USB Nervosa Thread Decrapifiers, pro interfaces, and bears oh my

In my setup I compared AES to SPDIF from my RN3.
I was all set to compare the TOSLINK as well (I have a Lifatec glass fiber cable) but the optical output from the RN3 is ADAT not TOSLINK.

Even so I did compare the toslink straight from my Mac into the PWD dac.
While the TOSLINK was better than straight USB as soon as any form of decrapifier for USB was introduced, it immediately eclipsed the TOSLINK.
But then the SPDIF eclipsed the even further tweaked USB, which was then eclipsed by the AES.

These last 2 steps in comparing SPDIF & AES was a much greater step than any of the preceding steps, and granted the RN3 plays a HUGE role in this SQ upstep.
But then AES ain't no slouch seeing as how it avoids a slew of technical 'issues' that SPDIF has (which they brought upon themselves).

And AES does respond well to all silver balanced true 110Ω cables as well.

JJ

 

Thanks bixby. Do you recommend glass over plastic? I am considering the Supra ZAC optical cable but it is only plastic.

 

What do people think of this digital only sound card on ebay:

http://www.ebay.com/itm/Elfidelity-...675286?hash=item41b05bb896:g:jGoAAOSwgZ1XrE-E

Looks interesting, supports both unbalanced output via Toslink, RCA, and BNC and balanced output via TRS. Has a PXO clock at 15ppm, is that any good?

May be better than an Asus Xonar Essence STX II as a digital transport?

I have the Elfidelity USB power filter and that definitely cleaned up my USB signal compared to straight USB from the motherboard.

 

I have only used crappy plastic so no experience with great plastic, but it should exist. And there is poor glass as well, unfortunately.

 

What I use: https://www.bluejeanscable.com/store/digital-audio/index.htm

 

Many of us have also tried and liked the Lifatec glass optical cable. I used one with great success from my PS3 into my integrated before I upgraded my setup. More discussion here.

 

Thanks for the advice and links on Toslink cables, especially the Blue Jeans quote. I vaguely remember reading somewhere else about single fibers being preferred to multiple for optical digital use.

It is a bit of a pain that the Essence STX II requires a mini-toslink, it limits my choices somewhat. I think I will acquire the Supra ZAC and give it a try. I could always purchase the Lifatec cable at a later date when the pound is a bit stronger.

If I took chance on that Elfidelity card are linked to, I would need a regular Toslink adapter or use my current IXOS cable.

I'm also considering comparing Toslink to a good quality digital coax cable, probably the Supra Trico. I have an ordinary RCA cable I could try before then but it will have to wait because I'm currently using it and its twin to connect the Modi Multibit and Jotty.

 

Has AES and I2S hdmi:
http://www.psaudio.com/directstream-memory-player/

 

About that Elfidelity card: it could be great, or it could be terrible. I have tried various SPDIF sources and the worst of them (my mobo onboard and iBasso DX90) are slightly worse than straight USB. If that card is in the same category then that's £100 down the toilet (I'm gathering you are in the UK).

My two penn'orth: rather than chase potential bargains like that, either (a) stop here or (b) chase an actual bargain. Lurk classifieds until an RME or a Lynx comes up at a good secondhand price. If you can score one at < price of your Modi Multibit you are set against future DAC upgrades and won't have to think about digital source for a very long time.

 

Thanks for the good advice Garns, that is what I will do.

I don't think the Elfidelity card will be any better than the Essence STX II card I currently have. The quoted 15ppm jitter doesn't really mean a great deal without a center frequency. I found this site for calculating jitter: https://www.jitterlabs.com/support/calculators/ppm

If we assume a center frequency of 10khz, this equates to a jitter of 3ns. Assuming a center frequency of 20khz gives a jitter of 1.5ns .

According to this site: http://www.stereophile.com/content/asus-xonar-essence-ststx-soundcards-measurements , the original Essence STX has a jitter of 1.48ns on the Toslink output. I have the Essence STX II which has improved jitter performance over the original, therefore I will stick with what I have If I see an RME card going for a good price, I will acquire it and make a comparison with the Essence STX II .

 

Since discovering for myself the sizeable improvements wrought by SPDIF vs USB into the Gungnir Multibit I have been really bugged to know exactly what the root cause of all this is. @atomicbob did some nice measurements back here which narrowed things down to jitter and noise. Here's something I stumbled across by chance today which surprised me, and which suggests that jitter is a key player.

My portable rig is an iBasso DX90 feeding a Chord Mojo over SPDIF coax. The Mojo has always sounded pretty tizzy through it. I generally use Rockbox on the DX90, but today some spidey-sense made me dual-boot into the stock firmware. To my great suprise there was a clear improvement on the Mojo, audible even on the train. The improvement is exactly the same nature that I get going from SPDIF to USB on the Gungnir Multibit: bass is improved, soundstage more precise, transients clearer, layering is better, treble much less tizzy.

