USB audio cracked... finally!

[mansr]

4 hours ago, PeterSt said:

It is not so difficult to see how a lacking ground elsewhere can make a USB cable fail or keep on working.

A proper cable works whether or not there is a separate ground path between the components.

 

4 hours ago, PeterSt said:

In this case we must be talking about the most easy principle of the current going where the path is the easiest (the path of the least resistance)

There is no such principle. Current flows through all available paths in inverse proportion to their resistance.

[PeterSt]

26 minutes ago, mansr said:

A proper cable works whether or not there is a separate ground path between the components.

 

???

 

27 minutes ago, mansr said:

There is no such principle. Current flows through all available paths in inverse proportion to their resistance.

 

???

 

I keep on saying : you are always right. I hope it is useful to some.

[pkane2001]

20 minutes ago, mansr said:

A proper cable works whether or not there is a separate ground path between the components.

 

That's true, but USB ground is often required to be connected, even if the +5v supply is not needed by the device. The differential signal pair of wires is often used in reference to the ground connection. If the ground is disconnected in the cable, such a device will not detect USB signal. Having built a number of USB cables with various configurations, I've found that disconnected USB ground works with about 50% of my tested devices.

 

<rant soapbox=on>

And speaking of curiosity, how many arguing here have actually attempted to measure their USB cables? Or blind test them? Or build their own? I define curiosity as the desire to learn the truth, the inner workings of things, seeking an explanation. A blind acceptance of what you're told is the opposite of curiosity. Shouting down someone asking questions is the opposite of curiosity. And certainly, personally attacking someone who doesn't seem to agree with your world view is the opposite of curiosity.

 

Paraphrasing the Most Interesting Man in the World: "Stay curious my friends"

</rant>

[PeterSt]

34 minutes ago, pkane2001 said:

1 hour ago, mansr said:

A proper cable works whether or not there is a separate ground path between the components.

 

That's true, but USB ground is often required to be connected, even if the +5v supply is not needed by the device.

 

In case you missed it, this is not about any separate ground path etc. etc. Mansr just seems to like to read things which were never written.

In this case it is about a transformer (like from a PC) which is not connected to ground. Now be creative on how the return path goes. According to mansr this is everywhere, so we don't need to worry about anything. The latter is of course my own conclusion, but I guess that happens when you can't have a clue about why people say things.

[PeterSt]

We just put out a next small batch of 10 Lush'es and I kind of hope it is the last in a quantity which disturbs (haha). So with a few hours spare I just asked to apply an additional pair of measurement wires to the DAC I am going to use to measure. And then I am really going to do it ...

Finally. So stay tuned.

 

[pkane2001]

1 minute ago, PeterSt said:

We just put out a next small batch of 10 Lush'es and I kind of hope it is the last in a quantity which disturbs (haha). So with a few hours spare I just asked to apply an additional pair of measurement wires to the DAC I am going to use to measure. And then I am really going to do it ...

Finally. So stay tuned.

 

 

Really appreciate your support, Peter! Even if the measurements do not reveal anything obvious, they will still be useful to continue the conversation about what can and cannot be measured and possibly what other measurements can be applied.

 

[PeterSt]

Thank you, Paul.

[PeterSt]

Oh, I very coincidentally found something else yesterday. It is almost to ridiculous to tell about :

 

I suppose that everybody at least once ran into pure static (noise) of the worst kind. Say that this can be compared to white noise, although undoubtedly someone (or everyone) will tell me that this can't be white noise for several reasons. Point is, because of the random nature of the digital data passed through the D/A conversion, the frequency is also evenly spread - at least that is my idea about it. Now, because this sound is window breaking (depending on the gain of the power amp and with the notice that static as such can not be attenuated), this brings a certain overall perceived (!) frequency spectrum. And what I try to tell to begin with, is that everybody who has receives this with a certain amplification chain, will perceive this nature today the same as tomorrow (rather never because it is so ear-shattering loud).

 

And so I am used to this sound myself which maybe occurs for two times a year. It always sounds the same.

But not so yesterday ...

 

Yesterday I had it for the first time with the Lush cable, and quite similar to what "we" perceive from music and a shift in frequency from highest to "lower" with certainly a more profound mid, this static now came across as completely "dark". Say that all the highest frequency "white" turned into a fair "brown". Much more powerful, actually more loud, but also not so much hurtful.

 

I know, everybody who tells me that this "A-B" can not be done, is inherently right. With two times a year and now one occasion of another situation ... how to do it. Still I am confident in my judgment. Something is "crazily" different and the USB cable is doing it.

