter, which reduces common-mode ground-loop noise on the cable.
Steve N.
Empirical Audio |
AudioL - what I mean by receiver is the chip that is receiving the digital differential signal on the USB cable. Common mode noise on this cable can exceed the CMR or common mode rejection of this receiver chip, and usually does, causing jitter in the receiver chip.
Steve N.
Empirical Audio |
The only server that delivers the same SQ for wav and FLAC files that I have heard is the Antipodes. Lots of intelligent S/W customization in that server to optimize SQ.
Steve N.
Empirical Audio |
