NOS1a : A new "tweak"

[PeterSt]

Some will love this, some may not at all. Personally I think I do because what else would be left for tweaking ...

USB cables now matter.


Most of us know that the last larger "noise" subject in order was about the USB interface and how to tweak that for the better. This was not per sé about better ("SQ") quality USB cables, but the way the interface performs as a whole and how things are related to timing. Also, the very very last subject I recall was about the length of the cable of inherent same quality.

Never I saw a real relation myself between that and XXHighEnd settings, or in other words, I never saw anyone writing "with USB Interface tweak XYZ you better set dial A in XXHE to 0.7". So not.

Maybe it has been too much in between the lines, but I already had my doubts about Windows 8  vs. Windows 7 not making a difference anymore, but which I dedicated to the PC's being different for it over here. Btw this attitude did not change although I must say I did not spend mucht time on it anymore. What did change though is my attitude to the PC itself not making a difference, which in itself I now like to dedicate to USB again. Again ...

The above summarized : XXHighEnd hence software does not influence anymore, but that other thing in parallel, USB ... I think it still does. But it also does more than before now ...

This latter needs more experience than I can apply on my own. I mean, I was told a week or so back and only yesterday felt like having some serious time to try it myself. For me, and this one evening with one swap only, it would be too soon to really state it is true, but combined with what I was told - yes. And it doesn't even look to be marginal.
Fun for me anyway : the sound improved from using that other cable and accidentally this is the one supplied with the NOS1(a). Otherwise I have always been using a USB3 cable with converter (not the best they say, but I didn't pereceive a difference from it anyway).

What has changed is the again better resolution the NOS1a shows hence it is more easy to detect differences. So for me myself - I have been listening to the NOS1a for maybe 10 weeks by now, and the change of USB cable (just one random attempt) made me say "wow, so there's even more improvement around the corner !".

Yes, and this while it has always been such a nice gag that for the NOS1 the cables did not make a difference. So maybe I am a bit sad as well. Or not at all - still to decide.

I promised myself to measure the performance of the USB interface because with the (ultra) fast analogue scope I am sure I can do/see a few things. When this is going to happen for real, don't ask me. But it should now be at a more "technical interface level", or IOW a 100 times more easy than when XXHighEnd would also be influencing in parallel.

There is one more thing of importance :
Because I generalized the both XXHighEnd and USB interface to being one and the same (maybe against all odds because as said I always have been thinking that both are not about the same influence), all NOS1's we receive have had reconnected their internal famous ground wire to the PSU screw. This has not been really thought over, but it looked more "back to basic and all good". By now I don't know that for sure, but I do know that my own wire never was reconneced, as well as that I still use the isolated Silverstone. All I want to say with this is that while people disconnected that wire in their NOS1 they may like to do that again in the NOS1a because this *is* 100% USB related.
Meanwhile we will keep on reconnecting them during the upgrade from NOS1 to NOS1a (to have things consistent for everybody).

Summarized

Not many people will expect that now they can try different USB cables, just like I myself never made an attempt. It is thus also not said that at this moment with your NOS1a you have the best achievable sound (like in my own case as it seems).
At this moment I think that your best (or cheapest) option is to try different lengths. Thus not more expensive but just different lengths and otherwise just "other" hence what you have laying around.
It is a hunch that the shorther the better it could be. Only a hunch.

Happy tweaking !
Peter

[pedal]

all NOS1's we receive have had reconnected their internal famous ground wire to the PSU screw. This has not been really thought over, but it looked more "back to basic and all good". By now I don't know that for sure, but I do know that my own wire never was reconneced

SUGGESTION: Make it switchable. For the sake of convenience.

[PeterSt]

HA !
And why not. The middle switch is free ...

[Scroobius]

Oh Dear I thought I was finished with tweaking. So maybe Dexa clocks on the USB interface will still make a difference? I removed the Dexa from the PC (while my NOS was being upgraded) so later today I will put it back and see what happens (the 24MHz USB Dexa remains in my NOS1a).

Also I still have my super short 25cm USB cable which for various reasons was removed but I shall put that back in today and report back with findings.

How can I explain to Ros that this is more important than gardening?

Paul

[AlainGr]

Hi Paul,

You asked Peter not to remove the Dexa clock from the NOS1 while being upgraded to the NOS1a ?

