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

[FredM]

The ‘decent’ power supply seems to be the catch.. it seems 3 variables are discussed at the same time: 

- At what level does a standard NUC with RAM OS perform, using a standard power supply?

- what is the increase when using a  better power supply (ie SBooster, 1.2, ..)

- what is the increase when using an extreme power supply?

 

 

 

 

[lmitche]

2 hours ago, FredM said:

The ‘decent’ power supply seems to be the catch.. it seems 3 variables are discussed at the same time: 

- At what level does a standard NUC with RAM OS perform, using a standard power supply?

- what is the increase when using a  better power supply (ie SBooster, 1.2, ..)

- what is the increase when using an extreme power supply?

 

 

 

 

Fred,

 

Having spent three months working through various NUC configurations, DC power quality is one aspect, and not the most important aspect, impacting NUC SQ. Galvanic isolation, vibration control, software configuration are three additional factors that greatly impact SQ. There are more.

 

SQ is like nothing else I have heard. We are in the end game.

 

Larry

[auricgoldfinger]

18 hours ago, Dev said:

 

Cool. What's the price on TLS PSU ? Though 3-5 months is still a lot of lead time for building a psu but I agree its far better than waiting 18-24 months on PH if it sounds as good. I have been using mojo audio Illuminati psu in my (and my friends) system for sometime and have been quiet content but I do not know if the TLS or PH will outperform it. It would be interesting to know how all these PSU rank.

 

The price depends on your specifications, so you would need to contact Adrian for that information.

 

The wait time is reasonable when you consider that manufacture and delivery of a custom transformer takes about 6 weeks.  In my situation, cryogenic treatment of the transformer added another 2-3 weeks.

 

[FredM]

24 minutes ago, lmitche said:

Fred,

 

Having spent three months working through various configurations, DC power quality is one aspect, and not the most important aspect, impacting NUC SQ. Galvanic isolation, vibration control, software configuration are three additional factors that greatly impact SQ. There are more.

 

SQ is like nothing else I have heard. We are in the end game.

 

Larry

 

Hi Larry, 

Thank you, yes I understand that that there is more then a keen power setup. I should have used more words in my reaction, like: “..seem to be the catch in the confusion”. I was trying to say that it could appear that a normal NUC with RAM OS only would outperform ie a mR or SE. Therefore I wanted to break this down into a couple of power setup variable to create a common reference, regarding the power setup.

 

I’m fully with you that RAM OS combined with a good power supply is not all it takes. The list with additional improvement area’s goes on and on. I’m certainly not an expert but from what I’ve been reading everything matters, using OCXO clocks, position/distance clocks, internal wiring, fuse, etc..(Which all goes above my hat)

 

To create some kind of common idea/reference, it would be nice to get a rough impression on how a NUC with only a RAM OS (ie loaded from an external USB stick) and a good power supply (ie 1.2, SBooster,.) would perform in comparison to a known other product. I’m presume these two improvements are doable for most ‘novice’ users. Tweaking ie the bios setting will fit nicely in the ‘expert’ department.. ?

 

 

[89reksal]

1 hour ago, FredM said:

To create some kind of common idea/reference, it would be nice to get a rough impression on how a NUC with only a RAM OS (ie loaded from an external USB stick) and a good power supply (ie 1.2, SBooster,.) would perform in comparison to a known other product. I’m presume these two improvements are doable for most ‘novice’ users. Tweaking ie the bios setting will fit nicely in the ‘expert’ department.. ?

 This would be of great interest to myself and I'm sure others. I've been trying to get as much information as I can on the SOtM sms-200 products over the last few weeks and am very close (or at least I thought I was) to making a final decision between the sms-200 Neo and the sms-200 ultra Neo.

In so many threads there's been talk of more diy type solutions, such as NUCs, possibly outperforming the sotm products and others. Just reading that alone has made me give serious consideration to the sms-200 over the ultra, since this area seems to be changing so fast and the sotm products, other than the Neo upgrades are already one or two years old which is an eternity in modern digital. At least if something new comes along next week, I will have saved some money by choosing the basic sms-200 Neo. That logic may be flawed but I'm pretty confused by what path to take right now.

It would be great to get a general idea of what would be needed to diy a NUC system that would be able to compete with sotm and other comparable products. I have zero experience with Linux but am willing to try and learn what I can and from reading the AudioLinux site (this one?  http://www.tophifi.it/  )

it looks like support is available, even up to doing a remote login if needed.