What I actually say is that whatever one can doubt about with playing music and expectation bias, is not the case with this "test", assumed I am correct in the perceived difference and not my ears suddenly failing on me. So this is shattering hurting (but not the loudest) vs not hurting at all (but way louder).

 

... I now actually think : if *that* can't be measured at the output, then I don't know what can.

It is only that when I start thinking about how to incur for this "signal" for real, I have problems with a good repeatable signal. This will be the same problem as that people will start to (or should start to) say that such a signal is not random, because ...

 

Peter

[Jud]

1 hour ago, pkane2001 said:

<rant soapbox=on>

And speaking of curiosity, how many arguing here have actually attempted to measure their USB cables? Or blind test them? Or build their own? I define curiosity as the desire to learn the truth, the inner workings of things, seeking an explanation.

 

Paraphrasing the Most Interesting Man in the World: "Stay curious my friends"

</rant>

 

Including with regard to the efficacy of listening tests and how to make these better, hopefully.

 

No way I could build anything like either of the two USB connections I’m currently using.  It also would be pretty well impossible to perform at home the sorts of listening tests I would want to do.  Measurements would be interesting, though I’m not sure I’m equipped to do anything very useful there, either.  (Happy to try any tests or measurements anyone is curious about, though probably not until next year, as I’m in the middle of arrangements for moving a couple of thousand miles.)

[mansr]

3 minutes ago, pkane2001 said:

That's true, but USB ground is often required to be connected, even if the +5v supply is not needed by the device. The differential signal pair of wires is often used in reference to the ground connection. If the ground is disconnected in the cable, such a device will not detect USB signal. Having built a number of USB cables with various configurations, I've found that disconnected USB ground works with about 50% of my tested devices.

A USB cable without a ground connection is flawed. It is required by the spec. You're right that in many cases an alternative ground path exists between the components, allowing even such a faulty cable to work. Relying on this is a bad idea. As you may know, the USB data lines are terminated to ground through a 45 Ω resistor. Suppose the only ground connection between the USB source and DAC happens to be through an analogue signal cable. Now what do you think will happen when the USB data lines are driven? To find out, I set up an iFi Nano DAC and a Tascam ADC powered from a linear lab supply with floating outputs and connected to different laptops running off battery. In this arrangement, I recorded a 1 kHz tone at -60 dB with three different USB ground connections:

1. None. This is possible if the ground wire is disconnected after the USB handshake has completed. In this configuration, the only connection between the USB source and the DAC is the USB data lines.
2. Normal. The USB ground wire is connected while the 5V is replaced by the lab supply.
3. Via audio ground at the ADC.

Here's the result:

When there is a ground connection, we get the typical 8 kHz packet noise and its harmonics. With the ground connection provided through the audio signal cable, these spikes are significantly higher than with a the USB ground wire attached. With no ground connection, the packet noise is gone.

[PeterSt]

54 minutes ago, mansr said:

With no ground connection, the packet noise is gone.

 

... which to me reads like a "and that's it". But that is not it at all, because it doesn't explain much.

So an addition :

 

And when there's decent isolation (with connected ground through the cable) the 8KHz is gone too.

 

Now it should be easier to explain the fact of USB noise disappearing when ... eh ...

Think of what "noise" in this case actually is.

[mansr]

15 minutes ago, PeterSt said:

... which to me reads like a "and that's it". But that is not it at all,

No, not at all. Firstly, such a cable doesn't work unless you temporarily connect the ground wire during USB enumeration. Secondly, it only helps if there is no other ground connection between the devices. In practice, this is rarely the case.

 

15 minutes ago, PeterSt said:

because it doesn't explain much.

So an addition :

 

And when there's decent isolation (with connected ground through the cable) the 8KHz is gone too.

It is unclear from my test what effect a USB isolator would have in the presence of an external ground connection between the components.

 

15 minutes ago, PeterSt said:

Now it should be easier to explain the fact of USB noise disappearing when ... eh ...

Think of what "noise" in this case actually is.

Not sure what you're trying to say there.

[PeterSt]

18 minutes ago, mansr said:

Quote

Now it should be easier to explain the fact of USB noise disappearing when ... eh ...

Think of what "noise" in this case actually is.

Not sure what you're trying to say there.

