Real USB loopback effects

[widemediaphotography]

Hi,
I wanted to focus attention on the loopback problems that the USB ports of a common PC are said to be affected by and for this reason there is a tendency, in my opinion, to propose UPnP systems via ethernet because they are galvanically isolated, or to believe that a streamer is better than a PC, because the streamer would also work as a galvanic isolator (In my opinion a streamer is basically a pc). Therefore the questions are these:
1) Is it sufficient to use a split USB cable (data+power), or use a USB cable with the +1 pin disconnected to eliminate any loopback problems of the USB port? If not, why?
2) Does the asynchronous transmission of an audio stream via USB require an analog carrier wave to ensure the transmission of digital data. Is this carrier associated with other current and therefore voltage, even if the DAC is equipped with its own power supply? Could this current create loopback problems?
Thanks

 

[pma]

If you want perfection and no loop currents, avoiding even capacitive currents, use optical fibre. It is the only non compromised way.

 

[widemediaphotography]

If you want perfection and no loop currents, avoiding even capacitive currents, use optical fibre. It is the only non compromised way.

I don't think a double electrical/optical/electrical conversion is necessary for a single meter of audio connection with the double result of introducing jitter and limiting the files to 24/192 KHz. and make the clock travel together with the data. Toslink is an old standard from 1980, when chips were expensive!

 

[DVDdoug]

Usually it's called a "ground loop".

I think the data transmission needs a ground (at both ends). But it can be optically-isolated with separate-isolated ground on each end (and the USB power can be transformer-isolated). A USB isolator is probably the best solution if you have a ground-loop problem.

If you think you've got ground-loop noise through the USB, unplug the USB connection to see if the noise goes away.

Does the asynchronous transmission of an audio stream via USB require an analog carrier wave to ensure the transmission of digital data. Is this carrier associated with other current and therefore voltage, even if the DAC is equipped with its own power supply?

No.

 

[widemediaphotography]

Mine is a technical question, to understand what happens beyond real practice. I simply need to raise the volume of the amplifiers to the maximum to verify that both the headphones and the speakers emit real zero noise even when I move the mouse on my audio-pc connected to the SMSL DO300 DAC
However, these are just my experiences, not based on quantitative but only qualitative phenomena.
If the USB cable does not carry the 5V voltage, because it has the pin disconnected, or it is a split cable, could the loopback phenomenon exist? Is it the same if I have pin +1 connected?

 

[Dunring]

I've got two different USB optical cables running and its flawless playback with a silent noise floor and zero ground loops.

Fibbr and also the Phoossno, for $49 the Fibbr is the best price currently. It provides power (designed for VR headsets) as well, not like the Corning Micro-B which is data only.

Amazon.com

The Yottamaster USB hub is $26 with 10 percent off currently. It accepts USB C for data and supplemental power up to 3 amps.

Amazon.com

I've got a system without built in optical, so it's the easiest solution. I have three SMSL DO300 DACs, so I can tell you for certain it will work with those Optical cable doesn't work with adapters though, you need a device like a DAC or hub with USB C for inputs. I like the hub since then everything runs conveniently from it with short USB cables.

 

[somebodyelse]

1) Is it sufficient to use a split USB cable (data+power), or use a USB cable with the +1 pin disconnected to eliminate any loopback problems of the USB port? If not, why?

Because it doesn't meet USB specs. It might work. It might fail in interesting ways. The proper way to do it is to use a USB isolator.

 

[widemediaphotography]

Because it doesn't meet USB specs. It might work. It might fail in interesting ways. The proper way to do it is to use a USB isolator

 

[solderdude]

Mine is a technical question, to understand what happens beyond real practice. I simply need to raise the volume of the amplifiers to the maximum to verify that both the headphones and the speakers emit real zero noise even when I move the mouse on my audio-pc connected to the SMSL DO300 DAC
However, these are just my experiences, not based on quantitative but only qualitative phenomena.
If the USB cable does not carry the 5V voltage, because it has the pin disconnected, or it is a split cable, could the loopback phenomenon exist? Is it the same if I have pin +1 connected?

The problem you have is caused by a ground loop (leakage current) in the ground connection. Disconnecting the +5V won't do anything to improve it.

The only ways would be to apply proper grounding, lower ground leakage currents, use another DAC with better ground layout inside the DAC or using an USB isolator (a real one)

 

[MC_RME]

I've got two different USB optical cables running and its flawless playback with a silent noise floor and zero ground loops.

Fibbr and also the Phoossno

Please note that not noticing ground-loop effects is the norm. That does not mean you don't have ground loops. Especially with these cables as they include copper for USB operating voltage and ground, so both sides are connected and not galvanicaly isolated at all. The optical fibre is only there to allow more length as copper data lines and reduce diameter and weight.

And thanks for pointing out these cables. I did not know that we have a newer and cheaper generation of these available now (compared to the first generation Corning cables).

 

[widemediaphotography]

My Audio-PC includes inside the cabinet an Topping HS01, but if I remove it I don't notice any difference. This could mean 2 things. I have no loopback issues, the Topping might alspo just be snake oil.
However, by leaving only the data connection in a USB cable, disconnecting the power part, how would the cable transmit a loopback? (many people use a small piece of electrical tape on the +1 pin). A cable without the power supply part is simply defined as not in specification, because the standardized and approved one contains it, but is this enough to declare that the cable could not work? According to science, it either works or it doesn't

 

[solderdude]

Some DACs want to see or use the +5V for its USB receiver or to auto-switch inputs or power the entire device. When this is the case you cannot use the 'cut the +5V wire trick' and have to replace it with another +5V (splitter cable). To insert 5V this is always with respect to the ground, which thus is still connected, and thus the leakage loop path also is not removed. 'Cleanness' of the +5V is rarely a problem anyway.
The problem here is that the leakage currents come through the ground wire/shield.

