USB audio cracked... finally!

[rickca]

6 minutes ago, PeterSt said:

Net, the ISO-Regen in the chain should be for the better, but only when at the PC side.

OK thanks.  What prompted my question is that @lmitche reported that he got the best results doing precisely that, and I started thinking about why.

[PeterSt]

2 minutes ago, rickca said:

What prompted my question is that @lmitche reported that he got the best results doing precisely that, and I started thinking about why.

 

Aha ! I didn't catch that one. I will reread his post about it.

 

Anyway, assumed that lmitche himself is right, we can see that audio is not really voodoo and that some things can be reasoned out.

(and as a matter of fact I like to reason out everything and all in advance - sometimes in aftermath as long as there's consistency in happening + story)

[PeterSt]

11 hours ago, lmitche said:

The USPCB > ISO REGEN(LPS-1) > Lush Cable chain is just sounding extraordinary here

 

So that, right ? And the USBPC is at the PC side I suppose ... 

 

Thus yes, that would "kill" the regeneration, but retain the isolation. Similar to the NOS1a/G3 where the USB isolation is at the other end (in the DAC). So on that matter, the isolation "anywhere" seems sufficient.

 

Funny ...

[greenleo]

2 hours ago, PeterSt said:

 

Slower rise (/fall) time of the single wave.

 

 

At first sight, yes. But when looking closer, we see that USB3 is not utilized for its speed when it is about a USB2 transmission (and with D/A converters it is always that (AFAIK)). So USB3 has a larger capacity. It's the car which can do a 200 and therefore accelerates fast when going from 60 to 70. The car doing a 100 can also accelerate from 60 to 70, but not as fast. It struggles more.

Thank you for your reply, Peter.

 

For paragraph1 of your original text:

JS of Uptone did mentioned that the rising time and falling time of USB is tricky, in the sense that no good if too fast or too slow.  Hence, paragraph1 means you have found the range of optimal rising time and falling time of USB in audio application.

 

For the 2nd paragraph of your text.  Does it means that a USB 3 port will use even smaller power to deliver the same number of bytes?  Then less electrical noise with e generated, hence better SQ?  I understand your analogy.  However, I find it indispensable that if the analogy may be expressed in physical termed as well.

 

 

 

[marce]

If you plug a USB 2 device into a USB3 socket ALL data transfer will be at the standard USB 2 speeds and down the USB 2 wires D+/D-. USB3 its self uses extra wires within the cable, an RX diff pair and a TX diff pair (also an extra GND connection). So USB2 is always going to work as USB2.

[PeterSt]

38 minutes ago, greenleo said:

For the 2nd paragraph of your text.  Does it means that a USB 3 port will use even smaller power to deliver the same number of bytes?  Then less electrical noise with e generated, hence better SQ?

 

Hi - Well, the smaller power can be so, but it does not work like that really (also see Marce's post please). So, USB2 will be in order as if it were USB2 for real but with the difference that a USB3 "chip" will be dealing with it. And now all is different because the signal goes through that chip, is generated by the chip etc. etc. and although I can design-in such a chip, I am not able to predict that it will sound different, let alone sound better. So this is just empirical finding with the notice that it shouldn't make an audible distance at all. But it does.

 

Generally I'd think that the over-capacity of the USB3 chip will net cause less power to be drawn. To see this happening more excessively, we can today look for USB2 chips doing a best job, find stuff which is 10 years old only, and seeing that it produces more heat (needs better cooling) than any modern USB3(.x) chip which is "infinitely" faster. So we also deal with better quality chips today, compared to say 10 years ago.

Also, it gets again more interesting if we see that those chips too carry jitter and PLL's to tame the jitter. These days you won't even notice because it is all on-board of the chip. But 10 years ago it was outboard, even requires outboard buffer memory and the signal again passes through all this stuff. And again, it should not matter for sound, but apparently it does. 

[PeterSt]

34 minutes ago, marce said:

So USB2 is always going to work as USB2.

 

Does that also count for the clock speed (24Mvs25M) ?

[rickca]

It is interesting to note that the ISO REGEN uses a USB3 hub because beta testers found it sounded better.  Alex notes, however, that he turns off the superspeed stuff.

[Jud]

16 minutes ago, rickca said:

It is interesting to note that the ISO REGEN uses a USB3 hub because beta testers found it sounded better.

 

I think Alex and John also found it to sound better (perhaps expected it to).  Larry and I independently confirmed this without knowing what if any differences existed between the two test samples we each received.

[Jud]

4 hours ago, PeterSt said:

Slower rise (/fall) time of the single wave.

 

Not understanding how this would avoid changing zero crossing times and thus creating jitter.

[PeterSt]

51 minutes ago, rickca said:

I asked earlier if there is a good introduction to the design of the current NOS1a G3 product.  It's difficult for a novice to your world to follow the evolution of your technology.  In particular, I don't even understand what is new and different in G3.

 

 

I suppose it is seriously off topic, but since you asked twice I will try to give a nice answer.

