We shouldn't be able to do this with sighted subjective listening, but we did...

[Superdad]

I find it perplexing that there are such widespread issues with USB!

 

Different, but no more so, than with S/PDIF (which its own host of designed-in limitations).

 

But the notion of USB "noise" is a fallacy. There is crap that gets in on the ground plane, but in the end everything--signal integrity, ground-plane noise, packet-data noise, etc. (affected by computer activity, PS, USB chip signal pattern and amplitude, USB cable impedance, etc.) all only make a difference because of how they manage to modulate activity and GPN of the USB receiver INSIDE the DAC, in turn affecting the PS,ground, and to a lesser extent the data lines, resulting in variations at the DAC's master clock.

 

No voodoo there.

[esldude]

Different, but no more so, than with S/PDIF (which its own host of designed-in limitations).

 

But the notion of USB "noise" is a fallacy. There is crap that gets in on the ground plane, but in the end everything--signal integrity, ground-plane noise, packet-data noise, etc. (affected by computer activity, PS, USB chip signal pattern and amplitude, USB cable impedance, etc.) all only make a difference because of how they manage to modulate activity and GPN of the USB receiver INSIDE the DAC, in turn affecting the PS,ground, and to a lesser extent the data lines, resulting in variations at the DAC's master clock.

 

No voodoo there.

 

No voodoo, and no measurements of this effect. How convenient.

[alfe]

The difference is that I am not selling anything, or receiving any commission from saying what I think about the Regen, based on my own experience with the device using an improved power supply for it. In fact, in HFC Forum I have suggested that the Intona Isolator, which is a more recent product may be a better alternative than just the Regen, although I also said that quite a few report further SQ improvements when using the Regen in tandem with the Intona Isolator.

 

Yes data cleaners, free shower for your data after each connection

[mmerrill99]

I find it perplexing that there are such widespread issues with USB!

Why?

[Jud]

I find it perplexing that there are such widespread issues with USB!

 

My thought is that it's due to the current widespread use of USB, combined with many (most?) audiophiles' natural search for the next relatively inexpensive tweak that brings real or imagined sonic benefits. I'm old enough to remember there were cable debates and improvement gizmos for SPDIF when that was the most popular interface.

[lucretius]

As always with only two data points it could as likely be coincidence as not.

 

However, the audio world is rife with such situations. Compression is one. Just a little compression will seem to make everything better to very nearly everyone. A moderate amount even will seem best by the great majority.

 

When done skilfully a little reverb of the right type also will seem far better to nearly everyone.

 

These are colorations pretty much no one is immune to even when generally preferring precise uncolored playback gear.

 

So what you are describing is interesting, but hardly unique nor difficult to believe nor requiring coincidence to be so.

 

It also shows why in my opinion, we should strive for the most accurate playback possible, and then color it to taste without guilt about doing so. It is mixing up accuracy and preference that I dislike.

 

I don't have enough experience to comment in this thread. But what you say reminds me of this article: 9 Sound Design Tricks To Hack Your Listeners Ears

[Sam Lord]

Jud, I haven't bothered to read all this, so maybe someone else has brought it up. You might be getting improved linearity by the biasing effect of one or more steady tones. The tones reduce crossover notch distortion. The effect only works in Class AB and perhaps class H amplifiers, but not in Class D or Class A, or any design with an output transformer. I don't recall experiencing this but it could certainly happen.

[Jud]

Jud, I haven't bothered to read all this, so maybe someone else has brought it up. You might be getting improved linearity by the biasing effect of one or more steady tones. The tones reduce crossover notch distortion. The effect only works in Class AB and perhaps class H amplifiers, but not in Class D or Class A, or any design with an output transformer. I don't recall experiencing this but it could certainly happen.

 

Class A amp here, I'm fairly sure.

 

By the way, I changed something in my system and removed the distortion, and my preference remains the same.

[Sam Lord]

On second thought I was wrong about the output xformers, as they don't eliminate CN distortion to a class AB circuit.

[jabbr]

Why?

 

I can understand how devices like the Regen and Intona help the USB SI. I don't understand why the type of cable used to feed these devices would make any difference.

[One and a half]

I can understand how devices like the Regen and Intona help the USB SI. I don't understand why the type of cable used to feed these devices would make any difference.

 

Yes, the drama is most boutique and exotic cables are not 90 Ohms, so the transmission goes through another quirk, often unpredictable, hence making the difference. The rest of the differences are connector impedance variations and continuity with majority of cable specs being spin.

[mmerrill99]

I can understand how devices like the Regen and Intona help the USB SI. I don't understand why the type of cable used to feed these devices would make any difference.

How cables can affect USB SI is through mismatches between the differential signals of USB (D+ & D-). Each of these signals are carried in separate wires within the cable & each wire's impedance to ground wire is the crucial factor in having a balanced differential signal. It's the accuracy of this balance that allows the USB receiver to cancel differential noise in these wires - the greater the impedance balance, the greater the differential receiver can be effective in this task. The more imbalance, the less cancellation of this noise which affects the USB SI.

 

Some other factors which might play a possible role in USB cables & USB SI:

- skew differences between D+ & D- i.e different lengths in these wires.

- ground loop through shield connection (Shield current induced noise)

 

And there are probably some others I'm forgetting

[jabbr]

...