Thanks to all the early adopters for so freely sharing their experiences and advice in this and other threads.

 

[TheAttorney]

On 10/24/2018 at 9:42 PM, romaz said:

.... As I shifted my lone DR rail to my NUC, I was forced to use an SR (single regulated) rail from my SR7 on my tX-USBultra and I was quite surprised/disappointed to find that my tX-USBultra was now resulting in a clear step backward with respect to dynamics and tonal richness and smoothness.  It's not like I was powering the tX-USBultra with a cheap switching PSU but that's how the DR rail made the SR rail sound like and this highlights the importance of PSUs. 

 

It is both a blessing and a curse that us audiophiles can be perfectly happy with a great product (e.g. the SR SR7), then hear something better (DR SR7) and suddenly the original isn't good enough anymore. As you once said, we can't un-hear what we've just heard, even though the original doesn't suddenly stop being a great product that may beat all other rivals.

 

Over a period, I've being trying a few external tweaks to my SR SR7 (don't get excited, nothing wild or innovative here). When I get some more time, I'll report my impressions on the SR7 thread, but for now, what struck me was that your comment "a clear step backward with respect to dynamics and tonal richness and smoothness" is just what I heard in each case when I removed two of those tweaks - to the point I didn't want to listen to my system untl they were reinstated.

 

The differences were probably objectively tiny, and probably nothing as big as your DR change, but it's a curse for me not to be able to un-hear these things. It also begs the question: Would various improvements to the SR7 bring the SR version closer to the DR version? Or would both versions benefit equally? On the basis that the less "bad stuff" there is flying around inside a power supply, the less there is for the 2nd regulator to improve upon.

[seeteeyou]

http://www.tophifi.it

Quote

 

Audiolinux PinkFaun version 1.0.1 (not for sale, available with PinkFaun Streamers)
Audiolinux ram version 0.7 (not for sale, in testing with some commercial streamers)

 

 

I guess it's a good time to pop the question now, how many of us would be interested in a crowd funded motherboard project that's led by Adrian?

 

Chassis - CHECK

 

OCXO - CHECK

 

LPS - CHECK

 

OS - CHECK

 

Of course he might be too busy since he's got his own business already. Or maybe he'll consider funding everything on his own if that particular minimum were still valid at the moment?

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=153&tab=comments#comment-724987

On 10/2/2017 at 8:09 PM, romaz said:

Finding a suitable motherboard was extremely difficult and it became s a matter of finding the motherboard with the least compromises. DFI was willing to design a motherboard for me to my specifications, however, it would have cost $1,500 minimum to get the project off the ground.

 

BTW, finding an OCXO for USB 3.0 / Ethernet should be fairly easy since they're sharing the same 25MHz frequency under most circumstances. However, the system clock itself could be a different story (e.g. 14.318MHz) and therefore we'll have to go for sCLK-EX + external 10MHz reference clock with degradation to boot (plus the importance of acquiring an excellent clock cable like Habst etc.)

 

https://www.head-fi.org/threads/review-comparison-of-5-high-end-digital-music-servers-aurender-n10-cad-cat-server-totaldac-d1-server-auralic-aries-audiophile-vortex-box.787020/page-79#post-14120317

 

In that case, Colpitts oscillators (like Neutron Star 2) might provide us the flexibility that we're looking for since they do offer a rather oddball frequency 19.2MHz for the system clock of Raspberry Pi

 

https://www.newclassd.com/index.php?page=200

 

A separate PSU for its recommended voltage @ 15V might be somewhat tricky, though. We can't do double-regulated LT3045 since its maximum output is also 15V. That means we'll have to do something like one DR LT3045 with 9V plus another DR LT3045 with 6V.

 

The maximum output of LT3094 is -20V but unfortunately they're asking for +15V

 

 

Now I'm wondering where did Paul Pang source his OCXO for modifying motherboards.

[str-1]

10 hours ago, romaz said:

For those who are in the SR7 queue and are willing to spend the money for the very best, based on demand, Paul Hynes is now offering the option of a Teflon circuit board to reduce dielectric absorption in the regulator circuitry and Vishay Z Foil ultra precision bulk foil resistors which reduce resistor noise and thermal drift to very low levels.  Paul has not yet told me what this upgrade will cost but I suspect it won't come cheap although based on how he described their impact, it will be worth every penny.