 

The most important part about the G3 upgrade is its all new new output stage. The output stage should be "ridiculously good" because it was derived from the poweramp prototype I created for the ever meant NOS2 (no NOS2 will be forthcoming any more). This poweramp was meant to do 8 channels of 72W RMS each, and it measured better than the old gain stage of the NOS1(a). Not everything is about figures of course, but for me it counts that if the THD ain't right, it won't sound good as well. Also, because we talk about the D/A converter, all what's wrong in there will be amplified (by say 20dB). Back to the poweramp, this measured 0.00059% of THD (1 KHz) at 18V into 3 Ohms with a totally flat noise line as is common from the NOS1(a) itself.

 

Because that poweramp version was so smacking good (measuring - I never listened to it) I thought that a lousy 2.25V RMS into 5V should not make that same amp worse. Only less heatsink. 

What also had to happen was that the I/V (current to voltage) conversion had to be integrated and furthermore the "amplifier" had to be set to unity gain (which is zero gain and thus actually no amplification) so it became a buffer with as much current possibility as I liked. So yes, I myself played for 9 months or so with 1 Ampere of current possibility (and say that we never know how much is when (for a 0.0001sec) needed).

You could say it became a simple buffer, but way oversized (OK, most NOS1a owners also use the 5GHz (over 160m capable Blaxius) interlinks, so we for dead-sure are completely crazy).

 

Because the sound of the B'ASS (which is how the gain stage is named) is so drastically different and actually nothing to pus through the throat of any NOS1a customer who listens to "his sound" for 5 years or whatever it is, the biggest feature of the B'ASS is that it can be configured to only have active the good old I/V conversion and bypass the "buffer". So, people would pay for something radically new (610 euros IIRC) but can luckily shut it off.  And they'd really have the old situation, if you imagine that the chips which require burn-in were moved over from the old I/V board (if you must listen for a month to a non-burned in NOS1 after knowing how the burned in sounds, you will look for a high bridge).

 

While with the buffer being active in an indeed no feedback design, there's also the possibility to configure it with feedback. From there a gag emerged, because the feedback "allows" for a quite high 2nd harmonic which people may love. But only when the SE (Single Ended) output is used, because else it is canceled by the differential setup (sorry - haha).

 

So now the 3 in G3 starts to emerge, because the B'ASS stage actually comprise of 3 very different gain stages, sounding totally different as well.

 

In the end the 3 is about other elements because it took me so long to approve the B'ASS and make a decently working board which would also fit in there somewhere (with crazy capacitance buffers) and which is set up fully double mono, that I allowed myself to work on something like the Phisolator. Well, no need to talk about that long, but that was done under way and it ended up inside of the NOS1a/G3, although this was at first not planned.

 

This is 2.

3 is the fact that someone started to  ask Crystek to produce 24MHz "femto" oscillators and while I found their secret stock I also bought it all. So if someone somewhere ran out - blame me.

But don't tell anyone, because this "3" is the secret upgrade on the USB interface, so nobody knows.

 

To get back on topic, what not everybody knows is that I was a bit slow with the Phisolator thing, that is, its due final version which would be outboard. I was so slow with it all, that when all was set for the G3 upgrades for everybody and I sat down to "re-do" the Phisolator and just had named it Resolator, some bloke named Alex C put forward an ISO-Regen (which would do the same as the Resolator, even with largely the same components). On the spot (Phasure forum) I announced that I then better did something else, and via via the Lush came from that.

I hope we can stay on-topic !

[marce]

Its periodic switching the slope of the wave does not matter to some extent,  as long as its the same from wave to wave...

[PeterSt]

That is very good, because it doesn't.

Jud, anything what triggers anywhere, triggers on edges (slopes). OK, maybe a scope can do it on zero crossing.

Now don't tell me your question won't change because of this. 

[PeterSt]

1 minute ago, marce said:

Its periodic switching the slope of the wave does not matter to some extent,  as long as its the same from wave to wave...

 

Marce, I see you actually never "quote" as such. This may make your post fall out of context, like in this case (because of my (larger) post in between by now).

If you drag the text you want to quote, you will see a small balloon popping up allowing you to quote that conveniently.

[marce]

Missed the zero crossing, only a major concern when you wanna work out RMS voltages for sine waves, digital switching dosen't use zero crossing...

[marce]

Yes, I should have added the quotes.

 

[manisandher]

9 minutes ago, PeterSt said:

I hope we can stay on-topic !

 

I've pasted Rickca's question and your response in the 'NOS1 G3' thread.

 

Mani.

[Boleary3]

On 7/20/2017 at 1:22 PM, Boleary3 said:

Though the Lush has a very, very pleasing sound it loses the "air" of the Clairixa. To my ears, in my system, in my house the Lush imparts a smooth, rounded and detailed sonic signature, but I have to turn up the volume to get the  bloom I so love with the Clairixa. Though the sound at higher volume is good, its just too damn loud for me. I am used to what is for me better sound, at a bit lower volume.

