A novel way to massively improve the SQ of computer audio streaming

[elan120]

14 hours ago, seeteeyou said:

 

We still dunno how ClocksEye^2 would pan out yet, maybe it doesn't hurt to ask you something like this.

 

Since all 4 clock outputs of sCLK-EX don't seem to have much space around them, do you think it's worth the risks to change those U.FL connectors to something else?

 

For instance, we could add one of those Micro BNC to BNC adapters as follows and we might be able figure a way to use something that's as good as Habst afterwards

 

http://www.coax-connectors.com/products/connectors/bnc-micro-(hd-bnc)75-ohm.aspx

 

Differences as shown below

 

https://www.canford.co.uk/Products/48-5022_COAX-CONNS-67-005-B66-FA-MICRO-BNC-12G-UHD-Male-cable-crimp-group-X

From the size standpoint, it is going to be challenging.  U.FL max width is 3mm, where Micro BNC is twice as much at 6mm, it could be done, but will be a very tight fit.  Other than the size, it is not clear if there is impedance matching issue involved.

[elan120]

Big thanks to both @austinpop and @lmitche for this Gotham JSSG360 cable idea!

I am slow in adapting this idea, but finally made the same cable, and think this is a great upgrade.  Now I need to find time making a 24pin ATX cable as well.

 

Here are few detail pictures I took during cable making:

 

Pins crimped:

 

Pins inserted to connector prior to heat-shrink cable shields:

 

Both cables completed:

 

8 Pin cable installed:

 

4 Pin cable installed:

 

 

[elan120]

Good tip Kyrill.  Any recommendations regarding the metalized fabric?  Perhaps I will try some Mu-Metal tapes or EMI shielding tapes as an experiment.

[elan120]

Just had to share another Gotham JSSG360 cable experience.  After the success from replacing my music PC CPU 12V power cable to Gotham JSSG360, I got very motivated to continue this path, and last night, I replaced SATA power cable to Gotham JSSG360 as well, after letting it burn-in overnight, I was very pleasantly surprised today how much SQ improvement this cable yielded.  Since these are not difficult cables to make, if you have not try them, I would highly recommend having them replaced, you will likely be as pleasantly surprised as I am.

 

Below are few pictures of this cable:

 

Pins crimped.  Note, my SSD only require 12V input, so the 4 available wires only 5V and 12V pins were used. 

 

SATA connector with pins inserted:

 

Completed cable:

 

Cable installed:

 

 

 

[elan120]

14 minutes ago, RickyV said:

 

Is there a website? @lmitche

 

https://nuckleheadaudio.com/

 

[elan120]

1 hour ago, mourip said:

 

Thanks so much for this series. It is very helpful. I downloaded the image last night and am embarking down this new trail.

 

As an aside. RUFUS is now on version 3.3.1400 and the screen choices are somewhat different. I did not see a selection for "DD Image".

 

Best,

 

Paul

For "DD" option, you will need to download 2.xx version instead, and the latest version being 2.18.

 

When you scroll down to "DOWNLOAD" section of front page, click on "Other versions", it will take to the page where you can download version 2.18.

 

[elan120]

13 minutes ago, Johnseye said:

 

The Y cable would be parallel right? 

 

Is "serial Y cable" a good name?

 

[elan120]

3 minutes ago, Superdad said:

 

Both @lmitche and I are happily powering his AL-NUCs with a single UltraCap LPS-1.2 set to 12V. 

Nothing in these NUCs actually runs from even 12V—it all gets stepped down, by multiple DC-DC switching regulators, to 3.3V, 1.8V, 1.1V, etc.  No advantage to running from 19V.

 

[Hope this time I have phrased these facts in such a way that avoids @seeteeyou from shouting “shameless promotion.”  Larry made a nice report in this very thread a couple of weeks ago, but it quickly got buried by one of Roy’s long posts and responses to same. Clearly it got overlooked.]

I plan on trying  both LPS-1.2 and JS-2 set at 12VDC and see if there are differences...hopefully next week.