 

Well, I imply the opposite of what I really said. So if you say that USB noise disappears when the ground is not there (after handshake) and I say the USB noise is not there with ground connected but galvanically isolated, what does it actually tell about the source of that noise. So the "noise" as such is current draw (to receive the packets (hey, or send them)). And thus the question, what actually causes the packet noise to not being present. And especially, where is it not present ? (we know, in the output, but ...)

 

And the general message was : these things are clear when stated (it is the truth so nothing wrong there) and accepted (we can measure it and maybe even hear it (better SQ)) but the WHY is way more interesting because then we might be able to make some real progress.

[barrows]

28 minutes ago, mansr said:

Secondly, it only helps if there is no other ground connection between the devices. In practice, this is rarely the case.

Thanks for your insights here.  But I would suggest that this is often the case.  For example, my source has a USB output which is floating in regards to power supply, chassis, and AC, and my DAC has an isolated floating USB input (USB ground and shield/shell has no connection to DAC chassis, AC ground, signal ground, etc)

From your measurements, I would expect that by having this isolation, I would have no 8 KHz packet noise on the DAC side regardless of the USB cable ground status.

 

One interesting thing I have noted with some USB (audiophile) cables:  All have the ground line hooked up, but not all have a low impedance ground path through the shells of the connectors.  The USB cable spec (if I remember correctly) calls for the shells to be connected to shield and ground.  I forget how my LUSH cable measured from shell to shell, but I think it was high impedance.  Indeed, i also recall (I think) John Swenson suggesting that just putting some R on the USB ground connection cleaned up some noise (could have been a ground loop issue of course) at the receiver side.

 

But it looks to me that a truly floating/isolated USB input should solve this little issue on its own, regardless of cable.

 

I have checked a few high end DACs which claim USB isolation, only to find the USB input ground to be common to the analog output ground, hmmm.  I would not call that isolated myself.

[PeterSt]

25 minutes ago, mansr said:

Secondly, it only helps if there is no other ground connection between the devices. In practice, this is rarely the case.

 

FWIW : My situation (the Phasure NOS1) won't bear no ground in the USB cable (at handshake). But of course - as you say yourself - there are other ground connections via a few back doors. Without really knowing, I feel that the shield is of more importance here and that capacitive loading of some sort messes up the signal. And no gnd means no shield either.

This is guessing because I never tried to have that part wrong on purpose. Just like the Lush is just "normal" on these aspects.

I think this is no subject. That of the USB noise is though (to really understand).

[jabbr]

I think that in terms of isolation, noise etc it is better to look at circuits in terms of impedance rather than (DC) resistance.

[PeterSt]

7 minutes ago, barrows said:

 

I have checked a few high end DACs which claim USB isolation, only to find the USB input ground to be common to the analog output ground, hmmm.  I would not call that isolated myself.

 

Hi Barrows - It is quite easy to make this mistake, as you will not see or measure this when you observe the DAC alone. At least in our case (NOS1a) this is so. It goes (interestingly) wrong when all IC connections are made (including USB cable) and the connection is made through the back door ... (with the notice that over an isolator you measure no connection)

NOS1a owners all know about "the famous black wire" which almost explicitly made that connection, but which (now laugh) is present to remove the USB noise. No, was, because with isolation (the "a" version) the USB noise is removed by that but this famous black wire made the connection now (and largely violated isolation, exactly as you say/saw). And so a switch popped up which now cuts this connection.

[mansr]

29 minutes ago, PeterSt said:

I say the USB noise is not there with ground connected but galvanically isolated

That's a contradiction. Something can't be both connected and isolated at the same time.

[PeterSt]

4 minutes ago, mansr said:

That's a contradiction. Something can't be both connected and isolated at the same time.

 

You are correct. But in this case we formally have the ground (and power if you want) at both sides of the isolator, which is a means to legally have no ground connected (it is a functional part of the active isolation device). So functionally the ground connection is there (read : handshake will work nicely and no ground is stolen from elsewhere).

 

But I see I am at the start of answering my own question about how (merely "that") no physical ground connection eliminates USB noise. It should be the same situation as in your test (after handshake). But I still don't have the answer.

I really don't because as told in the previous post, I can remove the noise without isolation (think merely about guiding away the noise). And so I only have the question because of the outcome of your test.

Side note : we also have the outcome of the Intona (commercial) tests/graphs regarding this, but these are heavily overblown if you ask me (and does not remove the noise completely anyway).

 

Otherwise never mind.

