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

[Bruce Orr]

1 hour ago, austinpop said:

Merry Christmas and Season's greetings to all!

 

Thanks to all of you who contribute you knowledge and experience on this thread. I have learned so much from here in the last 2 years, and I can honestly say it has transformed the sound quality of my system.

 

Thanks and Happy Holidays to Rajiv, Eric, Romaz and all the other CAs for sharing their time, knowledge and experiences. My SQ compared to original has improved more than I could imagine over the last year. 

 

While more a beneficiary than a contributing experimenter, I hope to soon share some thoughts on SQ impact when I:

- replace an SOtM modded Netgear108 with the new SOtM switch (with upgraded capacitors + silver) 

- replace the pigtail from the tx-USB ultra’s clock and REF10 with rhe REF10/Habst directly cabled to the SOtM switch.

 

In the new configuration , the SOtM tx-USB Ultra and new switch will both have the CLk and a direct REF10/Habst Input. 

 

I’ll try to share what my ears hear as best I can!

 

Bruce

[afrancois]

5 hours ago, rickca said:

@afrancois I've updated my post with documented SOtM changes re Roon in 4.4 firmware.

 

I'm confused about your configuration.  Are you running a NUC/AL server and an sMS-200Ultra endpoint?  Have you tried a NUC/AL endpoint for comparison with the sMS-200Ultra?

Thx, for the info about the 0.4.4 version. I sure hope that the difference I'm hearing is not due to a simple change of DC cable on my server. But then again, who knows!

 

I just listened again and my system has some sort of calm and neutrality that's almost uncanny. 

Version 0.4.4 can't be downloaded from the website I see. I did the upgrade from within Eunhasu.

 

When I upgrade to 0.4.5 I will put 0.4.4 aside (so I will be using a new sd card)

[afrancois]

2 hours ago, Advieira said:

@rickca @afrancois When Im trying to burn the new image on Sotm Micro USB Card, in OSX even in Windows, dont recognize and ask for format. Do I need format to burn the upated image?

I assume you are following the guideline from SOtM. http://docs.sotm-audio.com/doku.php?id=en:eunhasu:burn_sdcard_image

 

Just ignore the message to format upon insertion of the SD card.

 

Burning the image will do the necessary formatting and partitioning for you.

 

It's possible that I have used Rufus instead of imageUSB. Just be careful not to erase any data on your PC's HD's

[jean-michel6]

Merry Christmas and Happy holiday season .

 

Thank's to all to make this thread a really fun place to learn and experiment on computer audio !

 

A special thank to @austinpopfor his great contributions and talented moderation of this thread 

[seeteeyou]

On 11/29/2018 at 10:06 AM, austinpop said:

However, interestingly, music on NVMe Optane SSD was almost as good as in RAM.

 

https://www.provantage.com/intel-ssdpel1d380gax1~7ITEE0NW.htm

https://ark.intel.com/products/148607/Intel-Optane-SSD-905P-Series-380GB-M-2-110mm-PCIe-x4-20nm-3D-XPoint-

 

Intel Optane 905P 380GB M.2 NVMe SSD Review The Best

https://www.servethehome.com/intel-optane-905p-380gb-m-2-nvme-ssd-review-the-best/3/

Quote

Intel Optane 905P Sub 1ms Latency

 

ASRock's Optane 905P M.2 RAID Defies Limits

https://www.tomshardware.com/news/asrock-x299-vroc-optane-905p,37225.html

 

At 110 mm it's too long for NUC7i7DNBE internally unless an extension of some sort is added, though SQ might be affected

 

 

OTOH, Baby Canyon (only i5 and i7 models) and Bean Canyon do have Thunderbolt 3 ports. Therefore we could get an enclosure like this and end up with 760GB, both drives could be powered externally so maybe they might sound even better when compared to RAM?

 

https://item.taobao.com/item.htm?id=582968066568

 

[Volfram]

On 12/20/2018 at 12:52 AM, sandyk said:

Personally, I would like to see much larger value output capacitors at the output, but there are space limitations to consider.

Reminds me of large, in size, caps used in Naim product like hicap DR.

Don't know about the output spec tho.

The downside of Naim product is recapping, meaning sending em back to the supplier for service, is preferred every 10-15 yrs.

Could it be related to the use of large caps?

Maybe thats why other comanies shy away from them?

[seeteeyou]

While I can't vouch for the trustworthiness of these sellers, we could actually pay less than $1 / GB for 960GB Optane 905P now

 

https://ark.intel.com/products/129834/Intel-Optane-SSD-905P-Series-960GB-1-2-Height-PCIe-x4-20nm-3D-XPoint-

 

6,399 RMB (about $927.5)

https://item.taobao.com/item.htm?id=576244302092

 

6,000 RMB (about $870)

https://item.taobao.com/item.htm?id=582436015241

[austinpop]

Hi All,

 

I have had Adrian's DS-1 streamer in the house for the last couple of weeks, and have just posted my impressions here: https://www.computeraudiophile.com/ca/reviews/quick-take-the-linear-solution-ds-1-network-streamer-r762/

 

 

For regulars on this thread, feel free to continue the discussion here, as I suspect the article thread will read by folks who may not have the context of the ongoing discussion.