[elan120]

Another BIG Gotham JSSG360 cable.

 

I finally completed my BIG Gotham JSSG360 cable for my PC over the past weekend, and this time it is the 24pin ATX cable for motherboard.  This cable is more challenging to make, mainly due to much more cables to sort out, but aging, very rewarding in return.  After having almost all of the cables replaced in my system, I am still surprised this one cable (or a bunch) can help lift the sound quality up this much.

 

Below is a picture of the cable connected to the motherboard ATX24pin connector, and I would recommend this to anyone without hesitation.

 

[elan120]

Bridged Ethernet Ports

 

While trying different NUC setups, I decided to give the following a test drive, and the result is the best SQ so far in my system.

 

I am in early stage of incorporating NUC to my system, and still have more configurations to try, but since this setup is quite simple, you guys might want to give it a try.

 

[elan120]

NUC NUC here I come...my early stage of NUC impression.

 

After seeing many successful NUC setup in the past weeks, I decided to give NUC a try also, and with the help from Larry, I have mine up and running few days ago, and can say my WOW moment came soon after setup was completed and started listening few of my favorite tracks.

 

My system prior to setting up NUC as an endpoint consist of a recent build PC with AMD Threadripper 1950X to run HQPlayer, upsample all music files to DSD512 with non-2s filters.  This system is clock centric, where a sCLK-OCX10 connected sCLK-EX clock board inside the PC provide clocking signals to MOBO, ethernet ports, JCAT Net Card, and tX-USBexp.  Signal output from tX-USBexp is connected to txUSBUltra directly that is sCLK-OCX10 connected as well, and the output from txUSBUltra goes into Singxer SU-1 that is also clocked by sCLK-EX inside of txUSBUltra, finally, SU-1 is connected to Holo Spring L3 DAC.  I have spent considerable amount of efforts tweaking this system such as different power supplies, cables, isolation...etc., and was very pleased with the result to the point that I feel the end of tweaking is near.  However, after having many posts from well respected CAers about the NUC setup, I contacted Larry to help improve my learning curve since I am quite green in Linux.  It didn't take Larry long to get the NUC going in my system, and compare to my prior setup, I can say comfortably that this IS easily the best and most realistic sound I have extracted from my system to date, and this is still early in testing different setup configurations, which I can only imaging what it will be like after more tweaks are done.

 

As many have said it before, try it!  This is quite a breakthrough discovery that based on the relatively low entry cost, it should be quite an easy decision to make.  I am very glad I did, and I will be busy over the year end holidays enjoy my system.

[elan120]

8 minutes ago, jaspal kallar said:

 

Thanks for this, most interesting.

 

What's NUC model?

 

Where is your music stored eg NAS or internal drive?

 

Clocking centric aside for a rather simplistic view, your chain prior to the NUC was:

 

PC (AMD Threadripper 1950X to run HQPlayer, non-2s, to DSD512,tX-USBexp) >  txUSBUltra > Singxer SU-1 > Holo Spring L3 DAC

 

How does the chain look now with the NUC as the endpoint?  I presume PC is the server?

 

I have Intel J5005 (Pentium), which is in NUC7PJYH

 

All my music files are stored in NAS that is being extracted via either wifi or bridged configuration.

 

The new chain currently is similar to the chain prior other than moving the USB cable from tXUSBexp to NUC, and still using PC as server.

[elan120]

10 minutes ago, jean-michel6 said:

Thank you for this interesting report. 

If i understand well you went from a very optimized single pc set up to a dual pc set up the endpoint becoming the nuc , correct ?

If my understanding is right you did benefit from 2 improvement 

- going to dual pc set  up which is always a significant boost in sq 

- using an NUC under al as endpoint . 

Did you still use the txUSB ultra or tx usb ultra or did you went from nuc usb directly to singxer su-1 ?

 

I am already in dual pc set up with two supermicro x10sba board . I intend soon to test also the impact of running a NUC under al as an endpoint since there is so many good feedback.  