Thanks for this update, Roy, and great to see you posting again.  I am in the queue and I think I recall Paul mentioning these upgrades when I last spoke with him a couple of months ago, although at that time they were just possibilites he was thinking about.  I’ve still to finalise my order so I will seek his advice on their suitability for my needs.

 

Can I also take this opportunity to ask what spec your tX-USBultra is.  I know from past posts that it will have a clock input for use with the Mutec REF 10 but does it also have upgraded internal dc wiring (copper or silver) and EVOX caps?  I will soon be upgrading my tX-U for clock input (I think I’ll go for 75 ohms) and probably also the EVOX caps but I am unsure whether to upgrade internal wiring to the higher purity copper or to silver.

[FredM]

55 minutes ago, seeteeyou said:

how many of us would be interested in a crowd funded motherboard project that's led by Adrian?

 

There are already 2 keen streaming projects on the shelve: researched, developed, tuned, build and available for individuals or groups to support / buy ?

 

As an alternative -if you’re keen to add a personal touch/experiment- I guess one can always ask if specific adjustments are possible. 

 

[tims]

5 hours ago, TheAttorney said:

 

It is both a blessing and a curse that us audiophiles can be perfectly happy with a great product (e.g. the SR SR7), then hear something better (DR SR7) and suddenly the original isn't good enough anymore. As you once said, we can't un-hear what we've just heard, even though the original doesn't suddenly stop being a great product that may beat all other rivals.

 

Over a period, I've being trying a few external tweaks to my SR SR7 (don't get excited, nothing wild or innovative here). When I get some more time, I'll report my impressions on the SR7 thread, but for now, what struck me was that your comment "a clear step backward with respect to dynamics and tonal richness and smoothness" is just what I heard in each case when I removed two of those tweaks - to the point I didn't want to listen to my system untl they were reinstated.

 

The differences were probably objectively tiny, and probably nothing as big as your DR change, but it's a curse for me not to be able to un-hear these things. It also begs the question: Would various improvements to the SR7 bring the SR version closer to the DR version? Or would both versions benefit equally? On the basis that the less "bad stuff" there is flying around inside a power supply, the less there is for the 2nd regulator to improve upon.

Looking forward to your report.  I changed the power cord to my SR7 to a diy version of VH audio's flavor 4 power cord and the results was positive, so I'm interested in what you have done to yours to improve SQ.

https://www.venhaus1.com/diymains.html  (flavor 4 not described but similar to flavor 1 but using starquad cable)

 

[the_doc735]

12 hours ago, Dev said:

 

May not still beat the NUC + Audiolinux combo. The difference in processor architecture, their cache systems, pipelines, noise profiles can play a significant role in the sound quality. Too many variables at play and we can only be certain from subjective listening.

I don't think subjective listening alone could settle this debate.

[Superdad]

7 minutes ago, the_doc735 said:

I don't think subjective listening alone could settle this debate.

 

Then you probably will want to skip this forum thread. 

[Bricki]

13 hours ago, austinpop said:

 

Hi Roy,

 

This seems to swing the pendulum back to the best approach being having the NUC modified by SOtM with an sCLK-EX board in the chassis, that can reclock all 3 - the Ethernet, USB, and the system clock.

 

For the record - Roy know this well - I did try an early NUC with Adrian's Dream OS. The difference was that I didn't have access to an SR-7, and the NUC was provisioned with an SSD to load the OS. So, while not quite the optimal config that Larry and Roy have now found, this should have still displayed the benefits of the ramboot OS that Roy is talking about. Powered by an SR-4 set at 12V, this NUC (running Roon Bridge, with SE as Roon Server) did not outperform my SE direct.

 

Since I trust Roy's ears, I've got to believe that it's the PSU that makes the difference - along with using the USB stick to boot. I am waiting to confirm this with a DS-1 when Adrian sends me one.

The other difference I can see between the ds1 and the nuc that Larry and Roy have been using is the latency time that Larry measured was 4-5 us and the ds1 with dream os has on the tls website processor latency as 12 us and os latency at 10 us.... I think Roy's original findings were with an early version of the ds1 with dream os?? So it would be interesting to know if Roy was able to compare Larry's celeron end point nuc (with his trimmed audiolinix) with the ds1 with dream os.... 

 

Or perhaps there are more differences in in the processor architecture like the smart cache and each core being able to use all the level 1 cache (if 1 core active) like what Roy was talking about earlier. 