 

But you can decide.

[sunny_time_99]

On 12/19/2018 at 9:04 PM, Dev said:

 

Below is the interesting part and key difference from LPS-1(.2). Any thoughts from the PSU experts ?

 

Why are you using the super capacitor power bank as smoothing capacitors and do you not switch between one bank while loading the other?

There are several reasons why we do not do this. First it means extra switching electronics in the supply lines, second it means that you are using only half of capacity of the super caps at a time. Third and maybe most important, it means that the final regulator does not get a constant voltage at its input, meaning different voltage drop, and therefore slowly varying power dissipation and temperatures all the time. This influences the output characteristics of the regulator, and the output power quality. Because in our supplies we use a choke smoothing and pre-regulation before the supercapacitor banks, noise residue will be minimal and also the necessity for switching.

Its so funny we allready tried this in 2014 /2015 with first 3.000F boostcaps, than boostcap plus regulated psu, than with regulator behind, than switching F350 cells with a rpi doing the switching job... Than with choke smoothing and regulator... A polymer in the output behind the regulator... And and and...

more can be found here: (german language only) https://www.diy-hifi-forum.eu/forum/showthread.php?9501-Auf-der-Suche-nach-meinem-idealen-Netzwerkplayer-mein-Leidensweg/page2

That was much earlier than anyone else in the internet.

But in the end, we allways found out that there are much better ways without ultracaps. It was a nice way... But today i will not use ultracaps anymore in my system. I just found 2 pics from 2014/2015 on my mobile phone...but i can find pictures from every step. From time to time a friend uses these old ultracap psu to make sure how far we are now away from that point.

 

But i still like the way ultracaps indicate mistakes in your wireing ?

 

Regards

 

Sunny

[seeteeyou]

Quote

You think $1599 seems like a very fair price? Come now.

 

Ignore the chassis, RAM, the OS, and ongoing technical support for now. While three Connor-Winfield OH4 Series OCXO (better ones cost even more) alone might cost less than 200 bucks, we're potentially approaching the level of SQ that's comparable to what sCLK-EX + Mutec REF 10 + Habst + international shipping could offer.

 

$1599 for a Roon endpoint, maybe add another Reference 1 LPS for $1550. That's still cheaper than a brand new REF 10 while it's most likely "more endgame" than any kinda endgame that's mentioned earlier on this thread here.

 

I did try to ask Intel about the clocks of NUC7i7DNBE but apparently they're asking us to sign their CNDA first

 

https://forums.intel.com/s/question/0D50P0000490YJnSAM/frequencies-of-crystals-on-nuc7i7dnbe?language=en_US

 

There's certainly another way to spend that $1599 for even better SQ, pay someone to modify the clocks of NUC7i7DNBE by replacing them with stuff like Neutron Star 2 since we really don't have too many noteworthy choices when it comes to 25MHz. Of course the tricky part would be fitting all clocks + their own LPS inside the same chassis.

 

Maybe wait for @romaz to come back here and we'll find out how NUC + sCLK-EX + REF 10 + Habst would sound.

[rickca]

5 minutes ago, seeteeyou said:

three Connor-Winfield OH4 Series OCXO

So there are 3 separate OCXO at different frequencies in the DS-1? 

[sunny_time_99]

My opinion befor spending so much money for something new... Is allways important to find out which part is responsible for the better sq. Thats what am i doing for years now. Changing one endpoint to another will never end the game, because all retail stuff got build in other bottlenecks. There is nothing available on the market without a bottleneck. I am using 6 neutron star 2 in my setup until now and some neutrino star...The using of neutron stars is realy a big step in sq. But i am not using their power supply. The neutron star power supply sounds... Like the psu i build 3 years ago. The clock in a system should have the best psu. For that reason i build a kaskarde of my psu with 2 regulators at minimum.

 

Regards

[seeteeyou]

17 minutes ago, rickca said:

So there are 3 separate OCXO at different frequencies in the DS-1? 

 

Under most circumstances both USB 3.0 and Ethernet should be 25MHz but it better be verified by looking inside DS-1 where the clocks are located.

 

I can't tell what the 3rd one is since the system clock might not be 14.318MHz. And then it's gonna cost more to source an OCXO if the frequency weren't all that common.

[rickca]

14 minutes ago, seeteeyou said:

 

Under most circumstances both USB 3.0 and Ethernet should be 25MHz but it's better be verified by looking inside DS-1 where the clocks are located.

 

I can't tell what the 3rd one is since the system clock might not be 14.318MHz. And then it's gonna cost more to source an OCXO if the frequency weren't all that common.