[mansr]

27 minutes ago, barrows said:

Thanks for your insights here.  But I would suggest that this is often the case.  For example, my source has a USB output which is floating in regards to power supply, chassis, and AC,

How does it accomplish this? I'm not saying it isn't true, but it's definitely not common. What is your source?

 

27 minutes ago, barrows said:

and my DAC has an isolated floating USB input (USB ground and shield/shell has no connection to DAC chassis, AC ground, signal ground, etc)

What DAC is this? How does the signal get from the USB receiver to the DAC chip?

 

27 minutes ago, barrows said:

From your measurements, I would expect that by having this isolation, I would have no 8 KHz packet noise on the DAC side regardless of the USB cable ground status.

Why don't you measure it?

 

27 minutes ago, barrows said:

One interesting thing I have noted with some USB (audiophile) cables:  All have the ground line hooked up, but not all have a low impedance ground path through the shells of the connectors.  The USB cable spec (if I remember correctly) calls for the shells to be connected to shield and ground.

This is all I could find (quickly) in the USB spec:

Quote

6.8 USB Grounding
The shield must be terminated to the connector plug for completed assemblies. The shield and chassis are bonded together. The user selected grounding scheme for USB devices, and cables must be consistent with accepted industry practices and regulatory agency standards for safety and EMI/ESD/RFI.

 

27 minutes ago, barrows said:

I forget how my LUSH cable measured from shell to shell, but I think it was high impedance.  Indeed, i also recall (I think) John Swenson suggesting that just putting some R on the USB ground connection cleaned up some noise (could have been a ground loop issue of course) at the receiver side.

In case it wasn't clear, in my test setup the shells are not connected either.

 

27 minutes ago, barrows said:

But it looks to me that a truly floating/isolated USB input should solve this little issue on its own, regardless of cable.

One way that might work would be to have the USB receiver bus powered and connected to the separately powered DAC proper only through differential pairs with a high impedance towards ground.

 

27 minutes ago, barrows said:

I have checked a few high end DACs which claim USB isolation, only to find the USB input ground to be common to the analog output ground, hmmm.  I would not call that isolated myself.

Nor would I.

[mansr]

5 minutes ago, PeterSt said:

You are correct. But in this case we formally have the ground (and power if you want) at both sides of the isolator, which is a means to legally have no ground connected (it is a functional part of the active isolation device). So functionally the ground connection is there (read : handshake will work nicely and no ground is stolen from elsewhere).

Right, but this merely moves the problem to the downstream side of the isolator. It still requires a ground return from the DAC.

 

5 minutes ago, PeterSt said:

But I see I am at the start of answering my own question about how (merely "that") no physical ground connection eliminates USB noise. It should be the same situation as in your test (after handshake). But I still don't have the answer.

I really don't because as told in the previous post, I can remove the noise without isolation (think merely about guiding away the noise). And so I only have the question because of the outcome of your test.

Side note : we also have the outcome of the Intona (commercial) tests/graphs regarding this, but these are heavily overblown if you ask me (and does not remove the noise completely anyway).

If I had one of those, or some other isolator, I'd put it in the chain and see what happened. Someone else will have to do that.

[elcorso]

 

I could do some measurements, but this is not my goal now ...

 

I do a lot of experimentation, of course. It's part of my 'audiophile' life, but just to try to get the SQ I like, not necessarily the best, but the best for my taste ...

 

And being the most important thing (to me) to listen to music! Good ears? I will never really know, because they are the only ones I have and with whom I was born ...

 

Roch

[marce]

4 hours ago, barrows said:

 

I have checked a few high end DACs which claim USB isolation, only to find the USB input ground to be common to the analog output ground, hmmm.  I would not call that isolated myself.

The structure of an isolated ground island separated from the main ground via a ferrite bead would be connected at DC and thus at 0 oms DC, but would have some high frequency resistance, not galvanic though. The best results are full galvanic isolation with filtering at both sides of the isolating device for power and ground (they all have some internal capacitance.)

[marce]

The USB single ended signals will find a return path where they can if the actual USB return (ground!) path is broken. If this happens it is not good from an EMC point of view. This can often be via the PE and chassis connections meaning the noise created by the greatly enlarged USB signal loop area can happily couple everywhere.

[mansr]

22 minutes ago, marce said:

The USB single ended signals will find a return path where they can if the actual USB return (ground!) path is broken. If this happens it is not good from an EMC point of view. This can often be via the PE and chassis connections meaning the noise created by the greatly enlarged USB signal loop area can happily couple everywhere.

I believe my test showed something along those lines.