[elan120]

txUSBUltra with Gotham JSSG360 DC cable mod:

 

After the success from replacing stock DC cable for sCLK-EX clock inside my PC with Gotham JSSG360 cable, my motivation for trying the same in txUSBUltra got extremely high, my curiosity is now well satisfied after some time spent this afternoon.  This cable once again has elevated my system with improved SQ and I would highly recommend to give it a try.

 

Below are few pictures I took while doing the cable replacement.  Hopefully they will be useful if you plan on doing this mod as well.

 

Stock cable and tx-USBhubIn removed:

 

 JSSG360 from sCLK-EX to txUSBhub installed:

 

txUSBhub reinstalled:

 

JSSG360 from external DC input to sCLK-EX installed, ready to reinstall back to original txUSBUltra enclosure:

 

 

[Bricki]

12 hours ago, lmitche said:

Fred,

 

Having spent three months working through various NUC configurations, DC power quality is one aspect, and not the most important aspect, impacting NUC SQ. Galvanic isolation, vibration control, software configuration are three additional factors that greatly impact SQ. There are more.

 

SQ is like nothing else I have heard. We are in the end game.

 

Larry

Hi Larry 

 

For the celeron nuc that you configured for Roy, were there any other changes made besides loading audiolinux off a usb stick in headless ramboot mode and configured as a roon end point? 

 

And if connected to a dac that doesn't require usb power (or Iso regen) and no other usb peripherals what is the minimum psu requirements? I assume it's going to require more than the 12v 1.1a of the lps1.2? Have you tried to boot it off the lps1.2? Are you still powering yours with your sigma 11?

 

Also, what vibration control are you using? 

 

Thanks ?

Brendan 

[Superdad]

2 hours ago, Bricki said:

And if connected to a dac that doesn't require usb power (or Iso regen) and no other usb peripherals what is the minimum psu requirements? I assume it's going to require more than the 12v 1.1a of the lps1.2? Have you tried to boot it off the lps1.2? Are you still powering yours with your sigma 11?

 

I don't want to steal Larry's thunder, but he e-mailed me a couple of hours ago that he is playing his NUC overnight on an UltraCap LPS-1.2 to see how it holds up. 

As for the ISO REGEN, it draws just 40mA (that's 0.04A) from of 5VBUS current from whatever USB line is connected to its input.  Hardly a budget buster in possible LPS-1.2>NUC scenario...

[Dev]

21 hours ago, romaz said:

 

This would be an interesting comparison to make and I think we would all be interested in such a shootout.  With regards to a NUC, aside from power supplies from Paul Hynes, TLS and Mojo Audio, you can also get a customized PSU from Sean Jacobs, Kenneth Lau, and Paul Pang.  While not customizable,  there is the excellent JS-2, the latest HDPlex with LT30451 regs, and well regarded offerings from Teradak and SBooster.

 

 

Thanks! I would have gotten a SR7 by now if not for the wait time. I am not that patient enough. No disrespect to other PSU manufacturers but we probably need someone to build PSU at the PH SR7 level and without the exorbitant wait time as it seems to have set the gold standard in most of the reviews from its owner I have read. Another member confirmed TLS to be at par. Kenneth Lau is very popular with Lumin users but I haven't seen it getting compared to PH. I would like to pull the trigger for my next PSU upgrade but figuring out which one is becoming a bit tricky ?

[kelvinwsy]

On 10/26/2018 at 9:34 AM, kelvinwsy said:

oh i did upgrade the two Sata data cables to a high gauge Japanese type.. This help in the SQ Jump I guess

Completed the JSSG for the 24 pin ATX and did the 2x4pin CPU PCie as well as the 2x6/8pin GPU PCie .. Fitted Ferrite chokes on every wire and must be under the JSSG LAYERS !

SQ goes up.. evident in Live albums where the hall ambience noises are clearer, more definite stereo positioning width, depth and height! 

Center vocals and sounds of accompanying instruments on left and right more precise in image lucidity and solidity

Well recommended tweak

[the_doc735]

what about this as a novel way of improving sound quality?

 

(nas>router>FMC>)>microrendu 1.4>ISO Regen>Ultradigital?

(in my case, because my DAC USB input isn't as good as the i2s input)...

...downside - requires a lot of PSU's!