I did a test with the Gungnir Multibit when I got home. Previously I have observed DX90 SPDIF source, using Rockbox firmware, on a par with straight USB. I tried again using also the stock firmware and obtained

RME 9632 >>> DX90 (stock) >> DX90 (Rockbox) ~~ USB

So here we have exactly the same hardware delivering quite different performance with different firmware. I am disinclied to believe that changing firmware would change the inherent noise of the digital signal, so that leaves jitter. My thinking is that the limited hardware makes the different loads of the software yield different jitter performances. I found a believable explanation that supports this over at head-fi (here):

The stock firmware is very stripped down. Rockbox firmware is not very optimised. So this seems to make sense. If the above is correct, this seems like this would be a great test-case for isolating the effect of jitter on the sound.

 

And to add yet another aspect to this.
This only seems to apply if the computer that feeds the dac is on for an extended period of time.
And thus far I can only say this applies to Mac but I suspect it also applies to pc's as well, especially windows OS.

My system runs 24/7 and after 6-8 hrs, sometimes like 12 hrs, when the amount of ram being used and a system cache file grows to many GB's size…
A complete system reboot will yield a noticeable change in SQ, one that I hear as 'better'.
And at the same time I will also reboot my PWD dac and the RN3, which also seem to help just a touch more.

Then just before I start playing a track I'll purge that cache file.

The sum total of all of this is a very significant improvement in overall SQ.

Crazy, huh…?

JJ

 

@johnjen

Maybe the effect you are describing is related to:
https://en.wikipedia.org/wiki/Resource_leak
https://en.wikipedia.org/wiki/Memory_leak
https://en.wikipedia.org/wiki/Software_aging
https://en.wikipedia.org/wiki/Fragmentation_(computing)

Before I didn't believe that this effect affected something with embedded firmware like Master7 DAC, Yggdrasil DAC and Mutec MC3+USB. But my experience with flashing a 24hr/7day running 1yr Mutec (which finally rebooted) has changed my opinion on this effect. Now once in a month or so, I do power cycle my audio systems to perform some kind of "refresh".

You should also do this reboot to your ISP modem/Fiber ONT and Wireless Router once a week or more. This will greatly improve internet performance.

Well there's also some like bug head emperor programmer who will go to the extreme step of rewriting the audio file before playback.

 

I found on the Yggdrasil, USB gen3 was better than the DX90's coax output with the Lurker firmware. However, I can see the dx90's coax output being better than the USB inputs on portable dacs like the Mojo or Gamma2.

 

I have tested three portable transport with my Mutec MC3 USB+Yggdrasil:

In order of (subjective) Ranking:
1) Sony ZX2 (Good Frequency Balance but very lacking in depth/stereo width compared to Win10PC USB+Wyrd)
2) IPAD Pro 9.7 + Apple CCK (OK Frequency balance, very bad imaging, slight rolled off treble but is listenable)
3) DX90 Coax (Too warm sounding, also feels like it has quite high jitter, some kind of smearing)

 

This is not about a specific piece of gear but I would love to see eye patterns or square waves on disc and USB transport reviews in the future. I know this requires a scope but it would certainly help my (our) understanding of differences.

 

Yeah I see evidence of all those memory issues taking place.
The OS continually increases the amount of ram being used, as does the cache file, and Media Center as well.
I attribute it to the OS not doing a through job at releasing the used memory that has been 'released'.
And these have been 'known' issues with windows for quite a while as well.

And I also shutdown the modem and router daily as well, but I will also try doing so during the computer and dac reset just to see if they also contribute to a SQ change.

Still the question remains how does resetting the allocated memory change the SQ.
As the objectivists are wont to say, bits are bits and these bits are at the very front end of their journey to the dac and will be re-clocked multiple times after leaving the computer and even getting to the dac where they will be re-clocked yet again.

A puzzle, at least in my mind.

JJ

 

For most users the OS not immediately purging used memory is a good thing, on the theory that it doesn't do you any good if it's not being used. All that used but not active memory is seen as a cache by your system.

The act of freeing memory is fast but not free so if the system has very little actual free memory and running a timing critical task that requires a lot of memory then I could see it being a problem, especially if you're also upsampling or converting pcm to dsd. You might want to see if your playback software has an option to load the track to memory completely before starting playback.

 

Media Center is set to play from memory and there is a fair amount of data manipulation (sample rate conversion and DSP functions) going on in real time.
And the latest iteration of Media Center loads multiple albums into memory in advance, so standard HD access isn't an issue.

As for resource limitations like insufficient ram etc., that really isn't an issue since I'm running with 16GB of ram (I never see even half of that being used) and use SSD's on the PCIe bus for the boot volume.

So there must be something about the renderer (sox) inside Media Center and the OS with its memory management that somehow affects the bit stream or sumpt'n along those lines…

At this point whenever the SQ isn't 'up to snuff' I'll reset everything and it usually helps out, and sometimes by quite a bit.

JJ

 

From USB nervosa to computer nervosa.

But I confess: I did once have a situation where I too would say that, with ram full in use (Firefox mostly) music sufferred. And that under Linux too, not even Windows. The *nix approach to only taking memory back when it needs it is long-established, way back to grand-daddy Unix --- but I sometimes wonder if Linux implements memory management as well as Unix did. But hey, I never used Unix for playing music!

Any comments from our engineers?