 

When I did the above comparisons between the Clairixa and the Lush I forgot that I had my entire system running on 115 volts and not 120. I am currently auditioning a Pure Power 3000+ power re-generator. I have everything in my system plugged into it. Highly recommended! I had the PP set to output 115 volts because that's where the Clairixa sounded its absolute best. The hard bright spots that show up with the Clairixa cable when a female voice (say Patty Griffin) goes from 0 to 100 are much more controlled at 115 volts. Thus, with the Clairixa,  I was able to get lush qualities plus the "air" of  clairixa, without any noticeable loss of transient speed at 115 volts. This weekend I reset the PP to 120 volts and have been gobsmacked by the sound with the Lush. No longer any need to bump the volume with the Lush in place. What is really remarkable about the sound of the Lush is that it is both smoother sounding and, at the same time, everything, instruments and voices, are more solid sounding. Everything has more "presence", thus everything sounds more natural. After listening for several hours I again reset the PP to 115, inserted the Clairixa and was shocked to hear how the airiness of female voices suddenly sound hollow compared to the solidness of the Lush. Incredible achievement.

[mmerrill99]

1 hour ago, marce said:

Its periodic switching the slope of the wave does not matter to some extent,  as long as its the same from wave to wave...

Is it fully periodic? Yes, USB frames are sent every 1mS & USB microframes every 125uS (high speed) but there is a variable bit pattern being sent & so each microframe will be different. Are we not looking at the same weakness as was  uncovered in SPDIF receivers - Inter symbol interference (ISI) & the Dunn jitter test was specifically designed to examine?

[PeterSt]

23 minutes ago, Boleary3 said:

This weekend I reset the PP to 120 volts and have been gobsmacked by the sound with the Lush. No longer any need to bump the volume with the Lush in place. What is really remarkable about the sound of the Lush is that it is both smoother sounding and, at the same time, everything, instruments and voices, are more solid sounding.

 

Hi Brian,

 

Well that is great.

Of course over here it sounds 2x better again, at 240V.

I hear that in Australia they get well over 250V at times. So it is time to start packing ...

 

Anyway thank you again for sharing. Not sure yet what to think of this though. All I know is that USB is bonded to te mains for ground.

 

Quote

everything, instruments and voices, are more solid sounding.

 

Yes. And I am confident that nobody who ever heard this, can understand what this really means. If you hear it you will undoubtedly think "ahaaaa, so THAT is how low jitter sounds !". ... But never mind how special it is, it is not low jitter (I wouldn't put all my money on it though).

 

On another note I am a kind of confident that this kind of sound also implies the more forwardness (which includes more depth automatically). Something like "less spread" and thus "more coherent". Not sure if this makes sense and maybe you can improve my thinking and especially wording, Brian.

Thanks.

[mmerrill99]

A measurement of the emi emissions from the cable (at the receiving USB port), like I posted JosephK had taken with Wireworld cables, I would imagine would be interesting to show differences between Clairixa & Lush & other cables

[Jud]

1 hour ago, marce said:

Missed the zero crossing, only a major concern when you wanna work out RMS voltages for sine waves, digital switching dosen't use zero crossing...

 

Ahh - thanks.  So Figure 1 in this paper (taken from a 1996 Hawksford & Dunn paper) would not be a problem?  http://www.amr-audio.co.uk/large_image/MAC OSX audio players & Integer Mode.pdf

 

Or am I thinking about something else entirely than @PeterSt is talking about?

[PeterSt]

Jud, I assume you looked at this picture :

 

and assumed the blue horizontal line to be zero. This is not so, as zero would be at the mouse pointer (just a bit under the lowest level).

The blue horizontal line is what is says : the threshold of where a 0 starts to be a 1 on the rising edge and where the 1 starts to be a 0 on the falling edge.

 

I did not look into the text, but without text and only looking at the picture, this is DC Offset.

And yes, DC Offset, which is most often not stable at all, causes jitter in this case.

 

DC Offset ?

Now that ! 

 

 

 

[bibo01]

55 minutes ago, mmerrill99 said:

A measurement of the emi emissions from the cable (at the receiving USB port), like I posted JosephK had taken with Wireworld cables, I would imagine would be interesting to show differences between Clairixa & Lush & other cables

Funny you say that. I was thinking to get a Lush just to send it to my friend (JK).

[Jud]

28 minutes ago, PeterSt said:

Jud, I assume you looked at this picture :

 

and assumed the blue horizontal line to be zero. This is not so, as zero would be at the mouse pointer (just a bit under the lowest level).

The blue horizontal line is what is says : the threshold of where a 0 starts to be a 1 on the rising edge and where the 1 starts to be a 0 on the falling edge.

 

I did not look into the text, but without text and only looking at the picture, this is DC Offset.

And yes, DC Offset, which is most often not stable at all, causes jitter in this case.

 

DC Offset ?

Now that ! 

 

 

 

 

Sorry, I was thinking "threshold-crossing" but saying "zero crossing," so that understandably created confusion.  So back to the original question: if the slope in this figure is altered, the threshold-crossing point is altered, yes?  Or is this not the slope you are talking about altering?

 

DC offset, yes.  Same with noise on ground (if that is in fact separate from what you are referring to as DC offset - maybe not).