 

Which NUC did you try the dual core or the four core as @lmitchesays  i think that the four core is better with more body especially in the mid I think. 

I have done dual PC setup (sort of) in the past with RPi3 and my other desktop PC, until this optimized PC surpassed the SQ, and now with this NUC setup, it once again, surpassed the single PC.  I still have txUSBUltra in the signal chain, but plan on compare without due to some reporting here that it became the bottleneck, but will still use it to provide clocking signals to SU-1.

 

I am using Pentium (J5005), which is a quad core CPU.  I also have heard i7 could be a bit better still, but no first hand experience though, and this could be a project I will shoot for later.

 

[elan120]

2 minutes ago, greenleo said:

Very interesting, the more powerful the endpoint, the better the SQ, contradicting to what we thought previously that higher power implies higher electrical noises.

 

Maybe it's worthwhile if CAers try adding RAM to the same NUC, check the SQ and report.  SQ wise it should not be better, may be equal or worse than the original.  If so, the theory of the electrical noise still works.  Also, the theory of the quicker CPU implies shorter latency still works.  It would be interesting to find the sweet point as well.

That would be an interesting experiment to try, and I will give it a go hopefully sometime next week.  In the meantime, I also plan on trying out RAMs with different CL to see if that will offer different results.

[elan120]

A good switch between mentioned in the past post seems to be quite important to the NUC setup.  I had some open moment earlier, decided to do a quick comparison between WiFi, bridged, and bridged with my OCXO clocked switch, and like having OCXO clocked switch in the loop much better.  It adds quite a bit of dynamic and additional details, making the already crazy good NUC setup another clear notch up in SQ.  Now I am wondering what will a good switch like sNH-10G or upcoming EtherRegen be able to further contribute.

[elan120]

1 hour ago, austinpop said:

• PSUs for the NUC:
  • I ran the following: sPS-500, LPS-1.2, SR-4, and finally SR-7 DRXL.
  • The sPS-500 quickly was superseded by the LPS-1.2, which sounds wonderful. On careful listen, the SR-4 is subtly better. This is actually closer than when I ran these PSUs on my tX-U. With the NUC, the SR-4 provides an inky blackness and tames a bit of the harshness I alluded to earlier. The LPS-1.2 is more expansive, but just a smidge more fatiguing. Again - please remember, these are comparative statements only. The LPS-1.2 is by no means fatiguing.
  • Then came the SR-7 DRXL. Man - I am astounded every time I hear this PSU! How is it possible to improve on such excellence as the LPS-1.2 and SR-4?! But it is! With the SR-7, I finally understood what Roy has been hearing. There is clear daylight now between the NUC+SR-7 and my SE. The SR-7 adds even more dynamics, but along with it a refinement and ease, that is hard to describe, but easy to hear.

Great review as always, thank you for sharing.

 

Are all the PSU tests using the same voltage output at 12V?  Wonder if higher voltage output will perform better in your setup.

 

The quick test I ran between 12V and 19V show 19V is a tad better.

[elan120]

35 minutes ago, austinpop said:

 

HyperX Kingston Technology Impact 8GB Kit (2 x 4GB) 2400MHz DDR4 CL14 SODIMM Laptop Memory HX424S14IBK2/8

Could this RAM be a concern?  I was all set to purchase this exact RAMs also for their lower CL rating until I check Intel website (see below).  Although, not exactly the same model (2X16 instead of 2X4), but the incompatible warning isn't comforting.

 

Link:  http://compatibleproducts.intel.com/ProductDetails/ExportPeripheralInfo?moduleName=Intel® NUC&productType=Kits&productName=Intel® NUC Kit NUC7i7DNKE

 

[elan120]

44 minutes ago, sig8 said:

With NUC7i7DNBE being unavailable, what is our next equivalent choice?

Try NUC7i7DNHE, as this has the same board.

[elan120]

1 hour ago, elan120 said:

Try NUC7i7DNHE, as this has the same board.