Some DAC's also don't mind if the ground wire is disconnected. A condition, in that case, is that there cannot be a too high DC voltage difference between both 'grounds'.
That could, in some cases, be a solution. (data signaling is balanced so does not really need a ground).

 

[KSTR]

data signaling is balanced so does not really need a ground

That's a common misconception:
The data transfer mechanism for the payload data is differential but in the initial phase, communication is single-ended and ground-referenced.
The receiver circuit can only handle a limited amount of common-mode voltage, actually +- one Volt or so. When the common-mode voltage -- the difference voltage of the two system's grounds -- is too large then communication breaks, receivers running into clipping. More importantly, any stronger common-mode glitch can instantly kill the USB transceiver circuits and this happens (ask RME service department, i.e.).

Removing the ground connection from USB is absolutely *FORBIDDEN*. It may or may not work, for some time.
The 5V-line is not restricted, though, as it is just there for optional bus-powering a device. But sometimes it is used to power the transciever chip even when a device is self-powered.

 

[solderdude]

Yep, that's what I said. Added the word 'too' as it is missing.

A condition, in that case, is that there cannot be a too high DC voltage difference between both 'grounds'.
That could, in some cases, be a solution.

Fully agreed that this is an undesirable solution but in some rare cases it 'works' in lowering a ground loop issue.

 

[somebodyelse]

My Audio-PC includes inside the cabinet an Topping HS01, but if I remove it I don't notice any difference. This could mean 2 things. I have no loopback issues, the Topping might alspo just be snake oil.

It's not snake oil according to Archimago's testing but there have been reports of compatibility issues in some threads such as:
https://www.audiosciencereview.com/...topping-hs02-coming.41614/page-6#post-1551085
https://www.audiosciencereview.com/...ds/e1da-cosmos-adc.27038/page-50#post-1628637
If I've understood correctly that means some devices just don't work with the HS01, not that it fails to isolate.

When you say you can't hear a difference if you remove the HS01, does that mean that you can't hear any ground-related noise (usually hum or 'USB noise' when moving mouse, changing GPU load etc.), or that the noise is still present with the HS01? If no noise you don't have a problem to fix. If there is noise with or without the HS01 then either it's broken or the source of the noise isn't that USB connection.

 

[wwenze]

Well, we are definitely digging into the details here, something shouldn't be done based on design, yet something also works when done.

Put simply, it's like saying a car shouldn't be driven without a seatbelt, or that chain that discharges to ground. But people still do that anyway.
(Don't drive a car without seatbelt, and don't break the ground connection)

The protective function of the ground is clear in USB design. You can't have the receiver's ground floating at, say, half AC voltage, while you tell it to read a differential input at +2V and +1V respectively, it tells you "huh, both are below 0V"

However reality works differently due to details. Nothing can be truly floated, it will always be coupled to something else, and can be modeled as an impedance, or in some cases just straight up a resistor is in the connection. This will eventually couple the source and receiver to an acceptable voltage. Take a look at PCM2706's datasheet for example:

R2 1.5 kΩ connects D+ to VDD. So unless your ground current is measured in amps, D+ would never be more than a volt away from the PCM2706's VDD. D- will become D+ minus whatever the source wants it to be. To be specific, if the PCM2706's equipment is the floated one, then it's its VDD that will follow the source's D+. Then the ground will follow as a function of VDD.

But this gives us another problem: If common-mode noise is caused by unequal ground voltages, but removing ground connection cannot be done, but the indirect connection to ground happens anyway so it still kinda works, but this also means the common-mode noise still makes its way to the PCM2706 so nothing has changed.

FWIW, I have tried that myself by taping the power pins on the source side and connecting a power adapter on the load side. It works, but it didn't solve the problem I wanted it to solve.

 

[Roland68]

Removing the ground connection from USB is absolutely *FORBIDDEN*. It may or may not work, for some time.
The 5V-line is not restricted, though, as it is just there for optional bus-powering a device. But sometimes it is used to power the transciever chip even when a device is self-powered.

Which ground connection do you mean?
The metal connector housings, which are connected via the shielding, or the - wire of the 5v line?

 

[solderdude]

Which ground connection do you mean?
The metal connector housings, which are connected via the shielding, or the - wire of the 5v line?

Pin-4 which is the - for the power supply and the reference for the data lines (for 4-pin USB connections)

 

[Human Bass]

Jitter has been solved and 24/192 is more than plenty.

But USB works perfectly fine most of the time.

I don't think a double electrical/optical/electrical conversion is necessary for a single meter of audio connection with the double result of introducing jitter and limiting the files to 24/192 KHz. and make the clock travel together with the data. Toslink is an old standard from 1980, when chips were expensive!

 

[phofman]

I don't think a double electrical/optical/electrical conversion is necessary for a single meter of audio connection with the double result of introducing jitter and limiting the files to 24/192 KHz. and make the clock travel together with the data. Toslink is an old standard from 1980, when chips were expensive!

I know it was not stated clearly and could have been understood as you did, but @pma actually meant optical USB isolation, not toslink.