Let me rephrase my question.  Are there 3 OCXO in the DS-1 or a single OCXO with derived frequencies as required by ethernet/USB/system clock?  I'm not asking about the required frequencies.

[seeteeyou]

8 minutes ago, rickca said:

Let me rephrase my question.  Are there 3 OCXO in the DS-1 or a single OCXO with different derived frequencies as required by ethernet/USB/system clock?

 

I could only trust Rajiv to know what he's talking about, he mentioned all 3 in his article so my own interpretation would be separate ones. The location of each corresponding clock ain't THAT close to each other so deriving different frequencies from the same clock should be less than ideal due to the degradation of relatively long clock cables.

 

At the beginning TLS was also running into stability issues when they could only get one OCXO working for either USB 3.0 or Ethernet but not both simultaneously. I dunno how they managed to take care of everything and therefore once again I could only refer to what Rajiv wrote as mentioned above.

[rickca]

4 minutes ago, seeteeyou said:

At the beginning TLS was also running into stability issues when they could only get one OCXO working for either USB 3.0 or Ethernet but not both simultaneously.

That is precisely what prompted my question.  And of course I trust Rajiv to get the right answer.  I'm just asking for clarification.

[austinpop]

38 minutes ago, rickca said:

Let me rephrase my question.  Are there 3 OCXO in the DS-1 or a single OCXO with derived frequencies as required by ethernet/USB/system clock?  I'm not asking about the required frequencies.

 

I think it would be best if Adrian @allaboutacoustic answered this question himself.

[Music Enthusiast]

If manufacturers started making DAC's good enough for all of these tweaks to be ineffective, would that upset you guys? 

[seeteeyou]

12 minutes ago, Music Enthusiast said:

If manufacturers started making DAC's good enough for all of these tweaks to be ineffective, would that upset you guys? 

 

That will only come in the form of a marketing ploy of some sort, they could shape our perception all they want but ultimately it's pretty much like a miracle and a half if there were such thing as complete immunity to ALL source components.

 

The "ineffective" part could be tricky to define since not all ears / gears are created equal, therefore we don't even know where to draw a line in the sand IMHO.

 

Not to mention that should be more about synergy between both upstream and downstream, I just really don't understand what the "ineffective" part would actually mean.

[Dutch]

Edit; post removed as I realized it’s off topic.

[Music Enthusiast]

7 minutes ago, seeteeyou said:

 

That will only come in the form of a marketing ploy of some sort, they could shape our perception all they want but ultimately it's pretty much like a miracle and a half if there were such thing as complete immunity to ALL source components.

 

The "ineffective" part could be tricky to define since not all ears / gears are created equal, therefore we don't even know where to draw a line in the sand IMHO.

 

Not to mention that should be more about synergy between both upstream and downstream, I just really don't understand what the "ineffective" part would actually mean.

So you're saying it's impossible to engineer a DAC to not be influenced from upstream tweaks? 

[Dutch]

On 12/13/2018 at 5:37 PM, austinpop said:

Please note that discussion needs to stay focused on direct listening experiences with audio experiments discussed here.

 

@Music Enthusiast see above quote from first post.

[sunny_time_99]

45 minutes ago, Music Enthusiast said:

So you're saying it's impossible to engineer a DAC to not be influenced from upstream tweaks? 

That is definitely true.

 

And the dac is only one part in the whole game. And my listening experience  is that it has not the biggest influence on the total sq. I focused on the dac for years and build dacs with all available dac chips in the last years but today i would only say that the dac makes 20-25% of the total sq. There are many possibilities to get a better sq without  spending money for a new dac

 

Regards

[RickyV]

5 hours ago, sunny_time_99 said:

My opinion befor spending so much money for something new... Is allways important to find out which part is responsible for the better sq. Thats what am i doing for years now. Changing one endpoint to another will never end the game, because all retail stuff got build in other bottlenecks. There is nothing available on the market without a bottleneck. I am using 6 neutron star 2 in my setup until now and some neutrino star...The using of neutron stars is realy a big step in sq. But i am not using their power supply. The neutron star power supply sounds... Like the psu i build 3 years ago. The clock in a system should have the best psu. For that reason i build a kaskarde of my psu with 2 regulators at minimum.

 

Regards

 

Hi sunny time 

very interesting i have one neutron star 2 in my system with the stock power supply. What 3 regulators are you using? Was it a substantial or a more modest improvement?

Also I have two neutrino’s clocks but am not sure if a crystek 575, 957 wouldn’t be better.

 

rick

[RickyV]

3 hours ago, sunny_time_99 said:

That is definitely true.

 

And the dac is only one part in the whole game. And my listening experience  is that it has not the biggest influence on the total sq. I focused on the dac for years and build dacs with all available dac chips in the last years but today i would only say that the dac makes 20-25% of the total sq. There are many possibilities to get a better sq without  spending money for a new dac

 

Regards

 

What do you think the streamer makes on total sq?