Alain

[PeterSt]

Hey Alain,

Paul can speak for himself, but it never was a subject. So we can also twist this into : we never asked Paul to let it stay or to remove it. So whatever it does or does not, it should not influence the upgrade itself.

Regards,
Peter

[manisandher]

... I was told a week or so back...

I came to the conclusion that USB cables now matter greatly (with the NOS1a) about 3 weeks ago. I wasn't even looking for differences but just stumbled upon them. I was experimenting with where I could place my PC and NOS1a, and had pretty much resigned myself to having them both sitting exactly between my speakers in my main room. I then decided that it would be better to have the PC in the corner of the room, which would require a 3m (or so) USB cable. I got out my (very) expensive 3.5m USB cable, plugged it it and immediately thought WTF?... this sounds way worse than the 'standard' 1m cable I had been using!

I then went through a bunch of USB cables (all cheap printer cables) of various lengths, and they all seemed to have a slightly different sound. Unfortunately, with the way things were situated, I couldn't try a 0.5m length - maybe Paul (Scroobius) can share his findings with short lengths when he tries one. But what I did try was a long 5m USB cable. And you know what? It sounded way better than any other cable I had tried. So with my setup, there isn't a clear inverse correlation between length and SQ. If I had to hypothesize, I'd say, generally, the shorter the USB cable, the better... BUT...

... Very long USB cables (getting to the edge of the USB spec) may sound even better. I know they are very different animals, but I've found exactly the same thing with spdif/AES3 cables. The longer (~5m) cables sound better than shorter ones. For these types of digital cables, the practice actually matches the theory - they are terminated (or should be) so RF reflections come into play. As for whether the same theory would apply to USB cables, I have no idea.

Oh and BTW, I get absolutely zero USB errors showing in the USB Control Panel with the 5m USB cable.

I think Peter's advice of trying a few 'standard' (i.e. non-audiophile, non-expensive) USB cables of different lengths is bang on. So my strong recommendation is to buy a bunch of 'standard' cables from 0.5m to 5m lengths and give them a go. I mean, why not? The outlay should be no more than 20 Euros max, right?

Mani.

[manisandher]

And another point while I remember it...

I also tried the mobo USB port vs. the Silverstone card. The Silverstone card was much better than the mobo... but only with its bracket still attached.

So in summary, the best sound I achieved was this:

- SS card with its bracket attached (i.e. non-isolated)
- 5m 'standard' USB cable

Both these findings were total shocks for me and the exact opposite of what I expected (so no placebo here).

Oh and in case it matters, I have never touched the black cable within the NOS1(a) itself.

Mani.

[PeterSt]

Thank you for your help and considerations Mani ...

Peter

[manisandher]

Thanks Peter.

Here's something that John Swenson wrote on CA:

"First off, USB is not one signaling standard, there are currently 3 in use, the one in use for a large percentage of DACs at the moment is the high speed standard, so I will cover just that.

HS mode runs at 480 Mbs, at this speed over normal human cable runs(measured in meters not mm) the "signal integrity" (SI) at the receiver can vary GREATLY. The spec allows for a very wide range of SI at the receiver, this means the receiver circuit has to be good enough to recover the data over this wide range of SI.

This receiver is NOT just a simple logic gate input. It uses a multiphase clock at 480MHz, some use 8, some 9 and the ones I worked with use 12 phases. This means there are twelve clocks, each at 480MHz, but slight delays between clocks. These different phases are generated by digitally controlled delay lines with 1/2ps resolution, you read that right, one half pico second resolution. Each of these clocks controls a very fast ADC with a few bits resolution. Thus every "bit" on the incoming analog signal is sampled 12 times, the digital representation of these analog values go into some DSP circuitry that tries and figure out where the "edge" actually is amongst all the noise, jitter and reflections on the wire.

If the SI is pretty good this determination is pretty easy, if the SI is not good, it has to work harder to figure out where the edge is. Part of this process is continuously tweaking the frequency of the clocks AND adjusting the exact delays of each of those phases to get the best data to the DSP block. Even though the clocks are 480MHz, the rate at which the frequency and delays are adjusted is right in the middle of the audio range. The worse the SI the more tweaking goes on.

Lest someone thinks this is USB only, this sort of thing is the basis for ALL high speed serial interfaces. USB HS is actually the simplest because it has to be dirt cheap, others do even more complex feedback on the input parameters. You really don't want to know what goes on in a thunderbolt receiver!