Thanks for the added information.  You inspired me to check into the difference as well and have found the following:

 

LinK:  https://www.intel.com/content/dam/www/public/us/en/documents/product-briefs/nuc-kit-nuc7i7dnke-nuc7i7dnhe-board-nuc7i7dnbe-brief.pdf

 

[elan120]

43 minutes ago, kyrill said:

" But Jussi's image wins on performance for me - when playing new tracks there are pops just at the beginning of tracks with Audiolinux (and nearly every Linux distro I've had running NAA before). "bold added

 

This rather alarming for me. Do more audiolinux users have these pops or clicks with teh beginning of a new track?

No pops here.

[elan120]

13 minutes ago, Dutch said:

 

I do believe he mentioned he gets those when running HQplayers NAA deamon so this does not need to apply to Roon, LMS etc.

I am not sure what the difference is, but when I tried either AL with HQP NAA selection or HQP NAA image, both running with ramroot setup, no pops was observed.

[elan120]

18 minutes ago, Em2016 said:

 

Assuming everything stays in DSD, it's likely the quieter couple of pops between tracks has always been there with Audiolinux but maybe you never noticed it... as mentioned a few times, it's nowhere near as loud as the classic PCM to DSD transition pops. Easier for me to hear with headphones but they are there with speakers too with Audiolinux.

 

When you do your testing you may find these quieter pops with Audiolinux (and everything else) disappear with Jussi's bootable image - that's what I found.

 

I'm also at DSD512 for everything - it's possible it's less of an issue at lower sample rates with Audiolinux (and others) but I haven't bothered testing because with Jussi's bootable image it's 100% pop free at DSD512 (for me).

 

 

This is interesting, as I have not experienced any pops with either NAA images, and this is running DSD512 (non-2S), dead quiet between tracks with headphones.  Wonder what the difference is.

[elan120]

2 minutes ago, Em2016 said:

 

When you say either NAA images, do you mean Jussi's bootable?  And also with Audiolinux?

 

Yes, this is correct.  I ran Jussi's ramroot image first, was happy with the result, and changed to AL HQP NAA later.  Either case, no pops in my setup upsample all music files to DSD512 with non-2s filters with headphones.  The only pops I got is switching between PCM and DSD, but this is not the image issue anyway. 

[elan120]

1 minute ago, Em2016 said:

 

Noted and thanks. Do you hear any SQ difference between Jussi's ramroot image and AL ramroot HQP NAA?

 

 

I didn't detect any when I first ran them about couple weeks ago, but since I have few setup changes after switched to AL to improve the playback, I will take some time to test the two again to make sure.

[elan120]

20 minutes ago, ted_b said:

As soon as I get my new home's audio room built I plan on having Larry build me a NUC for HQPlayer NAA (I already have a powerful upsampling server, currently running Win10 but likely running whatever you guys tell me, aka AL ramboot; Note: I ran HQPlayer embedded and started that thread in my past life, but then went back to single pc Windows for whatever reason).  But, I'd like to ask those of you (johnseye, elan120, Em2016, etc) who are instead running Jussi's own bootable Linux ramroot NAA image (and comparing it favorably) two things:

1) would my endpoint NUC decision be any different if I focused on Jussi's bootable ramroot NAA image instead of AL (i.e smaller, larger, different RAM, etc) as my goal;

2) what are you folks running as an upsampler machine, and how connected to the ramroot-NAA? 

Thx

Ted

Ted,

 

1)  Using NUC build by Larry with standard spec'ed for Roon bridging, you can test Roon bridging and HQP NAA setup in AL image as well as Jussi's ramroot NAA image with the same hardware.

 

2)  My current updampler is a Windows PC I recently build running DSD512 non-2s filters (mostly poly-sinc-xtr), and connected to ramroot NAA first via wifi, later bridged, and now bridged with a OCXO clocked ZyXEL switch.

 

Hope this is clear enough to your 2 Q's