[vortecjr]

Playing from RAM:) You should know that Linux is already doing this by default. It uses RAM to speed disk operation for buffers and cache. 

[LTG2010]

11 hours ago, Bricki said:

I assume it's going to require more than the 12v 1.1a of the lps1.2? Have you tried to boot it off the lps1.2?

If you limit the RAM to say 4GB then the LPS 1.2 might work well.

[seeteeyou]

33 minutes ago, LTG2010 said:

If you limit the RAM to say 4GB then the LPS 1.2 might work well.

 

FYI - there's a "trick" to find the right 4GB DDR4 sticks with VERY low current draw that's quite a bit below 1A

 

4GB, 512Mx 64 SO-DIMM: HMA851S6AFR6N
https://www.skhynix.com/product/filedata/fileDownload.do?seq=7951

Quote

IDD4RA = 816 mA
IDD4RB = 816 mA
IDD5B = 784 mA

 

Hynix 4GB PC4-19200 DDR4-2400MHz non-ECC Unbuffered CL17 260-Pin SODIMM Memory Module Mfr P/N HMA851S6AFR6N-UH

https://www.newegg.com/Product/Product.aspx?Item=9SIADRH7AA1567

 

HYNIX HMA851S6AFR6N-UH 4GB 1Rx16 DDR4 19200 PC4-2400 NONECC LAPTOP MEMORY RAM

https://www.ebay.com/itm/HYNIX-HMA851S6AFR6N-UH-4GB-1Rx16-DDR4-19200-PC4-2400-NONECC-LAPTOP-MEMORY-RAM-/323481179145

 

Hynix 4GB PC4-19200 DDR4-2400MHz non-ECC Unbuffered CL17 260-Pin SoDimm Memory Module Mfr P/N HMA851S6AFR6N-UH
https://www.amazon.com/Hynix-PC4-19200-DDR4-2400MHz-Unbuffered-HMA851S6AFR6N-UH/dp/B071DVVJNL

 

4GB, 512Mx 64 SO-DIMM: HMA851S6CJR6N
https://www.skhynix.com/product/filedata/fileDownload.do?seq=8990

Quote

IDD4RA = 756 mA
IDD4RB = 756 mA
IDD5B = 784 mA

 

Hynix HMA851S6CJR6N-UH 4GB SODIMM DRR4 2400MHZ 1RX16 4CHIPS

https://www.neweggbusiness.com/Product/Product.aspx?Item=9SIV1PV77V2081

 

SK Hynix 4GB DDR4 Laptop RAM PC4-2400T 2400MHz 260pin HMA851S6CJR6N-UH

https://www.ebay.com/itm/SK-Hynix-4GB-DDR4-Laptop-RAM-PC4-2400T-2400MHz-260pin-HMA851S6CJR6N-UH-/123416740843

 

FMS Compatible with HMA851S6CJR6N-UHN0-AC Replacement for Hynix 4 Gb Pc4-19200 2400 Mhz So-Dimm Memory

https://www.amazon.com/FMS-Compatible-HMA851S6CJR6N-UHN0-AC-Replacement-Pc4-19200/dp/B07JH7X3LM

 

SK hynix HMA851S6CJR6N - VK Non ECC PC4-2666V 4GB DDR4 at 2666MHz 260pin SDRAM SODIMM Single Kit Laptop Memory - OEM
https://www.amazon.com/SK-hynix-HMA851S6CJR6N-PC4-2666V-2666MHz/dp/B07GH7RRTM

[rafa]

23 hours ago, seeteeyou said:

Now I'm wondering where did Paul Pang source his OCXO for modifying motherboards.

 5V from molex plug.

[seeteeyou]

23 minutes ago, rafa said:

 5V from molex plug.

 

I just checked his page, then I realized that's only listing 24MHz for USB 2.0 and 25MHz for Ethernet respectively

 

http://ppaproduct.blogspot.com/2014/02/clock-upgrade-module-for-main-board.html

 

No love for 14.318MHz then, maybe Neutron Star 2 could support that but not sure.

[seeteeyou]

Gosh, Cybershaft OP20 ($5,000) turned out to be a quantum leap when compared to OP17

 

http://www.hiendy.com/hififorum/forum.php?mod=viewthread&tid=122411

 

They're only talking about the bare minimum when OP20 was tested, stock clock cable plus an ordinary Acoustic Revive Power Sensual 18000 that's nothing to write home about.