All of this processing to track the input signal generates lots of noise on the power traces and ground plane of a board. Even with very careful circuits and board layout some of this gets through to the DAC chip and its clock, which CAN change the output signal. Note: none of this actually changes the bits, that input circuit IS good enough to recover the actual bits. It is the consequences of the heroic measures taken by the receiver to get those bits in the face of poor SI that can affect the sound.

Measuring this stuff is not easy, frequencies are high and amplitudes are low. I don't have anywhere NEAR good enough test equipment at home to do this justice (we are talking about at least $50K, I can't afford that!!) and if I tried to use the stuff at work, I wouldn't be working, so a book full of graphs proving all this is not going to be coming from me anytime soon.

Gordon just got some equipment that CAN show this so sometime in the future you might be seeing some data from him.

I hope this helps a little bit in understanding how a USB cable may change the sound of a DAC.

John S."

I think the NOS1a has eliminated this as a cause for differences in SQ between different USB cables. But I'm posting this as a reminder of how complicated this all can be.

Mani.

[pedal]

Thanks Peter.

Here's something that John Swenson wrote on CA:

Great quote, thank you! WHEN did he post it?

[AlainGr]

Hi Pedal,

It was posted last month (June 12)

http://www.computeraudiophile.com/f8-general-forum/new-nordost-heindall-2-universal-serial-bus-industry-standard-cables-connectors-and-communications-protocols-between-computers-and-electronic-devices-19782/index5.html#post327648

Alain

[Scroobius]

Hi Alain,

You asked Peter not to remove the Dexa clock from the NOS1 while being upgraded to the NOS1a ?

I left the Dexa in when I took my NOS over to Peters. First of all I simply could not be bothered to put the old (cheapo) clock back in. The Dexa certainly would not have any adverse affect on the sound and likely would have no effect at all on the NOS1a upgrade (proven I think by the tests Peter carried out on the upgraded unit). But read on I am glad left it in.

So this hot afternoon I tried a couple of things.

USB Cables

Unlike Mani I do not have any expensive USB cables. However, I do have different lengths of USB cheapo cables.

I have to say I really struggle to hear any significant difference between 5m and 25cm. Maybe, just maybe, I might prefer shorter (25cm) but I am not confident I would pass a double blind test. Certainly I do not hear any important differences. Before the NOS1 upgrade I could hear a clear difference and 25cm was always better then.

Dexa Clocks

All my listening since NOS1a returned has been via a mobo USB3 and excellent the sound quality has been. I do not have a standard PCIe USB3 card to try as all those that I have ended up with the clocks removed. But I tried the Dexa driven PCIe Silverstone card. In short Dexa's at both ends of the USB (PC and NOS1a) is better. But to get this in perspective the improvement is nothing like as big as before. Then it was a night and day improvement. Now it is like taking the very finest sound and applying the finest polish for the slightest of improvements. But actually the more you listen to it the better it gets. But lets be clear this is a small improvement but which is clearly there.

Earthing

I quickly tried lifting the USB ground wire - I have never heard any differences doing this before and nothing has changed I heard no difference this time. But maybe that is because I use the Dexa's. But who knows?

Anyway for sure I will not be selling the Dexa's just yet.

With the sound quality I am getting now I get the distinct impression that there is even more stunning sound just waiting to get out - so now I am in even more eager anticipation of my new Orelino's which Bert is now working on. At least I hope he is (ha ha). Maybe with the Orelino's I will hear more of a difference. Can't wait!!!!

Paul

[Nick]

Hi Pedal,

It was posted last month (June 12)

http://www.computeraudiophile.com/f8-general-forum/new-nordost-heindall-2-universal-serial-bus-industry-standard-cables-connectors-and-communications-protocols-between-computers-and-electronic-devices-19782/index5.html#post327648

Alain

Fascinating Alain,

And I guessed the input might be a pll !

12 phase 480mhz clocks..... no wonder a decent clock and targeted on board bypass capacitance helps so much around the interface board and USB chip !

Also might trow some light on why clock speed and matching are important, will have to think in this.

Nick

[manisandher]

Unlike Mani I do not have any expensive USB cables. However, I do have different lengths of USB cheapo cables.

I have to say I really struggle to hear any significant difference between 5m and 25cm.

Hi Paul, have you tried any lengths between 0.25m and 5m? If you have, and all USB cables still sound pretty much the same... get yourself a more transparent pair of speakers... haha.

Mani.