 

Some other factors which might play a possible role in USB cables & USB SI:

- skew differences between D+ & D- i.e different lengths in these wires.

- ground loop through shield connection (Shield current induced noise)

 

And there are probably some others I'm forgetting

 

Sure but cable technology is very well understood and @ TI connectivity USB cable (just an example from a major manufacturer) should be fine up to the point where the signal is reconstructed.

 

I mean 10gbe cables are "easy" and cat 8/40 gbe cables exist so is the USB protocol that much more difficult than Ethernet?

[Jud]

Sure but cable technology is very well understood and @ TI connectivity USB cable (just an example from a major manufacturer) should be fine up to the point where the signal is reconstructed.

 

I mean 10gbe cables are "easy" and cat 8/40 gbe cables exist so is the USB protocol that much more difficult than Ethernet?

 

I kind of doubt it's anything as "high tech" as protocols. I'm betting it's simply good old electrical stuff (e.g., noise on ground) in different folks' systems. Some people can incur and get rid of plainly audible noise like ground hum by changing cables (I have), so why can't changing cables change the amount and/or route of electrical noise in the system?

[mmerrill99]

Sure but cable technology is very well understood and @ TI connectivity USB cable (just an example from a major manufacturer) should be fine up to the point where the signal is reconstructed.

 

I mean 10gbe cables are "easy" and cat 8/40 gbe cables exist so is the USB protocol that much more difficult than Ethernet?

Sure, cable technology that ensures digital data delivery is well understood & there's something like 20% tolerance on impedance in the USB standard. Does this mean that differences in the areas I nominated do not exist between cables? Would these differences result in different qualities of USB SI?

 

Regen or Intona don't improve or change digital data delivery so one could argue that thsi is well understood & therefore there's no need for these devices which is effectively what you are saying about USB cables.

 

If you accept Regen & Intona can improve SI then cables can also affect USB SI positively or negatively

[Superdad]

I can understand how devices like the Regen and Intona help the USB SI. I don't understand why the type of cable used to feed these devices would make any difference.

 

Despite Merrill's best efforts, he missed calling attention to the key point of why USB cables before a REGEN, Intona, iFi hub, etc. still make a difference (and likely why even an Ethernet DAC like the Merging NADAC uses an isolator cube in front of its renderer board):

 

As most know, the REGEN (and similar) works by improving signal integrity (right at the DAC's USB input if used as we prescribe) which in turn allows the DAC's USB input PHY (physical interface) chip to generate less bursty activity INSIDE the DAC, and the SQ benefit follows from there.

 

But the hub chip inside the REGEN has its own PHYs and protocol engine, which themselves generate packet noise on ITS power and ground planes. So the REGEN itself is also sensitive to the SI of the signal fed to it, which is why good USB cables and specialty USB host boards feeding it still make a difference—maybe just not as much.

A lot of time was spent on the design and board layout to minimize this packet noise but it is still there. The impedance of the power delivery network (PDN) over a broad range of frequencies determines the amplitude of the packet noise produced by the hub chip. The REGEN’s frequency optimized PDN is what makes it such a good sounding source.

 

And as alluded to above, all Ethernet interfaces have a PHY as well, and while the small isolation transformer (typically built into every RJ-45 Ethernet jack) does help with noise, the sensitivity to and effect of bursty packet data activity must still be getting through--otherwise people would not hear improvements by using fiber media converters with their gear.

 

Lastly, I wish to restate that I cringe a little every time someone refers to USB (or other interfaces) as being "noisy." Except for crappy/noisy 5VBUS lines from a computer, most of what else is riding along on the data lines is not the problem. Even if it is "noise", it is NOT getting INTO the DAC. The effect on the DAC by what's upstream of the USB input (cables, regens, USB host card, etc.) is all because of variation in signal integrity (see eye-pattern) and impedance match.

 

The two exceptions being 5VBUS cleanliness (if you DAC uses the 5VBUS then stuff can ride in there), and the grounds. The latter is where galvanic isolation really helps, since almost no matter what you do in your system, there are leakage currents running everywhere, and ground-plane noise is incredibly pernicious.

 

Happy 4th-of-July to those here in the states. Carry on a have a lovely weekend everyone else.

 

--Alex C.

[jabbr]

...

 

And as alluded to above, all Ethernet interfaces have a PHY as well, and while the small isolation transformer (typically built into every RJ-45 Ethernet jack) does help with noise, the sensitivity to and effect of bursty packet data activity must still be getting through--otherwise people would not hear improvements by using fiber media converters with their gear.

 

Couple points:

 

1) yes it's good that fiber interfaces are available for Ethernet.

2) doesn't solve the packet switching issue -- even perfectly timed perfect bits arriving at the interface will create some degree of packet switching noise. Solutions to that are the next stage of isolation

 

... The effect on the DAC by what's upstream of the USB input (cables, regens, USB host card, etc.) is all because of variation in signal integrity (see eye-pattern) and impedance match.

 

eye pattern tests would go a long way toward demonstrating (or not) any difference between $2 and $500 cables.

[Superdad]

eye pattern tests would go a long way toward demonstrating (or not) any difference between $2 and $500 cables.

 

Wasn't Hi-Fi News running those with their USB cable reviews for a while? Impedance testing should be done at the same time--as given the wide variations I have seen in boutique cable construction, I suspect there are a lot of wildly out-of-spec cables being sold.